Catalytic Conversion of Lignin for the Production of Aromatics

Catalytic Conversion of Lignin for the Production of Aromatics

ISBN: 978-90-393-6001-9

Printed by: Gildeprint Drukkerijen - www.gildeprint.nl

Catalytic Conversion of Lignin for the Production of Aromatics

De Katalytische Omzetting van Lignine voor de Productie van Aromatische Chemicaliën

(met een samenvatting in het Nederlands)

Proefschrift

ter verkrijging van de graad van doctor aan de Universiteit Utrecht op gezag van de rector magnificus, prof.dr. G.J. van der Zwaan, ingevolge het besluit

van het college voor promoties in het openbaar te verdedigen op woensdag 11 september 2013 des ochtends te 10.30 uur

door

Anna Louise Jongerius

geboren op 1 oktober 1985 te Zaandam

Promotor: Prof.dr.ir. B.M. Weckhuysen Co-promotor: Dr. P.C.A. Bruijnincx

This research was funded by the Smart Mix Program of the Netherlands Ministry of Economic Affairs and the Netherlands Ministry of Education, Culture and Science within the framework of the CatchBio Program.

Chapter 1: General Introduction

Chapter 2: Lignin: Structure, Chemistry and Catalysis

Part I: Lignin Depolymerization

Chapter 3: Lignin Solubilization and Aromatics Production by Liquid- Phase Reforming and Hydrogenation

Chapter 4: Stability of a Pt/γ-Al2O3 Catalyst in Lignin Liquid-Phase Reforming Reactions

Chapter 5: Lignin Depolymerization by Alkaline Hydrogen Peroxide Treatment

Part II: Hydrodeoxygenation of Lignin Model Compounds

Chapter 6: CoMo Sulfide-Catalyzed Hydrodeoxygenation of Lignin Model Compounds

Chapter 7: W2C and Mo2C-Catalyzed Hydrodeoxygenation of the Lignin Model Compounds Guaiacol

Part III: Combined Depolymerization and Hydrodeoxygenation of Lignin

Chapter 8: Liquid-Phase Reforming and Hydrodeoxygenation as a Two- Step Route to Aromatics from Lignin

Chapter 9a: Summary and Concluding Remarks Chapter 9b: Nederlandse Samenvatting

List of PublicationsDankwoordCurriculum Vitae

7

21

79

99

123

149

173

195

213225

237239243

Contents

Chapter 1

General Introduction

Chapter 1

8

1.1 Introduction

With the depletion of fossil fuels as a source for fuels, chemicals, and energy, the fraction of energy and chemicals supplied by renewable resources such as biomass can be expected to increase in the foreseeable future. Furthermore, in an effort to limit the greenhouse effect by reducing the CO2 that is released into the environment, several governments have passed legislation mandating increases in energy and chemical production from renewable resources, especially biomass. The U.S. Department of Agriculture and U.S. Department of Energy have, for instance, set ambitious goals to derive 20% of transportation fuels and 25% of U.S. chemical commodities from biomass by 2030. [1] The European Union has set a mandatory target of 20% for the renewable energy’s share of energy consumption by 2020 and a mandatory minimum target of 10% for biofuels for all member states. [2] These goals have contributed to the current, intensified interest in the development of technology and processes for biomass valorization. Such processing of biomass to value-added products and energy is envisaged to take place in so-called biorefineries. In direct analogy to current petroleum refineries, which produce fuels and chemicals from crude oil, a biorefinery is a facility that produces multiple products including fuel, power, and bulk or fine chemicals from biomass. It is important to note that the economic necessity for a biorefinery to produce chemicals in addition to biofuels has been strongly advocated. [3] Indeed, the integrated production of fuels and commodity chemicals is deemed necessary to justify construction of the biorefinery and allow a high energy impact as well as proper return on investment. [4]

The recent ‘shale gas revolution’ that is currently predominantly taking place in the US but might soon be followed by other countries, is impacting the implementation of renewable alternative feedstocks. The production of this new source of fossil fuels increased from only 1% of the total US natural gas production in 2000 to 20% in 2009. [5] With shale gas production expected to expand over the coming years, the prices of fossil fuel-based energy will not increase as dramatically as predicted and as a result the urgency to find alternatives to replace fossil fuels as an energy feedstock is decreasing, at least from an economic perspective. This shift in feedstock will also have important ramifications for commodity chemicals production. Currently, the commodity chemicals that form the basis of the chemical industry, i.e. ethylene, propylene, butadiene, benzene, toluene and xylenes (BTX), are produced via cracking crude oil and further conversion of the naphtha fraction. The shift to lighter feeds for the crackers will affect the product composition after cracking and as a result


9

the availability of bulk chemicals. Although ethylene and propylene can be obtained via steam cracking of natural gas, the production of in particular butadiene and BTX from lighter feeds such as (shale) gas is very limited. [6] Indeed, the increased use of natural gas in the US has already resulted in an increase in butadiene prices. [7] To address the expected shortages of these important chemicals, new, dedicated routes for the on-purpose production of butadiene or BTX, preferably from renewable resources, are highly desired. The development of routes for the production of liquid fuels and chemicals from renewable resources therefore has lost none of its urgency and demands even more attention.

Although all kinds of biomass can in principle be converted into fuels and chemicals, the use of edible crops for these purposes cannot be justified in a world with our current population density. To prevent competition with the use of land and resources used for the production of food, it is recommended to use so-called second-generation biomass for the production of fuels and chemicals. Lignocellulosic biomass, a typical example of such a second-generation feedstock, consists mainly of non-edible cellulose, hemicellulose and lignin. It can be harvested as a waste product from food-producing crops such as sugarcane and cornstalks, or can be obtained from wood and purpose-grown crops such as switch grass and wild perennial vegetation able to grow on non-arable land. [8] Separation of lignocellulosic biomass into its components and processing the components to value-added products and energy will take place in biorefinery operations, as discussed above. Again in analogy with the petrochemical refineries, efficient use of the renewable resource demands that all components of lignocellulosic biomass are valorized for the biorefinery to be economically viable. The carbohydrate fraction of lignocellulose can be used for commercial production of fuels (e.g. bioethanol) and chemicals (e.g. 1,3-propanediol or succinic acid) by means of fermentation or chemocatalytic conversion. The development of commercial processes for the conversion of lignin lags behind, however. Currently operating biorefineries such as the commercial pilot plant of POET-DSM [9] and the integrated biorefining process of Roquette [10] receive and process enormous quantities of biomass, generating large amounts of lignin; valorization of also this component, for instance by converting it into fuels and chemicals, is imperative for economic profitability. The Borregaard biorefining process provides a successful example of the integration of lignin valorization in the biorefinery. Borregaard already produces lignin-based products such as binding and dispersing agents and is the main supplier of synthetic vanillin (3-methoxy-4-hydroxybenzaldehyde). [11] Vanillin is used extensively in foods and perfumes because of its flavor but also finds use in medicinal applications

Chapter 1

10

or as a platform chemical for pharmaceuticals production. [12, 13] Nevertheless, the market for vanillin is very small compared to the total amount of available lignin. To achieve complete integration of lignin in the biobased economy, more focus on the development of processes for lignin valorization to bulk chemicals and fuels is clearly desired.

Lignin is a natural, polyaromatic and amorphous polymer that acts as the essential glue that gives plants their structural integrity. It is a main constituent of lignocellulosic biomass (15-30% by weight, up to 40% by energy). [1] A detailed overview of lignin chemistry, structure and pretreatment methods is given in Chapter 2, in addition to an overview of catalytic lignin conversion processes. As of 2004, the pulp and paper industry alone produced 50 million tons of extracted lignin, yet the existing markets for lignin products remain limited and are concerned with low value products such as dispersing or binder applications in asphalt, cement or polymers. High value products such as vanillin can only be obtained in relatively low yields with high process costs and serve markets of limited volume. [14] As a result, only approximately 2% of the lignins that are available from the pulp and paper industry are currently used commercially with the remainder simply being burned as a low value fuel. [15] Nevertheless, lignin holds considerable potential as a renewable resource for the sustainable production of fuels and bulk chemicals. [1, 16] With its unique aromatic structure and chemical properties, liquid fuels and a wide variety of bulk and fine chemicals, particularly aromatic compounds, which will become less available with the increased use of shale gas, can potentially be obtained from lignin. Although routes have been reported for the transformation of sugars into aromatics, for example via a Diels-Alder reaction of sugar-derived furanics, [17] and commercial processes are being developed for the production of p-xylene from isobutanol [18] or biomass-derived alcohols and aldehydes, [19] lignin is the most obvious candidate to become the major aromatic resource of a future bio-based economy. Utilisation of the full potential of this large resource is hampered, however, by the current lack of efficient technologies that both depolymerize and lower the oxygen content of lignin, but leave the aromaticity intact.

The realization of biorefinery schemes with fully integrated lignin valorization processes requires the development of catalytic technology to perform the desired depolymerization of lignin. New approaches and strategies will have to be developed to achieve this. Cleavage of the primary linkages of lignin will ultimately result in formation of monomeric aromatic compounds, which depending on the cleavage


11

method employed can be to different extents functionalized with hydroxyl, allylic alcohol, aldehyde, ether, or carboxylic acid substituents. These monomeric compounds are then susceptible to an extensive array of subsequent transformations, principally either reductive in nature, forming less functionalized aromatic or aliphatic hydrocarbons, or oxidative in nature, resulting in aromatics with increased or specifically targeted functionality. An oxidative lignin valorization route requires the development of catalysts that introduce specific aromatic alcohols, aldehydes, acids, and other functionalized aromatics. The fine chemicals with high oxygen content and high oxidation state functional groups that can be potentially obtained from lignin in this way, currently require several synthetic steps for their production from the non-oxygenated hydrocarbons found in petroleum. The reductive route, on the other hand, requires the development of catalysts that partially or completely remove the alcohol, aldehyde, ether, and acid substituents from the monomers formed after lignin depolymerization with the aim of obtaining simple aromatics, such as benzene, toluene, xylene, or (alkylated) phenolics. These simple aromatic building blocks can then be further converted, using existing technology, to a wide variety of useful chemicals. A graphical depiction of the transformation of lignin to current or novel

OH

phenol benzenetoluenexylene

OH

OH

O

NH

O

NH2H2N

O OHO

HO

O

OH

O

HO

O

OH

O

OH

H2N NH2

CO2H

NCOOCN

NO2O2N

OHHO

OH O

OH OH

H2N NH2NO2O2N

HOO

O

OH

R R R

OH

O

OH

O

O + + +

HOO

O

OH

NewTechnology

CurrentTechnologyLignin

p-coumarylcompounds

polymers, dyes, resines, pharmaceuticals

OOH

OH

OHO

O

OOOH

OH

HO

O

O

HO

NewTechnology

coniferylcompounds

sinapylcompounds

vanillin vanillic acid

Methanol nitrophenols aminophenols

cyclohexanol cyclohexanone

bisphenol A

isophthalic acid

terephthalic acid

benzoic acid

diaminotoluene

dinitrotoluene

toluene diisocyanate

cyclohexane cyclohexanol

phenol cyclohexanone

styrene caprolactam

cumene adipic acid

1,6-diaminohexane

HO

OH

HO

OH

HO

OHO O O

p-coumaryl alcohol coniferyl alcohol sinapyl alcohol

Figure 1.1: Valuable products that can be potentially obtained from lignin based on the development and integration of new and current technology (adapted from Bozell [21] and Koutinas et al. [22]). The sinapyl, conyferyl and p-coumaryl alcohol building blocks are depicted in blue, green and red, respectively.

Chapter 1

12

aromatics is shown in Figure 1.1. The lignin-derived chemicals can find application in the production of plastics and other polymers, or can be used as pigments, dyes, resins, and many other products. [20, 21]

1.2 Strategies for the conversion of lignin

As indicated, a large array of aromatic products, (functionalized) phenolics and BTX in particular, can be potentially obtained from lignin. Catalytic lignin conversion processes nonetheless face a number of difficulties that will have to be overcome in order to become efficient and viable. First, the nature of lignin by which it acts as a crosslink between the carbohydrate polymers and gives strength and rigidity to plant cell walls also causes the polymer to be relatively recalcitrant towards chemical transformations. Furthermore, in contrast to cellulose, which consists of a single type of monomer and one type of linkage, lignin is a heterogeneous polymer that is constructed from three different building blocks, depicted in Figure 1.1, and several kinds of linkages. Depending on the plant species and the growth season, the monomer ratios and the number of polymer branches can vary. In addition, the different pretreatment methods employed to separate lignin from the carbohydrate components of lignocellulosic biomass also introduce chemical functionalities that affect lignin reactivity and solubility. While dissolution of lignin is critical to obtain a good interaction with the catalyst, the solubility of lignin is highly dependent on the pretreatment method that was used for isolation of lignin from lignocellulose. Indeed, no solvents are able to dissolve all the different lignins currently available. Finally, the conversion of the heterogeneous lignin polymer with its varying composition will likely yield a product mixture with a broad distribution. This oxygen-rich mixture can furthermore be relatively unstable and some of its components can be prone to repolymerization under reaction conditions.

In general, (catalytic) conversion of lignin into chemicals can be approached in two different ways (Figure 1.2). First, there is the approach in which lignin is directly converted into the desired products using only a one-step process. This approach typically involves a catalytic depolymerization route that yields products that are highly functionalized. Alternatively, following a two-step approach, one can first produce a bio-oil, for example by a thermal process such as catalytic pyrolysis. This bio-oil then contains a mixture of monomeric and oligomeric oxygen-rich compounds, which in a second catalytic conversion step can be deoxygenated and further


13

depolymerized to produce the targeted aromatic compounds. The advantage of a one-step process is that it is easier and cheaper to perform than two separate steps. In a two-step process, however, the second step can be independently chosen depending on the targeted end products and in addition, complete the depolymerization of any remaining dimers, trimers and higher oligomers thus allowing for an improved yield and flexibility.

Several groups have reported one-step conversions of lignin or even woody biomass using oxidative, reductive or redox-neutral reactions. A comprehensive overview of the catalytic lignin conversion reactions reported up to 2010 is given in Chapter 2 and some examples of catalytic lignin conversions that have appeared since are highlighted here. Oxidative routes generally target the formation of products such as vanillin, which could be obtained, for instance, in yields up to 6.3 wt% from kraft lignin using a CoCl2 catalyst in methanol/water. [23] Alternatively, syringaldehyde or benzoquinones can be produced in a 21% total yield after oxidation of organosolv beech lignin over Mn(NO3)2 in ionic liquids. [24] Higher yields of aromatics are generally obtained, however, using reductive routes, which are typically operated at higher

One-step approach

Lignin

Lignin-oil

Aromatic chemicals

Step 1: Depolymerization

Step 2: e.g. Hydrodeoxygenation

Figure 1.2: The conversion of lignin in a one-step or two-step approach. The latter approach involves a depolymerization and, e.g., a hydrodeoxygenation reaction.

Chapter 1

14

temperatures and higher pressures. Reactions under hydrogen atmosphere as well as more sustainable methods where the relatively expensive hydrogen is produced in situ have been reported. Song et al., for instance, reported combined selectivities for propylguaiacol and propylsyringol of up to 97% directly from birch sawdust at a conversion of 50% using a nickel on activated carbon catalyst with ethylene glycol as solvent. The reaction is suggested to place via fragmentation of the lignin polymer and hydrogenolysis of the fragmented species over the Ni catalyst using the solvent as a hydrogen donor. [25] Similar phenolics were obtained in 47% yield based on the lignin content from birch wood under 60 bar hydrogen pressure using a nickel and tungsten carbide on activated carbon catalyst, again with ethylene glycol as the solvent. [26] The combined depolymerization and hydrodeoxygenation of organosolv switch grass lignin over Pt/C with formic acid as the hydrogen source resulted in a 76% reduction in lignin molecular weight combined with a 50% reduction of the O/C ratio after 20 h. Monoaromatic products such as guaiacol and phenol could be identified after 4 h reaction in yields up to 21%. [27] Additionally, redox-neutral strategies based on a hydrogen borrowing approach and using homogeneous ruthenium/phosphine-based catalysts have also been proposed. It was shown that model compounds mimicking typical lignin ether linkages could be converted with a 99% yield of monomers, but these reactions still need to be demonstrated on real lignin feeds. [28]

A disadvantage of the one-step approach is that it is more difficult to both depolymerize the lignin polymer and at the same time remove or add chemical functionalities in order to end up with a product mixture of a narrow enough product distribution. By performing the conversion of lignin in two separate steps it is possible to overcome these problems by using the second step to narrow the product distribution and complete the depolymerization process. The two-step approach thus first yields a depolymerized lignin or bio-oil from either pure lignin or from lignocellulosic material. Catalytic pyrolysis of lignin is a common method to produce such a bio–oil and many results have been reported. [29] Apart from the liquid oil fraction, which typically consists of oxygen-rich (alkyl-substituted) aromatic monomers, dimers and trimers, also char and gases are formed. [30-32] Addition of zeolite catalysts has been shown to increase oil yields and decrease char formation by removing oxygen, resulting in products with a higher C/O ratio. [33] Pyrolysis oil yields and the selectivity towards different products can be controlled by changing operating conditions such as temperature, time and reactor loading, [34] or by altering catalyst variables such as zeolite Si/Al ratio or acidity. [33, 35] When the formation of high amounts of char can be prevented, oil yields of over 70% are


15

possible. [33] Other methods to obtain bio-oils from lignin include depolymerization in supercritical solvents [36] and base-catalyzed depolymerization, which consists of heating a solution of lignin and inorganic base in water at high pressures. [37, 38] Depending on the lignin source, depolymerization oils are complex mixtures of many compounds of which some have been identified as (alkylated) phenol, guaiacol and syringol as well as demethylated catechol-type molecules. [37-39] Typical monomeric oil yields for base-catalyzed processes are around 20% and were found to increase in the presence of boric acid, which acts as a capping agent for lignin monomers by preventing recondensation reactions. [40] The adverse toxicological and health effects make the use of boric acid less desirable from an industrial perspective, however.

Regardless of the method used for their production, lignin-derived bio-oils typically consist of an oxygen-rich mixture of aromatic molecules with high water content, low vapor pressure and high viscosity. Upgrading of these oils by removal of oxygen functionalities in a second step is essential, both for the production of fuels and chemicals. An effective method for the removal of oxygen from oxygen-functionalized aromatic compounds is by hydrodeoxygenation (HDO). Several different classes of catalysts have been reported for the hydrodeoxygenation of lignin-derived compounds; a comprehensive overview of the literature up to 2010 on HDO reactions of lignin and lignin model compounds is again given in Chapter 2. Catalyst systems that have been studied most extensively on both lignin pyrolysis oils and lignin model compounds such as guaiacol are sulfided CoMo/Al2O3 and NiMo/Al2O3 catalysts. These traditional hydrodesulfurization (HDS) catalysts were originally developed for the removal of sulfur and nitrogen from conventional oils and also proved to be useful for the removal of oxygen. [41, 42] An important advantage of these catalysts is that they are able to reduce the oxygen content without loss of aromaticity, [43] with up to 60% combined selectivity of (methylated) phenolics being reported from guaiacol. [42, 43] A downside is that in the presence of water they are known to deactivate through the loss of sulfur from the active site. Although the CoMo catalyst can be completely reactivated by resulfidation, this still means that the product stream will be contaminated with sulfur. [44] A more stable, sulfur-free alternative for molybdenum or nickel sulfides are supported noble metal catalysts. Pd, Pt, Ru but also Ni-based catalysts are more active and operate at lower temperatures than the hydrotreating catalysts. The downside of this class of catalysts is their very high hydrogenation activity and complete hydrogenation of the aromatic ring occurs prior to removal of the oxygen-containing functional groups. As a result, 80-100% selectivity towards cyclohexanes from guaiacol and phenols has been reported. [45-

Chapter 1

16

49] Although aromatic compounds can be obtained by ring hydrogenation of lignin pyrolysis oils followed by a zeolite-catalyzed upgrading step, [51] a direct HDO step that retains the aromaticity is more desirable from a hydrogen economy point of view. Recently, a third type of HDO catalysts was shown to be active in the HDO of guaiacol. High conversions, phenol yields up to 50% and a low ring hydrogenation activity were reported using supported molybdenum nitrides. [51-54]

1.3 Aim and outline of the thesis

The work described in this PhD thesis aims to develop an efficient two-step valorization route for the production of bulk aromatic chemicals from lignin and to gain fundamental insight into the catalytic steps involved. The oxygen-rich, polymeric nature of lignin demands that such a process consists of steps that lower the degree of polymerization and subsequently lower the oxygen content of the lignin polymer. Fundamental insight is furthermore needed into the performance of and changes to a catalyst to determine what makes a good catalyst for these reactions. In view of this, a two-step lignin valorization process consisting of both depolymerization and hydrodeoxygenation steps is presented.

Chapter 2 outlines the structure and chemistry of lignin and highlights the differences between lignins from different sources. It also provides a comprehensive overview of the state of the art of heterogeneously-catalyzed lignin conversion processes available at the start of this PhD project with particular emphasis on hydrogenation and hydrodeoxygenation routes.

Part I of this thesis is concerned with the first step in the envisaged lignin conversion scheme and describes two different approaches to lignin depolymerization (Figure 1.2). A new process for the solubilization and depolymerization of lignin in ethanol/water over a Pt/Al2O3 catalyst is introduced in Chapter 3. The stability under reaction conditions of the Pt/Al2O3 catalyst used in Chapter 3 is extensively studied in Chapter 4, focusing on the stability of the support, the Pt nanoparticles and the deposition of a carbon-rich material on the catalyst. In Chapter 5 a screening of reaction conditions for the oxidation of lignin with H2O2 under alkaline conditions is presented.

The HDO of model compounds mimicking typical components found in a lignin depolymerization stream is studied in Part II. A reaction network for the HDO of a


17

library of model compounds over a sulfided CoMo/Al2O3 catalyst is presented in Chapter 6. The HDO of the lignin model compound guaiacol is studied in Chapter 7 with tungsten and molybdenum carbides supported on carbon nanofibers as catalysts.

Part III, Chapter 8 aims to combine the depolymerization and deoxygenation steps in the conversion of lignin by performing a lignin depolymerization reaction described in Chapter 3 and using the product of this reaction to perform HDO reactions with the catalysts described in Chapters 6 and 7, leading to a two-step catalytic process for the production of aromatic chemicals from lignin.

In Chapter 9 a summary of the key findings of this PhD thesis is given, together with some concluding remarks.

1.4 References

[1] R. D. Perlack, L. L. Wright, A. F. Turhollow, R. L. Graham, B. J. Stokes, D. C. U. S. Erbach, US Department of Energy, Biomass as Feedstock for a bioenergy and bioproducts industry: the technical feasibility of a billion-ton annual supply, 2005. [2] Renewable Energy Road Map 2007, http://europa.eu/legislation_summaries/energy/renewable_energy/ l27065_en.htm.[3] Y. H. P. Zhang, Ind. Microbiol. Biotechnol. 2008, 35, 367-375.[4] J. J. Bozell, G. R. Petersen, Green Chem. 2010, 12, 539-554.[5] P. C. A. Bruijnincx, B. M. Weckhuysen, Angew. Chem. Int. Ed. 2013, Accepted.[6] E. McFarland, Science 2012, 338, 340-342.[7] GBI research report GBICH0076MR. [8] K. Butterbach-Bahl, R. Kiese, Nature 2013, 493, 483-485.[9] http://www.poetdsm.com.[10] http://www.roquette.com/biorefining-with-renewable-resources.[11] www.borregaard.com.[12] S. R. Rao, G. A. Ravishandkar, J. Sci. Food Agric. 2000, 80, 289-304.[13] M. B. Hocking, J. Chem. Educ. 1997, 74, 1055-1059.[14] C. O. Tuck, E. Pérez, I. T. Horváth, R. A. Sheldon, M. Poliakoff, Science 2012, 337, 695-699.

Chapter 1

18

[15] R. J. A. Gosselink, E. de Jong, B. Guran, A. Abächerli, Ind. Crops. Prod. 2004, 20, 121-129. [16] J. J. Bozell, J. E. Holladay, D. Johnson, J. F. White, Top Value Added Candidates from Biomass, Volume II: Results of Screening for Potential Candidates from Biorefinery Lignin; Pacific Northwest National Laboratory: Richland, WA, 2007.[17] C. L. Williams, C. C. Chang, P. Do, N. Nikbin, S. Caratzoulas, D. G. Vlachos, R. F. Lobo, W. Fan, P. J. Dauenhauer, ACS Catal. 2012, 2, 935-939.[18] www.gevo.com.[19] www.virent.com.[20] Y. S. Kim, H. M. Chang, J. F. Kadla, J. Wood Chem. Technol. 2008, 28, 1-25.[21] J. J. Bozell, J. E. Holladay, D. Johnson, J. F. White, Top Value Added Candidates from Biomass, Volume II: Results of Screening for Potential Candidates from Biorefinery Lignin, 2007.[22] A. A. Koutinas, C. Du, R. H. Wang, C. Webb, Introduction to Chemicals from Biomass, Wiley, 2008, p 78.[23] H. Werhan, J. M. Mir, T. Voitl, P. R. van Rohr, Holzforschung 2011, 65, 703-709.[24] K. Stärk, N. Taccardi, A. Bösmann, P. Wasserscheid, ChemSusChem 2010, 3, 719-723.[25] Q. Song, F. Wang, J. Cai, Y. Wang, J. Zhang, W. Yu, J. Xu, Energy Environ. Sci. 2013, 6, 994-1007.[26] C. Li, M. Zheng, A. Wang, T. Zhang, Energy Environ. Sci. 2012, 5, 6383-6390.[27] W. Xu, S. J. Miller, P. K. Agrawal, C. W. Jones, ChemSusChem 2012, 5, 667-675.[28] J. M. Nichols, L. M. Bishop, R. G. Bergman, J. A. Ellman, J. Am. Chem. Soc. 2010, 132, 12554-12555. [29] M. Brebu, C. Vasile, Cell. Chem. Technol. 2010, 44, 353-363.[30] R. Y. Nsimba, C. A. Mullen, N. M. West, A. A. Boateng, ACS Sustainable Chem. Eng. 2013, 1, 260-267.[31] P. R. Patwardhan, R. C. Brown, B. H. Shanks, ChemSusChem 2011, 4, 1629- 1636.[32] M. Kosa, H. Ben, H. Theliander, A. J. Ragauskas, Green Chem. 2011, 12, 3196- 3202.[33] Z. Ma, E. Troussard, J. A. van Bokhoven, Appl. Catal. A: Gen. 2012, 423-424, 130-136.[34] J. Gan, W. Yuan, Appl. Energy 2013, 103, 350-357.[35] H. Ben, A. J. Ragauskas, ACS Sustainable Chem. Eng. 2013, 1, 316-324.


19

[36] R. J. A. Gosselink, W. Teunissen, J. E. G. van Dam, E. de Jong, G. Gellerstedt, E. L. Scott, J. P. M. Sanders, Bioresource Technol. 2012, 106, 173-177. [37] J. M. Lavoie, W. Baré, M. Bilodeau, Bioresource Technol. 2011, 102, 4917-4920.[38] A. Toledano, L. Serrano, J. Labidi, J. Chem. Technol. Biotechnol. 2012, 87, 1593-1599.[39] A. Vigneault, D. K. Johnson, E. Chornet, Can. J. Chem. Eng. 2007, 85,906-916.[40] V. M. Roberts, V. Stein, T. Reiner, A. Lemonidou, X Li, J. A. Lercher, Chem. Eur. J. 2011, 17, 5939-5948.[41] Q. Bu, H. Lei, A. H. Zacher, L. Wang, S. Ren, J. Liang, Y. Wie, Y. Liu, J. Tang, Q. Zhang, R. Ruan, Bioresource Technol. 2012, 124, 470-477.[42] V. N. Bui, D. Laurenti, P. Afanasiev, C. Geantet, Appl. Catal. B: Environ. 2011, 101, 239-245.[43] Y. C. Lin, C. L. Li, H. P. Wan, H. T. Lee, C. F. Liu, Energy Fuels 2011, 25, 890-896.[44] M. Badawi, J. F. Paul, S. Cristol, E. Payen, Y. Romero, F. Richard, S. Brunet, D. lambert, X. Portier, A. Popov, E. Kondratieva, J. M. Goupil, J. El Fallah, J. P. Gilson, L. Mariey, A. Travert, F. Maugé, J. Catal. 2011, 282, 155-164.[45] C. Zhao, J. He, A. A. Lemonidou, X. Li, J. A. Lercher, J. Catal. 2011, 280, 8-16.[46] Z. Zhao, S. Kasakov, J. He, J. A. Lercher, J. Catal. 2012, 269, 12-23.[47] C. R. Lee, J. S. Yoon, Y.-W. Suh, J.-W. Choi, J.-M. Ha, D. J. Suh, Y.-K. Park, Catal. Commun. 2012, 17, 54-58.[48] A. Gutierrez, R. K. Kaila, M. L. Honkela, R. Slioor, A. O. I. Krause, Catal. Today 2009, 147, 239-246.[49] H. Ben, Wei, Mu, Y. Deng, A. J. Ragauskas, Fuel 2013, 103, 1148-1153.[50] T. P. Vispute, H. Zhang, A. Sanna, R. Xiao, G. W. Huber, Science 2010, 330, 1222-1227.[51] C. Sepúlveda, K. Leiva, R. Garcia, L. R. Radovic, I. T. Ghampson, W. J. DeSisto, J. L. García Fierro, N. Escalona, Catal. Today 2011, 172, 232-239.[52] I. T. Ghampson, C. Sepúlveda, R. Garcia, B. G. Frederick, M. C. Wheeler, N. Escalona, W. J. DeSito, Appl. Catal. A: Gen. 2012, 413-414, 78-84.[53] I. T. Ghampson, C. Sepúlveda, R. Garcia, J. L. García Fierro, N. Escalona, W. J. DeSisto, Appl. Catal. A: Gen. 2012, 435-436, 51-60.[54] I. T. Ghampson, C. Sepúlveda, R. Garcia, L. R. Radovic, J. L. García Fierro, W. J. DeSisto, N. Escalona, Appl. Catal. A: Gen. 2012, 439-440, 111-124.

Chapter 2

Lignin: Structure, Chemistry and Catalysis

Based on: J. Zakzeski, P. C. A. Bruijnincx, A. L. Jongerius, B. M. Weckhuysen, “The Catalytic Valorization of Lignin for the Production of Renewable Chemicals “ Chem. Rev. 2010, 110, 3552-3599.

Chapter 2

22

2 .1 Introduction

This Chapter provides an overview of the catalytic lignin valorization literature available at the start of this thesis and covers the literature up to the end of 2009; more recent developments are described in Chapter 1. The aim is to present the different approaches and strategies that have been reported for catalytic lignin conversion with a focus on the catalytic production of valuable and useful bulk and platform chemicals. The literature on the topic was found to be scattered, and focused primarily on engineering or biological aspects of lignin rather than specifically on the catalytic aspects of the conversion or on catalyst development. The latter two topics are, however, essential for efficient and selective lignin valorization processes. [1] In addition, given the different aims and focus, reports on catalytic lignin conversion involve a wide range of conditions, solvents, catalysts, and model compounds.

Reviews on biomass valorization have focused almost exclusively on cellulose with often only a paragraph devoted to lignin (see below). [2-4] A general review of chemical transformations of biomass by Corma and coworkers focuses on chemical intermediate platforms derived from saccharides, vegetable oils and animal fats, and terpenes. [3] Mäki-Arvela and coworkers, on the other hand, published a review on the synthesis of fine and specialty chemicals from wood and other biomass with a focus primarily on products obtainable from cellulose. [4] Most relevant to lignin valorization, Amen-Chen and coworkers published a review on the production of monomeric phenols obtained by the mostly non-catalytic thermochemical conversion of biomass. [5] More recently, Dumesic and coworkers wrote a review on the conversion of lignocellulosic biomass to chemicals over bi-metallic catalysts. [6] Also reviews devoted entirely to lignin have appeared since 2009; Hicks reviewed the various strategies towards C-O bond transformations in lignin [7] while Pandey and Kim reviewed thermochemical methods for the conversion of lignin. [8]

In this Chapter, we first introduce the lignin macromolecule and its various structural differences, linkages and isolation methods. Next, strategies for lignin dissolution are discussed. Finally, we focus on those strategies for catalytic lignin conversion that bear relevance to the approach taken in this thesis. These strategies include, among others, (hydro)cracking, lignin reduction reactions, aimed at making fuels or bulk aromatic and phenolic compounds, and lignin oxidation reactions, aimed at making functionalized aromatics for the production of fine chemicals. For a more comprehensive overview of catalytic lignin valorization strategies we refer to the review article that this Chapter is based on [Chem. Rev. 2010, 110, 3552-3599].


23

2.2 Lignin structure

Lignin is one of the three major components of lignocellulosic biomass, the other two components being cellulose and hemicellulose. [9] Lignin is a three-dimensional amorphous polymer consisting of methoxylated phenylpropane structures. [10] In plant cell walls, lignin fills the spaces between cellulose and hemicellulose, and it acts as a resin that holds the lignocellulose matrix together. [11] Crosslinking with the carbohydrate polymers then confers strength and rigidity to the system. Figure 2.1 depicts a schematic representation of lignin in biomass, highlighting the location and

Figure 2.1: Schematic representation of the location and molecular structure of lignin in lignocellulosic material. Adapted from Ritter. [11]

Chapter 2

24

molecular structure of lignin. [1] Considerable work has been done on the detailed structural characterization of

this complex natural polymer and an understanding of both structure and function is evolving as a result. Advances in spectroscopy [10] coupled with oxidation/reduction, [12, 13] ozonation, [14, 15] photochemical degradation, [16] thermogravimetic analysis, [17] and computational studies [18] have elucidated many of the salient structural features, constituents, and linkages of lignin. The combination of wet chemical methods and, more recently, advanced NMR methods have led to further identification and quantification of the various moieties, end groups, and linkages. These studies have resulted in an improved structural characterization of lignins, yet some uncertainty still remains. Here, we give a short description of the general structural characteristics of lignin and refer to leading references for a more detailed account. [19-23] Although the exact structure of protolignin, the untreated lignin found in plants (also known as ‘native lignin’), is still unknown, the biosynthesis of lignin is thought to involve the radical polymerization of three primary monomers: p-coumaryl, coniferyl, and sinapyl alcohols (Figure 2.2). [10] Polymerization by random phenol radical-radical coupling reactions under chemical control then leads to the formation of lignin vascular in plants. [19] The composition, molecular weight, and amount of lignin differ from plant to plant, with lignin abundance generally decreasing in the order of softwoods > hardwoods > grasses. [24]

2.2.1 Lignin structure and linkagesDuring the lignification process and the processes applied to extract lignin from the

plant material, several different types of linkages are formed between the monolignol building blocks. The linkages, individually depicted in Figure 2.3, include β-O-4, 5-5, β-5, 4-O-5, β-1, dibenzodioxocin and β-β linkages. The most abundant linkage in lignin

Figure 2.2: The three monolignols, the building blocks of lignin.


25

is the β-O-4 linkage, comprising more than half of the linkage structures of lignin. [10, 23] The β-O-4 ether bond is readily cleaved; indeed, the cleavage of these bonds during alkaline pretreatment constitutes the principle pathways in which the lignin is depolymerized. [10] The fragmentation of these linkages tends to lead to the generation of more water-soluble compounds containing free phenolic groups. [10] The carbon-carbon bonds in lignin are some of the most difficult bonds to break, and many of these linkages tend to survive the typical pretreatment processes. [10] The development of catalysts capable of cleaving these more recalcitrant linkages (particularly the aryl-aryl, i.e. 5-5, linkages) is a considerable challenge that has not yet been adequately addressed. Moreover, although carbon-carbon linkages are present in native lignin, additional carbon-carbon bonds can be formed during lignin pretreatment, such as in alkali-promoted condensation reactions during kraft pretreatment. [10] Other studies have suggested that a significant amount of the 5-5 linked structural units present in lignin are actually part of so-called benzodioxocin structures. The β-5 linkage is often found as part of a five-membered ring linking two aromatic structures via both a carbon-carbon and a carbon-oxygen bond. α-O-4 linkages have also been suggested, however, modern NMR experiments do not confirm the presence of non-cyclic α-O-4 moieties. Instead, it was suggested that these α-O-4 linkages are present only as part

Figure 2.3: Common lignin linkages and their numbering schemes.

HOO

HO

O

OO O

OO

O

OO OH

O

O

O OH

HOO

OO

O OH

OArHO

OO

HO

HO

OO

O

O

O

OHO

O

O

O OO

O

OO

OO

412

3 4

5

5

5

5

4

4

44

3

3

3

2

2

2

1

1

1

1

γ

α

β

α

α α

α β

ββ

β

γ

γ

γ

Dibenzodioxocin

Spirodienoneβ-O-4 5-5 β-5

4-O-5 β-1 β-β

Chapter 2

26

of dibenzodioxocin or phenylcoumaran structural units. The 4-O-5 aryl-aryl ether linkage does occur in lignin, mainly as the result of oligomer-oligomer couplings, and leads to further branching of the polymer.

The relative abundance of the various linkages in softwoods, including spruce, and hardwoods, such as birch and eucalyptus, is given in Table 2.1. As indicated, the dominant linkage in both softwood and hardwood is the β-O-4 linkage, consisting of approximately 50% of spruce linkages and 60% of birch and eucalyptus linkages. The identification and quantification of the various structures and linkages in lignin is a considerable challenge even with advanced NMR techniques as the lignin molecule is complex and very sensitive to isolation techniques. [23] In particular, distinguishing between completely etherified, semi-etherified, and completely non-etherified 5-5 moieties is very difficult, as a result the presence and relative abundance of these structures in lignin is somewhat uncertain. [21] The composition of lignin softwood, hardwood and grasses varies also in the relative abundance of the p-coumaryl, coniferyl, and sinapyl alcohol monolignols. Coniferyl alcohols constitute approximately 90% of softwood lignin, whereas roughly equal proportions of coniferyl alcohol and sinapyl alcohol appear in hardwood lignin, although many exceptions are known. [25] Grassy lignins have a relatively heterogeneous composition with an average coniferyl, sinapyl, p-coumaryl ratio of 1:0.8:0.6. [26, 27] The additional methoxy groups on the

Table 2.1: Common linkages and approximate abundance per 100 C9-units connecting the phenyl-propane units in softwood and hardwood lignin. The structures are shown in Figure 2.3.

Linkage β-O-4 5-5 β-5 S 4-O-5 β-1 D β-β

Softwood

Spruce [171-172] 45-50 19-22 9-12 nd 4-7 7-9 nd 2-4

Spruce [173] nd 22 12 nd nd 2 5 2.5

Spruce [174] 45 24-27a 9 nd nd 1 7 2

Spruce [21] 45 24-27 9 2 nd 1 7 6

Hardwood

Birch [171-172] 60 9 6 nd 6.5 7 nd 3

Eucalyptus grandis [175] 61 6 3 nd 9 1 <1 3

Eucalyptus grandis [22] 61 3 3 5 9 2 nd 3

Paulownia fortunei [20] 62 nd 11 3 nd 1 2 12

Grasses

Bamboo [27] 41 nd 4 nd nd 1 nd 6

nd = not determined, a = (etherified 19; phenolic 5-8), S = Spirodienone, D = Dibenzodioxocin


27

sinapyl aromatic rings prevent formation of 5-5 or dibenzodioxocin linkages, thus causing the hardwood lignin polymer to form more linear structures compared to softwood.

2.3 Lignin dissolution and pretreatment

The dissolution of lignin is critically important for its efficient valorization in a biorefinery and pretreatment of lignocellulosic biomass is needed in order to separate the lignin fraction from the carbohydrate fractions. Solubilization of lignocellulosic biomass as a whole remains a challenge because of the particular properties of the structures of cellulose, hemicellulose and lignin. Indeed, it has been recognized that the insolubility of wood in common solvents has severely inhibited efforts to valorize wood and its components. [28] Lignin solubility issues arise from the complex three-dimensional lignin network in wood, in which lignin is interlinked with the other lignocellulosic components and thus binds the entire wood architecture together (Figure 2.1). [28] The complex lignocellulosic structure serves to protect the plant species from microbial attack and provides resistance to the elements, yet it also makes the material recalcitrant to degradation by chemical reaction or fermentation to useful products. [29] Complicating the issue, the lignocellulose structure and composition varies significantly and depends on factors such as the plant species, plant parts, and growth conditions. [29] Non-covalent hydrophobic interactions between the aromatic rings in lignin may also hinder reactions between the lignin macromolecule, catalysts and reactants. [30, 31] These interactions are species-dependent and are found most commonly in softwood lignins, less in hardwood lignins, and not at all in straw lignins. [30, 32]

Disrupting this complex structure and separation into its components by dissolution or other physical processes reduces the resistance of the material to chemical reaction. [33] The efficient separation of wood into its main components, i.e. lignin, cellulose, and hemicelluloses, is therefore an important step in any biorefinery operation. [34] The identification and development of environmentally benign, cheap, convenient, and recyclable solvents and separation methods is currently the subject of many investigations. Cuprammonium hydroxide, DMSO/SO2, and DMSO/TBAF are some of the solvents commonly used in biomass dissolution. [35] DMSO/NMI was also found to effectively dissolve ball-milled wood, from which the lignin and carbohydrate fractions were separated by precipitation with dioxane/water.

Chapter 2

28

[36] Eckert and coworkers reported the use of CO2-expanded organic solvents to extract high value chemicals, such as vanillin, syringol, and syringaldehyde, from lignin. [37] The advantages of using CO2-expanded solvents include the ability to tune product solubility, transportability, and polarity of several standard solvents, and the quantity of solvent required to dissolve a given amount of biomass is reduced. [37] The separation and isolation of product components is facilitated by the ability to decrease component solubility such that the products fall out of solution when desired. [37] Ehara and coworkers used supercritical water to fractionate wood from Cryptomeria japonica and Fugus crenata into water-soluble and water-insoluble components via β-O-4 cleavage. [38] The water-insoluble components, which consisted mostly of lignin-derived products rather than carbohydrate-derived products, were treated with methanol, the components were found to have more phenolic hydroxyl groups than lignin in original wood. [38]

Ionic liquids have recently become very popular solvents for the dissolution of biomass. Ionic liquids are salts with melting points below an arbitrary set point of 100 ˚C. They have tunable physical properties based on the choice of cation and anion pair, and a negligible vapor pressure. [39] In a pioneering study by Rogers and coworkers, several ionic liquids, in particular 1-butyl-3-methylimidazolium chloride, were found capable of dissolving up to 10 wt% cellulose. [40] Hydrogen bonds formed with non-hydrated Cl-ions disrupt the intermolecular hydrogen bonding in the cellulose structure, allowing the dissolution of the cellulose. [39] Since this study, many researchers have investigated the dissolution of biomass, including lignin, using ionic liquids. [39, 41-43]

As indicated above, the pretreatment of lignin is an important initial step in any biorefinery operation. The pretreatment separates the principal components of biomass and related materials, degrades the extended polymer to small compounds, and occasionally causes other chemical transformations, such as the incorporation of sulfur. It is important to stress that the structure of the isolated lignin stream depends on the isolation method employed. Isolation/pretreatment methods that result in consistent types of lignin of high quality and purity next to high cellulose yields are highly desirable. A critical analysis of pretreatment technologies was published by Dale and coworkers, in which they divided the various pretreatment technologies into four categories: physical pretreatment (i.e. ball milling), solvent fractionation (including the organosolv process, which is described in more detail below, along with phosphoric acid fractionation and the use of ionic liquids), chemical pretreatment (acidic, alkaline, and oxidative), and biological treatment (using predominantly fungi).


29

[44] Gaspar and coworkers published a review focusing on polyoxometalates for the treatment of wood pulps. The feasibility of using these catalysts for delignification in the pulp and paper industry as an environmentally friendly replacement of chlorine-based chemical treatments has also been discussed. [45]

Pretreatment of lignin that takes place in the biorefinery will likely be aimed at separating all fractions of lignocellulosic material and obtaining pure feedstocks (i.e. organosolv); however, also lignin that was pretreated in the pulp and paper industry (i.e. kraft or lignosulfonate) is a likely candidate for further valorization. Each pretreatment method has both advantages and disadvantages, which will be discussed below. In addition, the various lignin pretreatments use different conditions (temperature, pressure, solvent and pH range) and degradation techniques that alter the chemical structure and linkages of the protolignin in different ways and to different extents. The choice of lignin pretreatment thus influences the types of high-value products that can be obtained from a lignin valorization process and is an important consideration for the integration of lignin in a biorefinery operation.

2.3.1 Kraft lignin processThe most common chemical pretreatment method is the kraft lignin process, which

employs high pHs and considerable amounts of aqueous sodium hydroxide and sodium sulfide at temperatures between 150-180 ˚C for about two hours to degrade lignin in a step-wise process. [46] Kraft lignin is currently produced commercially, for instance by MeadWestvaco, the world’s largest producer of kraft lignin, and on a pilot-scale by the LignoBoost technology, a process owned by Metso Corporation, in which lignin is extracted from pulp mill black liquor. [47, 48] The kraft lignin process, however, is highly energetically integrated and the pulp mill as a whole depends on intended use of the lignin as fuel for process heating. Increases in process efficiency can result in production of lignin in amounts that exceed the energy needs of the plant. This excess lignin can then serve as a feedstock for a biorefinery operation. [46] Ragauskas and coworkers have detailed the process chemistry surrounding kraft pulping including a description of the primary linkages in lignin and the ways in which these linkages are disrupted during the kraft process. [10] This review includes an overview of lignin degradation and condensation reactions, and it provides a discussion of the nature of residual lignin in kraft pulps. [10] The structural changes that occur to residual lignin on the cellulose fibers as a result of the subsequent chemical bleaching step during the kraft pulping process are described by Gierer et al. [49, 50] It is important to note that 5-5 linkages are highly refractory as they typically survive and are even

Chapter 2

30

formed during the kraft pulping process. Pulping is caused by nucleophilic attack on electron-deficient conjugated and carbonyl structures, while the bleaching of residual lignin is caused by electrophilic attack of electron-rich centers in aromatic nuclei and unsaturated, ring-conjugated side chains. [49, 50] A model structure for a kraft pine lignin fragment is presented in Figure 2.4. [46, 51] An NMR study of technical kraft lignin found most of the structures present in milled wood lignin. In addition, some new types of functional groups and linkages are introduced during the process, such as stilbenes, which are formed from the cleavage of α-aryl ether linkages of phenylcoumaran structures. [52] Contrary to earlier reports, neither diphenylmethane structures nor vinyl aryl ether structures could be detected by NMR spectroscopy. [52]

2.3.2 Lignosulfonate processThe sulfite pretreatment method that yields lignosulfonates is also relatively

common in the pulp and paper industry, and lignosulfonate lignin processes were reviewed by Lin and coworkers. [53] The lignosulfonate process is conducted between

Figure 2.4: Model depicting structural features characteristic of kraft pine lignin. [46, 51] *Contrary to earlier reports, a study showed no evidence for the presence of either diphenylmethane or vinyl aryl ether linkages in kraft lignin. [52]


31

pH 2-12 using sulfite with usually either calcium or magnesium as the counterion. [46] The product is typically soluble in water and in some highly polar organics and amines. [46] Lignosulfonate feedstreams derived from sulfite lignin treatment exhibit a higher average molecular weight and higher monomer molecular weights than kraft lignin as a result of incorporation of sulfonate groups in the struture. A model depicting the structural features of lignosulfonate lignin is depicted in Figure 2.5. Lignin obtained from a lignosulfonate process is currently commercially produced by Borregaard and can be used in a variety of products ranging from animal feed to construction materials. [54]

2.3.3 Organosolv processOrganosolv lignin is obtained by treatment of wood or bagasse, the fibrous residue

that remains after plant material (e.g. sugarcane) is crushed to extract juice or sap, with various organic solvents. [46] The Alcell process, no longer operational but previously

Figure 2.5: Model depicting structural features characteristic of lignosulfonate lignin. [46, 51]

Chapter 2

32

demonstrated at a Repap Alcell pilot plant, is the most well-known organosolv process, and it involved dissolution of lignin in either ethanol or ethanol/water mixtures. [55-57] Lignol Energy Corporation in Canada modified the pretreatment developed at the Repap Alcell pilot plant and began operation of a pilot plant to again produce organosolv lignin of high purity and potentially of high value. [58] CIMV developed a biorefinery based on the organosolv separation process in which the cellulose and hemicellulose, as well as the lignin can be obtained separately and used in various products. [59] The principal advantages of the organosolv process is that it forms separate streams of cellulose, hemicelluloses, and lignin, allowing valorization of all components of biomass. The process is generally considered environmentally friendly because it does not require the sulfides or harsh conditions used in the kraft or lignosulfonate processes. Organosolv lignin therefore typically has a very low sulfur content and is of higher purity than lignins obtained from other methods, which has important ramifications for its valorization to high-value chemicals. The principle disadvantage of the process is the high cost of solvent recovery. As a result of the solvent treatment, organosolv lignins generally have a high solubility in organic solvents.

2.3.4 Other processesThe steam explosion process involves steam impregnation under pressure followed

by rapid pressure release, which separates the lignocellulosic components and also ruptures linkages within the lignin structure. [60] In a typical treatment, wood or bagasse is exposed to steam at 13.8-34.5 bar and 180-230 ˚C for 1-20 min before the rapid pressure release. [46] The lignin molecular weight distribution in this process can be similar to the organosolv process. In addition, this process similarly uses no sulfur, and obtaining a separate cellulose stream is also possible. [46] Li and coworkers applied the steam explosion pretreatment process to both softwoods and hardwoods. [61] They noted that the process with SO2 pre-impregnation allowed efficient extraction of lignin from hardwood, but only low fractionation efficiencies were observed with softwoods. [61]

Several other methods for pretreating and isolating lignin are available, including the ammonia fiber explosion (AFEX) process [62] and the hot water process. More detailed information about these processes can be found in contributions by Bozell et al. [46] and Kamm et al. [63] The dilute acid process provides effective separation of the lignin from the other biomass streams but suffers from low yields and also corrosion of equipment by the acidic environment. [46] The alkaline oxidation process uses O2 or H2O2 to degrade lignin, which is then easy to recover. [46] The process suffers from slow delignification rates, which is the principle reason that the process


33

is not used extensively. Both the dilute acid and the alkaline oxidation method provide lignin fragments with similar weight distributions as the organosolv lignin. [46]

2.4 Catalytic lignin transformations

Following the biomass pretreatment, the lignin polymer is susceptible to a wide range of chemical transformations to convert it to valuable chemicals. As indicated above, the fragmentation reactions can be divided into lignin cracking or hydrolysis reactions, catalytic reduction reactions, and catalytic oxidation reactions. For lignin reductions, typical reactions involve the removal of the extensive functionalities of the lignin subunits to form less substituted monomeric compounds, such as phenols, benzene, toluene, or xylene. These simple aromatic compounds can then be hydrogenated to alkanes for fuel applications or used as platform chemicals for the synthesis of bulk and fine chemicals using technology already developed by the petrochemical industry. For lignin oxidations, lignin is converted to more complex platform chemicals with a greater degree of functionalization or converted directly to target fine chemicals (see Figure 1.1 of Chapter 1).

The complexity and variability of the lignin structure have prompted the use of several simplified, low molecular weight lignin model compounds to study catalytic lignin valorization. The use of lignin model compounds serves several primary purposes. The first purpose is that they contain linkages that resemble those found in the lignin polymer and thus their reactivity provides insight into the degradation and reaction of the polymer structure as a whole. A second reason for using model compounds is that similar molecules are often found in lignin degradation streams after depolymerization of the lignin polymer; methods for their valorization to high-value chemicals are therefore relevant for the development of two-step conversion processes. Third, the use of model compounds presents fewer analytical challenges relative to the complex lignin polymer and the large amounts of products that can be obtained therefrom. Because often only one type of linkage is contained in the model compound, analysis of the reaction paths, and thus catalytic performance, is similarly simplified. Several examples of lignin model compounds that are often used are depicted in Figure 2.6 and 2.7.

Chapter 2

34

Figure 2.6: Examples of monomeric lignin model compounds.

2.4.1 Lignin catalytic cracking and hydrolysisDisruption of the complex lignin polymer into smaller subunits is an important

part of a two-step approach to lignin valorization. The smaller fractions produced better resemble the model compounds and target products depicted above, and they expose various functional groups on the aromatic rings to further catalytic transformations. Amen-Chen and coworkers published a review of the production of monomeric phenols by thermochemical lignin conversion. [5] Several routes to phenolic compounds were described, including the pyrolysis of monomeric, dimeric, and trimeric compounds, in addition to the effects that different conditions have on forming methane, methanol, and various other compounds from biomass. [5] Several transition metal-catalyzed processes were also reviewed, including kraft lignin pyrolysis by ZnCl2. [5] Dorrestijn and coworkers published a review detailing the pyrolysis of lignin with a brief discussion of pyrolysis, catalytic hydrogenation and oxidation. [64] Britt et al. studied flash vacuum pyrolysis of methoxy-substituted β-O-4 lignin model compounds in order to provide mechanistic insight into the relevant reaction pathways. [65] The reactions were dominated by free radical reactions, molecular re-arrangements, and concerted eliminations. [65] Misson and coworkers


35

Figure 2.7: Examples of models for typical lignin linkages.

investigated the pretreatment of Empty Palm Fruit Bunches with NaOH, H2O2, and Ca(OH)2 before the lignin was subjected to catalytic pyrolysis over Al-MCM-41 and H-ZSM-5 to give phenolic yields of 90 wt% and 80 wt% yield, respectively. [66] Some more recent examples of catalytic lignin pyrolysis are discussed in Chapter 1. Li and coworkers studied the depolymerization/repolymerization of lignin during steam treatment of aspen wood. [67] They found that addition of a carbonium ion scavenger, such as 2-naphthol, suppresses the repolymerization reaction to give a more uniform and more easily extractable lignin of low molecular weight. [67] As indicated above, controlling the repolymerization rate is important for efficient lignin valorization.

2.4.1.1 CrackingCracking is a practice commonly employed in petroleum refineries to convert

higher-boiling hydrocarbons into more valuable products by C-C bond cleavage. [68] Fluid catalytic cracking is among the most important of catalytic processes, contributing between 20% and 50% of the blending components in the gasoline pool of a refinery. The process uses highly-optimized zeolites as catalyst to achieve the C-C bond cleavage in an acid-catalyzed reaction. In the hydrocracking process, the catalytic cracking of heavy oil fractions is combined with a hydrogenation/hydrogenolysis step; reactions are in this case run under elevated partial hydrogen pressure. The catalysts used in hydrocracking are predominantly bifunctional, combining a support active in cracking with a (noble) metal for the hydrogenation reaction. The hydrogenation catalyst is typically composed of cobalt, tungsten, palladium, or nickel, and the cracking component typically consists of zeolites or amorphous silica-alumina with various compositions. [68] Lignin can also be treated with hydrocracking catalysts, resulting in cleavage of the β-O-4 bond and some of the more unstable carbon-carbon bonds. [68] The resulting lower molecular aromatic compounds are then susceptible

Chapter 2

36

to further conversion to valuable products. Several catalysts crack lignin into low molecular weight compounds. Huber and

Corma included two examples of the catalytic cracking of lignin in their review of bio- and petrochemical refineries. [69] Sharma and Bakhshi reported on the catalytic cracking of pyrolytic lignin [70] or bio-oil produced by liquefaction [71] using H-ZSM-5 as a catalyst and temperatures in the range of 340-410 ˚C in a fixed bed reactor. The products were distilled, and the maximum amount of organic distillate was 30 wt% of pyrolytic lignin, and nearly 60 wt% of the bio-oil was obtained as useful chemical byproducts. [70] H-ZSM-5 and H-mordenite produced more aromatic than aliphatic hydrocarbons from fast pyrolysis bio-oil, whereas H-Y, silicalite, and silica-alumina produced more aliphatic than aromatic hydrocarbons. [72] H-ZSM-5 was most effective for the production of an organic distillate fraction and aromatic hydrocarbons relative to the other zeolites investigated. [73] Similarly, H-ZSM-5 produced the highest yield of a deoxygenated liquid fraction and aromatic and naphthenic compounds relative to other zeolites. [74] Chantal and coworkers used H-ZSM-5 to hydrotreat pyrolytic oil and found that the percentage of coke formed is mostly dependent on the flow rate of the oil, whereas the percentage of unreacted tar is a function of both temperature and flow rate. [75] The presence of methanol in the oil decreased coke formation. [75] Gayubo and coworkers investigated the effects of temperature and space-time on the conversion of model compounds mimicking products obtained from the flash pyrolysis of vegetable biomass using H-ZSM-5. [76] Alcohols (principally propanol and butanol) transform into gasoline hydrocarbons and light olefins similarly to ethanol and methanol, yet phenol and 2-methoxyphenol were less reactive and led to coke formation. [76]

The product distribution observed in these cracking reactions is thought to result from a series of reactions whereby non-volatile compounds are first cracked to heavy volatile compounds, which are subsequently cracked to volatile alkyl aromatics and ultimately to coke and gas. [70] The two main reaction pathways are either thermal, resulting in the formation of light and heavy organic compounds and polymerization to form char, or thermocatalytic, resulting in a range of processes including deoxygenation, cracking, cyclization, aromatization, isomerization, and polymerization. [72] The performance of the catalysts indicated above is strongly dependent on the structural characteristics of the catalyst, including framework and the presence and strength of acid sites. Hydrocarbon formation, for example, occurs more readily with H-ZSM-5 compared to silicalite. [72] These catalysts have similar frameworks but differ in that acid sites are present in the former but absent in the


37

latter. [72] The formation of aromatic compounds and coke are possibly linked to catalyst structure since aromatic products are observed with the zeolite catalysts but only low quantities of aromatics are observed with amorphous silica-alumina, and the effectiveness of the catalysts in reducing coke formation decreased with increasing pore size. [72] Excess water was found to have an adverse effect on catalyst performance by decreasing the number of acid sites in the catalyst. [70] Char and tar formation, which are thought to occur via the polymerization of heavy and non-volative bio-oil components, are likely temperature related, and the zeolite catalysts typically reduce char and tar formation more readily at elevated temperatures. [70, 72] Caution is required, however, to avoid decomposition of the lignin fragments of bio-oil, which is also favored at elevated temperatures. [70]

Other catalysts than zeolites were also reported to be active for the catalytic

cracking of biomass. Sheu and coworkers performed a kinetic study on the upgrading of pine pyrolitic oil produced from southern pine sawdust and bark in a trickle bed reactor using Pt/Al2O3-SiO2 and sulfided CoMo/Al2O3, Ni-W/Al2O3, and NiMo/Al2O3. [77] The reactions were conducted between 350 to 400 ˚C and at 51.7-103.4 bar. [77] Two models for oxygen removal and for compositional changes in the bio-oil were developed, and it was found that Pt/Al2O3-SiO2 had the best hydrotreating ability of the catalysts analyzed. [77] Supported or non-supported Pt-modified superacid catalysts, such as Pt/SO4

2-/ZrO2, Pt/WO42-/ZrO2, or Pt/SO4

2-/TiO2 and similar combinations, were demonstrated as effective hydrocracking catalysts at 350 ˚C and 103.4 bar H2. [78] Products included predominantly C1-C2 alkyl-substituted phenols and methoxyphenols or C3-C4 alkyl-substituted phenols depending on the methanol/lignin ratio in the preceding, mild base-catalyzed depolymerization step. [104] UOP LLC patented a process for the treatment of lignin and cellulosic biomass to produce aromatics to be used for fuel applications, as well as precursors for the chemical industry. [79] The lignocellulosic biomass is first dispersed with e.g. glycerol after which it is treated under a hydrogen atmosphere (300-400 ˚C, 33-68 bar H2) with a metal-loaded large-pore zeolite or a sulfided NiMo hydroprocessing catalyst claimed to produce various phenols and aromatics, amongst other products.

2.4.1.2 HydrolysisLignin hydrolysis also has been the focus of several investigations. Karagoez and

coworkers described the use of Rb and Cs carbonate solutions to treat pine sawdust to form phenolic compounds. [80] They found that more oil, consisting of a mixture of oxygenated lignin products, was produced with Rb2CO3 than with Cs2CO3 as a

Chapter 2

38

catalyst. [80] The base catalysts hindered char formation and favored the formation of (methylated) catechols and 2-methoxyphenyl products. [80] Thring explored the depolymerization of Alcell lignin by alkaline hydrolysis. [81] Between 7 and 30% conversion of Alcell lignin was obtained to yield a concentration of 4.4% phenols mostly consisting of syringol (2.4%). [81] Several years later, Miller and coworkers performed alkaline hydrolysis of Alcell lignin using KOH in supercritical methanol or ethanol. [82] Only 7% of the ether-insoluble material was left in the KOH/methanol solution after 10-15 min at 290 ˚C. [82] The reaction was favored by strong bases, and combinations of bases gave either positive synergistic effects, such as with NaOH and Ca(OH)2, or negative synergistic effects, such as with LiOH or CsOH with Ca(OH)2, as indicated by the relative decrease in insolubles. [82] Model compound studies indicated that the principle route for lignin depolymerization was through solvolysis of the ether linkages. [82] Nenkova and coworkers described the alkaline depolymerization of technical hydrolysis lignin and poplar wood sawdust. [83] Isolated products from extraction with toluene included several high-value products commonly obtained from lignin oxidation, such as 2-methoxyphenol, 4-hydroxy-3-methoxybenzaldehyde, 2,6-dimethoxyphenol, and 1-(4-hydroxy-3-methoxyphenyl)ethanone. [83]

Several examples of lignin fragmentation by supercritical water were reported. Supercritical water has several advantageous properties that make it suitable for use as a solvent for lignin valorization. It is, for example, completely miscible with light gases, hydrocarbons, and aromatic compounds. [84, 85] Reactions with biomass containing a relatively high water content are possible without the need to dry the feedstock, and several organic decomposition and formation reactions, such as oxidations and hydrolysis, have been reported to occur without a catalyst. [84, 85] In addition, supercritical water has a relatively low viscosity, high diffusivity, and a dielectric similar to many organic solvents, but with the advantage of thermal stability. [86] The separation of organic products formed during reactions from the water is done with relative ease. [86] The principle disadvantages include the relatively high temperatures and pressures required to reach supercritical conditions (Tc = 374 ˚C, Pc = 221 bar), [71] and char formation can be problematic. Wahyudiono and coworkers used supercritical water in a batch reactor to decompose the lignin model compound catechol, where it was found that manipulating the temperature and pressure of the supercritical water controlled the reaction rate to forming phenol. [84, 85] Watanabe and coworkers used supercritical water as a solvent for the NaOH and ZrO2-catalyzed partial oxidative gasification of organosolv lignin, where it was found


39

that the presence of both ZrO2 and NaOH greatly increased the H2 yield. [87]

2.4.2 Lignin reductionThe catalytic hydrogenation of lignin and its model compounds has been studied for

many years and is the subject of several publications. With regard to reductive lignin depolymerization, the emphasis of the reported studies is mainly on the production and upgrading of bio-oils and fuels, although the production of phenols as chemical commodity is also considered. The replacement of petrochemical-based routes for the production of bulk aromatic compounds, such as benzene, toluene, xylene (BTX) as well as phenol, by renewable routes has nonetheless received relatively little attention. [88] Since approximately 60% of all aromatics produced by a typical integrated chemical production center are first generation unfunctionalized aromatics, the conversion of biomass and lignin in particular to BTX therefore deserves more attention. [88]

In this section we review those studies that are primarily concerned with the production of phenols and/or aromatics from lignin or lignin model compounds via catalytic hydrogenation or hydrodeoxygentation.

Most reports on the hydrotreatment of biomass-derived feeds are focused on either bio-oil production or upgrading (hydroprocessing) because chemical conversion is required to turn such bio-oils into useful transportation fuels. Hydrotreatment is employed to increase the thermal stability and volatility of the oil and to reduce viscosity through oxygen-removal and lowering of the molecular weight. [89] (Fast) pyrolysis studies are generally aimed at maximizing production of liquid products. In this respect it has been found that reducing conditions, i.e. the presence of hydrogen or hydrogen-donating compounds, are beneficial and lead to higher yields and to less coke formation in pyrolysis processes. [90] Studies dealing with the direct conversion of biomass to bio-oils (by fast or slow pyrolysis, liquefaction, etc), i.e. those that do not deal directly with lignin or lignin-related models, are not included. A review on wood/biomass pyrolysis for bio-oil production is available. [91] For general developments in the field of catalytic hydroprocessing of bio-oils the reader is referred to work by Elliott [89] or to Briens et al., [92] and Behrend et al. [93] Furimsky has published a review on catalytic hydrodeoxygenation from a broader perspective. [94] An early review on mechanistic aspects, reaction pathways and kinetics of catalytic hydroprocessing, including hydrodesulfurization, hydrodenitrogenation, and hydrodeoxygenation, is also available. [95] The actual composition of bio-oils is complex, and a multitude of compounds have already been identified. [91] In fact, component analysis and determination of chemical reactivity of the identified components for upgrading

Chapter 2

40

purposes is an active field of research. Indeed, some studies are concerned with lignin-related model compounds and aim to further understand the processes governing bio-oil upgrading. As such, they provide important insights into the chemical pathways for the conversion of lignin into valuable chemicals. Most of these efforts are concerned with the hydrotreatment of (a multitude of) oxygen-containing model compounds rather than actual lignin or bio-oil feeds. In terms of catalyst development, an optimal catalyst for the conversion on lignins into phenols or aromatics should have the following characteristics: high conversion at modest temperatures to minimize char formation and competitive thermal condensation reactions, high selectivity to phenols to prevent excessive hydrogen consumption, tolerance to water, the ability to deal with various lignin streams, and possibly the capability for dealkylation (side chain removal).

2.4.2.1 Heterogeneous catalysts for lignin reductionEarly studies on lignin heterogeneous hydrogenation were mainly aimed at

structure elucidation of the complex lignin polymer. A catalytic reaction of hardwood lignin with hydrogen was reported by Harris and coworkers as early as 1938, in which lignin was found to react with hydrogen over copper-chromium oxide. [96] The rather harsh hydrogenation conditions led to full reduction of the aromatic rings to yield mainly some monomeric (substituted) propylcyclohexanols and methanol. Other early studies on lignin hydrogenation also included the use of Raney-Ni as a catalyst, in which syringol and guaiacol components were isolated. [97, 98] Pepper et al. studied the influence of a number of catalysts (i.e. Raney-Ni, Pd/C, Rh/C, Rh/Al2O3, Ru/C and Ru/Al2O3) on softwood lignin (spruce wood) hydrogenation. A significant amount of the original lignin was converted into the monomeric products 4-propylguaiacol and dihydroconiferyl alcohol under mild conditions (34 bar, 195 ˚C), with Rh/C giving the highest yield. [99] The Pd/C-catalyzed reaction yielded mainly dihydroconiferyl alcohol (representing some 24% of the lignin), while 4-propylguaiacol was found in addition to dihydroconiferyl alcohol with Rh/C (together accounting for some 34% of the original lignin). The observed product distribution thus implies that Rh/C is capable also of cleaving other linkages than those cleaved by Pd/C. Reaction conditions, such as catalyst loading, hydrogen pressure and pH of the medium, were optimized. [100] The nature of the obtained products was influenced by catalyst loading, as higher loadings resulted in over-hydrogenation and degradation reactions, as well as by variation of the pH. Hardwood lignin (aspen wood), on the other hand, gave mainly the corresponding syringyl and guaiacyl compounds bearing a propyl or propanol group with Rh/C, accounting for about 40 wt% of the original lignin. These


41

results clearly reflected the differences in building block composition of the hard- and softwood lignin polymers. [101] Since these initial reports, many more examples of catalytic hydrogenation have been reported, and a summary of the catalytic systems used for lignin model compound hydrogenation is given in Table 2.2 and the data for lignin hydrogenation as summarized in Table 2.3.

Workers at the Japanese Noguchi Institute worked on lignin liquefaction for phenol production and discovered an active catalyst in the early 1950s. This discovery led to the patented Noguchi process, in which it was claimed that a mixture of C6-C9

monophenols could be obtained upon hydrogenolysis in yields as high as about 40%. [102] An iron(II) sulfide catalyst with a co-catalyst of at least one sulfide of copper, silver, tin, cobalt, chromium, nickel, zinc, or molybdenum (e.g. Fe-S-Cu-Zn in a ratio of 10:12:1:1) was used, and the reaction was conducted in a solvent such as lignin tars and phenols at 250-450 ˚C with an initial hydrogen pressure of 152-456 bar. The high yields of monomeric phenols were in part caused by alkylation of the phenolic solvent during the process, but nonetheless a lignin-derived phenol yield of 21% was obtained. The process was extensively evaluated in a multitude of its variants but suffered from difficulties in reproducibility regarding the production of high yields of monophenols. Although it was concluded that the process remained the best one for lignin liquefaction to that date, [103] the process was economically unattractive because of the kind of lignin used, the relatively low economic value of the monophenol product mixture, and the loss of phenol itself when used as a solvent. Around the same time, Inventa A.-G. patented a similar process consisting of decomposing lignin into distillable products containing a substantial amount of phenols using iron sulphate as the hydrogenation (pre)catalyst. [104]

Urban et al. later claimed a modification and improvement on the Noguchi process, affording up to about 45% cresols and about 65% monophenols from alkali lignin from the kraft process. Cresol yield was substantially increased by the addition of methanol, which is important since cresols might be the source of maximum economic return from the liquefaction of lignin. The catalyst is generated in situ and consists of ferrous sulfide with smaller amounts of other metal sulfides as promoters. [105]

2.4.2.2 CoMo- and NiMo-based hydrodeoxygenation catalystsHeterogeneous catalyst systems that have been studied most extensively for lignin

hydrogenation are conventional cobalt- and nickel-promoted molybdenum catalysts. Indeed, already in 1970 Alpert and Shuman patented a process for the production

Chapter 2

42

of chemicals from lignin using a CoMo/Al2O3 catalyst. [106] The initial interest in the hydrocracking and hydrodeoxygenation activity of these catalysts rested on the fact that synthetic oils, either from coal or biomass, can have an oxygen content well in excess of 10% and can even approach 50% for biomass feeds. [94] Application of biomass-derived hydrocarbons requires the removal of oxygen from the feed. [93] The well-established hydrotreating catalysts originally developed for the removal of sulfur (HDS) and nitrogen (HDN) from conventional oil feed for purification and upgrading processes proved a useful lead for the removal of oxygen (HDO) from biomass-derived product streams.

Indeed, these conventional catalysts are the most studied systems also for reductive lignin conversion. Elliot published an early study in which a variety of commercial catalysts (CoMo, NiMo, NiW, Ni, Co, Pd, and CuCrO) were screened for phenol hydrogenation/hydrodeoxygenation activity; the sulfided CoMo catalyst provided the best results, giving the highest yield of benzene (34%) at 400 ˚C. [107] The superiority of the CoMo system in terms of hydrodeoxygenation activity with retention of aromaticity would later be confirmed in many of the earlier studies. With the aim of obtaining high-quality gasoline, an advanced process concept was eventually developed by the same group in which hydroxyaromatic compounds were converted into monoaromatics by dehydroxylation, while avoiding ring saturation. [108]

As discussed in Chapter 1, two strategies can be discerned if one aims for the production of valuable bulk chemicals from lignin. In the first one-step approach, various (oxygenated) aromatics can be produced by depolymerization of lignin itself. In this case, the product stream will still have a fairly high oxygen content and products will generally resemble the original building blocks of lignin, i.e. consists of molecules containing substituted phenol, guaiacyl, and syringyl moieties (Figure 2.8). In a two-step approach, catalyst systems will also need to be developed for the conversion and further upgrading of product streams of already degraded lignin. Irrespective

Figure 2.8: Syringyl, guaiacyl and phenol-derived moieties.


43

of the way in which the macromolecule was depolymerized, these systems should be able to convert the obtained mixtures of smaller oxygenated aromatic fragments into e.g. alkylated benzenes or phenol. It is important to note that in this case the goal is to keep the aromaticity of the feed intact. This is in contrast with typical hydrodesulfurization and hydrodenitrogenation processes, in which the heteroatom is usually removed after full hydrogenation of the aromatic component. Many of the studies on model compounds that mimic bio-oil components are relevant with respect to the second route and will also be discussed below.

Initial studies focused on the hydrogenolysis and hydrocracking of the carbon-oxygen bonds of simple aromatic model compounds, such as phenol (1), o-cresol (5), anisole (2), and guaiacol (4) (Figure 2.6). [109, 110] Some basic aspects of (substituted) phenol HDO were noted. First of all, the reaction can follow two mechanistically different paths, as HDO can be accomplished by either direct deoxygenation of phenol leading to aromatics or by ring hydrogenation followed by deoxygenation, the latter route being favored at higher pressures. The position of substituents on the ring was found to influence reactivity, as ortho-substitution led to lower activity, which was ascribed to steric hindrance. [110]

Bredenberg et al. reported that phenol and o-cresol proved to be quite stable under mild hydrocracking conditions (sulfided NiMo/SiO2-Al2O3 catalyst, 250-350 ˚C, 50 bar H2). A strong influence of temperature was observed for guaiacol, not only on conversion but also on product selectivity. Depending on the temperature chosen, guaiacol was mainly converted into phenol (at more elevated temperatures) and/or catechol (3, Figure 2.6), illustrating the possibility for control over selectivity in this process. Anisole mainly yielded phenol, o-cresol and 2,6-dimethylphenol with constant selectivities over a 250-350 ˚C temperature range. Oxygen-free aromatics only appeared at higher temperatures (over 300 ˚C). The lower reactivity of guaiacol compared to anisole was attributed to a stronger interaction between the substrate and the support. At higher temperatures rapid deactivation of the catalyst was observed, thought to be caused by the large amounts of water released, the loss of sulfur and excessive coke formation. Below 250 ̊ C catalyst activity remained constant for over 50 h, though. [109] These and other studies demonstrated the feasibility of oxygen removal at conditions far less than required for thermal fragmentation and deoxygenation. [111]

Similar observations were made by Hurff and Klein for a sulfided CoMo/Al2O3catalyst

Chapter 2

44

(at 250 ̊ C, 34.5 bar H2) as guaiacol conversion also mainly yielded catechol and phenol, after a primary demethylation reaction was followed by dehydroxylation. Further hydrodeoxygenation of phenol yielded benzene and cyclohexane. [112] Anisole gave phenol as the only primary product, with subsequent conversion to benzene and cyclohexane. No ring methylation was reported in this case. Guaiacol conversion was much faster than for anisole, indicating that the electronic effect of the ortho-methoxy substituent is stronger than any steric hindrance it may cause. Guaiacol was also found to be more prone to coke formation than anisole. At a slightly higher temperature and pressure (325 ˚C, 50 bar H2) excellent conversion of guaiacol and good selectivity to phenols (77%) were reported. In this case, the hydrodeoxygenation reaction was found to coincide with significant ring methylation. [113] A strong temperature dependence was observed as catechol rather than phenol became the major product at 275 ˚C. A comparison between the different methoxyphenol isomers led to the suggestion that the different reactivities are the result of different adsorption modes on the catalyst surface, with the guaiacol isomer adsorbed in an inclined rather than flat mode.

Kallury et al. tested the hydrodeoxygenation activity of a NiMo/Al2O3 catalyst on a number of substrates, including phenol, catechol, guaiacol and syringol. [114] Catechol proved to be more reactive than phenol itself at 350 ˚C, with loss of one hydroxyl group to give phenol as the major product. The addition of methanol to the reaction mixture resulted in significant amounts of ring-methylated products. The addition of methanol and of water was, however, found to limit the activity of the catalyst. The deactivating effect was suggested to arise from competitive adsorption and blocking of the active sites. Water formation was also thought to be responsible for the reduced deoxygenation ability, associated with the molybdenum sites, of the catalyst over several runs. Interestingly, the hydrogenation properties of the catalyst, evidently due to nickel, were not affected. The results obtained with guaiacol resembled those of catechol. Alkylated phenols are also detected without the addition of methanol, which again illustrates that cleavage of the aryl methyl ethers is a facile process. Syringol is equally reactive and demethylation and dehydroxylation proved rather efficient. The ring hydrogenation activity of the NiMo catalysts was limited at 350 ˚C under these conditions. [114]

2.4.2.3 Support effects in CoMo and NiMo HDO catalystsIt was noted that the studies of Bredenburg and Kallury were done with poorly

sulfided catalysts and therefore might not present optimal results. [115] Laurent et al. attempted to address this issue by reporting a comparison of the HDO of guaiacol


45

with both CoMo and NiMo catalysts. [115] Importantly, they found that the alumina support itself also showed catalytic activity, as 37% of the guaiacol substrate was converted to catechol with alumina alone. [115] A comparison was made between CoMo and NiMo catalysts using a mixture of reagents typical of bio-oil composition, which included guaiacol. The results confirmed that catechol was formed first as the primary product, followed by dehydroxylation to phenol (300 ˚C, 70 bar H2). No significant methylation was observed in this study, however. A higher activity was found for the NiMo catalyst, but the CoMo catalysts showed a higher selectivity for the production of catechol and phenol. [115] Indeed, side reactions are more pronounced with the NiMo catalyst, which was also reflected by a poorer mass balance at similar conversion. The impossibility to close the mass balance for both catalysts was attributed to the formation of heavy products or coke. Given the propensity of guaiacol and catechol-like compounds to form polycondensation products and coke along with the relatively strong interaction of these compounds with the common alumina support, other neutral supports, such as carbon and silica, were subsequently tested as well. [116] Although both alternative supports show 3-6 times lower activity compared to alumina, further confirming the involvement of acid sites in guaiacol conversion, the carbon-supported catalyst produced phenol faster than the alumina-supported one, resulting in a phenol/catechol ratio seven times greater than that for the alumina-supported catalyst. [116] The latter result indicates that phenol might be produced directly from guaiacol with this catalyst by elimination of the methoxyl group. Furthermore, the results clearly indicated that coking of guaiacol takes place on the acid sites on alumina. On the other hand, guaiacol conversion was lower for the silica- and carbon-supported catalysts, confirming that acid sites are involved also in the steps leading to product formation. The involvement of acid sites is further corroborated by the fact that guaiacol conversion is highly inhibited by ammonia. [117] De la Puente et al. noticed negligible coke formation also on an activated carbon-supported CoMo catalyst. Activated carbons differing in the nature and amount of functional groups on the surface were tested, but rather similar results were obtained in the guaiacol HDO reactions. Moderate conversions of guaiacol (35% at 280 ˚C, 70 bar H2) led predominantly to the formation of catechol and, to a lesser extent, phenol. [118] The phenol/catechol ratio was found to depend on the extent of oxygen-functionalization of the amphoteric support. The non-modified, least acidic carbon gave a higher yield of phenol, further confirming the involvement of acid-sites in the guaiacol-to-catechol-to-phenol route, but not in the direct conversion of guaiacol to phenol. Ferrari et al. showed that the nature of the carbon support and the impregnation order of the metals (CoMo or MoCo) both have an influence on catalyst

Chapter 2

46

activity and selectivity in the hydrodeoxygenation of various substrates, but no clear trends emerged for guaiacol hydrodeoxygenation. [119]

Support effects in hydrotreating catalysts are well known and have been extensively studied for hydrodesulfurization and hydrogenation processes. [120] For these hydrotreating catalysts, it has been well established that the use of supports other than alumina (e.g. carbon, titania, mixed oxides, zeolites or clays), can lead to enhanced catalytic properties. For hydrodeoxygenation purposes, this topic has been less well explored.

The origin of the the almost exclusive use of alumina as support can be ascribed to its very good textural and mechanical properties and its relatively low cost. [120] It is well known, however, and clearly confirmed by the previously discussed results that the support is not an inert carrier and that other supports should be explored as well. An additional reason for the use of alternative supports, in particular with respect to the conditions encountered in HDO, is the possible instability of alumina in the presence of high levels of water as is also shown in Chapter 4. [121] Alumina is known to be metastable under hydrothermal conditions, for instance, and partially transforms into boehmite under processing conditions. A limited number of studies on support influence on HDO activity and selectivity have been reported. For instance, the (Lewis) acidity of the alumina support was shown to be an important characteristic for the observed demethylation activity with the (substituted) guaiacol substrates. In addition to this reaction taking place on the metal, it is also thought to occur on the support surface (albeit via a different mechanism). Indeed, anisole demethylation activity (and subsequent ring methylation of the phenol product) could be partly blocked by selective poisoning of the acid sites of the support with pyridine. [122] The hydrodeoxygenation and hydrogenation selectivity was not markedly affected, however.

MgO was also tested as a support for CoMo catalysts, [123] with the aim of promoting the dispersion of the (acidic) MoO3 precursor on the basic support and inhibiting coke formation. Hydrotreatment of phenol was conducted using a sulfided CoMo/MgO catalyst in supercritical hexane (350-450 ˚C, 50 bar H2). Additionally, phosphorus-doped analogues (CoMoP/MgO) were also tested. Phosphorus doping has been commonly used in attempts to improve the activity of MoS2-based hydrotreating catalysts in hydrodenitrogenation and hydrodesulfurization studies, but information on its effect on CoMo catalysts for hydrodeoxygenation is limited. Both systems


47

proved to be effective for phenol hydrodeoxygenation, with CoMoP/MgO giving superior activity and yielding mainly benzene and some cyclohexyl-aromatics as the products. The MgO-supported catalysts also showed good resistance to coking. [123]

2.4.2.4 CoMo and NiMo-catalyzed HDO of more complex model compoundsPetrocelli et al. expanded the scope of the hydrodeoxygenation of lignin model

compounds by studying the hydrotreatment of 4-methylguaiacol (8), 4-methylcatechol (7), eugenol (10), vanillin (9), o,o’-biphenol (11), o-hydroxydiphenylmethane (13), and diphenyl ether (14) (Figure 2.7 and 2.8) over a sulfided CoMo/Al2O3 catalyst (at 250-325 ̊ C, 69 bar H2). The latter three substrates mimic some of the more thermally stable linkages found in lignin. [111] For the monoaromatic compounds, hydrodeoxygenation proceeded predominantly by demethylation followed by dehydroxylation to a monohydroxyl-substituted intermediate, which finally undergoes dehydroxylation to an aromatic hydrocarbon; saturation is observed but only to a minor extent. The methyl and propyl substituents of the reactants and products appeared to be quite stable. This observation is in agreement with previous findings of Odebunmi et al. after hydrodeoxygenation of cresol using a CoMo/Al2O3 catalyst. [124] Encouragingly, the results show that substituted guaiacols and catechols readily react to form thermally stable phenols during hydrodeoxygenation at 300 ˚C, with possible yields of total single-ring phenols of about 60%. Importantly, char formation was greatly reduced in comparison to pyrolysis. Reasonable activity was observed towards hydrocracking of the interaromatic ring bonds in the di-aromatic substrates with dehydroxylation both preceding and following breakage of the interunit link, although the C-C bond in o,o’-biphenol proved stable under the conditions employed.

2.4.2.5 Hydrodeoxygenation over other transition metal catalystsKoyama reported an extensive comparison of iron and molybdenum catalysts

in the hydrogenation of various model compounds containing different kinds of ether bonds. The hydrocracking of lignin model dimers using Fe2O3S, Fe2O3/Al2O3S, NiOMoO3/Al2O3, and MoO3/TiO2S between 340-450 ˚C was described. [125] The Mo catalysts significantly increased the bond cleavage between the aromatic rings of 4-hydroxydiphenyl ether, diphenyl ether, and diphenylmethane, whereas the Fe2O3/Al2O3S catalysts only slightly promoted bond-cleaving between the aromatics of these compounds. [125] The higher activity of the molybdenum-based catalysts led to the conclusion that these catalysts are more likely to give higher monophenol and benzene yields in lignin hydrocracking processes. [125]

Shabtai et al. presented a thorough investigation into the activities of first-, second-

Chapter 2

48

and third-row transition metals as possible promoters for supported molybdenum sulfide catalysts. [126] Although their initial interest was in the preparation of new catalysts with improved hydrodeoxygenation activity for coal- and peat-derived liquids, the hydrodeoxygenation results of the model compound studied are also relevant for lignin product streams. The systematic study consisted of C-O bond hydrogenolysis of diphenyl ether of a series of sulfided M-Mo/Al2O3 catalysts (M = Cr, Fe, Co, Ni, Ru, Rh, Pd, Re, Ir, or Pt, at 350 ˚C, 138 bar H2). The corresponding M/Al2O3 catalysts (i.e. without molybdenum) were also tested. The CoMo, RhMo, and RuMo catalysts showed the highest hydrogenolysis activity, in that order, although considerable ring hydrogenation activity was also observed with CoMo (see Figure 2.9). The RuMo catalyst provided the highest selectivity for hydrogenolysis. The NiMo catalyst yielded the lowest hydrogenolysis selectivity, mainly because of high ring hydrogenation activity.

2.4.2.6 Direct hydrodeoxygenation of ligninRatcliff et al. studied a sulfided CoMo/Al2O3 catalyst for the hydrodeoxygenation

of both model compounds and organosolv lignin in a batch reactor with the aim of

Figure 2.9: Variation in C-O hydrogenolysis activity (k1) for MMo/Al2O3 catalysts as a function of periodic table position of M, adapted from Shabtai et al. [126]

VI VII VIII1 VIII2 VIII30

5

10

15

20

25C-

O h

ydro

geno

lysis

acti

vity

(k

₁),cm

³/g m

in

Periodic Table Position

1st row

3rd row

2ⁿd row

Co

NiPdPt

Ir

Rh

Ru

Fe

MoReCr


49

converting the obtained (substituted) phenols into phenyl methyl ethers, which can be blended into gasoline. [127] Various parameters were tested and 1-methylnaphthalene was used as the solvent, which, as was later discovered, acted as a reagent rather than as an inert solvent. Phenol yield, although low overall (10 wt%), could be considerably improved if the reaction was carried out under flow rather than batch conditions. A significant amount of char was formed as well in all the lignin hydrodeoxygenation experiments (14 wt% of lignin charged). The model compound 4-propylguaiacol could, depending on the temperature used, be converted into catechols (< 300 ˚C), phenols, or saturated and aromatic hydrocarbons (> 400 ˚C). The use of a NiMo catalyst on a more acidic support, a phosphated alumina, resulted in higher dealkylation activity and subsequently higher yields of cresols and phenol.

Related to this study is a patented hydrocracking process in which kraft lignin is used to produce monoaromatic phenol-containing products from lignin-containing feedstocks. The Hydrocarbon Research Institute’s (HRI) Lignol process combines a hydrotreatment step in an ebullated catalyst bed reactor with a subsequent thermal dealkylation step. Using a catalyst comprised of a (Co or Ni promoted) iron or molybdenum oxide on alumina, a 37.5% yield of phenols is claimed (wt% based on organic content of lignin). [128] The yield of monophenols by the HRI process is quite high and corresponds to approximately 60% of the aromatic rings making up the lignin molecule. It exceeds the conversion obtained in the Noguchi process, although the HRI results have not been independently confirmed.

Shabtai et al. patented a two-stage, catalytic reaction process for the conversion of lignin into a reformulated hydrocarbon gasoline product with a controlled amount of aromatics. [129] The (wet) lignin material is first subjected to a base-catalyzed depolymerization step in a supercritical alcohol, followed by a two-step hydroprocessing reaction to produce the reformulated hydrocarbon gasoline mixture. Of particular interest, the first hydroprocessing treatment of the depolymerized lignin products, primarily methoxy-substituted alkylphenols, entails an exhaustive hydrodeoxygenation step using a sulfided CoMo/Al2O3 catalyst system. The hydrodeoxygenation step (at 350-375 ˚C, 97–152 bar H2) yields mainly a mixture of alkylated benzenes but hardly any benzene itself (undesired for gasoline application as it is a known carcinogen). In yet another patent, lignin is also subjected to a base-catalyzed partial depolymerization first, followed by stabilization through partial hydrodeoxygenation using a sulfided catalyst system M-Mo supported on Al2O3 or on activated carbon (M = Co, Rh, Pd, Ru, Pt). These catalysts, which were already

Chapter 2

50

Entr

yCa

taly

stSu

ppor

tRe

acti

on C

ondi

tion

sLi

gnin

M

ajor

Pro

duct

sCo

nver

sion

Not

esRe

f

T (˚

C)P

(bar

)t (

min

)(%

)

1Cu

CrO

none

260

220

1080

ligni

nm

etha

nol,

4-n-

prop

ylcy

cloh

exan

ol,

4-n-

prop

ylcy

cloh

exan

edio

l, gl

ycol

7096

2Cu

CrO

none

250

200

300

hydr

ol li

gnin

3-cy

cloh

exyl

-1-p

ropa

nol,

4-n-

prop

ylcy

cloh

exan

ol,

3-(4

-hyd

roxy

cycl

ohex

yl)-

1-pr

opan

ol

1297

3Ra

ney

Ni

none

173

200

360

map

le w

ood

mea

l4-

ethy

lsyr

ingo

l, 4-

etha

nols

yrin

gol

2798

4Ra

ney

Ni

none

195

3430

0sp

ruce

woo

d m

eal

dihy

droc

onife

ryl a

lcoh

ol,

4-n-

prop

ylgu

aiac

ol16

99

5R

hca

rbon

195

3430

0sp

ruce

woo

d m

eal

dihy

droc

onife

ryl a

lcoh

ol,

4-n-

prop

ylgu

aiac

ol34

99

6R

hA

l 2O 319

534

300

spru

ce w

ood

mea

ldi

hydr

ocon

ifery

l alc

ohol

, 4-

n-pr

opyl

guai

acol

1399

7Pd

carb

on19

534

300

spru

ce w

ood

mea

ldi

hydr

ocon

ifery

l alc

ohol

, 4-

n-pr

opyl

guai

acol

2499

8R

hca

rbon

195

3430

0as

pen

woo

d m

eal

nsns

100

9Fe

Sano

nec

250-

450

152-

456

60-1

20lig

nin

phen

ols,

benz

enes

nsg

102

10Fe

Sno

nec,

d37

5-42

550

-150

60kr

aft l

igni

nm

onop

heno

ls C

6-C9

nsh

105

11Co

Mo

Al 2O 3

c,e

400-

450

705-

60or

gano

solv

lign

inin

solu

ble

resi

due

ns12

7

12Co

Mo

Al 2O 3

c,e

400-

450

7060

orga

noso

lv li

gnin

inso

lubl

e re

sidu

e/ph

enol

sns

i12

8a W

ith

coca

taly

st. b M

= R

u, C

o, C

u, Ir

, Re,

Pd,

Fe,

Rh,

Pt,

or N

i. c S

ulfid

ed. d P

rom

oter

can

be

used

. e γ-A

l 2O 3. f Wit

h Cr

O on

alu

min

a. g C

onti

nuou

s flo

w. h

Not

spe

cifie

d. i P

yrid

ine

pois

oned

. j Dim

ethy

ldis

ulfid

e in

feed

. k 1/3

act

ive

as N

iMo.

Tabl

e 2.

2: H

eter

ogen

eous

cat

alys

t sys

tem

s for

the

hydr

ogen

atio

n an

d hy

drod

eoxy

gena

tion

of li

gnin

.


51

Entr

yCa

taly

stSu

ppor

tRe

acti

on C

ondi

tion

sLi

gnin

M

ajor

Pro

duct

sCo

nver

sion

Not

esRe

f

T (˚

C)P

(bar

)t (

min

)(%

)

13M

oA

l 2O 334

0-45

034

-170

hde

poly

mer

ized

lig

nin

phen

ol, c

reso

ls,

alky

lphe

nols

, al

kylb

enze

nes

ns12

8

14Co

Mo

Al 2O 3

350-

370

100-

150

hde

poly

mer

ized

lig

nin

tolu

ene,

eth

ylbe

nzen

e,

xyle

nes,

trim

ethy

lben

zene

s, al

kylb

enze

nes

ns12

9

15M

b-M

oA

l 2O 320

0-30

035

-138

5-15

depo

lym

eriz

ed

ligni

nph

enol

sns

126

16N

iMo

Al 2O 3 c

400/

370

100/

180

hor

gano

cell

ligni

nph

enol

, cre

sols

, al

kylp

heno

ls, x

ylen

ols,

guai

acol

nsj

132

17ze

olite

A37

010

0h

orga

noce

ll lig

nin

phen

ol, c

reso

ls,

alky

lphe

nols

, xyl

enol

s, gu

aiac

ol

nsj

132

18Pd

acti

vate

d ch

arco

al38

010

015

orga

noce

ll lig

nin

oils

1513

3

19Fe

2O 3no

ne38

010

015

orga

noce

ll lig

nin

oils

1713

3

20Ra

ney

Ni

none

380

100

15or

gano

cell

ligni

noi

ls53

133

21N

iMo

SiO 2-A

l 2O 338

010

015

orga

noce

ll lig

nin

oils

5313

3

22N

iMo

zeol

ite38

010

015

orga

noce

ll lig

nin

oils

1713

3

23N

iMo

SiO 2-A

l 2O 3, or

zeol

itej

400

100

40or

gano

cell

oils

49-7

113

4

24N

iWSi

O 2-Al 2O 3

c , SiO 2-

Al 2O 3-P

O4c

300-

450

35-2

40h

ligni

nph

enol

ics

hk

150

25Pt

carb

on20

040

240

ligni

nm

onom

ers,

dim

ers

4214

9

26M

ono

ne40

070

-100

65lig

nin

oils

ns15

4a W

ith

coca

taly

st. b M

= R

u, C

o, C

u, Ir

, Re,

Pd,

Fe,

Rh,

Pt,

or N

i. c S

ulfid

ed. d P

rom

oter

can

be

used

. e γ-A

l 2O 3. f Wit

h Cr

O on

alu

min

a. g C

onti

nuou

s flo

w. h

Not

spe

cifie

d. i P

yrid

ine

pois

oned

. j Dim

ethy

ldis

ulfid

e in

feed

. k 1/3

act

ive

as N

iMo.

Chapter 2

52

demonstrated to be effective C-O bond [126] and C-N bond [130] hydrogenolysis catalysts, were also used as lignin hydrocracking/ring hydrogenation catalysts. [129] Shabtai et al. reported conversion of over 95% of methoxyphenols and benzenediols to phenol-based products having a single oxygen moiety. Finally, the product is converted into bio-fuel via further refining steps. [131] A general drawback of processes such as those mentioned above is that they suffer from coke formation, which limits the capacity and lifetime of the catalyst and can cause blocking of the reactor. [90]

Meier et al. subjected organocell lignin to catalytic hydrocracking using a lignin-derived slurry oil with the objective of maximizing phenol yields. [132] Up to 12.8 wt% of a mixture of mono-phenols and little coke formation was obtained using a spent, conventional NiMo hydrocracking catalyst. Of the parameters studied, hydrogen pressure proved the most important. A number of lignins were also subjected to catalytic hydropyrolysis using various different catalysts in a gas-solid type reaction to exclude any influence of the solvent or pasting oil on the origin of the degradation products. Of the catalysts tested, NiMo/Al2O3SiO2 and Pd/C gave the highest yields of liquid products (oil) and the least amount of char formation. The NiMo catalyst gave the best results in terms of phenol production, while mostly alkylated cyclohexanones were obtained with Pd/C. [133] Several different catalysts were studied in the conversion of five different softwood and hardwood kraft lignins and one softwood organocell lignin into oil-like products. Highest yields were obtained for the organocell lignin with a catalyst mixture of sulfided NiMo/Al2O3SiO2 and Cr2O3/Al2O3 resulting in at most 10 wt% of the lignin feedstock converted into alkylbenzenes and phenols, in addition to unidentified products. [134]

2.4.2.7 Influence of process parametersIn some cases, the low sulfur content of bio-oils or lignin product streams demands

the addition of a sulfiding agent to the feed to maintain the sulfidation degree and consequently the activity of the catalysts. [94] The addition of the sulfiding agent affects the hydrodeoxygenation process and differences have been observed for different classes of substrates on how particular pathways and catalyst activity are influenced by such additives. Whereas a promoting effect was observed for aliphatic oxygenates, hydrodeoxygenation activity of phenolic compounds was generally found to be suppressed on sulfided NiMo and CoMo catalysts. [110, 135-137] The addition of H2S, for instance, suppressed direct hydrogenolysis of phenol due to competitive adsorption. [135] Şenol et al. compared the influence of increasing concentrations of H2S on phenol hydrodeoxygenation with both sulfided CoMo and NiMo catalyst under identical conditions. [135] On both catalysts the HDO conversion of phenol decreased


53

with increasing H2S concentration in the feed. Hydrodeoxygenation of phenols is generally accepted to proceed via two parallel reaction pathways: the direct hydrogenolysis route involving cleavage of the C-O bond giving aromatic products, or ring hydrogenation prior to C-O bond cleavage (hydrogenation-hydrogenolysis) yielding saturated hydrocarbons. The yields of both aromatics and saturated hydrocarbons dropped upon introduction of H2S, but not to the same extent. The decrease in molar ratio of aromatics to saturated hydrocarbons was minor with the NiMo catalyst, indicating that both pathways were equally affected. The addition of H2S to phenol hydrodeoxygenation over a sulfided CoMo/Al2O3, on the other hand, suppressed the direct hydrogenolysis route but not the combined hydrogenation-hydrogenolysis route. This observation supports the idea that the two reactions take place at different sites. Although only a minor pathway in the absence of H2S, the latter route thus becomes more important with increasing concentrations of the sulfiding agent as the direct hydrogenolysis route becomes increasingly blocked. The inhibiting effect of H2S was interpreted in terms of competitive adsorption of phenol and H2S on the catalytic sites, i.e. coordinatively unsaturated sites associated with the MoS2 phase. Interestingly, the addition of H2S did not solve the deactivation problem of the sulfided CoMo catalyst, hinting at the influence of the formation of coke and high molecular weight compounds on catalyst performance. [137] Laurent et al. also noted that hydrogenolysis was affected more than hydrogenation upon addition of H2S for both CoMo and NiMo catalysts. [136] As expected, the CoMo catalyst activity is more sensitive to H2S as the dominant hydrogenolysis pathway becomes inhibited. It was also noted that the activity and selectivity of CoMo and NiMo catalysts in 4-methylphenol hydrodeoxygenation was not substantially affected by the presence of added water. Pretreatment of a sulfided NiMo catalyst with water under hydrotreating conditions did, however, result in a loss of two-thirds of the initial activity after 60 h, but the hydrogenolysis/hydrogenation selectivity remained unchanged. Partial recrystallization of the support into a hydrated boehmite phase was observed together with partial oxidation of the nickel sulfide phase into oxidized nickel species. [121]

Benzofuran (12, Figure 2.7) has been used as a common probe molecule to evaluate catalyst performance in hydrodeoxygenation reactions, and various studies using NiMo or CoMo catalysts have been reported, with the former being more active for this substrate. [138-144] As indicated above, the reduced benzofuran derivative 2,3-dihydrobenzofuran resembles some of the cyclic ethers found in lignin and is therefore studied as a model compound of this linkage. Different hydrodeoxygenation

Chapter 2

54

routes have been suggested depending on the catalyst and conditions used, but ethylcyclohexane is commonly obtained as the major product. Product distribution was found to depend strongly on temperature and hydrogen pressure employed. For sulfided NiMo/Al2O3, for instance, hydrogenation of the benzofuran heterocycle to 2,3-dihydrobenzofuran is followed by hydrogenolysis, producing 2-ethylphenol. Further hydrogenation/dehydroxylation leads to loss of aromaticity and ethylcyclohexane formation but only at higher temperatures. A reduced NiMo catalyst showed much higher hydrogenation activity, which resulted in ring saturation being favored over the hydrogenolysis route. [140, 141] The activity of Mo2N for benzofuran hydrodeoxygenation has also been investigated and molybdenum nitride was found to be an effective catalyst as rapid hydrogenation of the heterocyclic ring, followed by hydrogenolyis of the ether and release of the heteroatom yielded a mixture of alkylated aromatics, i.e. benzene, toluene and ethylbenzene in approximately equal amounts. [145]

Interesting effects of adding H2S to the feed for the sequential deoxygenation of benzofuran and its products were observed with CoMo and NiMo catalysts. The addition of H2S was found to have a major influence on benzofuran hydrodeoxygenation over sulfided NiMo-P/Al2O3, as benzofuran conversion increased significantly giving 2-ethylphenol as the major product. [142] The influence on each individual step of the reaction pathway was more subtle, however. Both Bunch et al. and Romero et al. found, for instance, that while H2S promoted the conversion of dihydrobenzofuran to 2-ethylphenol, it inhibited the conversion of 2-ethylphenol to ethylbenzene. [140-142] This phenomena can be explained by the role of H2S in filling vacancies in the MoS2

phase, leading to an interconversion of direct deoxygenation sites to hydrogenation sites. [142]

2.4.2.8 Noble metal based hydrodeoxygenation catalystsSome disadvantages that are associated with conventional hydrodeoxygenation

catalysts are possible contamination of products by incorporation of sulfur, rapid deactivation by coke formation, and potential poisoning by water. These issues arise especially with biomass feedstocks and thus have prompted efforts to explore alternative hydrogenation catalysts. [146]

Elliott et al. reported on the use of Ru/C and Pd/C for the catalytic hydroprocessing of guaiacol (among others), which was used as chemical model for bio-oil. The supported platinum-group catalysts are known to be more active than the sulfided Mo-based ones


55

and can therefore be used at lower temperatures, and the use of non-alumina supports such as carbon or TiO2 avoids the water instability associated with Al2O3. [147] The Ru-catalyzed reactions of guaiacol yield methoxycyclohexanol and cyclohexanediols at 150 ˚C, cyclohexanol at 200 ˚C, and gasification products at temperatures exceeding 250 ˚C. In contrast, Pd-catalyzed reactions yield methoxycyclohexanone at 150 ˚C, cyclohexanol and cyclohexane at 250 ˚C, and cyclohexane and a considerable amount of methanol at temperatures exceeding 300 ˚C. [150] Although different products are obtained for the two different metals, substrate hydrogenation and loss of aromaticity rather than hydrodeoxygenation is observed for both. Similarly, Zhao et al. showed that a combination of Pd/C (or Pt/C, Ru/C, Rh/C) and a mineral acid could completely convert phenolic bio-oil components (phenols, guaiacols, and syringols) to cycloalkanes and methanol, i.e. full hydrogenation and deoxygenation was achieved. [146] No direct hydrogenolysis of phenol to benzene was observed. This result is in stark contrast with sulfided CoMo catalysts, which mainly yield benzene with little formation of cyclohexane from phenol. Guaiacol and syringol substrates showed fast ring hydrogenation before removing oxygen functional groups to form cycloalkanes with high selectivity.

De Wild et al. reported on the hydrotreatment of a pyrolytic lignin oil fraction obtained from Alcell lignin using a Ru/C hydrogenation catalyst. Full reduction of the aromatic compounds was observed; cycloalkanes, alkyl-substituted cyclohexanols, cyclohexanol and linear alkenes were identified as the major products. This fast and full hydrogenation led to the conclusion that Ru/C is a too active catalyst for the desired conversion of the pyrolytic lignin oil to low molecular weight phenolics. [148]

A two-step process for the selective degradation of actual wood lignin over noble-metal catalysts has also been communicated. In a first step, catalytic cleavage of the C-O-C bonds (without disrupting the C-C linkages) in white birch wood lignin was achieved using a series of active carbon supported catalysts (i.e. Ru/C, Pd/C, Rh/C, and Pt/C), under modest H2 pressures and using acidified, near-critical water as the solvent. The Pt/C catalyst gave the best results and four monomers, namely guaiacylpropane, syringylpropane, guaiacylpropanol, and syringylpropanol, were identified as the main constituents of the product stream. Combined total yields of monomer and dimer products reached about 45 (close to the calculated theoretical maximum) and 12 wt%, respectively. In a second step, the products could be further hydrogenated over Pd/C with excellent yields to the corresponding fully saturated hydrocarbons for eventual application in transportation fuel production. [149]

Chapter 2

56

2.4.2.9 Non-conventional hydrodeoxygenation catalystsThring et al. used a NiW/SiO2Al2O3 catalyst for hydrocracking of solvolysis (Alcell)

lignin in the presence of tetralin, a hydrogen donor solvent at little or no hydrogen pressure. Reactions conditions were deemed insufficient to degrade the lignin to liquid and gaseous products, as conversion did not exceed 50% at the highest severity conditions. Recondensation appeared to dominate hydrocracking, which was attributed to insufficient amounts of hydrogen atoms released by the solvent to stabilize the lignin fragments. Recondensation poses a general problem in lignin degradation, as mentioned previously. Size exclusion chromatography studies showed that lignin was nevertheless increasingly depolymerized, but yields of monomeric products were very low. [68] A patent by Engel et al. claimed that hydrocracking of kraft lignin afforded phenolics in high yields using a supported NiW catalyst. Using a mildly acidic support and various additives high yields of phenols and cresols were obtained. [150]

In search of a stable and active non-sulfided hydrodeoxygenation catalyst, which would dispense with the need to possibly add a sulfur source to the feed and give less rise to coking, Yakovlev et al. tested a number of metals and supports for anisole hydrodeoxygenation activity. The support was again found to play a major role, and zirconia and ceria, which can have a valence change under hydrodeoxygenation conditions, proved the most effective because of possible additional activation of oxy-compounds on the support surface. Various supported rhodium catalysts performed well with good selectivity to aromatic products in some cases. The influence of the support was clearly illustrated by a comparison of NiCu/ZrO2 and NiCu/CeO2, as the former gave mostly aromatic products and the latter full conversion to cyclohexane. [151]

In an alternative approach, Filley et al. reported the reductive deoxygenation of guaiacol (as well as catechol) in the presence of the cheap reductant α-terpinene catalyzed by vanadium on alumina at atmospheric pressure. Phenol and methyl-substituted phenols were obtained in high yields and with excellent selectivity. [152] Nickel boride was also communicated as an effective catalyst for the preparation of 4-ethylguaiacol and 4-ethylsyringol from various wood meals at 180 ˚C and 130 bar H2 in basic medium. [153] The nickel boride catalyst was generated within the wood structure by impregnating the wood with a nickel salt, followed by reaction with sodium borohydride. Yields of phenolics of up to 69% showed that this catalyst


57

Entr

yCa

taly

stSu

ppor

tRe

acti

on C

ondi

tion

sLi

gnin

Mod

elM

ajor

Pro

duct

sCo

nver

sion

Not

esRe

f

T (˚

C)P

(bar

)t (

min

)Co

mpo

und

(%)

1Co

Mo

SiO 2-A

l 2O 330

0-50

010

0-20

0i

poly

cycl

ic a

rom

atic

sga

solin

e hy

droc

arbo

nsi

108

2N

iMo

SiO 2-A

l 2O 3b,c,

d30

050

hph

enol

C 6 hyd

roca

rbon

s2

109

3N

iMo

SiO 2-A

l 2O 3b,c,

d32

550

hph

enol

C 6 hyd

roca

rbon

s17

109

4N

iMo

SiO 2-A

l 2O 3b,c,

d32

550

ho-

cres

olph

enol

/C7 h

ydro

carb

ons

2610

9

5Co

Mo

Al 2O 3b

300

5025

04-

met

hylp

heno

lto

luen

e10

011

0

6Co

Mo

Al 2O 3b,

e32

569

101

4-m

ethy

lgua

iaco

lto

luen

e, c

reso

l iso

mer

s, m

ethy

lcat

echo

l98

111

7Co

Mo

Al 2O 3b,

e30

069

344

4-m

ethy

lcat

echo

lto

luen

e, c

reso

l, al

kylp

heno

l, m

ethy

lcyc

lohe

xane

9911

1

8Co

Mo

Al 2O 3b,

e30

069

240

euge

nol

prop

ylcy

cloh

exan

e,

prop

ylph

enol

, pr

opyl

guai

acol

, pr

opyl

cate

chol

100

111

9Co

Mo

Al 2O 3b,

e57

369

254

vani

llin

met

hylc

yclo

hexa

ne,

met

hylc

atec

hol,

cres

ol98

111

10Co

Mo

Al 2O 3b,

e30

069

443

o,o-

biph

enol

biph

enyl

, cy

cloh

exyl

benz

ene,

di

benz

ofur

an,

2-ph

enyl

phen

ol

9211

1

11Co

Mo

Al 2O 3b,

e30

069

361

o-hy

drox

ydip

heny

-lm

etha

nebe

nzen

e, c

yclo

hexa

ne,

tolu

ene,

phe

nol,

diph

enyl

met

hane

100

111

a M =

Ru,

Co,

Cu,

Ir, R

e, P

d, F

e, R

h, P

t, or

Ni.

b Sul

fided

. c Am

orph

ous.

d Pel

lets

. e γ-A

l 2O 3. f Sev

eral

car

bon

supp

orts

test

ed. g P

hosp

horu

s dop

ed. h C

onti

nuou

s flo

w.

i Not

spe

cifie

d. j S

ever

al s

olve

nts c

an b

e us

ed. k P

heno

l is s

olve

nt a

nd la

ter l

igni

n ta

rs. l R

eact

ion

mix

ture

wit

h th

ree

subs

trat

es. m

Flo

w e

xper

imen

t, pr

oduc

ts

rem

oved

. n L

igni

n w

as m

ixed

wit

h di

ffer

ent l

igni

n-de

rive

d sl

urry

oils

. o Inhi

bite

d by

H2S.

p Low

er p

ress

ure

and

tem

pera

ture

, les

s hyd

roge

nati

on. q H

2S in

fe

ed. r A

ddit

ion

of lo

wer

alip

hati

c alc

ohol

s inc

reas

es p

heno

lics y

ield

.

Tabl

e 2.

3: H

eter

ogen

eous

cat

alys

t sys

tem

s for

the

hydr

ogen

atio

n an

d hy

drod

eoxy

gena

tion

of li

gnin

.

Chapter 2

58

Entr

yCa

taly

stSu

ppor

tRe

acti

on C

ondi

tion

sLi

gnin

Mod

elM

ajor

Pro

duct

sCo

nver

sion

Not

esRe

f

T (˚

C)P

(bar

)t (

min

)Co

mpo

und

(%)

12Co

Mo

Al 2O 3b,

e30

069

379

phen

ylet

her

benz

ene,

cyc

lohe

xane

, ph

enol

9811

1

13Co

Mo

Al 2O 3b,

e25

0-32

534

400-

600

anis

ole

phen

ol, b

enze

ne,

cycl

ohex

ane

100

112

14Co

Mo

Al 2O 3b,

e25

034

1200

guai

acol

cate

chol

, phe

nol,

benz

ene,

cy

cloh

exan

e10

011

2

15Co

Mo

Al 2O 3b

275-

325

50i

o-m

etho

xyph

enol

phen

ols,

diox

ygen

co

mpo

unds

, oth

er

hydr

ocar

bons

23-9

911

3

16Co

Mo

Al 2O 3b

275-

325

50i

m-m

etho

xyph

enol

phen

ols,

diox

ygen

co

mpo

unds

, oth

er

hydr

ocar

bons

27-9

511

3

17N

iMo

Al 2O 3b

450

28i

cate

chol

phen

ol, b

enze

ne,

cycl

ohex

ane

9811

4

18N

iMo

Al 2O 3b

350

28i

guai

acol

benz

ene,

tolu

ene

9811

4

19N

iMo

Al 2O 3b

350

28i

syri

ngol

benz

ene,

tolu

ene,

tr

imet

hylb

enze

ne98

114

20Co

Mo

Al 2O 3b,

e28

070

150

4-m

ethy

lace

toph

enon

eet

hylm

ethy

lben

zene

100

j11

5

21Co

Mo

Al 2O 3b,

e28

070

150

guai

acol

phen

ol, c

atec

hol

57j

115

22N

iMo

Al 2O 3b,

e28

070

150

4-m

ethy

lace

toph

enon

eet

hylm

ethy

lben

zene

100

j11

5

23N

iMo

Al 2O 3b,

e28

070

150

guai

acol

phen

ol, c

atec

hol

65j

115

24Co

Mo

Al 2O 3b

280

70i

4-m

ethy

lace

toph

enon

eet

hylm

ethy

lben

zene

i11

6

25Co

Mo

Al 2O 3b

280

70i

guai

acol

phen

ols,

cate

chol

, hy

droc

arbo

nsi

116

26Co

Mo

Carb

onb,

f28

070

180

4-m

ethy

lace

toph

enon

eet

hylm

ethy

lben

zene

100

118

a M =

Ru,

Co,

Cu,

Ir, R

e, P

d, F

e, R

h, P

t, or

Ni.

b Sul

fided

. c Am

orph

ous.

d Pel

lets

. e γ-A

l 2O 3. f Sev

eral

car

bon

supp

orts

test

ed. g P

hosp

horu

s dop

ed. h C

onti

nuou

s flo

w.

i Not

spe

cifie

d. j S

ever

al s

olve

nts c

an b

e us

ed. k P

heno

l is s

olve

nt a

nd la

ter l

igni

n ta

rs. l R

eact

ion

mix

ture

wit

h th

ree

subs

trat

es. m

Flo

w e

xper

imen

t, pr

oduc

ts

rem

oved

. n L

igni

n w

as m

ixed

wit

h di

ffer

ent l

igni

n-de

rive

d sl

urry

oils

. o Inhi

bite

d by

H2S.

p Low

er p

ress

ure

and

tem

pera

ture

, les

s hyd

roge

nati

on. q H

2S in

fe

ed. r A

ddit

ion

of lo

wer

alip

hati

c alc

ohol

s inc

reas

es p

heno

lics y

ield

.


59

Entr

yCa

taly

stSu

ppor

tRe

acti

on C

ondi

tion

sLi

gnin

Mod

elM

ajor

Pro

duct

sCo

nver

sion

Not

esRe

f

T (˚

C)P

(bar

)t (

min

)Co

mpo

und

(%)

27Co

Mo

Carb

onb,

f28

070

180

guai

acol

cate

chol

, phe

nol

3511

8

28Co

Mo

Carb

onb,

f28

070

igu

aiac

olph

enol

, ben

zene

, cy

cloh

exan

ei

119

29Co

Mo

Al 2O 3b,

d,e

300

50i

anis

ole

phen

ol, o

-cre

sol,

benz

ene

7712

2

30Co

Mo

Al 2O 3b,

d,e

300

50i

anis

ole

phen

ol, o

-cre

sol,

benz

ene

50k

122

31N

iMo

Al 2O 3b,

d,e

300

50i

anis

ole

phen

ol, o

-cre

sol

9112

2

32N

iMo

Al 2O 3b,

d,e

300

50i

anis

ole

phen

ol, c

yclo

hexa

ne,

o-cr

esol

69k

122

33Co

Mo

MgO

b35

050

60ph

enol

cycl

ohex

ylar

omat

ics

1712

3

34Co

Mo-

PM

gOb

450

5060

phen

olbe

nzen

e,

cycl

ohex

ylar

omat

ics

9012

3

35Co

Mo

Al 2O 3b,

r35

069

icr

esol

sto

luen

e,

met

hylc

yclo

hexa

nei

124

36Fe

2O 3N

oneb

450

9850

dim

eric

spe

cies

benz

enes

, mon

ophe

nols

, di

mer

s3-

100

125

37Fe

2O 3A

l 2O 3b45

098

50di

mer

ic s

peci

esbe

nzen

es, m

onop

heno

ls,

dim

ers

12-1

0012

5

38N

iMo

Al 2O 3

450

9850

dim

eric

spe

cies

benz

enes

, mon

ophe

nols

, di

mer

s36

-100

125

39M

oTi

O 2b45

098

50di

mer

ic s

peci

esbe

nzen

es, m

onop

heno

ls,

dim

ers

36-1

0012

5

40N

iMo

Al 2O 3b,

e,g

250-

450

34i

4-pr

opyl

guai

acol

prop

ylph

enol

s, et

hylp

heno

ls, c

reso

ls,

phen

ol

50-1

00l

127

a M =

Ru,

Co,

Cu,

Ir, R

e, P

d, F

e, R

h, P

t, or

Ni.

b Sul

fided

. c Am

orph

ous.

d Pel

lets

. e γ-A

l 2O 3. f Sev

eral

car

bon

supp

orts

test

ed. g P

hosp

horu

s dop

ed. h C

onti

nuou

s flo

w.

i Not

spe

cifie

d. j S

ever

al s

olve

nts c

an b

e us

ed. k P

heno

l is s

olve

nt a

nd la

ter l

igni

n ta

rs. l R

eact

ion

mix

ture

wit

h th

ree

subs

trat

es. m

Flo

w e

xper

imen

t, pr

oduc

ts

rem

oved

. n L

igni

n w

as m

ixed

wit

h di

ffer

ent l

igni

n-de

rive

d sl

urry

oils

. o Inhi

bite

d by

H2S.

p Low

er p

ress

ure

and

tem

pera

ture

, les

s hyd

roge

nati

on. q H

2S in

fe

ed. r A

ddit

ion

of lo

wer

alip

hati

c alc

ohol

s inc

reas

es p

heno

lics y

ield

.

Chapter 2

60

Entr

yCa

taly

stSu

ppor

tRe

acti

on C

ondi

tion

sLi

gnin

Mod

elM

ajor

Pro

duct

sCo

nver

sion

Not

esRe

f

T (˚

C)P

(bar

)t (

min

)Co

mpo

und

(%)

41M

oA

l 2O 3b,e

250-

450

34i

4-pr

opyl

guai

acol

prop

ylph

enol

s, pr

opyl

benz

enes

, pr

opyl

hexa

ne, d

ealk

ylat

ed

prod

ucts

50-1

00l

127

42M

b or

M

b-M

oA

l 2O 3b,e

350

137

idi

phen

yl e

ther

and

na

phth

alen

e m

ixtu

reph

enol

, ben

zene

, cy

cloh

exan

e, te

tral

in,

deca

lin

i12

6

43Co

Mo

Al 2O 3b,

e25

075

iph

enol

benz

ene,

cyc

lohe

xane

im

135

44N

iMo

Al 2O 3b,

e25

075

iph

enol

benz

ene,

cyc

lohe

xane

im

135

45Co

Mo

Al 2O 3b,

e34

0-35

570

i4-

met

hylp

heno

lto

luen

e,

met

hylc

yclo

hexa

ne80

-100

136

46Co

Mo

Al 2O 3b,

e25

015

iph

enol

benz

ene

3613

7

47Co

Mo

Al 2O 3b,

e30

015

iph

enol

benz

ene

cycl

ohex

ane/

hexe

ne71

137

48Co

Mo

Al 2O 3b,

e25

015

ian

isol

eo-

cres

ol, x

ylen

ol, p

heno

l, be

nzen

e88

137

49Co

Mo

Al 2O 3b,

e30

015

ian

isol

eto

luen

e, p

heno

l, be

nzen

e,

o-cr

esol

9713

7

50N

iMoP

Al 2O 3b

350

35i

benz

ofur

andi

hydr

oben

zofu

ran,

et

hylp

heno

l, ph

enol

, et

hylb

enze

ne, t

olue

ne,

benz

ene

138

51Co

Mo

Al 2O 3b,

e36

070

ibe

nzof

uran

ethy

lphe

nol,

ethy

lben

zene

, et

hylc

yclo

hexe

ne,

ethy

lcyc

lohe

xane

139

a M =

Ru,

Co,

Cu,

Ir, R

e, P

d, F

e, R

h, P

t, or

Ni.

b Sul

fided

. c Am

orph

ous.

d Pel

lets

. e γ-A

l 2O 3. f Sev

eral

car

bon

supp

orts

test

ed. g P

hosp

horu

s dop

ed. h C

onti

nuou

s flo

w.

i Not

spe

cifie

d. j S

ever

al s

olve

nts c

an b

e us

ed. k P

heno

l is s

olve

nt a

nd la

ter l

igni

n ta

rs. l R

eact

ion

mix

ture

wit

h th

ree

subs

trat

es. m

Flo

w e

xper

imen

t, pr

oduc

ts

rem

oved

. n L

igni

n w

as m

ixed

wit

h di

ffer

ent l

igni

n-de

rive

d sl

urry

oils

. o Inhi

bite

d by

H2S.

p Low

er p

ress

ure

and

tem

pera

ture

, les

s hyd

roge

nati

on. q H

2S in

fe

ed. r A

ddit

ion

of lo

wer

alip

hati

c alc

ohol

s inc

reas

es p

heno

lics y

ield

.


61

Entr

yCa

taly

stSu

ppor

tRe

acti

on C

ondi

tion

sLi

gnin

Mod

elM

ajor

Pro

duct

sCo

nver

sion

Not

esRe

f

T (˚

C)P

(bar

)t (

min

)Co

mpo

und

(%)

52N

iMo

Al 2O 3b,

e36

020

ibe

nzof

uran

cycl

ohex

ane,

et

hylc

yclo

hexa

ne99

n14

0

52N

iMo

Al 2O 3e

180

55i

benz

ofur

anoc

tahy

drob

enzo

fura

n,

ethy

lcyc

lohe

xane

, di

hydr

oben

zofu

ran

97 (2

5% H

DO)

141

54N

iMoP

Al 2O 3b

340

70i

benz

ofur

andi

hydr

oben

zofu

rane

, et

hylp

heno

l, ph

enol

4814

2

55N

iMoP

Al 2O 3

340

70i

benz

ofur

anet

hylp

heno

l, ph

enol

, et

hylc

yclo

hexa

ne81

o14

2

56Co

Mo

none

b29

0-34

020

-100

hth

ioph

enes

, ind

oles

, ph

enol

scy

cloh

exan

esns

143

57Co

Mo

SiO 2b

290-

340

20-1

00h

fura

nes,

phen

ols

cycl

ohex

anes

ns14

3

58N

iMo

Al 2O 3b

250-

390

70i

dibe

nzof

uran

cycl

ohex

ane,

m

ethy

lcyc

lope

ntan

e,

cycl

open

tane

, ben

zene

, m

ethy

lcyc

lohe

xane

100

144

59Co

Mo

Al 2O 3b

250-

390

70i

dibe

nzof

uran

cycl

ohex

ane,

m

ethy

lcyc

lope

ntan

e,

cycl

open

tane

, ben

zene

, m

ethy

lcyc

lohe

xane

100

p14

4

60M

o 2N-

450

1i

benz

ofur

anbe

nzen

e, to

luen

e,

ethy

lben

zene

3514

5

61M

o 2N-

450

1i

benz

othi

ophe

neet

hylb

enze

ne40

145

62M

o 2N-

450

1i

indo

leto

luen

e, b

enze

ne,

ethy

lben

zene

2014

5

63Pd

carb

on25

050

30ph

enol

cycl

ohex

anol

, cyc

lohe

xane

100

146

a M =

Ru,

Co,

Cu,

Ir, R

e, P

d, F

e, R

h, P

t, or

Ni.

b Sul

fided

. c Am

orph

ous.

d Pel

lets

. e γ-A

l 2O 3. f Sev

eral

car

bon

supp

orts

test

ed. g P

hosp

horu

s dop

ed. h C

onti

nuou

s flo

w.

i Not

spe

cifie

d. j S

ever

al s

olve

nts c

an b

e us

ed. k P

heno

l is s

olve

nt a

nd la

ter l

igni

n ta

rs. l R

eact

ion

mix

ture

wit

h th

ree

subs

trat

es. m

Flo

w e

xper

imen

t, pr

oduc

ts

rem

oved

. n L

igni

n w

as m

ixed

wit

h di

ffer

ent l

igni

n-de

rive

d sl

urry

oils

. o Inhi

bite

d by

H2S.

p Low

er p

ress

ure

and

tem

pera

ture

, les

s hyd

roge

nati

on. q H

2S in

fe

ed. r A

ddit

ion

of lo

wer

alip

hati

c alc

ohol

s inc

reas

es p

heno

lics y

ield

.

Chapter 2

62

Entr

yCa

taly

stSu

ppor

tRe

acti

on C

ondi

tion

sLi

gnin

Mod

elM

ajor

Pro

duct

sCo

nver

sion

Not

esRe

f

T (˚

C)P

(bar

)t (

min

)Co

mpo

und

(%)

64Pd

carb

on25

050

304-

n-pr

opyl

guai

acol

cycl

oalk

anes

, cy

cloa

lcoh

ols,

met

hano

l10

0q

146

65Pd

carb

on25

050

304-

ally

lgua

iaco

lcy

cloa

lkan

es,

cycl

oalc

ohol

s, m

etha

nol

99q

146

66Pd

carb

on25

050

304-

acet

onyl

guai

acol

cycl

oalk

anes

, cy

cloa

lcoh

ols,

met

hano

l10

0q

146

67Pd

carb

on25

050

304-

ally

lsyr

ingo

lcy

cloa

lkan

es,

cycl

oalc

ohol

s, m

etha

nol

92q

146

68R

hSi

O 230

010

ian

isol

ens

3015

1

69R

hCo

Al 2O 3

300

10i

anis

ole

ns75

151

70R

hCo

SiO 2

300

10i

anis

ole

ns79

151

71R

hCo

SiO 2

300

10i

anis

ole

ns81

151

72Co

SiO 2

300

10i

anis

ole

ns6

151

73R

hZr

O 230

010

ian

isol

ens

9115

1

74R

hCe

O 230

010

ian

isol

ens

9515

1

75N

iSi

O 230

010

ian

isol

ens

4615

1

76N

iCr

2O3

300

10i

anis

ole

ns16

151

77N

iA

l 2O 330

010

ian

isol

ens

9515

1

78N

iZr

O 230

010

ian

isol

ens

6915

1

79N

iCu

Al 2O 3

300

10i

anis

ole

ns99

151

80N

iCu

ZrO 2

300

10i

anis

ole

ns60

151

81N

iCu

CeO 2

300

10i

anis

ole

ns10

015

1a M

= R

u, C

o, C

u, Ir

, Re,

Pd,

Fe,

Rh,

Pt,

or N

i. b S

ulfid

ed. c A

mor

phou

s. d P

elle

ts. e γ

-Al 2O 3. f S

ever

al c

arbo

n su

ppor

ts te

sted

. g Pho

spho

rus d

oped

. h Con

tinu

ous f

low

. i N

ot s

peci

fied.

j Sev

eral

sol

vent

s can

be

used

. k Phe

nol i

s sol

vent

and

late

r lig

nin

tars

. l Rea

ctio

n m

ixtu

re w

ith

thre

e su

bstr

ates

. m F

low

exp

erim

ent,

prod

ucts

re

mov

ed. n

Lig

nin

was

mix

ed w

ith

diff

eren

t lig

nin-

deri

ved

slur

ry o

ils. o In

hibi

ted

by H

2S. p L

ower

pre

ssur

e an

d te

mpe

ratu

re, l

ess h

ydro

gena

tion

. q H2S

in

feed

. r Add

itio

n of

low

er a

lipha

tic a

lcoh

ols i

ncre

ases

phe

nolic

s yie

ld.


63

Entr

yCa

taly

stSu

ppor

tRe

acti

on C

ondi

tion

sLi

gnin

Mod

elM

ajor

Pro

duct

sCo

nver

sion

Not

esRe

f

T (˚

C)P

(bar

)t (

min

)Co

mpo

und

(%)

82Pd

carb

on20

069

igu

aiac

olvo

lati

le h

ydro

carb

ons,

cycl

ohex

aned

iol,

2-m

etho

xycy

cloh

exan

ol

6614

7

83Ru

carb

on20

069

igu

aiac

ol2-

met

hoxy

cycl

ohex

anol

, cy

cloh

exan

ol10

014

7

84Pd

carb

on25

040

30-1

20m

onom

ers,

dim

ers

alka

nes,

met

hano

l95

-100

q14

9

85V

Al 2O 3b,

e35

0i

igu

aiac

ol(m

ethy

late

d) p

heno

l10

0r

152

a M =

Ru,

Co,

Cu,

Ir, R

e, P

d, F

e, R

h, P

t, or

Ni.

b Sul

fided

. c Am

orph

ous.

d Pel

lets

. e γ-A

l 2O 3. f Sev

eral

car

bon

supp

orts

test

ed. g P

hosp

horu

s dop

ed. h C

onti

nuou

s flo

w.

i Not

spe

cifie

d. j S

ever

al s

olve

nts c

an b

e us

ed. k P

heno

l is s

olve

nt a

nd la

ter l

igni

n ta

rs. l R

eact

ion

mix

ture

wit

h th

ree

subs

trat

es. m

Flo

w e

xper

imen

t, pr

oduc

ts

rem

oved

. n L

igni

n w

as m

ixed

wit

h di

ffer

ent l

igni

n-de

rive

d sl

urry

oils

. o Inhi

bite

d by

H2S.

p Low

er p

ress

ure

and

tem

pera

ture

, les

s hyd

roge

nati

on. q H

2S in

fe

ed. r A

ddit

ion

of lo

wer

alip

hati

c alc

ohol

s inc

reas

es p

heno

lics y

ield

.

Chapter 2

64

system might provide an easy to prepare, non-pyrophoric and cheaper alternative to noble metal catalysts, such as rhodium or palladium. In a somewhat similar approach, Oasmaa and Johansson impregnated kraft pine lignin with an aqueous solution of ammonium heptamolybdate and hydrotreated after drying at 80 bar H2 and 450 ˚C, with or without additives such as sodium hydroxide or carbon disulfide. High yields (61% of the original lignin) of low molecular weight oils were obtained, consisting mainly of phenols, benzenes, naphthalenes and cyclohexanes. The catalyst remained in the solid residue and was easily separated. [154]

As mentioned above, dihydrobenzofuran is studied as a model compound for some of the cyclic ether structures that can be found in lignin. A Cu-doped MgAl mixed metal oxide was found to transfer H2 equivalents from methanol to dihydrobenzofuran under supercritical conditions leading first to hydrogenolysis of the ether bond and subsequent hydrogenation of the aromatic ring, resulting the formation 2-ethylcyclohexanol. Significant amounts of methylethylcyclohexanols were also found in the product mixture. Methanol served as a relatively inexpensive source of in-situ production of H2 equivalents in this reaction. [155]

Finally, BASF has patented the use of supported or unsupported transition-metal carbides, tungsten carbide in particular, for the hydrogenation of lignin in a single stage process under relatively mild conditions (130-190 ̊ C, 70-140 bar H2). It is claimed that the catalyst can cope with both sulfur-rich as well as sulfur-poor lignin streams to yield mainly a mixture of low molecular weight oligomers, consisting of e.g. dimers and trimers of coniferyl and coumaryl alcohols. [156]

2.4.3 Lignin oxidationWhereas reductive reactions tend to disrupt and remove functionality in lignin to

produce simpler phenols, oxidation reactions typically tend to form more complex aromatic compounds with additional functionality. Many of these chemicals can either serve as platform chemicals used for subsequent organic synthesis, or they serve as target fine chemicals themselves. The catalytic processes employed in the oxidation of lignin, including their historical development in the pulp and paper industry, are considered below.

Heterogeneous oxidation catalysts have played an important role in the pulp and paper industry as a means to remove lignin and other compounds from wood pulps


65

in order to increase the quality of the final paper product. Table 2.4 lists a summary of lignin heterogeneous oxidation catalysts, reaction conditions, and results. The first examples include photocatalytic oxidation catalysts, which were designed to remove lignin from paper industry wastewater streams. The most common catalysts involve TiO2 [157] or supported precious metals, such as Pt/TiO2, [158] which were found to efficiently degrade lignin using ultraviolet light. The addition of small quantities of Fe2+ provided an increase in lignin photooxidation efficiency using TiO2 catalysts. [159] The use of UV light displaces the valence-band electrons in the TiO2, which iswas necessary to initiate the oxidation. [157] Other examples include Ni/MgO catalysts used in the gasification of lignin to H2, CH4, and CO2, [86] or methylrhenium trioxide catalysts immobilized on poly(4-vinyl pyridine) or polystyrene. [160] The latter catalyst was found to oxidize phenolic, non-phenolic, monomeric, and dimeric lignin model compounds in addition to sugar cane lignin and red spruce kraft lignin for treatment of kraft pulp. [160, 161] Vanillyl and veratryl alcohol were oxidized to the acids, aldehydes, and quinones in up to 49% yield with the balance forming polymeric products. [160, 161] Herrmann and coworkers also applied methylrhenium trioxide in the presence of H2O2 for the oxidation of isoeugenol and trans-ferulic acid to produce vanillin. [162] The catalyst was found to cleave the C-C double bond to yield either the aldehyde or the acid depending on the reaction conditions. The catalyst was found to deactivate through the formation of a perrhenate species after 1000 to 2000 cycles, but a one-pot method to reactivate the catalyst was described. [163] Sales and coworkers investigated the applicability of Pd/Al2O3 catalysts for the oxidative conversion of alkaline lignin extracted from sugarcane bagasse in both batch slurry and continuous fluidized-bed reactors. [164] Approximately 0.56 g vanillin and 0.50 g syringaldehyde were obtained from 30 g lignin at 120 ˚C after 2 h. [164] Bhargava and coworkers described the catalytic wet oxidation of ferulic acid using single-metal (Cu), bimetal (Cu-Ni, Cu-Co, and Cu-Mn), and multimetal (Cu-Ni-Ce) alumina-supported catalysts, Cu and Cu-Mn kaolin-supported catalysts, and multimetal oxide (Cu-Co-Mn and Cu-Fe-Mn) catalysts. [165] Cu-Ni-Ce/Al2O3 catalysts were most active but were susceptible to catalyst leaching. [165] Cu-Mn/Al2O3 was the most stable and was second to Cu-Ni-Ce/Al2O3 in terms of activity of the nine catalysts studied. [165] Citing the desire to replace “toxic” metal ions (i.e. Sr, Ce, Co, and Mn) in pervoskite-type oxides with “non-toxic” iron to avoid environmental pollution, Zhang and coworkers reported the use of the perovskite-type oxide LaFe1-xCuxO3 (x = 0, 0.1, 0.2) for the wet aerobic oxidation of lignin. [166] Improvements in aromatic aldehyde yields and conversion were reported, and the catalyst was stable after a series of recycling steps. [166]

Chapter 2

66

Entr

yCa

taly

stO

xida

ntSo

lven

tRe

acti

on C

ondi

tion

sLi

gnin

/ M

odel

Maj

orYi

eld

Conv

ersi

onN

otes

Ref

T (˚

C)P

(bar

)t (

h)Co

mpo

und

Prod

ucts

(%)

(%)

1Ti

O 2a

H2O

20a

6al

kali

ligni

na

ad

157

2Ti

O 2a

H2O

20a

6hu

mic

aci

da

ad

157

3Ti

O 2a

H2O

aa

1A

ldri

ch

com

mer

cial

lig

nin

pow

der

aa

158

4Pt

/TiO

2a

H2O

aa

1A

ldri

ch

com

mer

cial

lig

nin

pow

der

aa

158

5Fe

2+/T

iO2

aH

2O20

a.2

5A

ldri

ch s

ynth

etic

lig

nin

aa

d15

9

6Pd

/Al 2O 3

O 2N

aOH

/H

2Ob10

0-14

02-

102.

5al

kalin

e lig

nin

from

sug

ar

baga

sse

vani

llin

syri

ngal

dehy

de

p-hy

drox

yben

z-al

dehy

de

a16

4

7N

i/M

gOns

H2O

c40

0a

2or

gano

solv

lign

inca

rbon

gas

eshy

drog

enTH

F in

sulu

bles

45.4

99.2

74.0

86

8Cu

-Ni-C

e/A

l 2O 3

O 2H

2O10

01.

73 (O

2)2

feru

lic a

cid

degr

aded

pro

ucts

a16

5

9Cu

-Mn/

Al 2O 3

O 2H

2O10

01.

73 (O

2)2

feru

lic a

cid

degr

aded

pro

duct

sa

165

10La

Fe1-

xCuxO 3

(x=0

,0.1

,0.2

)O 2

NaO

H/

H2Ob

100

2 (O

2)3

enzy

mat

ic

hydr

olys

is o

f st

eam

-exp

losi

on

corn

stal

ks

p-hy

drox

yben

z-al

dehy

de,

vani

llald

ehyd

e,

syri

ngal

dehy

de

10-2

016

6

a N

ot s

peci

fied.

b 2

mol

/L. c

Sup

ercr

itic

al. d

360

nm

UV

light

.

Tabl

e 2.

4: H

eter

ogen

eous

cat

alyt

ic s

yste

ms f

or th

e ox

idat

ion

of li

gnin

and

its m

odel

com

poun

ds.


67

2.5 Concluding remarks

Lignin is a major component of lignocellulosic biomass from which several valuable chemicals can potentially be produced as indicated by the work initiated in the second half of the last century and the major renewed interest in its valorization that is seen today. Lignin valorization constitutes an important bottleneck as well as an opportunity in modern biorefinery schemes, as the molecular structure and composition of lignin pose unique challenges and offer unique routes to produce fine and bulk aromatics. In this chapter, we have presented some of the different approaches and strategies currently available for catalytic lignin valorization. The approaches shown are those relevant for the work described in this thesis, a more extensive overview including homogeneous catalysts as well as electrocatalysts can be found in the review on which this Chapter is based. Generally, lignin reduction processes produce bulk chemicals with reduced functionality, whereas lignin oxidation processes produce fine chemicals with increased functionality.

Considerable effort has already been devoted to developing a wide variety of catalytic routes specifically for lignin oxidation and reduction, yet several issues are apparent after review of these processes. First, there is a general lack of detailed information regarding the performance of catalysts in the valorization of actual lignin streams. This lack of information originates with the analytical challenges associated with the native lignin polymer itself, the influence of different pretreatments on this structure, [167-170] and the wide variety of compounds typically obtained from lignin degradation. Moreover, lignin streams can contain residual sugars, proteins, inorganic salts, and other potential poisons that generally complicate catalysis. Although important for understanding the chemistry of the lignin polymer and the possible chemicals that can be obtained from its conversion, a majority of the reported catalytic systems involve the use of pure lignin model compounds that are free from these complicating factors. Since catalyst materials will need to be eventually tested in an operational biorefinery, obtaining performance information with real feeds is important for the development of effective catalysts suitable for use in real biorefineries. In addition, if model compounds are used, we advocate the use of the most relevant model compounds (i.e. those with appropriate functionality in the same positions as found in the conferyl or sinapyl alcohols) for the development of catalytic processes. Use of these compounds is not only important for better understanding the lignin chemistry, but reactions with substrates with similar functional groups allow for proper comparisons in terms of catalytic activity and selectivity.

Chapter 2

68

2.6 Acknowledgments

Joe Zakzeski is acknowledged for his contributions to the publication on which this chapter is based.

2.7 References

[1] A. Sakakibari, Wood Sci. Technol. 1980, 14, 89-100.[2] J. N. Chheda, G. W. Huber, J. A. Dumesic, Angew. Chem. Int. Ed. 2007, 46, 7164-7183.[3] A. Corma, S. Iborra, A. Velty, Chem. Rev. 2007, 107, 2411-2502.[4] P. Mäki-Arvela, B. Holmbom, T. Salmi, D. Y. Murzin, Catal. Rev. Sci. Eng. 2007, 49, 197- 340.[5] C. Amen-Chen, H. Pakdel, C. Roy, Bioresour. Technol. 2001, 79, 277-299.[6] D. M. Alonso, S. G. Wettstein, J. A. Dumesic, Chem. Soc. Rev. 2012, 41, 8075-8098.[7] J. C. Hicks, J. Phys. Chem. Lett. 2011, 2, 2280-2287.[8] M. P. Pandey, C. S. Kim, Chem. Eng. Technol. 2010, 34, 29-41.[9] C. Hill, Wood modification: Chemical, Thermal, and Other Processes, Wiley, 2006.[10] F. S. Chakar, A. J. Ragauskas, Ind. Crops Prod. 2004, 20, 131-141.[11] S. K. Ritter, Chem. Eng. News 2008, 86, 10-17.[12] A. Holmgren, G. Brunow, G. Henriksson, L. Zhang, J. Ralph, Org. Biomol. Chem. 2006, 4, 3456-3461.[13] J. Parkas, G. Brunow, K. Lundquist, BioRes. 2007, 2, 169-178.[14] T. Akiyama, H. Goto, D. S. Nawawi, W. Syafii, Y. Matsumoto, G. Meshitsuka, Holzforschung 2005, 59, 276-281.[15] R. Wang, C.-L. Chen, J. S. Gratzi, Holzforschung 2004, 58, 622-630.[16] C. Crestini, M. D’Auria, Tetrahedron 1997, 53, 7877-7888.[17] X. Xie, B. Goodell, D. Zhang, D. C. Nagle, Y. Qian, M. L. Peterson, J. Jellison, Bioresour. Technol. 2009, 100, 1797-1802.[18] A. Beste, A. C. Buchanan, J. Org. Chem. 2009, 74, 2837-2841.[19] J. Ralph, K. Lundquist, G. Brunow, F. Lu, H. Kim, P. F. Schatz, J. M. Marita, R. D. Hatfield, S. A. Ralph, J. H. Christensen, W. Boerjan, Phytochem. Rev. 2004, 3, 29-60.[20] J. Rencoret, M. Gisela, A. Gutiérrez, L. Nieto, J. Jiménez-Barbero, Á. T. Martínez, J. C. del Río, Ind. Crops Prod. 2009, 30, 137-143.[21] E. A. Capanema, M. Y. Balakshin, J. F. Kadla, J. Agric. Food. Chem. 2004, 52, 1850-1860.[22] E. A. Capanema, M. Y. Balakshin, J. F. Kadla, J. Agric. Food. Chem. 2005, 53, 9639-9649.[23] T. Q. Hu, Characterization of Lignocellulosic Materials, Blackwell, 2008, p 148-170.


69

[24] D. V. Evtuguin, C. Pascoal Neto, J. Rocha, J. D. Pedrosa de Jesus, Appl. Catal. A: Gen. 1998, 167, 123-139.[25] B. Kamm, P. R. Gruber, M. Kamm, Biorefineries - Industrial Processes and Products, Wiley-VCH Verlag, 2006, Vol. 2, p 151-163.[26] L. S. Kocheva, A. P. Karmanov, D. V. Kuz’min, G. N. Dalimova, Chem. Nat. Compd. 2011, 47, 792-795.[27] J.-L. Wen, B.-L. Xue, F. Xu, R.-C. Sun, A. Pinkert, Ind. Crops Prod. 2013, 42, 332-343.[28] I. Kilpelaeinen, H. Xie, A. King, M. Granstrom, S. Heikkinen, D. S. Argyropoulos, J. Agric. Food. Chem. 2007, 55, 9142-9148.[29] Y.-H. P. Zhang, S.-Y. Ding, J. R. Mielenz, J.-B. Cui, R. T. Elander, M. Laser, M. E. Himmel, J. R. McMillan, L. R. Lynd, Biotechnol. Bioeng. 2007, 97, 214-223.[30] M.-L. Mattinen, S. Tapani, R. Gosselink, D. S. Argyropoulos, D. Evtuguin, A. Suurnäkki, E. de Jong, T. Tamminen, BioRes. 2008, 3, 549-565.[31] Y. Chen, S. Sarkanen, Phytochem. Rev. 2003, 2, 235-255.[32] A. Guerra, A. Gaspar, I. S. Contreras, L. Lucia, C. Crestini, D. S. Argyropoulos, Phytochemistry 2007, 68, 2570-2583.[33] M. Zavrel, D. Bross, M. Funke, J. Büchs, A. C. Spiess, Bioresour. Technol. 2009, 100, 2580-2587.[34] M. Delmas, Chem. Eng. Technol. 2008, 31, 792-797.[35] J. Vitz, T. Erdmenger, C. Haensch, U. S. Schubert, Green Chem. 2009, 11, 417-424.[36] M. Fasching, P. Schröder, R. P. Wollboldt, H. K. Weber, H. Sixta, Holzforschung 2008, 62, 15-23.[37] C. Eckert, C. Liotta, A. Ragauskas, J. Hallett, C. Kitchens, E. Hill, L. Draucker, Green Chem. 2007, 9, 545-548.[38] K. Ehara, S. Saka, H. Kawamoto, J. Wood Sci. 2002, 48, 320-325.[39] D. A. Fort, R. C. Remsing, R. P. Swatloski, P. Moyna, M. Guillermo, R. D. Rogers, Green Chem. 2007, 9, 63-69.[40] R. P. Swatloski, S. K. Spear, J. D. Holbrey, R. D. Rogers, J. Am. Chem. Soc. 2002, 124, 4974-4975.[41] S. Zhu, J. Chem. Technol. Biotechnol. 2008, 83, 777-779.[42] P. Wasserscheid, T. Welton, Ionic Liquids in Synthesis, Wiley-VCH Verlag, 2008, Vol. 1, 89-102.[43] Y. Pu, N. Jiang, A. J. Ragauskas, J. Wood Chem. Technol. 2007, 27, 23-33.[44] L. da Costa Sousa, S. P. S. Chundaway, V. Balan, B. E. Dale, Curr. Opin. Biotechnol. 2009, 20, 339-347.[45] A. R. Gaspar, J. A. F. Gamelas, D. V. Evtuguin, C. P. Neto, Green Chem. 2007, 9, 717-730.

Chapter 2

70

[46] J. J. Bozell, J. E. Holladay, D. Johnson, J. F. White, Top Value Added Candidates from Biomass, Volume II: Results of Screening for Potential Candidates from Biorefinery Lignin, Pacific Northwest Laboratory, 2007.[47] Metso LignoBoost: Lignin from black liquor, Metco, 2009.[48] M. N. Belgacem, A. Gandini, Monomers, Polymers and Composites from Renewable Resources, Elsevier, 2008.[49] J. Gierer, Wood Sci. Technol. 1985, 19, 289-312.[50] J. Gierer, Wood Sci. Technol. 1986, 20, 1-33.[51] W. G. Glasser, R. A. Northy, T. P. Schultz, Lignin: Historical, Biological, and Materials Perspectives, ACS Symposium Series 740, American Chemical Society, 1999, p 304- 320.[52] T. M. Liitiä, S. L. Maunu, B. Hortling, M. Toikka, I. Kilpeläinen, J. Agric. Food. Chem. 2003, 51, 2136-2143.[53] S. Y. Lin, I. S. Lin, Ullmann’s Encylcopedia of Industrial Chemistry, 5 ed., VCH, 1990, Vol. 15, p 305-315.[54] Borregaard Ingredients, 2009, http://www.borregaard.com.[55] B. Scholze, C. Hanser, D. Meier, J. Anal. Appl. Pyrolysis 2001, 58-59, 387-400.[56] B. Scholze, D. Meier, J. Anal. Appl. Pyrolysis 2001, 60, 41-54.[57] A. K. Varshney, D. Patel, J. Sci. Ind. Res. 1988, 47, 315-319.[58] P. Hughes, Cellulosic Ethanol - The Sustainable Fuel, Lignol Energy, 2009.[59] http://www.cimv.com.[60] B. K. Avellar, W. G. Glasser, Biomass Bioenerg. 1998, 14, 205-218.[61] J. Li, G. Gellerstedt, K. Toven, Bioresour. Technol. 2009, 100, 2556-2561.[62] T. H. Kim, J. S. Kim, C. Sunwoo, Y. Y. Lee, Bioresour. Technol. 2003, 90, 39-47.[63] B. Kamm, P. R. Gruber, M. Kamm, Biorefineries - Industrial Processes and Products, Wiley-VCH Verlag, 2006, Vol. 2.[64] E. Dorrestijn, L. J. J. Laarhoven, I. W. C. E. Arends, P. Mulder, J. Anal. Appl. Pyrolysis 2000, 54, 153-192.[65] P. F. Britt, A. C. Buchanan, M. J. Cooney, D. R. Martineau, J. Org. Chem. 2000, 65, 1376- 1389.[66] M. Misson, R. Haron, M. F. A. Kamaroddin, N. A. S. Amin, Bioresour. Technol. 2009, 100, 2867-2873.[67] J. Li, G. Henriksson, G. Gellerstedt, Bioresour. Technol. 2007, 98, 3061-3068.[68] R. W. Thring, J. Breau, Fuel 1996, 75, 795-800.[69] G. W. Huber, A. Corma, Angew. Chem. Int. Ed. 2007, 46, 7184-7201.[70] R. K. Sharma, N. N. Bakhshi, Energy Fuels 1993, 7, 306-314.[71] R. K. Sharma, N. N. Bakhshi, Bioresour. Technol. 1991, 35, 57-66.


71

[72] J. D. Adjaye, N. N. Bakhshi, Fuel Process. Technol. 1995, 45, 161-183.[73] J. D. Adjaye, N. N. Bakhshi, Fuel Process. Technol. 1995, 45, 185-202.[74] M. A. Jackson, D. L. Compton, A. A. Boateng, J. Anal. Appl. Pyrolysis 2009, 85, 226-230.[75] P. Chantal, S. Kaliaguine, J. L. Grandmaison, A. Mahay, Appl. Catal. 1984, 10, 317-332.[76] A. G. Gayubo, A. T. Aguayo, A. Atutxa, R. Aguado, J. Bilbao, Ind. Eng. Chem. Res. 2004, 43, 2610-2618.[77] Y.-H. E. Sheu, R. G. Anthony, E. J. Soltes, Fuel Process. Technol. 1988, 19, 31-50.[78] J. S. Shabtai, W. W. Zmierczak, E. Chornet, 2001, US Patent 6172272.[79] M. J. McCall, R. J. Allen, D. E. Mackowiak, 2008, WO2008/157165 A1.[80] S. Karagoez, T. Bhaskar, A. Muto, Y. Sakata, Fuel 2004, 83, 2293-2299.[81] R. W. Thring, Biomass Bioenerg. 1994, 7, 125-130.[82] J. E. Miller, L. Evans, A. Littlewolf, D. E. Trudell, Fuel 1999, 78, 1363-1366.[83] S. Nenkova, T. Vasileva, K. Stanulov, Chem. Natural Comp. 2008, 44, 182-185.[84] Wahyudiono, T. Kanetake, M. Sasaki, M. Goto, Chem. Eng. Technol. 2007, 30, 1113- 1122.[85] Wahyudiono, M. Sasaki, M. Goto, Fuel 2009, 88, 1656-1665.[86] T. Furusawa, T. Sato, M. Saito, Y. Ishiyama, M. Sato, N. Itoh, N. Suzuki, Appl. Catal. A: Gen. 2007, 327, 300-310.[87] M. Watanabe, H. Inomata, M. Osada, T. Sato, T. Adschiri, K. Arai, Fuel 2003, 82, 545- 552.[88] J. Van Haveren, E. L. Scott, J. Sanders, Biofuels Bioprod. Bioref. 2008, 2, 41-57.[89] D. C. Elliott, Energy Fuels 2007, 21, 1792-1815.[90] M. Kleinert, T. Barth, Chem. Eng. Technol. 2008, 31, 736-745.[91] D. Mohan, C. U. Pittman Jr., P. H. Steele, Energy Fuels 2006, 20, 848-889.[92] C. Briens, J. Piskorz, F. Berruti, Int. J. Chem. React. Eng. 2008, 6, 1-49.[93] F. Behrendt, Y. Neubauer, M. Oevermann, B. Wilmes, N. Zobel, Chem. Eng. Technol. 2008, 31, 667-677.[94] E. Furimsky, Appl. Catal. A: Gen. 2000, 199, 147-190.[95] M. J. Girgis, B. C. Gates, Ind. Eng. Chem. Res. 1991, 30, 2021-2058.[96] E. E. Harris, J. D’Ianni, H. Adkins, J. Am. Chem. Soc. 1938, 60, 1467-1470.[97] C. P. Brewer, L. M. Cooke, H. Hibbert, J. Am. Chem. Soc. 1948, 70, 57-59.[98] J. M. Pepper, H. Hibbert, J. Am. Chem. Soc. 1948, 70, 67-71.[99] J. M. Pepper, Y. W. Lee, Can. J. Chem. 1969, 47, 723-727.[100] J. M. Pepper, P. Supathna, Can. J. Chem. 1978, 56, 899-902.[101] J. M. Pepper, R. W. Fleming, Can. J. Chem. 1978, 56, 896-899.[102] K. Kashima, Y. Maeda, M. Oshima, 1964, Canadian Patent 700210..[103] D. W. Goheen, Adv. Chem. Ser. 1966, 59, 205-225.

Chapter 2

72

[104] J. Giesen, 1959, US Patent 2870133.[105] P. Urban, D. J. Engel, 1988, US patent 4731491.[106] S. B. Alpert, S. C. Shuman, 1970, Canadian Patent 851709.[107] D. C. Elliott, Prepr. Pap. - Am. Chem. Soc. Div. Pet. Chem. 1983, 28, 667-674.[108] E. G. Baker, D. C. Elliott, 1993, US Patent 5180868.[109] J. B. s. Bredenberg, M. Huuska, J. Raty, M. Korpio, J. Catal. 1982, 77, 242-247.[110] B. S. Gevert, J. E. Otterstedt, F. E. Massoth, Appl. Catal. 1987, 31, 119-131.[111] F. P. Petrocelli, M. T. Klein, Ind. Eng. Chem. Prod. Res. Dev. 1985, 24, 635-641.[112] S. J. Hurff, M. T. Klein, Ind. Eng. Chem. Fundam. 1983, 22, 426-430.[113] J. B. s. Bredenberg, M. Huuska, P. Toropainen, J. Catal. 1989, 120, 401-408.[114] R. K. M. R. Kallury, W. M. Restivo, T. T. Tidwell, D. G. B. Boocock, A. Crimi, J. Douglas, J. Catal. 1985, 96, 535-543.[115] E. Laurent, B. Delmon, Appl. Catal. A: Gen. 1994, 109, 77-96.[116] A. Centeno, E. Laurent, B. Delmon, J. Catal. 1995, 154, 288-298.[117] E. Laurent, B. Delmon, Appl. Catal. A: Gen. 1994, 109, 97-115.[118] G. De La Puente, A. Gil, J. J. Pis, P. Grange, Langmuir 1999, 15, 5800-5806.[119] M. Ferrari, B. Delmon, P. Grange, Carbon 2002, 40, 497-511.[120] M. Breysse, P. Afanasiev, C. Geantet, M. Vrinat, Catal. Today 2003, 86, 5-16.[121] E. Laurent, B. Delmon, J. Catal. 1994, 146, 281-291.[122] M. Huuska, J. Rintala, J. Catal. 1985, 94, 230-238.[123] Y. Yang, A. Gilbert, C. Xu, Appl. Catal. A: Gen. 2009, 360, 242-249.[124] E. O. Odebunmi, D. F. Ollis, J. Catal. 1983, 80, 56-64.[125] M. Koyama, Bioresour. Technol. 1993, 44, 209-215.[126] J. Shabtai, N. K. Nag, F. E. Massoth, J. Catal. 1987, 104, 413-423.[127] M. A. Ratcliff, D. K. Johnson, F. L. Posey, H. L. Chum, Appl. Biochem. Biotechnol. 1988, 17, 151-160.[128] D. T. A. Huibers, H. J. Parkhurst, 1983, US Patent 4420644.[129] J. S. Shabtai, W. W. Zmierczak, E. Chornet, 1999, US Patent 5959167.[130] J. Shabtai, Q. Guohe, K. Balusami, N. K. Nag, F. E. Massoth, J. Catal. 1988, 113, 206-219.[131] W. W. Zmierczak, J. D. Miller, 2006, WO2006/119357 A2.[132] D. Meier, J. Berns, O. Faix, U. Balfanz, W. Baldauf, Biomass Bioenergy 1994, 7, 99-105.[133] D. Meier, R. Ante, O. Faix, Bioresour. Technol. 1992, 40, 171-177.[134] A. Oasmaa, R. Alen, D. Meier, Bioresour. Technol. 1993, 45, 189-194.[135] O. I. Senol, E. M. Ryymin, T. R. Viljava, A. O. I. Krause, J. Mol. Catal. A: Chem. 2007, 277, 107-112.[136] E. Laurent, B. Delmon, Ind. Eng. Chem. Res. 1993, 32, 2516-2524.[137] T. R. Viljava, R. S. Komulainen, A. O. I. Krause, Catal. Today 2000, 60, 83-92.


73

[138] M. C. Edelman, M. K. Maholland, R. M. Baldwin, S. W. Cowley, J. Catal. 1988, 111, 243- 253.[139] C.-L. Lee, D. F. Ollis, J. Catal. 1984, 87, 325-331.[140] A. Y. Bunch, U. S. Ozkan, J. Catal. 2002, 206, 177-187.[141] A. Y. Bunch, X. Wang, U. S. Ozkan, Appl. Catal. A: Gen. 2008, 346, 96-103.[142] Y. Romero, F. Richard, Y. Renème, S. Brunet, Appl. Catal. A: Gen. 2009, 353, 46-53.[143] L. D. Rollmann, J. Catal. 1977, 46, 243-252.[144] V. LaVopa, C. N. Satterfield, Energy Fuels 1987, 1, 323-331.[145] H. Abe, A. T. Bell, Catal. Lett. 1993, 18, 1-8.[146] C. Zhao, Y. Kou, A. A. Lemonidou, X. Li, J. A. Lercher, Angew. Chem. Int. Ed. 2009, 48, 3987-3990.[147] D. C. Elliott, T. R. Hart, Energy Fuels 2009, 23, 631-637.[148] P. de Wild, R. Van der Laan, A. Kloekhorst, E. Heeres, Environ. Prog. Sustainable Energy 2009, 28, 461-469.[149] N. Yan, C. Zhao, P. J. Dyson, C. Wang, L.-t. Liu, Y. Kou, ChemSusChem 2008, 1, 626-629.[150] D. J. Engel, K. Z. Steigleder, 1987, US Patent 4647704.[151] V. A. Yakovlev, S. A. Khromova, O. V. Sherstyuk, V. O. Dundich, D. Y. Ermakov, V. M. Novopashina, M. Y. Lebedev, O. Bulavchenko, V. N. Parmon, Catal. Today 2009, 144, 362-366.[152] J. Filley, C. Roth, J. Mol. Catal. A: Chem. 1999, 139, 245-252.[153] B. Loubinoux, M. Heitz, G. Coudert, G. Guillaumet, Tetrahedron Lett. 1980, 21, 4991- 4994.[154] A. Oasmaa, A. Johansson, Energy Fuels 1993, 7, 426-429.[155] G. S. Macala, T. D. Matson, C. L. Johnson, R. S. Lewis, A. V. Iretskii, P. C. Ford, ChemSusChem 2009, 2, 215-217.[156] S. Kotrel, M. Emmeluth, A. Benöhr, 2009, WO2009/037284 A2.[157] E. Portjanskaja, K. Stepanova, D. Klauson, S. Preis, Catal. Today 2009, 144, 26-30.[158] Y.-S. Ma, C.-N. Chang, Y.-P. Chiang, H.-F. Sung, A. C. Chao, Chemosphere 2008, 71, 998- 1004.[159] E. Portjanskaja, S. Preis, Int. J. Photoenergy 2007, 2007, 76730-76736.[160] C. Crestini, M. C. Caponi, D. S. Argyropoulos, R. Saladino, Biorg. Med. Chem. 2006, 14, 5292-5302.[161] C. Crestini, P. Pro, V. Neri, R. Saladino, Bioorg. Med. Chem. Lett. 2005, 13, 2569-2578.[162] W. A. Herrmann, T. Weskamp, J. P. Zoller, R. W. Fischer, J. Mol. Catal. A: Chem. 2000, 153, 49-52.[163] W. A. Herrmann, R. M. Kratzer, R. W. Fischer, Angew. Chem. Int. Ed. 1997, 36, 2652- 2654.

Chapter 2

74

[164] F. G. Sales, L. C. A. Maranhão, N. M. Lima Filho, C. A. M. Abreu, Chem. Eng. Sci. 2007, 62, 5386-5391.[165] S. Bhargava, H. Jani, J. Tardio, D. Akolekar, M. Hoang, Ind. Eng. Chem. Res. 2007, 46, 8652-8656.[166] J. Zhang, H. Deng, L. Lin, Molecules 2009, 14, 2747-2757.[167] S. Baumberger, A. Abaecherli, M. Fasching, G. Gellerstedt, R. Gosselink, B. Hortling, J. Li, B. Saake, E. De Jong, Holzforschung 2007, 64, 459-468.[168] D. P. Koullas, E. G. Koukios, E. Avgerinos, A. Abaecherli, R. Gosselink, C. Vasile, R. Lehnen, B. Saake, J. Suren, Cellul. Chem. Technol. 2006, 40, 719-725.[169] C. Vasile, R. Gosselink, P. Quintus, E. G. Koukios, D. P. Koullas, E. Avgerinos, D. A. Abacherli, Cellul. Chem. Technol. 2006, 40, 421-429.[170] R. J. A. Gosselink, A. Abächerli, H. Semke, R. Malherbe, P. Käuper, A. Nadif, J. E. G. Van Dam, Ind. Crops Prod. 2004, 19, 271-281.

Part I

Lignin Depolymerization

Chapter 3

Lignin Solubilization and Depolymerization by Liquid-Phase Reforming and Hydrogenation

AbstractA novel catalytic process is presented for the valorization of lignin (exemplified

with kraft, organosolv and sugarcane bagasse lignin) using a mixture of ethanol and water as solvent. It was found that ethanol/water mixtures readily solubilize lignin when heated to 225 ˚C at 58 bar He pressure with little residual solids. The molecular weight of the dissolved lignins was shown to be reduced by Gel Permeation Chromatography and quantitative Heteronuclear Single Quantum Coherence NMR showed that typical ether linkages in lignin were cleaved. The use of liquid-phase reforming of the solubilized lignins over a Pt/γ-Al2O3 catalyst at 225 ˚C and 58 bar helium pressure is introduced to give a combined yield of up to 17.6% of monomeric aromatic oxygenates such as guaiacol and substituted guaiacols. Reduction of the kraft and organosolv lignin after dissolution in ethanol/water using a supported transition metal catalyst at 200 ˚C and 30 bar hydrogen yielded up to 6% of cyclic hydrocarbons and aromatics.

Based on: J. Zakzeski*, A. L. Jongerius*, P. C. A. Bruijnincx, B. M. Weckhuysen, “Catalytic Lignin Valorization Process for the Production of Aromatic Chemicals and Hydrogen“ ChemSusChem 2012, 5, 1602-1609.* Both authors contributed equally to this work.

Part I, Chapter 3

80

3.1 Introduction

Lignin is a natural amorphous three-dimensional polymer consisting of methoxylated phenylpropane structures that confers strength and rigidity to plants and protects the cellulose and hemicellulose from microbial attack. [1-5] Although the composition of lignin varies considerably from plant to plant, particularly with regards to the type and quantity of linkages in the polymer and the number of methoxy groups present on the aromatic rings, lignocellulosic biomass typically contains about 15-30% of lignin by weight and about 40% by energy. Therefore, lignin represents a valuable renewable resource and an immense opportunity for the sustainable production of chemicals and energy vectors. [2-10] Currently, as little as 2% of lignin is used in low-value products such as dispersants or binding agents with the remainder burned as a low-value fuel. [11] Based on its unique chemical structure, a wide variety of bulk and fine chemicals, particularly aromatics, and fuels can potentially be obtained from lignin provided that appropriate catalytic technology is developed.

Processes that derive relatively pure lignin streams from lignocellulosic biomass have already been developed with prominent examples including organosolv processes and the recently developed LignoBoost process. [12, 13] The latter technology produces extra pure softwood kraft lignin on a large scale as a byproduct of pulp and paper production. The availability of such lignin streams further encourages the development of lignin valorization technologies, in particular for the production of renewable, high-value chemicals. [14] Various routes have been explored to depolymerize such extracted technical lignins including pyrolysis, catalytic hydrogenation, oxidation, hydrolysis and hydrocracking. [1, 15-24] A two-stage catalytic process that utilizes a base-catalyzed depolymerization followed by selective hydrocracking with a super-acid catalyst to form hydrocarbon gasoline from lignin has also been reported. [25] The average total aromatic monomer yield is, with a few exceptions, in the range of 5-10%. [1, 26, 27] Most of these lignin conversion processes are operated at elevated temperatures (about 250-600 oC). Aqueous-phase reforming (APR) (Scheme 3.1a) of biomass-derived oxygenates provides an interesting alternative as the production of hydrogen and alkanes from various renewable oxygenates has been reported at temperatures (T < 265 oC) and pressures milder than those required for gasification or pyrolysis. [28] Recently, our lab reported the first use of APR at temperatures of 225 oC and He pressures of 29 bar for the conversion of lignin to monomeric aromatic chemicals by cleavage and further conversion of some of the major linkages in lignin, particularly the β-O-4 (Scheme 3.1b) and, to a lesser extent, the 5-5’ linkages. [29]

Lignin Solubilizatin and Aromatics Production by Liquid-Phase Reforming and Hydrogenation

81

Hydrogen and other light gases were obtained as useful byproducts, but the lignin APR system was plagued by extensive solid formation caused by lignin recondensation, which resulted in relatively low yields of isolated monomeric products. Further improvements in the process conditions of this lignin valorization process are desired to improve monomer yields and minimize recondensation.

As low monomer yield and substrate agglomeration are strongly associated with the limited solubility of the lignin in the aqueous medium used, changing the solvent composition should hold obvious advantages. Indeed, the exploration of different (combinations of) solvents for lignin valorization, such as (supercritical) acetone, ethanol or methanol, to aid in lignin solvolysis has been reported. In a recent example, a mixture of carbon dioxide/acetone/water was used to obtain a phenolic oil yield of 10-12%. [26] Similarly, solvent optimization has also aided some APR processes, as exemplified by the recycle and use of the actual APR product mixture as the solvent in a recent patent application [30] and the improvement in hydrogen yield from sorbitol APR upon addition of ethanol to the system. [31] The use of an ethanol/water mixture to enhance lignin dissolution and conversion is therefore a logical step to address the issues encountered in lignin aqueous phase reforming, i.e. solubility and agglomeration, especially since ethanol is already used in organosolv processes for solubilizing the lignin fraction of lignocellulosic biomass. [12] Furthermore, ethanol can be produced sustainably from the cellulose and hemicellulose fractions of lignocellulosic biomass, allowing further integration of lignin into a potential biorefinery operation.

Scheme 3.1: Aqueous-phase reforming (APR) of a: glycerol, b: the lignin β-O-4 linkage.

Part I, Chapter 3

82

In this Chapter, we show that different types of lignin, i.e. kraft, organosolv and bagasse lignin, can indeed be readily dissolved in ethanol-water mixtures at moderate temperatures without formation of agglomerates. This lignin solubilization is connected to a reduction of the molecular weight of the dissolved lignins due to cleavage of the lignin ether linkages as shown by quantitative 2D NMR and Gel Permeation Chromatography (GPC) analysis. This enhanced lignin dissolution increases the possibilities for further catalytic valorization to produce bulk platform and specialty chemicals. We have explored two routes, which illustrate the possibility of tuning the composition of the product slate depending on the reaction conditions and catalyst(s) employed. First, we introduce the liquid-phase reforming (LPR) of lignin using ethanol/water mixtures in an effort to prevent the formation of solids and increase the isolated monomeric aromatic product yields. The process operates under moderate conditions resembling those of APR with the exception that higher pressures (P > 55 bar) are required because of the higher vapor pressure of ethanol relative to water. Secondly, the reduction of solubilized lignin in the presence of noble metal catalysts and hydrogen is reported. The latter can in principle be renewably sourced via the APR of different fractions of lignocellulosic biomass, making the overall process fully integrated.

3.2 Results and Discussion

3.2.1 Solubilization of ligninA schematic depiction of the overall approach and the particular catalysts used

is presented in Figure 3.1. The use of ethanol/water, which is cheap and readily obtained from the sugar fraction of lignocellulosic biomass feed, as the solvent together with various combinations of homogeneous and heterogeneous catalysts for the valorization of lignin is explored. The results depicted in Figure 3.1 and Table 3.2 were obtained in the valorization of kraft lignin, one of the most recalcitrant forms of lignin yet produced in great abundance in the pulp and paper industry, although the process reported in this work was also readily applicable to other lignin sources (See Table 3.3 and 3.4). The clear benefits of using ethanol as co-solvent are visualized in Figure 3.2, illustrating the agglomeration/solubilization behavior of kraft lignin in water and ethanol/water solutions from 20 to 225 ˚C using an autoclave equipped with optical windows.

Agglomeration of solid material, which otherwise occurred for kraft lignin at


83

Figure 3.2: Solubilization of kraft lignin in water (top) and ethanol/water (bottom) from 20 to 225 ˚C, observed with an autoclave equipped with optical windows.

Figure 3.1: Valorization of lignin via reduction or liquid-phase reforming produces valuable aromatic chemicals. The yields and products shown are obtained with the kraft lignin.

OO

OHO

OHO

OH

OO

OHO

Reduction Liquid-Phase ReformingPt/Al2O3 + HPA

Pt/Al2O3 + NaOH

Pt/Al2O3 + H2SO4

Pt/Al2O3, Pd/C, Ru/C

LigninSolubilization

(>99%)

6.2% 17.6% 13.2% 12.8% Aromatic Isolated Yield

Most Abundant Products

Gases

LignocellulosicBiomass

O

OH

OOH

OH

OHO

OH

OOO

OH

OH

O

O

HO OH

O

O

HO

Lignin

Cellulose / Hemicellulose

O O

OH

O

OH

O

OH

HO

OH

HO

n

Lignin Pretreatment

Ethanol Fuels andBulk Chemicals

H2 CH4 C2H6 CH4 C2H6 H2 CH4 C2H6 C3H8

OHO

OHO

OHO

OHO

O

OHO

OHOH

Part I, Chapter 3

84

temperatures in excess of 200 ̊ C in aqueous solutions, [29] was completely suppressed in ethanol/water. Importantly, the lignin remained completely solubilized even after cooling to room temperature and no solid material was recovered. The organosolv and bagasse lignins also proved to be highly soluble with only about 7.9% and 9.4% of solids recovered, respectively. GPC measurements of the kraft lignin before and after ethanol/water treatment indicated a reduction in the molecular weight from 5300 to 3800 Da and a decrease in polydispersity from 6.7 to 5.5 during the solubilization process. In the cases of organosolv lignin and sugarcane bagasse lignin, a reduction in molecular weight from 4000 to 3000 Da and 3100 to 2950 Da, respectively, was observed. Moreover, quantitative HSQC (Q-HSQC) NMR spectra of the kraft and organosolv lignin, shown in Figure 3.3, before and after solubilization indicate a clear reduction in the number of β-β, β-5 and β-O-4 linkages present in the lignin structure following solubilization. The disruption of the latter directly leads to depolymerization and reduction of the molecular weight of the lignin. [6, 32, 33] Analysis of the NMR spectra of the bagasse and solubilized bagasse lignins was hampered by the many impurities that are present in the bagasse lignin, which did not allow information about changes in the linkages to be obtained.

Quantitative analysis of the HSQC NMR spectra, shown in Figure 3.3 and Table 3.1, indicate a reduction of around 50%, 20%, and 20% of the number of β-O-4 bonds, β-β bonds, and β-5 bonds, respectively, in kraft lignin and reductions of 70%, 20% and 35% in organosolv lignin. These numbers indicate that the cleavage of the lignin ether linkages aids in the solubilization process either by shortening the polymer chain or by increasing polarity. The use of alcohols or phenols, either for solvolysis of the ethers or as capping agents, has previously been demonstrated for lignin depolymerization. [34, 35] Similarly, the incorporation of ethanol into the isolated aromatic products after the catalytic lignin valorization process reported here (vide supra) suggests that ethanol reacts with these disrupted linkages and hinders lignin repolymerization, which in the absence of ethanol results in the formation of highly recalcitrant solids.

3.2.2 Liquid-phase reforming of ligninThe lignin solubilization process can now be combined in one pot with a

catalytic valorization step. A summary of the kraft, organosolv and sugarcane bagasse lignin conversion results, including the most abundant isolated monomeric aromatic products, gases formed, and byproducts obtained from the ethanol co-solvent, are provided in Tables 3.2, 3.3 and 3.4. The LPR reactions were conducted with the benchmark Pt/γ-Al2O3 reforming catalyst in the presence of acid (H2SO4


85

6

6

5

5

4

4

3

3

H (ppm

90 90

80 80

70 70

60 60

50 50

13C

(ppm

)

)

6

6

5

5

4

4

3

3

1H (ppm)

90 90

80 80

70 70

60 60

50 50

13C

(ppm

)

6

6

5

5

4

4

3

3

H (ppm

90 90

80 80

70 70

60 60

50 50

13C

(ppm

)

)

6

6

5

5

4

4

3

3

H (ppm

90 90

80 80

70 70

60 60

50 50

13C

(ppm

)

)

a) b)

c) d)

1

2

34

5

6

1

2

34

5

6

1

2

34

5

6

1

2

34

5

6

1

11

Figure 3.3: Expanded side chain region δC/δH 45-95/2-6 HSQC NMR spectrum of a) organosolv lignin, b) kraft lignin, c) solubilized organosolv lignin and d) solubilized kraft lignin. For identification of the numbered peaks see Table 3.1.

Part I, Chapter 3

86

Link

age

Kra

ft li

gnin

(n

orm

aliz

ed)

Solu

biliz

ed

kraf

t lig

nin

(nor

mal

ized

)

Dec

reas

e kr

aft

ligni

n (%

)

Org

anos

olv

ligni

n (n

orm

aliz

ed)

Solu

biliz

ed

orga

noso

lv

ligni

n (n

orm

aliz

ed)

Dec

reas

e or

gano

solv

lig

nin

(%)

1β-

O-4

Cα

19.7

9.2

5317

.36.

463

2β-

O-4

Cβ

(gua

iacy

l)13

.46.

949

6.7

1.4

78

3β-

O-4

Cβ

(syr

ingy

l)1.

70.

851

10.9

2.3

79

4β-

β C α

6.5

5.1

2213

.410

.522

5β-

β C γ2

12.6

10.2

1917

.815

.115

6β-

56.

34.

724

6.9

4.4

36

Mw

N/A

5270

3788

2839

6129

8925

Tabl

e 3.

1: Q

uant

itat

ive n

orm

aliz

ed N

MR

peak

are

as sh

owin

g the

pro

port

iona

l dec

reas

e in

the l

inka

ges o

f lig

nin

follo

win

g sol

ubili

zatio

n an

d th

e de

crea

se in

mol

ecul

ar w

eigh

t mea

sure

d by

GPC

.


87

or phosphotungstic acid) or base (NaOH) co-catalysts. In the case of the H2SO4 co-catalyzed reaction of kraft lignin, guaiacol was the most abundant isolated monomeric aromatic product followed by ethylcatechol and methanol, which are most probably derived from ethylguaiacol via acid-catalyzed hydrolysis of the aryl-alkyl ether. The total amount of isolated monomeric aromatic products obtained by the LPR of lignin exceeded by an order of magnitude that obtained in the previously reported lignin APR process, [30] with up to 17.6% of the original lignin mass converted to monomeric isolated aromatics. This yield is comparable to the highest isolated monomer yield in other lignin conversion processes and exceeds many of the typical state-of-the-art lignin valorization technologies (see Chapter 1 for the most recent developments). [22, 26, 27, 36] Another important difference with lignin APR is that the formation of solids was suppressed; indeed, no char formation was observed and the only solid material recovered at the conclusion of the reaction was the catalyst. During the LPR reaction, light gases, in particular H2 along with CH4 and C2H6, were also formed. No CO was detected as the LPR conditions favor the water-gas-shift reaction resulting in the detection of only CO2. The limited amounts of light alkanes formed can result from several reaction pathways that are available under the employed reaction conditions, including Brønsted acid-catalyzed dehydration and dealkylation reactions or Fischer-Tropsch and methanation reactions of H2 and CO or CO2. [37] In addition to the reaction of the lignin, 34% of the ethanol used in the process was converted to diethyl ether, a process catalyzed by H2SO4 at elevated temperatures. [38]

Substitution of H2SO4 for phosphotungstic acid resulted in higher yields of guaiacol and ethoxylated guaiacol products relative to the H2SO4 co-catalyzed reaction. The formation of the latter product provides evidence that the ethanol aids the valorization process by serving as a capping agent for the phenolic functionality of the depolymerized lignin, which is otherwise susceptible to repolymerization with other lignin molecules. Up to 13.2% of monomeric aromatics could be isolated with this catalyst/co-catalyst combination. Somewhat remarkably, no H2 was observed in the gas phase with HPA, although CO2 was detected, indicating a rapid consumption of any hydrogen formed, for instance, by a hydrogenolysis process. The light alkanes were nonetheless again observed in more or less the same quantities. The ethanol solvent was also significantly more stable in the presence of this co-catalyst, with reduced quantities of 2% diethyl ether detected at the conclusion of the reaction.

The hydrolytic degradation of alkaline lignin in water/ethanol with NaOH as a catalyst was previously reported to give lignin oligomers if phenol was used as

Part I, Chapter 3

88

a capping agent. [33] The use of an additional catalyst now allows for sequential degradation of such oligomers to yield monomeric aromatic compounds in one step. Indeed, the use of a catalyst combination of Pt/Al2O3 and NaOH resulted in a yield of approximately 12.8% of isolated monomeric products, with respect to the original kraft lignin mass; however, the product distribution differed from the acid-co-catalyzed reactions. Although guaiacol was still obtained in relatively high yields, benzyl alcohols were also obtained. The latter products were not observed when an acidic co-catalyst was used. H2 production was again observed in the presence of NaOH, with the basic conditions predictably reducing alkane production compared to the acidic conditions. The ethanol solvent also gave different byproducts, as small amounts of mainly 1-butanol (4%) were obtained via a Guerbet-type reaction under alkaline conditions. [39, 40] The acid and base-catalyzed reactions thus result in small changes in the final product distribution.

In traditional aqueous-phase reforming reactions of sugar-derived oxygenates, C-C bond cleavage is favored over C-O cleavage reactions to convert the oxygenates selectively to H2 and CO2. Acidic conditions lead to increased C-O bond cleavage by a hydrolysis pathway ultimately leading to the formation of alkanes as shown for the aqueous phase processing of sorbitol. [42] Similarly, the monomeric products obtained in the H2SO4-catalyzed reactions still often contain an intact propyl side-chain, whereas the side-chains in the products from NaOH-catalyzed reactions are at least shortened and ethyl- and methyl-substituted aromatics are mainly obtained. Furthermore, the acidic conditions favor ether hydrolysis, limiting the formation of ethyl ethers of the free phenolics. In the presence of NaOH, however, these phenolic groups are readily etherified, to give 4-ethoxy-3-methoxytoluene, for instance, as the main product from both organosolv and kraft lignin.

Differences in product composition were also observed for the organosolv or bagasse lignins compared to the kraft lignin. This result is to be expected, as the monolignol ratios differ in the three lignin samples. Indeed, syringol was among the major products obtained from organosolv hardwood lignin reactions and ethylphenol was the main product obtained from sugarcane bagasse lignin. For organosolv lignin, isolated from the lignocellulose with a water/ethanol mixture, relatively low total yields (9.0 and 4.1% respectively) of monomeric products were obtained using the H2SO4 and phosphotungstic acid co-catalysts. Product yields from sugarcane bagasse, at 15.5% (H2SO4) and 15.0% (phosphotungstic acid), better resembled the yields obtained with kraft lignin. The use of NaOH as co-catalyst did result in a relatively


89

Process Catalyst Co-Catalyst

Entry Most Abundant Isolated Monomeric Aromatic

Product

Yield (mg)

Selectivity to Aromatic

Products (%)

H2 (%)

Alkanes (%)

Liquid Pt/Al2O3 H2SO4 1 guaiacol 2,21 17,6 3,79 0,69

Phase 2 ethylcatechol 1,28

Reforming 3 vanillin 1,26

4 propylguaiacol 0,59

5 propenylguaiacol ethyl ether

0,54

Pt/Al2O3 HPA 1 guaiacol 4,09 13,2 0,00 0,29

2 4-ethoxy-3-methoxytoluene

1,15

3 propeneguaiacol 0,29

4 ethylguaiacol 0,28

5 methylguaiacol 0,26

Pt/Al2O3 NaOH 1 4-ethoxy-3-methoxytoluene

1,49 12,8 3,76 0,17

2 α-methyl(methylbenzyl) alcohol

1,44

3 guaiacol 1,17


5 methylbenzyl alcohol 0,80

Hydrogenation Ru/C 1 propylguaiacol 1,25 3,7 N/A N/A



4 guaiacol 0,33

5 hypovanillyl alcohol 0,26

Pd/C 1 propylguaiacol 1,51 4,8 N/A N/A



4 guaiacol 0,49

5 vanillin 0,22

Pt/Al2O3 1 propylguaiacol 1,51 6,2 N/A N/A



4 guaiacol 0,71


Ni/SiO2 1 vanillin 0,76 5,3 N/A N/A

2 guaiacol 0,52


4 cyclohexanone 0,04

5 propylguaiacol 0,03

Table 3.2: Most abundant isolated monomeric aromatic products, lignin conversion, useful gases, and ethanol-derived products obtained during the valorization of kraft lignin.

Part I, Chapter 3

90

high monomeric product yield of 11.6% for the organosolv lignin, whereas sugarcane bagasse only produced 2.2% aromatics in this case. The many impurities present in the bagasse lignin are apparently better tolerated under acidic than basic conditions. The relatively high H2 production from bagasse in the presence of H2SO4 is also notable. This increase can be attributed to the amount of remaining sugars in this technical lignin, which are readily reformed. For all other reactions, gas production was comparable to that of the kraft lignin.

3.2.3 Catalytic reduction of ligninRather than combining the catalytic step with solubilization in a one-pot approach,

the solubilized lignin can also be subsequently converted by subjecting it to a second catalytic conversion step allowing the use of different reaction conditions (e.g., temperature, reducing atmosphere). Different reduction catalysts in the presence of 30 bar H2 at 200 ˚C are used here in this second step with the solubilized lignins to obtain up to 6% monomeric products from kraft and organosolv lignin with a product distribution differing significantly from that obtained during the lignin LPR process. In contrast to lignin LPR, which tended to yield many aromatic compounds with the propyl side-chains shortened (e.g., methylguaiacol) or lacking (e.g., guaiacol), the most abundant compound obtained under the reductive conditions was propylguaiacol followed by ethylguaiacol.

Remarkably, ring hydrogenation products are hardly observed after the reduction reactions. Only small quantities of hydrogenated products, such as cyclohexanol and cyclohexanone, were detected in all reactions. A gas phase hydrogenation reaction of various guaiacol and anisole-type molecules over Pt/γ-Al2O3 in a continuous flow system also resulted in only trace amounts of ring hydrogenation products. [42] Several Pt/C catalysts were reported in water under hydrogenation conditions, however, to first hydrogenate the aromatic ring of various phenolic compounds before deoxygenation to cycloalkanes by hydrogenolysis took place under reaction conditions more comparable to our system. [43] As summarized in Tables 3.2, 3.3 and 3.4, the observed reactivity was similar for all the noble metal reduction catalysts tested. However, it is important to note that a Ni/SiO2 catalyst produced a different product profile with vanillin being the principle product obtained. In the case of organosolv lignin, syringol and propylsyringol were among the main products, whereas the sugarcane bagasse produced primarily phenol, alkylated phenolics and 5-hydroxymethylfurfural, which can be formed from residual sugars in the bagasse. These results are again in line with the monolignol composition and known impurities.


91

3.3 Conclusions

The use of ethanol/water mixtures to dissolve various lignin feedstocks greatly enhanced the solubility of lignin and consequently led to higher yields of aromatic chemicals in one-pot lignin liquid phase reforming reactions. The solubilization resulted in cleavage of various ether linkages in the macromolecule, as evidenced by HSQC NMR measurements, and a reduction in molecular weight, as revealed by GPC. The Pt/γ-Al2O3-catalyzed reforming reactions yielded up to 17.6% of monomeric guaiacol-type products when performed in the presence of H2SO4. These results show an increase of an order of magnitude in yield compared to the yields reported earlier for the aqueous-phase reforming of lignin and compares well to other reported lignin depolymerization processes. [42] Depending on the lignin source and the used co-catalyst, the product distributions changes and different light gases, such as hydrogen and methane, were formed. Char formation was not observed in any of the reactions. Reduction of solubilized lignin using transition metal catalysts led to the formation of alkyl-substituted guaiacol-type molecules with isolated yields of up to 6% for Pt/γ-Al2O3.

3.4 Materials and Methods

3.4.1 Lignins and catalyst materialsThe INDULIN AT kraft lignin (63.25% C, 6.05% H, 0.94% N, 1.64% S, 28.12% O by

difference), provided by ECN, was obtained from pine and is free of all hemicellulosic materials. The Alcell organosolv lignin (66.47% C, 5.96% H, 0.15% N, 27.43% O by difference) provided by Wageningen University, was obtained from hardwoods and isolated by the organosolv extraction method. The lignin from sugarcane bagasse (58.90% C, 4.90% H, 0.14% N, 1.53% S, 34.53% O by difference) provided by The Dow Chemical Company, was derived from Brazilian sugarcane. H2SO4 (Fisher Scientific, > 95%), phosphotungstic acid hydrate (Sigma-Aldrich, 99.995%) NaOH (Sigma-Aldrich, 97%). The catalyst materials, 1 wt% Pt/γ-Al2O3, 5 wt% Ru/C, 5 wt% Pd/C, 5 wt% Pt/γ-Al2O3, were obtained from Sigma-Aldrich and used as received. The 5 wt% Ni/SiO2 catalyst was prepared by homogeneous deposition precipitation using 30 g silica (Aerosol 300, Degussa), 7.9 g Ni(NO3)2 (Acros, 99%) and 4.85 g of urea (Acros, 99%) at 90 °C for 18 h in 1.5 L demineralized water. The dried catalyst was reduced under a hydrogen atmosphere at 700 ˚C. Helium (5.0), argon (5.0) and hydrogen (5.0) were purchased from Linde gas.

Part I, Chapter 3

92



Product

Yield (mg)


Products (%)

H2 (%)

Alkanes (%)

Liquid Pt/Al2O3 H2SO4 1 propeneguaiacol 0,30 9,00 4,46 0,69

Phase 2 ethylcatechol 0,28

Reforming 3 guaiacol 0,26

4 syringol 0,22

5 1-ethoxy-4-methoxybenzene 0,18

Pt/Al2O3 HPA 1 syringol 1,09 4,10 0,00 0,37

2 guaiacol 0,59

3 4-ethoxy-3-methoxytoluene 0,26

4 propenylguaiacol ethyl ether 0,19


Pt/Al2O3 NaOH 1 4-ethoxy-3-methoxytoluene 2,18 11,60 3,10 0,24

2 α-methyl(methylbenzyl) alcohol

1,24

3 propylbenzyl alcohol 1,03

4 syringol 1,00

5 methylacetophenone 0,95

Hydrogenation Ru/C 1 propenylguaiacol ethy ether 1,53 6,43 N/A N/A

2 propenylguaiacol 0,47

3 1-catechol-1-propanone 0,42



Pd/C 1 syringol 0,68 5,48 N/A N/A



4 syringolaldehyde 0,32


Pt/Al2O3 1 propenylguaiacol 0,38 3,22 N/A N/A


3 syringol 0,29



Ni/SiO2 1 5-methylfurfural 0,38 3,32 N/A N/A

2 syringolpropaldehyde 0,37

3 p-coumaric ethyl ether 0,28

4 2-propenylsyringol 0,26

5 syringol 0,23

None 1 syringolpropaldehyde 0,37 2,24 N/A N/A

Table 3.3: Most abundant isolated monomeric aromatic products, lignin conversion, useful gases, and ethanol-derived products obtained during the valorization of organosolv lignin.


93



Product

Yield (mg)


Products (%)

H2 (%)

Alkanes (%)

Liquid Pt/Al2O3 H2SO4 1 ethylphenol 2,46 15,5 6,03 0,83

Phase 2 1-ethoxy-4-methoxybenzene 1,38

Reforming 3 propenylguaiacol ethyl ether 1,16

4 phenol 0,75

5 syringol 0,42

Pt/Al2O3 HPA 1 ethylphenol 5,19 15 0 0,47

2 1,2-dimethoxybenzene 2,18

3 guaiacol 1,65

4 phenol 1,19


Pt/Al2O3 NaOH 1 ethylphenol 0,38 2,2 2,7 0,33

2 methylacetophenone 0,28

3 ethoxyphenol 0,27

4 4-ethoxy-3-methoxytoluene 0,26

5 phenol 0,23

Hydrogenation Ru/C 1 ethylphenol 0,17 1,2 N/A N/A


3 phenol 0,06

4 2-hydroxylpropionic acid ethyl ether

0,05

5 5-methylfurfural 0,04

Pd/C 1 vanillic acid 0,24 1,8 N/A N/A

2 ethylphenol 0,22



5 5-methylfurfural 0,07

Pt/Al2O3 1 ethylphenol 0,65 2,2 N/A N/A



4 propenylguaiacol ethyl ether 0,1


Ni/SiO2 1 5-hydroxymethylfurfural 0,14 2,2 N/A N/A


3 ethylphenol 0,13

4 ethylguaiacol 0,1


None 1 syringpropaldehyde 0,19 1,4 N/A N/A

Table 3.4: Most abundant isolated monomeric aromatic products, lignin conversion, useful gases, and ethanol-derived products obtained during the valorization of sugarcane bagasse.

Part I, Chapter 3

94

Figure 3.4: Schematic depiction of the windowed-autoclave.

3.4.2 Catalytic reactions The lignin solubilization studies

were conducted in a semi-batch 200 mL autoclave equipped with quartz windows (Figure 3.4), thermocouple, pressure gauge and transducer, magnetic driver (750 rpm), and back-pressure regulator set at 58 bar. Lignin samples were stored in a desiccator prior to use. During a typical treatment, 0.200 g kraft lignin was added to the autoclave with 100 mL H2O and 100 mL ethanol. The autoclave was then sealed, purged and charged with 58 bar He, and finally heated at approximately 4 ˚C/min to 225 ˚C.

The liquid phase reforming and reduction reactions were conducted in a semi-batch 40 mL Parr autoclave equipped with a thermocouple, a pressure transducer and gauge, a magnetic driver (750 rpm), and a back-pressure regulator set at 58 bar. Lignin samples were stored in a desiccator prior to use. During a typical liquid phase reforming reaction, 0.125 g lignin was added to the autoclave along with 0.125 g 1wt% Pt/γ-Al2O3, 5.5 g H2O, 5.5 g ethanol, and either 0.58 g H2SO4, 0.3 g phosphotungstic acid hydrate or NaOH. The autoclave was then sealed, purged with He, and then 58 bar He was charged to the autoclave. The autoclave was then rapidly heated to 225 ˚C in the course of about 15 min. Gas sampling was conducted using a dual-column Varian 490-GC micro gas chromatography unit. After the designated time (typically 1.5 h), the autoclave was cooled in an ice bath and vented.

Catalytic reduction of solubilized lignin was performed in a 25 mL stainless steel high pressure Parr batch autoclave equipped with a thermocouple, a pressure transducer and gauge, a magnetic driver (750 rpm). In a typical reaction, 12 mL of pre-dissolved lignin solution (0.120 g kraft in 6 mL H2O and 6 mL EtOH) was added to the autoclave along with 0.050 g of the specified catalyst (5 wt% Ru/C, Pt/γ-Al2O3, Pd/C, or Ni/SiO2). The autoclave was thrice purged with argon and charged with 30 bar H2, and heated to 200 ˚C. After a designated reaction time of 4 h, the reactor was


95

allowed to cool and then quenched with an ice bath.

At the conclusion of the reactions, the autoclaves were vented, the liquid phase was separated from the solids, and finely dispersed solids were isolated by centrifugation if necessary. Products contained in the liquid phase were isolated by three sequential extractions using approximately 9 g dichloromethane, the quantity of which was subsequently reduced using a rotary evaporator at 40 ˚C until dichloromethane no longer evaporated from the extracted solution.

3.4.3 AnalysisChemical composition of the isolated yields was determined by a Varian GC

equipped with a VF-5ms capillary column and a Shimadzu GC equipped with a CP-WAX capillary column, each equipped with an FID detector. Syringaldehyde was used as an internal standard. The quantity of unknown products was estimated using the response factor determined for guaiacol. Product identification was conducted using a Shimadzu GCMS-QP2010 equipped with either a VF-5ms or CP-WAX capillary column and by comparison with pure compounds when available.

HSQC NMR spectra were obtained using a Bruker Avance II 600 MHz spectrometer equipped with a 5 mm CPTCI 1H-13C/15N/2H cryogenic probe with z-gradients at 25 ˚C using the CPMG-QHSQC pulse program. [44] The lignins and lignin solutions were freeze-dried and dissolved in DMSO-d6 (40 mg/mL) (99.9% DMSO-d6, Cambridge Isotopes Laboratories), and chemical shifts were referenced to the residual DMSO signal (2.50/39.5) ppm. Peaks were assigned according to the literature and quantitative data were obtained by integration of separated peaks and normalization to the total integral of the aromatic area. δC/δH 100-137/4.8-7.8. Quantitative analysis was done using TopSpin version 2.1.

GPC measurements were performed on an alkaline SEC by Waters Alliance system equipped with a manually packed column (4.6 mm x 30 cm) with ethylene glycolmethacrylate copolymer TSK gel Toyopearl HW-55F according to the work of Gosselink et al. [45] Sodiumpolystyrene sulfonates (Mw range 891 to 976,000 Da) were used for calibration of the molar mass distribution. The lignins and lignin solutions were freeze-dried and dissolved in a 0.5 M NaOH solution at room temperature at a concentration of 0.5 wt% by gently shaking the solution for 12 h. GPC measurements were run at 40 ˚C with 0.5 M NaOH eluent at a flow rate of 1 mL/min and UV detection at 280 nm.

Part I, Chapter 3

96

3.5 Acknowledgments

Joe Zakzeski is acknowledged for his contributions to the publication on which this chapter is based. Jaap van Hal (ECN), Richard Gosselink (Wageningen University) and Matthijs Ruitenbeek (The Dow Chemical Company) are kindly acknowledged for supplying the kraft, organosolv and sugarcane lignin, respectively. We also thank Richard Gosselink and Jacinta van der Putten from Wageningen University and Hans Wienk from Utrecht University for help with the GPC and NMR measurements, respectively. NMR experiments were performed at SONNMRLSF at the Bijvoet Institute of Utrecht University.

3.6 References

[1] J. Zakzeski, P. C. A. Bruijnincx, A. L. Jongerius, B. M. Weckhuysen, Chem. Rev. 2010, 110, 3552-3599.[2] R. D. Perlack, L. L. Wright, A. F. Turhollow, R. L. Graham, B. J. Stokes, D. C. Er bach. U. S. Department of Energy, Biomass as Feedstock for a Bioenergy and Bioproducts Industry: The Technical Feasibility of a Billion-ton Annual Supply, 2005.[3] F. S. Chakar, A. J. Ragauskas, Ind. Crops Prod. 2004, 20, 131-141.[4] S. K. Ritter, Chem. Eng. News, 2008, 86, 59-68.[5] Y.-H. P. Zhang, S.-Y. Ding, J. R. Mielenz, J.-B. Cui, R. T. Elander, M. Laser, M. E. Himmel, J. R. McMillan, L. R. Lynd, Biotechnol. Bioeng. 2007, 97, 214-223. [6] E. A. Capanema, M. Y. Balakshin, J. F. Kadla, J. Agric. Food Chem. 2004, 52, 1850-1860.[7] D. V. Evtuguin, C. Pascoal Neto, J. Rocha, J. D. Pedrosa de Jesus, Appl. Catal. A: Gen. 1998, 167, 123-139.[8] E. Adler, Wood Sci. Techol. 1977, 11, 169-218.[9] D. N.-S. Hon, N. Shiraishi, Wood and Cellulose Chemistry, CRC Press, 1991, chapter 4. [10] Z. Z. Lee, G. Meshitsuka, N. S. Cho, J. Nakano, Mokuzai Gakkaishi 1981, 27, 671-678.[11] R. J. A. Gosselink, E. de Jong, B. Guran, A. Abacherli, Ind. Crops Prod. 2004, 20, 121-129.[12] T. N. Kleinert, 1971, US patent 3585104.[13] M. Nagy, M. Kosa, H. Theliander, A. J. Ragauskas, Green Chem. 2010, 12, 31-34.[14] P. A. Willems, Science 2009, 325, 707-708. [15] E. Dorrestijn, L.J.J. Laarhoven, I.W.C.E. Arends, P. Mulder, J. Anal. Appl. Pyrolysis 2000, 54, 153-192.[16] P. F. Britt, A. C. Buchanan, M. J. Cooney, D. R. Martineau, J. Org. Chem. 2000, 65, 1376- 1389.


97

[17] M. Misson, R. Haron, M.F.A. Kamaroddin, N.A.S. Amin, Biores. Technol. 2009, 100, 2867-2873.[18] R. W. Thring, J. Breau, Fuel 1996, 75, 795-800.[19] W. L. Schinski, A. E. Kuperman, J. Han, D. G. Naae, 2009, US patent 2009/0218062 A0218061.[20] P. de Wild, R. V. der Laan, A. Kloekhorst, E. Heeres, Environ. Prog. Sust. Energy 2009, 28, 461-469.[21] H. Werhan, J. M. Mir, T. Voitl, P. P. van Rohr, Holzforschung 2011, 65, 703-709.[22] V. M. Roberts, V. Stein, T. Reiner, A. Lemonidou, X. B. Li, J. A. Lercher, Chem. Eur. J. 2011, 17, 5939-5948.[23] P. R. Patwardhan, R. C. Brown, B. H. Shanks, ChemSusChem 2011, 4, 1629-1636.[24] M. Nagy, K. David, G. J. P. Britovsek, A. J. Ragauskas, Holzforschung 2009, 63, 513-520.[25] J. S. Shabtai, W. W. Zmierczak, E. Chornet, 2001, US patent 6172272 B1.[26] R. J. A. Gosselink, W. Teunissen, J. E. G. van Dam, E. de Jong, G. Gellerstedt, E. L. Scott, J. P. M. Sanders, Bioresource Technol. 2012, 106, 173-177.[27] For recent developments and more high yielding processes, see: Q. Song, F. Wang, J. Cai, Y. Wang, J. Zhang, W. Yu, J. Xu, Energy Environ. Sci. 2013, 6, 994-1007.[28] R. D. Cortright, R. R. Davda, J. A. Dumesic, Nature 2002, 418, 964-967.[29] J. Zakzeski, B. M. Weckhuysen, ChemSusChem 2011, 4, 369-378.[30] J. N. Chheda, J. B. Powell, 2011, International patent 2011/082000 A1.[31] A. V. Tokarev, A. V. Kirilin, E. V. Murzina, K. Eränen, L. M. Kustov, D. Yu. Murzin, J.-P. Mikkola, Int. J. Hydrogen Energ. 2010, 35, 12642-13649.[32] J. C. del Rio, J. Rencoret, G. Marques, J. Li, G. Gellerstedt, J. Jiménez-Barbero, Á. T. Martínez, A. Gutieérrez, J. Agric. Food Chem. 2009, 57, 10271-10281.[33] T. M. Liitiä, S. L. Maunu, B. Hortling, M. Toikka, I. Kilpeläinen, J. Agric. Food Chem. 2003, 51, 2136-2143.[34] Z. Yuan, S. Cheng, M. Leitch, C. Xu, Biores. Technol. 2010, 101, 9308-9313.[35] T. Voitl, P. R. von Rohr, ChemSusChem 2011, 1, 763-769.[36] N. Yan, C. Zhao, P. J. Dyson, C. Wang, L.-t. Liu, Y. Kou, ChemSusChem 2008, 1, 626-629.[37] R. R. Davda, J. W. Shabaker, G. W. Huber, R. D. Cortright, J. A. Dumesic, Appl. Catal. B: Environ. 2005, 56, 171-186.[38] X. Xu, C. P. De Almeida, M. J. Antal Jr., Ind. Eng. Chem. Res. 1991, 30, 1478-1485.[39] S.-M. Lee, M. O. Cho, C. H. Park, Y.-C. Chung, J. H. Kim, B.-I. Sang, Y. Um, Energy Fuels 2008, 22, 3459-3464.[40] M. León, E. Díaz, S. Ordóñez, Catal. Today, 2011, 164, 436-442.[41] G. W. Huber, R. D. Cortright, J. A. Dumesic, Angew. Chem. Int. Ed. 2004, 43, 1549-1551.

Part I, Chapter 3

98

[42] R. C. Runnebaum, T. Nimmanwudipong, D. E. Block, B. C. Gates, Catal. Sci. Technol. 2012, 2, 113-118.[43] H. Ohta, H. Kobayashi, K. Hara, A. Fukuoka, Chem. Commun. 2011, 47, 12209-12211.[44] S. Heikkinen, M. M. Toikka, P. T. Karhunen, I. Kilpeläinen, J. Am. Chem. Soc. 2003, 125, 4362-4367.[45] R. J. A. Gosselink, J. E. G. Van Dam, E. de Jong, E. L. Scott, J. P. M. Sanders, J. Li, G. Gellerstedt, Holzforschung 2010, 64, 193-200.

Chapter 4

Stability of a Pt/γ-Al2O3 Catalyst in Lignin Liquid-Phase Reforming Reactions

AbstractThe physicochemical stability of a 1 wt% Pt/γ-Al2O3 catalyst was tested in an

ethanol/water mixture at 225 ˚C and autogenic pressure, conditions at which it is possible to dissolve and depolymerize various kinds of lignin. Changes in the catalyst structure were studied by means of XRD, 27Al MAS NMR, N2 physisorption, TEM, H2

chemisorption, elemental analysis, TGA-MS and IR. In the absence of reactants the alumina support is found to transform into boehmite within 4 h, leading to a reduction in support surface area, sintering of the supported Pt nanoparticles and a reduction of active metal surface area. Addition of aromatic oxygenates to mimic the compounds typically obtained by lignin depolymerization leads to a slower transformation of the support oxide. These compounds were not able to slow down the decrease in dispersion of the Pt nanoparticles, however. Vanillin and guaiacol stabilize the aluminum oxide more than phenol, anisole and benzaldehyde do, due to the larger number of oxygen functionalities that can interact with the alumina. Interestingly, catalyst samples treated in the presence of lignin showed almost no formation of boehmite, no reduction in support or active metal surface area and no Pt nanoparticle sintering. Furthermore, in the absence of lignin-derived aromatic oxygenates, ethanol forms a coke-like layer on the catalyst, while oxygenates prevent this by adsorption on the support via coordination of the oxygen functionalities.

Based on: A. L. Jongerius, J. R. Copeland, G. S. Foo, J. P. Hofmann, P. C. A. Bruijnincx, C. Sievers, B. M. Weckhuysen, “Stability of Pt/γ-Al2O3 Catalysts in Lignin and Lignin Model Compound Solutions under Liquid Phase Reforming Reaction Conditions” ACS Catal. 2013, 3, 464-473.

Part I, Chapter 4

100

4.1 Introduction

The aqueous-phase reforming (APR) of biomass-derived oxygenates provides an interesting pathway for the production of hydrogen from various renewable oxygenates. APR reactions are run in aqueous solutions and have been reported at temperatures between 210 and 260 ˚C. [1] Most commonly, renewable oxygenates, such as glycerol and other polyols or sugars, are used, [2] but recently, our group reported the first use of APR for the conversion of lignin, using a Pt/γ-Al2O3 catalyst at 225 ˚C and a pressure of 29 bar for the production of monomeric aromatic chemicals by cleavage and further conversion of some of the major linkages in lignin. The lignin APR system, however, suffers from extensive formation of solids caused by lignin recondensation and relatively low yields of isolated monomeric products are obtained as a result. [3] It is known that solvent optimization and in particular the addition of ethanol can improve APR processes. [4, 5] The liquid-phase reforming (LPR) of lignin, discussed in Chapter 3, addresses the problem of lignin condensation by using ethanol/water as solvent to aid lignin solubilization leading to much improved monomer yields. The LPR process operates under conditions resembling those of APR with the exception that higher pressures are required (58 bar), because of the higher vapor pressure of ethanol. [6]

The stability of some common solid catalysts is limited under these harsh hydrothermal conditions. It is well established that γ-alumina supports, for instance, are thermodynamically unstable under these conditions and are easily hydrated in hot water. The Sievers group has recently reported on the stability of γ-alumina under APR conditions, showing that the conversion of γ-alumina into a hydrated boehmite takes place at 200 ˚C or above resulting in a reduction of both the surface area and the number of Lewis acid sites. γ-Alumina used as support for metal nanoparticles shows a lower rate of transformation compared to the bare support, but sintering of the metal nanoparticles is observed in this case. [7] In addition, the presence of 5 wt% oxygenates, such as glycerol and sorbitol, is found to greatly enhance the stability of the catalyst material compared to treatment in pure water. The conversion to boehmite is thought to be slowed down by blockage of the surface of the support, presumably by the formation of carbonaceous deposits on the alumina surface, thus preventing hydrolysis. The sintering of the supported Pt nanoparticles is reduced in runs with these oxygenates, but a large part of the metal surface area was blocked by the carbonaceous deposits. [8] There are to the best of our knowledge no investigations on the influence of ethanol, needed for lignin solubilization, or phenolics, formed in


101

lignin conversion processes, on the stability of γ-alumina and Pt/γ-Al2O3 catalyst materials.

In this Chapter, a 1 wt% Pt/γ-Al2O3 catalyst material, which has been intensively explored in the LPR studies of Chapter 3, is treated in ethanol/water under LPR conditions at 225 ˚C and autogenic pressure. The influence of different aromatic oxygenates that can be derived from lignin on catalyst stability was studied by adding phenol, anisole, benzaldehyde, guaiacol, vanillin, as well as actual organosolv lignin to the reaction mixture. [9] Various treatment times and guaiacol concentrations are used to gain insight in the stability of the catalyst material under relevant LPR conditions. The treated catalyst samples were extensively characterized with different techniques, such as X-ray diffraction (XRD), 27Al magic angle spinning (MAS) nuclear magnetic resonance (NMR), N2 physisorption, transmission electron microscopy (TEM), H2 chemisorption, elemental analysis, thermogravimetric analysis (TGA) and infrared spectroscopy (IR), to determine the extent of hydration of the support, the available metal surface area, the metal nanoparticle dispersion and the extent of carbonaceous deposit formation.


4.2.1 Structural changes of the aluminum oxide supportThe use of γ-Al2O3 as a support oxide has found widespread use in the field of

heterogeneous catalysis. Under the hydrothermal conditions that are often used for liquid-phase biomass conversion processes, γ-Al2O3, which is prepared via dehydration of gibbsite, will be rehydrated. Indeed, it was shown before that alumina is hydrated into boehmite in hot liquid water at conditions typically used in APR reactions and that this process is slowed down by the presence of metal particles on the support. [7, 10, 11] Furthermore, the presence of oxygenates, such as glycerol or sorbitol (5 wt%; 16 and 8 mmol in 30 mL H2O, respectively), stabilizes the catalyst material by slowing down the formation of boehmite. This effect was shown to be stronger with increasing number of hydroxyl functionalities. [8] The influence of ethanol and lignin-derived aromatic oxygenates on the stability of Pt/γ-Al2O3 under conditions that are used to dissolve and convert lignin has not been investigated before. In a first set of experiments a fresh 1 wt% Pt/γ-Al2O3 catalyst material was treated at 225 ˚C in ethanol/water and ethanol/water containing 4 mmol of different lignin-derived aromatic molecules. The catalyst was treated for various durations in ethanol/water

Part I, Chapter 4

102

and ethanol/water/guaiacol. X-ray diffraction, nitrogen physisorption and 27Al NMR spectroscopy were used to observe and identify structural changes in the support.

Figure 4.1 shows the XRD patterns of the 1 wt% Pt/γ-Al2O3 catalyst after various treatment times in ethanol/water and ethanol/water with 4 mmol guaiacol. The XRD pattern of the fresh Pt/γ-Al2O3 catalyst material shows the small, broad peaks of the γ-alumina support. Diffraction peaks corresponding to a crystalline phase were observed after treatment of the catalyst and the intensity of these diffraction peaks increased with increasing treatment time. This phase was identified as boehmite with the main peaks located at 2θ = 16.8˚, 32.9˚, 44.9˚, 57.5˚ and 57.9˚ corresponding to the (020), (120), (031), (051) and (200) lattice planes, respectively (Figure 4.1a). [12] The peak intensity leveled off after 4 h for the samples treated with only ethanol/water (Figure 4.3a). The rehydration to boehmite was much slower in the presence of 4 mmol (0.5 gram) of guaiacol, as the diffraction peaks associated with boehmite were much less intense and still increasing in intensity after 15 h (Figure 4.1b and 4.3a). The influence of ethanol itself on the transformation of γ-alumina can be deduced from a comparison with a sample treated in water alone. After 4 h in water at 225 ˚C, the boehmite peaks were one third more intense than for the sample treated in ethanol/water, indicating that ethanol also slowed down the formation of (larger) boehmite crystals, but only to a limited extent.

The transformation of the support causes changes in the pore volume and surface area. The untreated Pt/γ-Al2O3 catalyst has a BET surface area of 156 m2/g, which dropped below 40 m2/g within 4 h after treatment in ethanol/water and then leveled off. In line with the changes in BET surface area, the total pore volume decreased and the average pore size increased (Table 4.1). In comparison, in the presence of guaiacol the decrease was less pronounced as after 15 h the BET surface area was 76 m2/g and

5 15 25 35 45 55 65 752 θ (˚)

5 15 25 35 45 55 65 752 θ (˚)

a. b.

Untreated Untreated1 h2 h4 h6 h

10 h

1 h2 h4 h6 h10 h15 h

Figure 4.1: X-ray diffraction patterns of a 1 wt% Pt/Al2O3 catalyst treated at 225 ˚C for various treatment times in a) ethanol/water, b) ethanol/water with 4 mmol guaiacol.


103

still decreasing (Figure 4.3c). This increased stability of the support oxide was also observed for catalyst samples treated in water with glycerol and sorbitol present, for which no significant change in alumina surface area was observed in solutions of 5 wt% of the oxygenates (16 and 8 mmol in 30 mL respectively). [8] The surface area of the Pt/γ-Al2O3 catalyst in hot water at 200 ˚C was previously reported to increase during the first 6 h of treatment after which it decreased again. [7]

The hydration of γ-Al2O3 causes the conversion of tetrahedral aluminum sites to octahedral ones, which can be quantitatively monitored with 27Al MAS NMR spectroscopy. The 27Al MAS NMR spectrum of an untreated 1 wt% Pt/γ-Al2O3

catalyst shows resonances at 66 and 7 ppm, which correspond to the tetrahedrally

Catalyst treatment BET surface area (m2/g)

Total volume of pores (cm3/g)

Untreated 156 0.46ethanol/water 1 h 121 0.40

2 h 60 0.334 h 36 0.326 h 34 0.33

10 h 38 0.344 mmol guaiacol 1 h 158 0.41

2 h 139 0.404 h 110 0.366 h 97 0.36

10 h 88 0.3815 h 76 0.38

0.8 mmol guaiacol 4 h 71 0.388 mmol guaiacol 4 h 127 0.384 mmol phenol 4 h 57 0.364 mmol anisole 4 h 68 0.374 mmol benzaldehyde 4 h 67 0.374 mmol vanillin 4 h 139 0.410.5 g lignin 4 h 157 0.26

Table 4.1: N2 physisorption data of the 1 wt% Pt/γ-Al2O3 catalyst samples treated in an ethanol/water mixture with and without aromatic lignin model compounds.

Part I, Chapter 4

104

100 80 60 40 20 0 -20 --40 -60

untreated

1h

2 h

4 h

6 h

10 h

Chemical shift (ppm) 100 80 60 40 20 0 -20 --40 -60

untreated

1 h

2 h

4 h

6 h

10 h

15 h

Chemical shift (ppm )

a. b.

Figure 4.2: 27Al MAS NMR spectra of a 1 wt% Pt/γ-Al2O3 catalyst treated at 225 ˚C for various durations in a) ethanol/water, b) ethanol/water with 4 mmol guaiacol.

and octahedrally coordinated aluminum species, respectively (Figure 4.2). [13, 14] Linear combination of the 27Al MAS NMR spectra of γ-Al2O3 and boehmite allows the fraction of fully hydrated Al to be calculated. The untreated sample contained 26% tetrahedrally coordinated aluminum, comparable with previously reported results, whereas in pure boehmite all aluminum species are octahedrally coordinated. [15] The 27Al MAS NMR spectra show that γ-Al2O3 is gradually converted into boehmite when treated in ethanol/water with and without guaiacol (Figure 4.2a and 4.2b). This was indicated by the continuous decrease in intensity of the peak corresponding to tetrahedrally coordinated Al with time. In the presence of 4 mmol guaiacol, the rate of the hydration is slower compared to treatment without guaiacol (Figure 4.3e). However, complete conversion was reached after 6 h in both cases. This was much faster than the phase transformation of 1 wt% Pt/γ-Al2O3 treated in water at 200 ˚C, in which case a boehmite fraction of 0.6 was reported only after 10 h. [7]

All techniques show changes in the structure of the support that proceed over time. For samples that were treated in ethanol/water without guaiacol these changes appear to be complete after 4 h, while for the guaiacol-treated samples the changes seem to take place more slowly and continue for a longer time. The XRD and physisorption data show that for the guaiacol-treated samples even after 15 h recrystallization still proceeds, while the 27Al MAS NMR data show the conversion to octahedral aluminum to be complete after 10 h. This difference between XRD and NMR data was also observed before. [7] The initial hydration of the tetrahedral aluminum species to octahedral aluminum species is detected by NMR, but does not immediately result in the formation of a boehmite crystal lattice of sufficiently long range order


105

to be detected by XRD. After complete transformation to octahedral aluminum, the first small domains of boehmite crystals will still grow and as a result the physical properties of the support material continue to change.

Comparison of the samples treated for 4 h in the presence of different concentrations

0 2 4 6 8 10

Sign

al in

tens

ity (A

.U.)

Time (h)

BET

surf

ace

area

(m²/

g)

10.3 pt

Time (h)

0

50

100

150

0 2 4 6 108

BET

surf

ace

area

(m²/

g)

0

40

80

120

160

Exte

nt o

f hyd

ratio

n

0

0.2

0.4

0.6

0.8

1

a. b.

c. d.

f.

Sign

al in

tens

ity (A

.U.)

0

0.2

0.4

0.6

0.8

1

0 2 4 6 8 10

Exte

nt o

f hyd

ratio

n

Time (h)

e.

Figure 4.3: X-ray diffraction peak intensity at 2θ = 16.8˚ (a+b), BET surface area (c+d) and boehmite formation based on 27Al MAS NMR (e+f) of a 1 wt% Pt/γ-Al2O3 catalyst after treatment with and without guaiacol at 225 ˚C. (a,c,e ) Samples treated in ethanol/water and (a,c,e ) samples treated in ethanol/water with 4 mmol guaiacol. (b,d,f) Catalyst treated in ethanol/water at 225 ˚C with 4 mmol of various aromatic oxygenates for 4 h.

Part I, Chapter 4

106

of guaiacol shows that higher concentrations slowed down the loss of BET surface area and formation of a crystalline boehmite phase more (Figure 4.3b and 4.3d). Furthermore, the XRD patterns and BET surface area of samples treated with 4 mmol of different aromatic lignin model compounds indicate that the addition of guaiacol and vanillin, with two and three oxygen functionalities, respectively, resulted in less intense boehmite peaks and smaller loss of surface area than mono-oxygenated, phenol, anisole and benzaldehyde did. Only small differences are observed between the methoxy, phenolic and aldehyde functional groups (Figure 4.3b and 4.3d), indicating that the type of oxygen functionality does not influence the stabilization of the support much. This comparable reactivity suggests that the different molecules eventually form the same surface oxygen species. 27Al MAS NMR shows that the influence of the oxygenates on the rate of hydration of the alumina support is relatively small. The differences in extent of hydration observed are small compared to the variation observed in XRD peak intensity (Figure 4.3f). For phenol, anisole, benzaldehyde and 0.8 mmol of guaiacol, conversion was complete after 4 h. Only vanillin and 8 mmol guaiacol treated samples still have tetrahedrally coordinated aluminum left after 4 h with conversions of 0.90 and 0.92, respectively.

The increased stability of the support in the presence of guaiacol and other aromatic oxygenates indicates that these molecules prevent hydrolysis of the alumina. Most likely the oxygen functionalities coordinate to unsaturated aluminum sites on the support oxide thereby preventing water molecules to access these sites. Previously, the polyols glycerol and sorbitol were also shown to inhibit the transformation from γ-alumina into boehmite. Boehmite fractions of only 15% and 2% for 5 wt% glycerol and sorbitol were calculated from 27Al MAS NMR after 10 h reaction, which is less than for the samples treated in 4 mmol solutions of guaiacol after 4 h. [8] The increasingly strong inhibitory effect of polyols with growing carbon chain length is attributed to the increasing number of hydroxyl groups present in longer polyols, allowing stronger adsorption via the formation of multidentate surface species. [16] Moreover, a spatial separation between the different functional groups allows for the formation of more stable surface species. [17] Here a similar correlation is observed with the aromatic oxygenates, as mono-oxygenated species show only limited stabilization of the support compared to compounds that are able to form bidentate or multidentate surface species.

Remarkably, when the stability of the catalyst was tested in the presence of 0.5 g of organosolv lignin, no significant change in the diffraction pattern and surface area


107

of the original catalyst material was observed. The total pore volume and the average pore diameter both were reduced by half. The fraction of hydrated aluminum as observed by 27Al NMR was only 0.15. This together with the XRD results indicates that the hydration only took place to a very small extent and that the transformation into boehmite was completely blocked. The inherently multidentate mode of adsorption of the highly functionalized and relatively hydrophobic lignin macromolecule leads to a strong interaction with the support and ultimately prevents water from hydrating the surface.

4.2.2 Reactivity of ethanol and aromatic oxygenates on the alumina support

As was suggested before, the adsorption of oxygen-containing molecules on the alumina support, both by coordination of the molecule to specific sites on the alumina and by the formation of coke, may block the access of water molecules to these positions, thereby preventing hydration of the support. [8] The quantity and nature of these organic deposits will depend on the oxygenate(s) they are derived from and can be determined using techniques, such as thermogravimetric analysis (TGA) and IR spectroscopy, respectively.

Upon heating to 900 ̊ C, the untreated Pt/γ-Al2O3 catalyst material showed a gradual weight loss of up to 3%. All ethanol/water-treated samples as well as the samples treated in the presence of guaiacol and other model compounds showed a total weight decrease between 14 and 17% with a sharp decrease around 450 ˚C. The total loss of weight was similar for all samples regardless of the treatment time and guaiacol concentration with exception of the lignin-treated sample which showed a total weight loss of 22% over a temperature range between 250 ˚C and 500 ˚C. Weight loss started at lower temperatures for samples treated with higher guaiacol concentrations as well as samples treated with vanillin, resulting in a more gradual decrease in weight over a longer temperature range. In comparison, samples treated with only ethanol/water

150 250 350 450 550 650

Wei

ght l

oss (

%/˚

C)

Temperature (˚C)

lignin

ethanol/water

guaiacol

untreatedFigure 4.4: Thermogravimetric analysis of a 1 wt% Pt/γ-Al2O3 catalyst after treatment at 225 ̊ C with ethanol/water, 0.5 g (4 mmol) guaiacol and 0.5 g lignin.

Part I, Chapter 4

108

always lost most of their weight in a narrow temperature range around 500 ˚C (Figure 4.4). These results are comparable to those previously observed with polyols. [8] The nature of these deposits, which given the multiple peaks in TGA, were shown to be heterogeneous in nature in the case of the polyols, seems to be more homogeneous in the case of the guaiacol-treated sample.

Mass spectrometric analysis of the gases released during heating showed the formation of CO2 with maxima between 225 ˚C and 450 ˚C and an evolution of water resulting from dehydration of the boehmite around 500 ˚C (Figure 4.5a and b). As the samples were dried at 150 ˚C for 2 h before the TGA measurements, any remaining volatile aromatic molecules would have already been desorbed during this preheating stage and should not contribute to any detected weight loss in the TGA measurements. Samples treated in ethanol/water without model compound showed additional CO2 release at 520 ̊ C. In the presence of lignin and lignin model compounds, these deposits, which apparently do not contribute to the stabilization of alumina towards boehmite formation, are not formed. This release of CO2 at this relatively high temperature points to the formation of more stable carbonaceous deposits (Figure 4.5b). [18] This has also been observed after steam reforming reactions of ethanol over alumina-supported catalysts. [19] Samples that were shown to contain a higher amount of boehmite by XRD also released more water during TGA analysis and samples treated with guaiacol and vanillin released water at a lower temperature (485 ˚C). As the samples treated with lignin model compound solutions released most of the CO2 at temperatures lower

A.U

.

150 250 350 450 550 650

A.U

.

Temperature (˚C)

A.U

.

a.

b.

lignin

guaiacol ethanol/water

lignin

guaiacol

ethanol/water

Figure 4.5: a) H2O and b) CO2 released during thermogravimetric analysis of a 1 wt% Pt/γ-Al2O3 catalyst after treatment at 225 ̊ C with ethanol/water, 0.5 g (4 mmol) guaiacol and 0.5 g lignin.


109

than 410 ˚C and most water at a higher temperature, it is possible to compare the amount of weight loss due to CO2 and water release between the different samples (Table 4.2). With the exception of the sample treated with lignin, the majority of the weight loss was caused by release of water at temperatures above 410 ˚C. Increasing the amount of a model compound in solution also resulted in an increased amount of CO2 release during TGA (3, 4 and 6 wt% for the samples treated with 0.8, 4 and 8 mmol guaiacol, respectively). Samples treated in ethanol/water with phenol, anisole and benzaldehyde released only a small amount of CO2, which indicates that compared to guaiacol less of these molecules were strongly adsorbed on the support at the same molar concentration. This again points at the relatively weak monodentate adsorption of these molecules.

Longer treatment times did not lead to an increase in CO2 formation, but for both the ethanol/water and ethanol/water/guaiacol samples longer reaction times did lead to more release of water. The continued uptake of water is consistent with the growth of the boehmite phase over time as is corroborated by the XRD and physisorption data.

IR spectra of the treated samples were recorded to determine the nature of the carbonaceous deposits on the catalyst material. The original Pt/γ-Al2O3 catalyst shows a broad band in the region 3700-3000 cm−1 corresponding to the O-H stretching vibrations of surface hydroxyl groups and small peaks at 1537, 1458 and 1259 cm−1

Catalyst treatment Weight loss from CO2 (%) Weight loss from H2O (%)

untreated 1 1water 1 13ethanol/water 2 140.8 mmol guaiacol 3 124 mmol guaiacol 4 98 mmol guaiacol 6 8phenol 3 13anisole 2 12benzaldehyde 3 11vanillin 5 8lignin 17 4

Table 4.2: Weight loss, as determined by thermogravimetric analysis, attributed to the formation of CO2 in the temperature range 150-410 ˚C and the formation of water in the temperature range 410-650 ̊ C for the 1 wt% Pt/γ-Al2O3 catalyst samples treated under various conditions.

Part I, Chapter 4

110

originating from carbonate-like species. [20] After treatment in ethanol/water the spectrum showed an increased absorption in the O-H stretch area due to the formation of a hydrated boehmite phase. The spectrum of the sample diluted with KBr shows two major bands at 3299 and 3085 cm−1, which are characteristic for the OH stretching bands of boehmite. [21] New, broad bands appeared at 2096 and 1973 cm−1 as a result of both combination bands and overtones of Al-O resonances and a sharp peak at 1071 cm−1 is assigned to OH bending vibrations of boehmite (Figure 4.6). [21] In-situ diffuse reflectance (DRIFT) IR analysis of the guaiacol-treated sample in a nitrogen flow with D2O showed a decrease in the OH bands at 3670, 3325 and 1072 cm−1 and formation of new bands at 2707 and 2390 cm−1. No changes were observed in the vibrations around 2000 cm−1 confirming that these do not belong to OH vibrations (Figure 4.8). [22] No peaks characteristic for ethanol are observed, which could have been expected since ethanol is present in large quantities. Any ethanol deposits could be masked though by the strong boehmite bands in the CH stretch and CO stretching regions. Ethanol is known to form acetates on alumina above 200 ˚C, but no clear OCO stretching modes at 1585 cm−1 for organic acetates are found. [23]

The limited amount of hard coke formed in ethanol/water-treated samples that was observed in the TGA measurements is not visible in the IR spectrum. The vibrations associated with the newly formed boehmite phase can also be seen in spectra of the catalyst treated in guaiacol and phenol solutions, now with additional new peaks at 1579, 1464, 1424 and 1328 cm−1 (Figure 4.7), which can be assigned to aromatic ring vibrations (1579 and 1464 cm−1), phenolic OH or CH3 vibrations and a phenolic CO stretch, respectively. [22] This indicates that the molecules are adsorbed with intact aromatic rings, most likely via the oxygen groups. The absence of a guaiacolic OH group in the 1366 cm−1 region points towards the formation of a guaiacolate species that is directly bound to the alumina. Also, no peaks corresponding to the CH stretching modes of a methyl group were observed in the 2900 cm−1 region, indicating that demethylation has taken place for the most strongly adsorbed guaiacol-derived species, which remain after isolation of the spent catalyst to give bidentately bound catecholate species. While it has been shown for vicinal diols such as 1,2-ethyleneglycol or 1,2-propanediol that such bidentate coordination modes are not very favorable due to geometric constraints, [17] they have indeed been reported before for guaiacol on alumina. It was shown that the methoxy group of guaiacol can react with alumina at temperatures between 150 ˚C and 200 ˚C resulting in the formation of guaiacol molecules strongly anchored on the support as bidentate catecholates. [22]

Although it is likely that, under the relatively mild conditions applied here, many


111

400140024003400Wavenumber (cm-1)

a.

b.

32993085

2096 1973

3459

1071

1162

Figure 4.6: IR spectra of a 1 wt% Pt/γ-Al2O3 catalyst pressed into a KBr wafer a) before, and b) after treatment at 225 ˚C in ethanol/water with 4 mmol guaiacol.

100015002000250030003500Wavenumber (cm )-1

36702707

3325

2390

20702000

1072

Untreated sample

D2O treated sample

Dried over night

Figure 4.8: IR spectra in diffuse reflective mode of a 1 wt% Pt/γ-Al2O3 catalyst treated in ethanol/water at 225 ˚C for 4 h with 4 mmol guaiacol mixed with KBr, dried sample, sample treated in a gas flow with D2O and the D2O-treated sample after drying overnight.

130014001500160017001800Wavenumber (cm-1)

a.b.c.

1579 1464 13281424

d.

Figure 4.7: IR spectra of a 1 wt% Pt/γ-Al2O3 catalyst a) after treatment at 225 ˚C with 0.5 g lignin, b: after treatment at 225 ˚C with 4 mmol guaiacol, c) fresh catalyst.

Part I, Chapter 4

112

guaiacol molecules interact weakly with the support and coordinate to the alumina mostly with both the OH and the methoxy group still intact. Such species would be washed off the catalyst rather easily at the end, leaving only the more firmly anchored molecules for which deprotonation or deprotonation/demethylation has taken place. It should be noted in this respect that previous results under conditions where pure guaiacol is introduced to alumina show much higher surface coverages than were observed here for the ethanol/water/guaiacol-treated samples. This clearly indicates that guaiacol adsorption is reversible and that the adsorption of water and ethanol is still highly competitive with the adsorption of the aromatic oxygenates. Therefore, the stabilization of the support by these molecules will be limited and, as was shown in the previous section, will only slow down and not stop the transformation to boehmite. [22]

Large amounts of CO2 (17 wt% of the total weight of the original sample) were released during TGA analysis from the sample treated with lignin, consistent with the larger weight loss observed for this sample. Possibly, stronger adsorption due to multidentate interactions of the molecule increases the stability of lignin on the alumina surface and prevents it from washed away after the treatment. Formation of water occurred in the same temperature region as the CO2 release, indicating that the lignin adsorbed on the support still has a high oxygen content. This is also seen in the IR spectra where a broad OH stretch band with a maximum at 3530 cm−1 was observed for the catalyst treated in the lignin solution which can be attributed to the alcohol functionalities in the lignin macromolecule. A vibration corresponding to aliphatic CH stretching modes at 2935 cm−1 and bands at 1713 cm−1 and 1506 cm−1 corresponding to C=O stretch and aromatic ring vibrations were also observed, in addition to phenolic vibrations at 1579, 1464, 1424 and 1328 cm−1. Peaks corresponding to the hydrated boehmite phase were not observed in the OH stretch region, nor were the broad bands at 2085 and 1963 cm−1 or the OH bending vibration at 1072 cm−1.

A comparison of the spectrum of pure lignin with the lignin adsorbed on the catalyst (Figure 4.9) shows that some major peaks were still present after adsorption, although large changes in peak ratios and small shifts did occur in the fingerprint region. The major differences are a relative decrease of the C=O stretching vibrations at 1704 cm−1, the symmetric aryl stretching vibrations at 1607 cm−1 and the aromatic C-H in plane deformations at 1120 cm−1 and an increase of the asymmetric aryl stretching vibrations at 1516 cm−1. [24] Remarkably, a pure lignin sample loses weight at higher temperatures during TGA than the lignin adsorbed on the alumina (Figure 4.10),


113

indicating that the support is able to lower the lignin decomposition temperature.

4.2.3 Stability of the supported metal nanoparticlesThe transformation of the γ-alumina support into boehmite affects the stability

of the supported metal nanoparticles on the oxide surface. The mobility of the metal nanoparticles increases when the support surface changes, which in turn leads to sintering or encapsulation of the metal nanoparticles. In addition, added organics can lead to coke formation resulting in blockage of the particles. In both cases the accessibility of Pt during the reaction is drastically reduced. It is important to gain more insight in the stability of the supported metal nanoparticles as a decrease in dispersion will lead to loss of catalyst activity.

The platinum loading of the Pt/γ-Al2O3 catalyst and all samples treated in ethanol/

1000150020002500300035004000Wavenumber (cm-1)

a.

b.

1704

1607

1516 1462 1116

29352849

3410

3530

150 250 350 450 550 650

Wei

ght l

oss (

%/˚

C)

Temperature (˚C)

Organosolv lignin

Lignin treated sample

Figure 4.9: a) IR spectrum of a lignin KBr pellet, b) IR spectrum of a self-supporting wafer of a 1 wt% Pt/γ-Al2O3 catalyst treated in a lignin solution.

Figure 4.10:Thermogravimetric analysis of a 1 wt% Pt/γ-Al2O3 catalyst treated in ethanol/water at 225 ˚C for 4 h with 0.5 gram lignin compared to the TGA analysis of pure lignin.

Part I, Chapter 4

114

water with and without model compounds were determined by ICP-OES (Inductively Coupled Plasma – Optical Emission Spectroscopy). The untreated catalyst was found to have a platinum content of 0.95 wt%, whereas loadings between 0.80% and 0.85% were found for the ethanol/water treated samples with and without guaiacol and other model compounds. The sample treated in an ethanol/water mixture with lignin contained 0.79% Pt. Taking into account the uptake of water and deposition of carbonaceous deposits during the run, shown to be between 14 and 17% by TGA, it can be concluded that no significant amount of platinum was leached into solution during the treatment.

The TEM micrograph of the untreated catalyst shows well-dispersed Pt nanoparticles on the amorphous γ-Al2O3 support (Figure 4.11a). TEM micrographs of the catalyst treated for 4 h in ethanol/water and ethanol/water containing benzaldehyde show that the support changed to a uniform density with sharp edges, indicative of crystallinity. The Pt particles on the fresh catalyst have an average size of 2.0 nm with 16.0% of the particles having a diameter of less than 1 nm. Most particles (i.e., 48%) have a diameter between 2.0 and 3.0 nm. The average Pt particle size increased when the catalyst was treated for 4 h in mixtures of ethanol/water and ethanol/water with phenol, anisole, benzaldehyde, 0.8 mmol guaiacol, 8 mmol guaiacol, and vanillin (Figure 4.12). The degree of metal sintering is again not influenced much by the particular monoaromatic compound used, even though it did influence the relative change in surface area and boehmite formation. This increase in average particle size was also observed for a Pt/Al2O3 catalyst hydrothermally treated in aqueous glycerol and sorbitol solutions. However, for these hydrothermal treatments, sintering was found to be less severe when the polyols were present. [8] In contrast, with the aromatic oxygenates, larger increases in average particle size are seen compared to ethanol/water alone. The Pt particles sintered significantly during these treatments (Figure 4.11b and c). The untreated catalyst, and the catalyst treated with guaiacol (4 mmol) and lignin were the only samples that contained Pt particles with diameters less than 1 nm. In hydrogen chemisorption measurements, the untreated Pt/γ-Al2O3 catalyst showed a hydrogen uptake of 16 µmol/g, corresponding to an average particle diameter of 1.7 nm, which is consistent with the TEM measurements (average particle size of 2 nm). The hydrogen uptakes of samples treated in ethanol/water with and without guaiacol were 3 and 4 µmol/g, which are, however, much lower than expected based on the TEM results. After calcination at 400 ˚C the guaiacol treated sample still showed a hydrogen uptake of 4 µmol/g. According to TGA measurements, a treatment at this temperature should have removed any carbonaceous deposits that


115

cover the Pt particles. This indicates that some metal nanoparticles got encapsulated in the boehmite crystals and are not accessible anymore for catalysis. The XRD and physisorption data show that samples treated in ethanol/water without any aromatic model compound show a faster transformation of the Pt/γ-Al2O3 into boehmite. It can be expected that this faster change of the support also leads to faster encapsulation of the supported Pt nanoparticles, thereby decreasing the time they are prone to sintering. With the exception of the lignin-treated sample, all samples that showed a slow conversion to boehmite in XRD therefore show a larger average particle size by TEM.

Notably, the sample treated in lignin solutions exhibited only marginal changes in average Pt particle size and particle size distribution. The TEM micrograph of the catalyst treated for 4 h with lignin furthermore shows that the support retains an amorphous appearance (Figure 4.11d), similar to that of the untreated catalyst (Figure 4.11a) and in line with the XRD data. With 6 µmol/g the hydrogen uptake of the

25 nm

a.

50 nm

b.

50 nm

c.

50 nm

d.

Figure 4.11: a) TEM micrographs of an untreated 1 wt% Pt/γ-Al2O3 catalyst, b) treated at 225 ˚C in ethanol/water for 4 h, c) treated at 225 ˚C in a benzaldehyde solution and d) treated at 225 ˚C in a lignin solution for 4 h.

Part I, Chapter 4

116

lignin-treated sample was also lower than for the untreated catalyst. As a result, the average particle size of 4.0 nm calculated from the hydrogen chemisorption results was higher than the particle size of 1.9 nm obtained by TEM. These results indicate that the Pt surface on the lignin-treated sample is less accessible for molecular hydrogen as it is partially covered by the support or, more likely, a layer of adsorbed lignin. Formation of a layer of lignin can explain the increased stability of the catalyst in the presence of lignin, both by occupying the positions that would otherwise be hydrated but probably also by increasing the hydrophobicity of the surface. Glycerol and sorbitol were also shown to prevent part of the loss of active sites, but a 60% drop in accessible metal surface area was still observed, which was explained by the formation of coke. [8]

0%

20%

40%

60%

80%

100%

Perc

enta

ge o

f partic

les

4-5 nm

3-4 nm

2-3 nm

1-2 nm

0-1 nm

b.

Parti

cle

size

(nm

)

a.

0

1

2

3

4

5

6

Figure 4.12: a) Average diameter of the supported Pt nanoparticles, b) particle size distributions of an untreated 1 wt% Pt/γ-Al2O3 catalyst and of a 1 wt% Pt/γ-Al2O3 catalyst after 4 h treatment at 225 ˚C in solutions of various oxygenates.


117

4.3 Conclusions

Under LPR conditions in ethanol/water mixtures at 225 ˚C, the γ-alumina support of a 1 wt% Pt/γ-Al2O3 catalyst material is converted into boehmite, which causes the formation of a crystalline phase and loss of surface area. These severe changes in the support material also cause sintering and encapsulation of the Pt nanoparticles leading to loss of active surface area and catalytic activity. This transformation is slowed down in the presence of lignin-derived molecules, such as guaiacol, which are believed to adsorb on the alumina support via its oxygen functionalities. The ability of each of these molecules to stabilize the support is proportional to the number of oxygen functionalities it contains. Higher concentrations of guaiacol increase the stability of the 1 wt% Pt/γ-Al2O3 catalyst under LPR reaction conditions. Lignin itself shows an increased affinity for the alumina support resulting in adsorption of a large amount of lignin, the multidentate interactions between lignin and the support oxide in that way completely prevent the formation of a boehmite phase and sintering of the supported Pt nanoparticles, while retaining the metal surface area. The Pt/Al2O3 catalyst material is not stable in lignin conversion processes at the conditions required to dissolve lignin, however, and high lignin concentrations could be used to partially overcome this problem. It is important to note that other supports (e.g., titania, zirconia and carbon) might be more stable, but detailed studies need to be performed to assess the potential transformations of these materials under LPR reaction conditions.

4.4 Experimental section

4.4.1 Chemicals and catalyst materialsThe catalyst material, all model compounds and solvents were obtained

commercially and used as received: 1 wt% Pt/γ-Al2O3 catalyst (Sigma Aldrich, same as in Chapter 3), ethanol (Fisher, HPLC grade), phenol (Acros, 99%), anisole (Fluka, 99%), benzaldehyde (Fluka), guaiacol (Sigma), vanillin (Sigma Aldrich, 99%). Deionized water was purified in a MilliQ system. The Alcell organosolv lignin (66.47% C, 5.96% H, 0.15% N, 27.43% O by difference), provided by Wageningen University, was obtained from hardwoods and isolated by an organosolv extraction method. [25]

Part I, Chapter 4

118

4.4.2 Catalyst stability testsThe catalyst material stability tests were performed in a 40 mL Parr batch autoclave

equipped with a thermocouple, a pressure transducer and gauge and a magnetic driver (750 rpm). 1 g of 1 wt% Pt/γ-Al2O3 was suspended in 15 mL of deionized water and 15 mL of ethanol and, if applicable, 4 mmol of the model compound or 0.5 g lignin was added after which the mixture was heated to 225 ˚C. After the designated time, the autoclave was cooled in an ice bath and vented. The catalyst material was recovered by filtration and washed with water and ethanol.

4.4.3 AnalysisX-ray powder diffraction patterns were recorded on a Bruker-AXS D2 Phaser

powder X-ray diffractometer using CoKα1,2 with λ = 1.79026 Å. Measurements were carried out between 5 – 75° 2θ using a step size of 0.04° 2θ and a scan speed of 1 s. N2

physisorption isotherms were recorded with a Micromeritics Tristar 3000 at −196 ˚C. The samples were dried prior to performing the measurement for at least 16 h at 200 ˚C in a N2 flow. The surface area was determined using the Brunauer-Emmett-Teller (BET) theory. The total pore volume was defined as the single-point pore volume at p/p0 = 0.95.

27Al MAS NMR experiments were performed using a Bruker DSX 400 Spectrometer. The catalyst samples were packed into a 4 mm zirconia rotor and spun at a frequency of 12 kHz. The resonance frequency of 27Al was 104.2 MHz. A π/12 pulse was used for excitation with a recycling delay of 250 ms. Aqueous Al(NO3)3 was used as reference compound (δ = 0 ppm). The normalized 27Al MAS NMR spectra were fitted as a linear combination of the spectra of pure boehmite and γ-alumina to calculate the fraction converted to hydrated aluminum. [7,8]

Thermogravimetric analysis was performed with a Perkin–Elmer Pyris 1 apparatus. Typically, 15 mg of catalyst sample was dried at 150 ˚C for 2 h and heated with a ramp of 5 ˚C min−1 to 900 ˚C in a 10 mL min−1 flow of air. In parallel, evolved gas analysis was performed with a Pfeiffer Omnistar quadrupole mass spectrometer, which was connected to the outlet of the TGA apparatus. Ion currents were recorded for m/z values of 18 (H2O) and 44 (CO2).

For the IR studies the catalyst sample was finely grained and pressed to form a self-supporting wafer with a weight of 10-15 mg and a diameter of 13 mm. The wafer was placed in an IR transmission cell with CaF2 windows. The cell was evacuated to 10−6


119

bar and kept at 200 ˚C for at least 1 h after which it was cooled to 150 ˚C. IR spectra were obtained using a Perkin-Elmer 2000 FT-IR instrument with an optical resolution of 4 cm−1 and an accumulation of 25 scans from 4000 to 1000 cm−1. Additional IR measurements were carried out on catalyst samples dilluted with KBr to reduce signal adsorption. IR spectra of the KBr wafers were obtained without drying at RT in air using a Perkin-Elmer 2000 FT-IR instrument with an optical resolution of 4 cm−1

and an accumulation of 10 scans from 4000 to 400 cm−1. In situ DRIFTS measurements were performed with a Bruker Tensor 27 apparatus utilizing an HVC-DRP-3 diffuse reflectance reaction chamber with CaF2 windows and an MCT detector. The bottom of the sample cup was filled with silicon carbide and covered by a grid to minimize temperature gradients, on which the sample mixed with KBr was placed. Nitrogen gas flowed through the sample from top to bottom at a flow rate of 10 mL min−1, while the temperature was increased and kept at 200 ˚C for at least 1 h after which it was cooled to 50 ̊ C. The sample was exposed to nitrogen flow that was led through D2O for 30 min after which the sample was dried again in nitrogen over night. IR scans were recorded from 4000 to 1000 cm−1 at a resolution of 4 cm−1.

TEM images were recorded with a JEOL 100CX microscope at a 100 kV acceleration voltage. The samples were prepared by applying three drops of a catalyst in ethanol slurry on a graphene-coated, 200 mesh copper grid. The slurry was homogenized via a sonication bath prior to applying to the sample grid.

The Pt content was determined using a Spectro Arcos ICP-OES (Inductively Coupled Plasma – Optical Emission Spectroscopy) with a standard operating procedure. Samples were prepared by dissolving 125 mg of the catalyst sample in 6 mL aqua regia at 90 ˚C overnight. After evaporation of the aqua regia at 160 ˚C the samples were dissolved in 20 mL HCl at 90 ˚C.

H2 chemisorption measurements were performed using a Micromeritics ASAP 2020. Samples were dried at 100 ˚C for 60 min (ramp 10 ˚C min−1) in vacuum. Subsequently, the samples were reduced at 200 ˚C (ramp 5 ˚C min−1) for 2 h under H2 flow. Hereafter, the samples were degassed for 30 min under vacuum at 200 ˚C. Adsorption isotherms were subsequently measured at 50 ˚C. Average particle sizes were calculated as described by Scholten et al. [26] using the linear part of the adsorption isotherm.

Part I, Chapter 4

120

4.5 Acknowledgments

John Copeland, Gui Siu-Foo and Carsten Sievers from the Georgia institute of Technology are acknowledged for performing the TEM and 27Al MAS NMR analysis and their contributions to the chapter. We would like to thank Richard Gosselink (Wageningen University) for supplying the lignin and Helen de Waard and Ton Zalm (Faculty of Geosciences, Utrecht University) for the ICP-OES analyses.

4.6 References

[1] R. D. Cortright, R. R. Davda, J. A. Dumesic, Nature 2002, 418, 964-967.[2] A. Tanksale, J. N. Beltramini, G. M. Lu, Renew. Sust. Energy Rev. 2010, 14, 166-182.[3] J. Zakzeski, B. M. Weckhuysen, ChemSusChem 2011, 4, 369-378.[4] J. N. Chheda, J. B. Powell, 2011, International patent 2011/082000A1.[5] A. V. Tokarev, A. V. Kirilin, E. V. Murzina, K. Ernen, L. M. Kustov, D. Yu. Murzin, J.-P. Mikkola, Int. J. Hydrogen Energ. 2010, 35, 12642-13649.[6] J. Zakzeski, A. L. Jongerius, P. C. A. Bruijnincx, B. M. Weckhuysen, ChemSusChem 2012, 5, 1602-1609.[7] R. M. Ravenelle, J. R. Copeland, W. G. Kim, J. C. Crittenden, C. Sievers, ACS Catal. 2011, 1, 552-561.[8] R. M. Ravenelle, J. R. Copeland, A. H. Van Pelt, J. C. Crittenden, C. Sievers, Top. Catal. 2012, 55, 162-174.[9] E. K. Pye, J. H. Lora, Tappi J. 1991, 74, 113-115.[10] R. M. Ravenelle, F. Z. Diallo, J. C. Crittenden, C. Sievers, ChemCatChem 2012, 4, 492- 494.[11] W. C. Ketchie, E. P. Maris, R. J. Davis, Chem. Mat. 2007, 19, 3406-3411.[12] G. G. Christoph, C. E. Corbató, D. A. Hofmann, R. T. Tettenhorst, Clays Clay Miner. 1979, 27, 81-86.[13] C. Pecharroman, I. Sobrados, J. E. Iglesias, T. Gonzalez-Carreno, J. Sanz, J. Phys. Chem. B 1999, 103, 6160-6170.[14] G. Urretavizcaya, A. L. Cavalieri, J. M. P. Lopez, I. Sobrados, J. Sanz, J. Mater. Synth. Proces. 1998, 6, 1-7.[15] R. S. Zhou, R. L. Snyder, Acta Crystallogr. B 1991, 47, 617-630.[16] W. van Bronswijk, H. R. Watling, Z. Yu, Colloid Surf. A 1999, 157, 85-94.[17] J. R. Copeland, X. -R. Shi, D. S. Sholl, C. Sievers, Langmuir 2012, 29, 581-593.


121

[18] M. Henry, M. Bulut, W. Vermandel, B. Sels, P. Jacobs, D. Minoux, N. Nesterenko, S. Van Donk, J. P. Dath, Appl. Catal. A: Gen. 2012, 413, 62-77.[19] M. C. Sánchez-Sánchez, R. M. Navarro, J. L. G. Fierro, Int. J. Hydrogen Energ. 2007, 32, 1462-1471.[20] C. Morterra, G. Magnacca, Catal. Today 1996, 27, 497-532[21] Y. I. Ryskin, The Infrared Spectroscopy of Minerals, Mineralogic. Soc. Monograph 4, 1974.[22] A. Popov, E. Kondratieva, J. M. Gouil, L. Mariey, P. Bazin, J. P. Gilson, A. Travert, F. Maugé, J. Phys. Chem. C 2010, 114, 15661-15670.[23] H. E. Evans, W. H. Weinberg, J. Chem. Phys. 1979, 71, 1537-1542.[24] C. Heitner, D. Dimmel, J. A. Schmidt, Lignin and Lignans: Advances in Chemistry, CRC Press, 2010, chapter 4.[25] T. N. Kleinert, 1971, US patent 3585104.[26] J. J. F. Scholten, A. P. Pijpers, M. L. Hustings, Catal. Rev. Sci. Eng. 1985, 27, 151–206.

Chapter 5

Lignin Depolymerization by Alkaline Hydrogen Peroxide Treatment

AbstractAn oxidative lignin depolymerization reaction, inspired by cellulose pulp bleaching

processes, is presented. The depolymerization of organosolv lignin (2 wt%) is studied in 0.5 M NaOH with 0.5 mL H2O2 at room temperature leading to a decrease in lignin molecular weight of 25%. Higher peroxide concentrations and higher temperatures resulted in higher degrees of depolymerization (up to 55%), while no influence of base concentration on the extent of depolymerization was observed. Alkaline oxidation of lignins obtained with different pretreatment methods showed that the efficiency of the depolymerization depends on the type of lignin and the impurities that are present. Pretreatment of the organosolv lignin by solubilization in ethanol/water led to an increase in free phenolic OH groups, as evidenced by 31P NMR, after which the efficiency of the oxidation increased and a 32% decrease in molecular weight was obtained. Cobalt-catalyzed oxidation of lignin, aimed at increasing the concentration of aryl-α-ketone functionalities, which are thought to be required for the Dakin-like mechanism of depolymerization, actually led to a 50% increase of molecular weight. A combination of solubilization, Co-catalyzed oxidation and peroxide oxidation did not lead to an improved depolymerization of lignin. IR and 31P NMR analysis revealed that depolymerization by alkaline oxidation most likely does take place via a Dakin-like mechanism and leads to the introduction of carboxylic acid groups.

Part I, Chapter 5

124

5.1 Introduction

In the pulp and paper industry a method called bleaching is used to remove any lignin that is left on the cellulose fibers after pulping to ultimately obtain white paper. Chemical pulps from kraft or sulfite processes usually contain about of 1.5-4.5% residual lignin in the fiber whereas in mechanical pulps, this can be up to 30%. Together with other impurities, the residual lignin causes undesired coloration of the cellulose fibers. [1] Residual lignin highly resembles the original lignin, with all structural units and linkages still present except for the dibenzodioxocin linkage. [2] The ratio between different linkages may vary compared to native lignin with higher amounts of 5-5 and 4-O-5 and less of the β-O-4; [1,2] the content of free phenolic hydroxyl groups is also much higher in pulp lignin. [3] Several, usually oxidative delignifying processes are available to bleach cellulose fibers by removal or decoloration of the lignin. Chlorine dioxide is currently the most used bleaching agent for chemical pulps. Alternative processes include totally chlorine-free bleaching methods, such as peroxide bleaching and less common processes such as oxygen and ozone bleaching. [1]

The sole purpose of pulp bleaching is destruction of the lignin fragments to obtain cleaner and more valuable cellulose. Pulp bleaching has been studied extensively, with the purpose to increase the effectiveness of bleaching methods but with little interest in the fate of the lignin. The high structural resemblance between residual lignin and native lignin suggests that the bleaching methods used to remove lignin from chemical pulps could also be applied for the depolymerization of pure lignins. Peroxide bleaching is a particularly attractive bleaching method as the aromaticity of lignin is thought to remain intact and because the reaction takes place in basic solution, most pure lignins will be soluble under the reaction conditions. Although some oxidative depolymerization routes for lignin have been reported (see Chapter 1 and 2), only a few reports are available where peroxide bleaching methods have been applied to pure lignin. Kadla et al. showed that lignin degradation does take place and that carboxylic acid groups are introduced during the alkaline hydrogen peroxide oxidation. [6] The alkaline conditions that are used in peroxide bleaching reactions are applied to deprotonate hydrogen peroxide to the hydroperoxy anion, which can perform addition reactions on (conjugated) carbonyl groups. [4] Such hydroperoxy addition reactions already take place at room temperature and are usually not carried out at temperatures above 70 ˚C to prevent loss of cellulose. At temperatures above 100 ˚C increased amounts of quinones are formed that can be further oxidized to


125

ultimately lead to lignin degradation. [5] Also, H2O2 can break up homolytically into hydroxyl radicals, which react indiscriminately with various lignin functional groups thus lowering the selectivity of the reaction by effectively removing peroxide from the reaction mixture. [6]

Several reaction mechanisms have been proposed for the alkaline hydrogen peroxide bleaching of pulps. It was initially proposed that the bleaching occurs via a Dakin-type reaction mechanism (Scheme 5.1b). The hydroperoxy anion that is formed under basic conditions (Scheme 5.1a) is thought to react with a phenolic aryl-α-carbonyl functionality, cleave the lignin linkage and form an carboxylic and a phenolic end group. [5] As lignins generally contain relatively small amounts of such carbonyl functionalities, an alternative mechanism was proposed, i.e. a Dakin-like reaction (Scheme 5.1c), that does not require a ketone functional group. [5, 7] The first step in this Dakin-like reaction entails the formation of a quinone methide intermediate by dehydration, a reaction that only occurs to a limited extent at room temperature, but starts to become more dominant at temperatures above 100 ˚C. [1, 8] This intermediate then reacts with the hydroperoxy anion ultimately leading to C-C bond cleavage and formation of a phenolic and an aldehyde end group.

Scheme 5.1: a) formation of the hydroperoxy anion, b) the Dakin reaction, adapted from [1], c) Dakin-like reaction, adapted from [1] (L = lignin structure).

Part I, Chapter 5

126

Both the Dakin and the Dakin-like reaction require a phenolic OH group on the 1- position of the aromatic ring to be deprotonated under the basic conditions applied for the reaction. While phenolic OH groups will be deprotonated, in lignin most phenolic OH groups of the constituent monolignols end up as part of an ether linkage, such as the predominant β-O-4 (see Chapter 2). Hydrolysis of the lignin ether linkages therefore needs to take place before Dakin or Dakin-like reactions can occur to a significant extent.

As an alternative to the reduction reactions presented in Chapter 3, oxidative depolymerization reactions could also be of interest for the first step in a two-step lignin conversion process. During oxidation reactions, chemical functionalities are introduced on the aromatic molecules rather than removed, as is often the case during reduction reactions. This might allow highly functionalized products to be obtained directly from lignin using new oxidative routes , compounds which otherwise require several synthetic steps when made from petroleum feedstocks using old technology.

Here, we report on the use of a base-catalyzed oxidation for the depolymerization of various lignins. The influence of base and peroxide concentration on the molecular weight of the product has been investigated. Additional reactions were performed with the aim to enrich the lignin in free phenolic OH groups and aryl-α-ketones, functional groups required for Dakin-(like) reactions. Oxidized lignins were analyzed with IR, NMR and GPC to identify any chemical changes as a result of the oxidative depolymerization reaction.


5.2.1 Oxidation of organosolv ligninThe oxidation reactions were performed by dissolution of the lignin in aqueous

NaOH in glass vials at room temperature followed by addition of H2O2. Influence of daylight on the bleaching process was prevented by covering the vials in foil. Addition of the peroxide resulted in gas formation, oxygen can be formed during the disproportionation of hydrogen peroxide: HOOH + HOO- --> HO- + O2 + H2O. Transition metals impurities as well as higher temperatures will increase the decomposition of H2O2, leading to a lower degree of oxidation of the lignin. [9] Already after 30 min of oxidative treatment, no peroxide could be detected anymore. Light decoloration of the sample could be observed indicating that parts of lignin that absorb light in the visible


127

region, i.e. coniferaldehyde and quinone structures, are converted. [1] Gel Permeation Chromatography (GPC) analysis (Figure 5.1) showed a decrease in molecular weight (Mw) from 4000 to 3000 Da (25% reduction) and a more narrow weight distribution as evidenced by a change in polydispersity (PDI) from 6.2 to 3.8 (40% reduction). The chromatogram shows a decrease in the high molecular weight lignin molecules, eluting at the shortest retention times, indicating that the reaction results in shortening of the longest polymer chains (Figure 5.1). These results are in line with the mechanism of the Dakin reaction that takes place at the end of a polymer chain, removing one monomer at a time. Different results, however, were obtained by Kadla et al. in the depolymerization of kraft lignin with alkaline hydrogen peroxide at temperatures of 70-100 ˚C. They found the longest chains to remain intact and concluded that the reaction mainly takes place by decomposition of the shortest chains. [6] Blank reactions in the absence of H2O2, NaOH or both showed small increases in Mw of the lignin. The decrease in Mw and PDI over time shown in Figure 5.2 (blank reactions are depicted in Figure 5.3) illustrates that the drop in Mw and PDI is most pronounced in the first hour of the reaction, after which only small changes are observed. While after 5 h almost no changes in the Mw could be observed anymore, the decrease in PDI still continued. The GPC traces of the reaction mixtures that were oxidized for 0.5-5 h revealed the presence of larger amounts of small molecules with elution times between 10 and 11 min. At longer reaction times, this signal gradually disappeared until it completely vanished after 20 h of reaction. The disappearance of the smaller molecules that were initially formed can be caused by recondensation and has little effect on the calculated Mw of the reaction mixture; it could, however, explain the ongoing decrease in PDI also in reaction longer than 5 h.

5 7 9 11Retention time (min)

before oxidation

after oxidation

Figure 5.1: GPC chromatogram of organosolv lignin before and after a 20 h alkaline oxidation reaction.

Part I, Chapter 5

128

0

10

20

30

40

50

0 5 10 15 20

Decr

ease

(%)

Time (h)

Mw PDI

Figure 5.2: Change in molecular weight and polydispersity index of organosolv lignin over time.

Lignin depolymerization via Dakin-(like) reaction pathways is expected to take place at the end of the polymer chain, leading to the formation of aromatic monomers. The amount of these monomers formed after 20 h of lignin oxidation were determined by acidification of the reaction mixture with HCl followed by extraction with dichloromethane. GC and GC-MS analysis of the extracted products identified oxygen-rich aromatic lignin monomers accounting for 1.2% of the original weight of the lignin. The four different syringol and guaiacol-derived-oxidation products that were detected are shown in Figure 5.3. The major product, 4-hydroxy-3,5-methoxybenzoic acid (1), accounted for 1% of the original lignin. Larger amounts of mono-aromatic products can possibly be obtained from reactions of 0.5-5 h, since the GPC data of these reactions indicated larger amounts of small molecules. Similar compounds as well as aromatic monomers with two carboxylic acid groups and 5-5 and 4-O-5 linked dimers were observed by Kadla et al. after high temperature reactions. [6] Vanillic acid (2) and carboxylic acid-substituted vanillic acid were reported to be the predominant low molecular weight products; no exact yields in percentage of the original lignin were reported, however. Most likely, similar products are also formed during our

Figure 5.3: Monomeric aromatics formed during the alkaline hydrogen peroxide-catalyzed oxidation of organosolv lignin.


129

alkaline oxidation reaction of organosolv lignin, but these highly oxidized compounds or higher Mw dimers cannot be observed with our GC protocols.

The influence of NaOH and H2O2 concentration on the degree of depolymerization were determined by a set of experiments using different molar concentrations of NaOH (Figure 5.4a) with 0.5 mL peroxide, and a set of experiments using different volumes of peroxide (Figure 5.4b) in a 1 M NaOH solution. The amount of organosolv lignin and the total volume of the reactions were always 200 mg and 10 mL. Although NaOH is needed for the deprotonation of H2O2, hydrogen peroxide (pKa of 11.6) [6] is not the strongest acid in solution and it is expected that a minimum NaOH concentration is needed to deprotonate all lignin phenolic OH groups (pKa around 10) [10] before any H2O2 is deprotonated. According to 31P NMR measurements (vide infra), the organosolv lignin contains 0.65 mmol OH groups in total per 200 mg lignin. The minimal OH- concentration in a 10 mL solution then needs to be 0.65 M in order for any OH- to be available for peroxide deprotonation. Figure 5.4a shows that the degree of depolymerization did not change when the NaOH concentration was increased. At the higher NaOH concentrations of 2 and 4 M not all lignin dissolved and for the 4 M reaction a smaller decrease in molecular weight was observed compared to the other reactions. It should be emphasized that during GPC measurements, all lignin was dissolved completely. A larger influence of NaOH concentration was observed on the PDI with lower concentrations leading to a lower PDI of the depolymerized lignin. It was shown before that a lignin fragment has to contain at least 31 phenolic OH groups per 100 phenylpropane units in order to be completely soluble in alkaline solutions. [11] The NaOH concentrations applied here are high enough to ensure complete

-10

0

10

20

30

40

50

0 M 0.5 M 1 M 2 M 4 M

Decr

ease

(%)

NaOH concentration (mol/L)

-20

-10

0

10

20

30

40

50

0 0.1 0.2 0.3 0.5 1

Decr

ease

(%)

Volume H2O2 added (mL)

a. b.

Figure 5.4: Change in ( ) molecular weight and ( ) polydispersity of organosolv lignin during oxidation as a result of changing a) NaOH concentration and b) H2O2 volume.

Part I, Chapter 5

130

solubility of the lignin and it is not clear why the organosolv lignin was not soluble at the highest NaOH concentrations.

Variation of the volume of peroxide added to the reaction had a larger impact on the Mw decrease of lignin during the oxidation reaction. While addition of 0.1 mL of peroxide did not lead to lignin depolymerization, increasing the volume of peroxide from 0.1 to 1.0 mL led to the formation of lower average molecular weight lignin molecules with a decrease of 35% in Mw when 1 mL was added. A decrease in PDI could already be observed at the lowest volumes added and leveled off with higher peroxide volumes. Increasing the amount of peroxide to more than 1 mL proved to be difficult because of the formation of large amounts of gas from the reaction mixture leading to leakage of lignin solution from the reaction vial. Gradually dosing the peroxide by adding 0.2 mL every 30 min resulted in a minimal increase of the degree of depolymerization to 24% for a total of 1 mL compared to 19% for the standard 0.5 mL sample (Table 5.1, entry 6). Addition of more peroxide had no influence on the Mw (Table 5.1, entry 7).

Additional experiments were performed using higher reaction temperatures for the alkaline oxidation. Incidentally, it should be noted that a difference in the calculated molecular weights was observed between the results reported above and all of the data reported below. This is attributed to a repacking of the GPC column, which decreased the retention time, resulting in an increase in the calculated weight by 500 Da for all samples. For comparison, a standard sample of the depolymerization reaction with 1 M NaOH and 0.5 mL peroxide was included in every set of GPC measurements and the

Entry NaOH H2O2 T Mw decrease PDI decrease (M) (mL) (˚C) (%) (%)

1 0 0 RT -2 02 1 0.5 RT 19 233 1 0.5 50 28 374 1 0.5 200 55 275 1 2 RT 25 276 1 5 x 0.2 RT 24 297 1 10 x 0.2 RT 24 29

Table 5.1: Decrease of Mw and polydispersity of organosolv lignin after alkaline oxidation under various reaction conditions.


131

degree of depolymerization (i.e. % Mw decrease) can be compared to this reaction.

Performing the reaction at higher temperatures did result in increased depolymerization activity, at 50 ˚C the Mw decrease went up from 19% to 28% (Table 5.1, entry 3). It was already shown in Chapter 3 that at temperatures higher than 200 ˚C lignin solubilization in ethanol/water mixtures takes place by partial depolymerization of lignin. In addition quinone methide intermediates start to form in larger quantities in Dakin-like reactions at temperatures above 100 ˚C. [1, 8] Reactions at temperatures higher than 100 ˚C had to be run in an autoclave reactor and were only performed for 2 h instead of overnight as usual. At 200 ˚C the degree of depolymerization increased to 55% (Table 5.1, entry 4), which is much larger than the decrease in molecular weight that was observed in Chapter 3 after solubilization of the lignin in ethanol/water. This indicates that the oxidation reaction is much more efficient at higher temperatures. It is expected that higher degrees of depolymerization can be achieved in the presence of stabilization agents that prevent metal-catalyzed peroxide decomposition, as described by Kadla et al. [6]

5.2.2 Oxidation of various types of ligninDepending on the plant from which a lignin sample is obtained and the method

used for extraction, lignins can differ in their chemical functionalities, the amount of branching and in the nature and amounts of impurities. For example, hardwood type lignins are expected to be relatively more linear compared to lignins from softwood and grasses. Organosolv lignins are very clean lignins with low impurity and ash content, whereas lignosulfonates are rich in sulfonate functionalities and soda

Entry Source Pretreatment Mw decrease PDI decrease(%) (%)

1 wheat straw organosolv 24 272 wheat straw organosolv 12 273 unknown alkali 47 294 softwood Indulin kraft 29 275 sugarcane sugarcane bagasse 18 266 grass soda 97 hardwood Alcell organosolv 19 23

Table 5.2: Decrease of Mw and polydispersity of different types of lignin after alkaline oxidation treatment.

Part I, Chapter 5

132

lignins have a high sodium content. All these differences can contribute to changes in reactivity when lignin conversion reactions are performed. The standard alkaline oxidation treatment was therefore tested on 7 different lignin samples obtained from several sources and by varying extraction methods. The results are shown in Table 5.2.

The majority of the samples tested showed a Mw decrease after alkaline oxidation in the range of 10-20%; surprisingly, the Mw of the alkali-treated lignin obtained from Sigma-Aldrich decreased by 47%. Other lignins that were also obtained through an alkaline pulping method, kraft and soda lignin, did not show this high degree of depolymerization. The soda lignin actually did not show any weight loss and only a small decrease in PDI.

Of course there are many factors that contribute to the reactivity of a given lignin in this reaction. Metal impurities, such as Fe, Mn and Cu, catalyze the disproportionation of hydrogen peroxide resulting in a lowered degree of depolymerization. [4] In addition, the abundance of (conjugated) carbonyl groups and free phenolic OH groups differs for each type of lignin and for each lignin extraction method. The high reactivity of the alkali lignin might, for instance, be due to a higher amount of free phenolic groups in this lignin, whereas lack of depolymerization of the soda lignin might be caused by a significant amount of metal impurities. Metal-catalyzed decomposition of hydrogen peroxide can be reduced by adding a peroxide stabilizing agent or a chelating agent that binds the metals, for example diethylenetriamine penta(methylene phosphonic acid) (DTPMP) or diethylenetriaminepentaacetic acid (DTPA); the latter one is also used for the stabilization of hydrogen peroxide in pulp bleaching. [1, 6]

5.2.3 Oxidation of pretreated ligninAssuming that the oxidative depolymerization takes place via a Dakin or Dakin-like

reaction mechanism (Scheme 5.1b and c), the reaction is expected to proceed more efficiently if the lignin is enriched in units that contain both a ketone on the aryl-α-carbon position and have a free phenolic OH group. Alcohols rather than ketones are mostly found on the α-carbon position, however, and the phenolic OH groups are linked, most commonly via a β-O-4 linkage. Pretreating lignin before oxidation aimed at increasing the amounts of α-carbon ketones and free phenolic OH groups could then increase the efficiency of the oxidative depolymerization. In Chapter 3, we have shown that solubilization of lignin in ethanol/water at 225 ˚C and 58 bar He pressure leads to partial depolymerization (25% decrease of Mw) of the lignin macromolecule. This


133

decrease in Mw was attributed to cleavage of the lignin ether linkages, in particular the β-O-4 linkage. Furthermore, it was previously shown in our group that the Co-catalyzed oxidation of lignin and lignin model compounds did not lead to monomer formation but rather specifically resulted in oxidation of the OH group on the α-carbon to a ketone, as shown by in situ ATR-IR measurements. [12] These cobalt-catalyzed oxidations were performed with various cobalt complexes in the ionic liquid 1-ethyl-3-methylimidazolium diethylphosphate [EMIM] [DEP] under 5 bar oxygen pressure, with sodium hydroxide aiding the formation of the active species. The system was also active in alkaline water, but lower conversions were obtained. [12]

A set of experiments was performed to compare the influence of solubilization and cobalt-catalyzed oxidation on the alkaline hydrogen peroxide oxidation of organosolv lignin. Lignin solubilization was performed by heating 250 mg organosolv lignin in 50 mL ethanol/water to 225 ˚C. After reaching this temperature the solubilized lignin solution was cooled, filtered and freeze-dried to obtain the solubilized lignin. The changes in Mw and polydispersity of the treated lignins with respect to the parent lignin are shown in Table 5.3. The standard base-catalyzed oxidation and the solubilization step showed Mw decreases of 23% and 22%, which are comparable to the results reported above and in Chapter 3 (Figure 5.5; Table 5.3, entry 1 and 2). Subjecting the solubilized lignin to the alkaline oxidation treatment showed that the pretreatment step had a positive effect on the extent of depolymerization, as a total decrease in Mw of 32% was observed (Table 5.3, entry 3).

The method for Co-catalyzed oxidation was adapted from Zakzeski et al. [12]

Table 5.3: Decrease of Mw and polydispersity of organosolv lignin after different oxidation and solubilization treatments

Entry Treatment Mw decrease PDI decrease(%) (%)

1 Alkaline oxidation 23 212 Solubilization 22 53 Solubilization + Alkaline oxidation 32 274 Co-catalyzed oxidation -57 05 Co-catalyzed oxidation + Alkaline oxidation 1 236 Solubillization + Co-oxidation -16 147 Solubillization + Co-oxidation + Alkaline

oxidation24 35

Part I, Chapter 5

134

Co(salen) was used as the catalyst in aqueous sodium hydroxide with 5 bar oxygen at 80 ˚C. The Co-oxidized lignin was analyzed subsequently with IR and NMR spectroscopy to confirm formation of aryl-α-ketones. Although no change in Mw was expected in this oxidation step, the GPC measurements of the organosolv lignin after Co-catalyzed oxidation showed a remarkable increase of the Mw with 50% (Figure 5.5) (Table 5.3, entry 4). Influence of the Co-catalyzed oxidation on the alkaline oxidation reaction was tested after diluting the reaction mixture to the lignin and NaOH concentration needed for the base-catalyzed oxidation. Cobalt ions were removed with Chelex 100, which binds transition metal ions via an ion exchange mechanism in basic media. Not all of the cobalt could be completely removed, however, as elemental analysis of the dried solutions after treatment with Chelex 100 showed trace amounts of the metal to still be present. This remaining cobalt can influence the second oxidation step by catalyzing the disproportionation of H2O2. When the Co-catalyzed oxidation is followed by a base-catalyzed oxidation, the average molecular weight of the organosolv lignin is comparable to that of the untreated sample but with a much higher polydispersity (Table 5.5, entry 5).

Subjecting the solubilized lignin to a Co-catalyzed oxidation step resulted in a net increase of the polymer weight by 16% (Table 5.4, entry 6). Subsequent solubilization, Co-catalyzed oxidation and alkaline-oxidation was performed in order to obtain a lignin that is both enriched in free phenolic OH groups and α-ketones. The Mw of the lignin after the last oxidation step showed a decrease of 24% compared to the untreated organosolv lignin; this reduction can also be achieved with a single base-catalyzed oxidation, however. The results obtained by solubilization/alkaline

4 6 8 10 12 14Retention time (min)

solubilized

Co-oxidized

not treated

Figure 5.5: GPC chromatogram of organosolv lignin, organosolv lignin after Co-catalyzed oxidation and solubilized organosolv lignin.


135

oxidation indicate that increasing the amount of free phenolic OH groups in lignin has a beneficial effect on the degree of depolymerization after alkaline oxidation. The Co-catalyzed oxidation of lignin, however, caused an increase in Mw that made it impossible to measure any effect of the amounts of any additionally formed α-ketones on the alkaline-oxidation.

There are several possible explanations for the observed increase in Mw of lignin after the Co-catalyzed oxidation reaction. First of all, as the GPC measurements determine the hydrodynamic volume of the polymer rather than the actual weight of a polymer, changes of functional groups perhaps influence the behavior of the molecule in solution and alter the outcome of this analysis. If the oxidized lignin has a larger hydrodynamic volume than the parent lignin this would explain the increased molecular weight. To assess any changes in or introduction of functional groups during the oxidation step, the samples were analyzed by NMR and IR spectroscopy. The spectroscopic data, which are discussed in section 5.2.4, show an increase in oxygen-rich functional groups such as carboxylic acids and ketones after Co-catalyzed oxidation, indicating that oxidation of the lignin indeed took place. Other than a change in hydrodynamic volume, it is also possible that the Mw of the lignin increased by condensation or coupling reactions that do not occur in the absence of cobalt. The biosynthesis of lignin, that is growth of the polymer, involves radical-initiated coupling of the monolignols, [13, 14] a reaction that could also be facilitated by a first-row transition metal such as cobalt. Indeed, cobalt is able to generate phenoxy radicals in the presence of molecular oxygen and it was even shown before that Co(salen) complexes are able to form phenoxy radicals on lignin fibers under these conditions. [15] It t is therefore plausible that the increased Mw observed after the Co-catalyzed oxidation of lignin is caused by a mechanism similar to the reactions that take place during lignin synthesis in plants.

5.2.4 Chemical changesAfter the Co-catalyzed oxidation, solubilization and alkaline oxidation reactions,

the lignins were isolated by freeze-drying the reaction mixtures. The original organosolv lignin and the reacted lignins were characterized by IR and 1H and 31P NMR analysis to determine what chemical changes took place. After alkaline oxidation, the solution was neutralized with HCl to a pH of 7 or acidified to a pH of 1, after which the depolymerized lignin and the remaining NaCl salt were retrieved by freeze-drying the solution. If the base-catalyzed oxidation of lignin takes place via a Dakin-type reaction pathway (Scheme 5.1b), carboxylic acid groups should be formed. Depending on the

Part I, Chapter 5

136

pH of the solution before freeze-drying, the carboxylic acids present in the lignin can be deprotonated (pH > pKa) or protonated (pH < pKa). As most carboxylic acids have a pKa between 2.5 and 5, lignin isolated at pH 7 is expected to contain deprotonated carboxylic acids and as a result was still soluble in water. Lignin that was isolated at pH 1 was not soluble in water anymore, but could be dissolved in organic solvents.

The IR spectra of lignin samples after alkaline oxidation isolated at pH 7 and pH 1, a sample oxidized by Co(salen) isolated at pH 1 and a solubilized lignin sample were compared to the spectrum of the original organosolv lignin (Figure 5.6). The spectrum of the (deprotonated) lignin isolated at pH 7 (deprotonated, a) clearly shows two distinct peaks at 1597 and 1396 cm-1 belonging to carboxylate salts. Because of their intensity, these peaks dominate the spectrum and no other details could be observed. [16] The lignin samples isolated at pH 1 after both alkaline oxidation and Co-catalyzed oxidation (b and c) showed major changes in the IR spectrum compared to the original organosolv lignin (e). The lignin that was solubilized in ethanol/water (d) was almost identical to the original sample (e). This indicates that during solubilization, during which some ether linkages are broken resulting in the decreased molecular weight (Chapter 3), only functional groups that were already present in the pure organosolv sample are formed. Both oxidized lignins show a broadening of the OH stretching band in the 2800-3400 cm-1 region and formation of a weak band around 2630 cm-1. The OH stretching bands at these relatively low wavenumbers probably belong to more acidic protons and the band at 2630 cm-1 indicates the presence of carboxylic acids.

The most significant changes are observed in the region between 2000 and 1000 cm-1. The C=O stretch vibration band at 1707 cm-1 increased in intensity and shifted to higher wavenumbers (1724 cm-1) after both alkaline and Co-catalyzed oxidation. This carbonyl vibration is found at relatively very high wavenumbers and it is therefore unlikely that the shift is caused by formation of ketones. Any C=O stretch vibration bands at these high wavenumbers can possibly indicate the formation of carboxylic acids of which the carbonyl is not involved in a hydrogen bonding interaction. The absence of a dimer band at ~940 cm-1, which is characteristic for hydrogen-bonded carboxylic acid dimers, further supports this assignment. The observation of IR bands assigned to carboxylic acids supports the hypothesis that the reaction takes places via a Dakin reaction pathway. The Co-catalyzed oxidation was previously reported to result in the formation of ketones, but the high wavenumber of the C=O band in the IR spectrum of lignin after Co-catalyzed oxidation indicates that this sample also contains carboxylic acids. [12] The same shift to higher wavenumbers was observed


137

400900140019002400290034003900

A.U

.

16081515 1116

1463

15971393

a.

12251724

b.

c.

d.

e.

1707

1723

1216

1216

1612

1608

2630

11211515

1468

2630

wavenumber (cm ¹)-

Figure 5.6: IR spectra of organosolv lignin before and after oxidative treatment or solubilization, a) lignin isolated at pH 7 after alkaline oxidation, b) lignin isolated at pH 1 after alkaline oxidation, c) lignin isolated at pH 1 after oxidation by a cobalt complex, d) lignin solubilized in ethanol/water and e) original lignin.

to a lesser extent for the band at 1216 cm-1 in the alkaline oxidized sample, which has contributions from aromatic C-O stretch vibrations. This shift can probably be attributed to an increase in the amount of aromatic esters and phenolic alcohols. The C=O stretching band is a fairly strong vibration and for carboxylic acids the band is

Part I, Chapter 5

138

even stronger than for ketones and aldehydes. This resulted in a relative decrease in intensity of all other lignin vibrations, which are, nonetheless, still present in the oxidized lignin spectrum. Bands belonging to aryl ring stretching vibrations (1608, 1515 and 1463 cm-1) and aliphatic C-O vibrations (1116 cm-1) all decreased in intensity compared to the C=O stretching band. [1, 17] The observations listed above show that most of the chemical functionalities present in the original lignin probably remain present in the oxidized lignin; the spectra are, however, dominated by the presence of large bands originating from newly formed carboxylic acid groups.

31P NMR can be used to determine changes in the ratios between different functional groups in lignin after reaction of the sample with a phosphitylating agent. Untreated organosolv lignin, the oxidized lignins isolated at pH 1 and the solubilized lignin were phosphitylated using 2-chloro-4,4,5,5-tetramethyl-1,3,2-dioxaphospholane according to the method developed by Argyropoulos et al. [18] 31P NMR spectra of these samples were obtained using phosphitylated cholesterol as internal standard. Peaks can be attributed to specific functionalities present in lignin as is shown in Figure 5.7. Just as was seen in the comparison of the IR spectra of the various lignins, the 31P NMR spectrum of the solubilized lignin was almost identical to the spectrum of the original lignin. This confirms that no major changes in the lignin structure or the type of functional groups took place during the solubilization process. Solubilization resulted in an increase in the amount of free phenolic OH groups per gram of lignin with 10%, 25% and 8% for syringyl, guaiacyl and p-coumaryl functionalities, respectively. The data thus further confirms that solubilization takes place via breaking of lignin ether linkages, the β-O-4 linkage specifically, as was suggested in Chapter 3.

The presence of a large amount of salt in the lignin samples freeze-dried after oxidation unfortunately does not allow quantitative analysis of the peak areas for these measurements. Nonetheless, the abundance of specific functional groups relative to other functional groups could be calculated. The amount of carboxylic acids, which were present in low concentrations in the original lignin, were shown to increase in both the peroxide-oxidized lignin and the Co-catalyzed oxidized lignin. This shows that Co-catalyzed oxidation does not only oxidize alcohols to ketones, but also leads to the formation of carboxylic acids. A comparison of the results obtained here with the previous report on the Co-catalyzed oxidation of lignin clearly shows that more insight could be obtained with the use of multiple analytic techniques instead of one single measurement. Remarkably, the spectrum obtained from Co-catalyzed oxidized lignin does not show any signals corresponding to syringyl, guaiacyl or


139

Chemical shift (ppm)

Alkaline oxidation

Untreated organosolv lignin

internal standard

a b c d e

134135136137138139140141142143144145146147148149

Solubilization

Co-oxidation

Figure 5.7: 31P NMR spectra of phosphitylated organosolv lignin, solubilized organosolv lignin, oxidized organosolv lignin and organosolv lignin after Co-catalyzed oxidation. The areas highlighted represent the a) aliphatic-OH, b) syringyl-OH, c) guaiacyl-OH, d) p-hydroxyl-OH and e) carboxylic acid groups.

p-coumaryl phenolics anymore. These results suggest that during the oxidation all free phenolic OH groups have reacted, which could be the result of a Co-catalyzed oxidative radical coupling reaction, in a manner similar to lignin biosynthesis. [13, 14, 15] Such additional linkages will look very similar to the linkages that are present in the original lignin sample and would both explain the increased molecular weight as observed with GPC, as well as the similarity of the IR spectra.

In the peroxide-oxidized sample a comparison of the relative intensities of the peaks belonging to different functional groups showed a more than 10–fold increase in the amount of carboxylic acid groups relative to the syringyl, guaiacyl and p-hydroxyl functionalities. Furthermore, the relative amount of aliphatic OH groups doubled compared to the aromatic OH groups. This enormous increase in carboxylic acid functionalities, as also seen in the IR data further supports the proposed Dakin

Part I, Chapter 5

140

oxidation mechanism (Scheme 5.1b). Dakin and Dakin-like oxidation reactions, however, should also lead to the formation of new free phenolic OH groups, but no significant increase was observed. The oxidation of the hydroquinones formed to quinones could be a possible explanation for the lack of increase in free phenolic OH groups. [19] However, these quinones should show a typical double band in the IR spectra between 1690 en 1655 cm-1 and no such band was observed. [16]

1H and 2D 1H-13C HSQC NMR spectra of the untreated organosolv lignin, the oxidized lignins isolated at pH 1 and the solubilized lignin were recorded. Also here quantitative analysis proved impossible because of the salts present in the oxidized lignin sample. In accordance with the 31P NMR spectra, the 1H NMR spectra of the original organosolv lignin and the solubilized organosolv are very similar, again confirming that no significant chemical changes take place during the solubilization (Figure 5.8). Comparison of the spectra obtained from the oxidized samples to the original organosolv lignin (Figure 5.8) shows formation of well-defined new peaks (c and d), and an increase in intensity of peak a and b. Peaks a and c can be assigned to saturated alcohol and saturated carboxylic acid end groups, respectively, [20] further confirming the formation of carboxylic acid groups during the oxidative treatment and the increase in the amount of saturated alcohols. HSQC NMR of the peroxide-oxidized lignin did not show any obvious changes in the peaks corresponding to the ether linkages, but as quantification was not possible it is difficult to interpret any changes in this area. The peak corresponding to the β-5 α-carbon atom completely disappeared in the peroxide-oxidized lignin spectrum indicating a definite change in chemical structure of this linkage. Indeed the β-5 linkage has been known to form a phenylcoumaranone species in the presence of oxygen in alkaline solutions (Scheme 5.2). [21] In agreement with the 1H spectrum, new cross peaks were observed in the aliphatic area at (1.23, 28), (2.33, 33) and (1.57, 51) (1H, 13C) ppm corresponding to the observed saturated alcohol and carboxylic acid end groups that were also present in the 1H spectra.

Scheme 5.2: Formation of the phenylcoumaranone species, adapted from [21].


141

012345678910

Co-oxidation

Untreated organosolv lignin

Chemical shift (ppm)

H2O

DMSO

DMSO a

c

d

b

Solubilization

Alkaline oxidation

DMSO a

c

d

b

Figure 5.8: 1H NMR spectra of organosolv lignin and solubilized, peroxide-oxidized, and Co-oxidized organosolv lignin in dmso-d6.

5.3 Conclusions

The bleaching methods used in the pulp and paper industry were designed for the removal of lignin from cellulose pulps. Peroxide bleaching of lignin, which was originally developed to break down the small remaining amounts of lignin, can also be used for the depolymerization of pure, bulk lignin. A molecular weight decrease of 20% could be obtained in the base-catalyzed depolymerization of organosolv lignin with H2O2. After increasing the temperature to 200 ˚C or increasing the amount of

Part I, Chapter 5

142

peroxide in the reaction, the degree of depolymerization increased to 55 and 25%, respectively. The exact influence of temperature on the reaction speed and obtained products is not yet clear, optimization of the reaction temperature is expected to lead to higher degrees of depolymerization. Depending on the lignin source and the pretreatment method used, the depolymerization reaction is more or less effective. This is believed to be related to the presence of metal impurities or the amount of free phenolic OH groups. Increasing the amount of free phenolic OH groups in a lignin sample by solubilization in a mixture of ethanol/water led to an increased degree of depolymerization. Performing a Co-catalyzed oxidation intended to enrich lignin in α-ketone groups resulted in the formation of carboxylic acids as well as ketones. Side reactions increased the Mw of the lignin to such an extent that the increased concentration of α-ketone groups did not result in improved depolymerization. IR and NMR analysis of the oxidized lignin confirmed that the depolymerization of lignin takes place via a Dakin or Dakin-like reaction mechanism, although, the formation of quinones in the reaction product could not be confirmed.

5.4 Experimental

5.4.1 ChemicalsThe Alcell organosolv lignin (66.47% C, 5.96% H, 0.15% N, 27.43% O by difference),

provided by Wageningen University, was obtained from hardwoods and isolated by an organosolv extraction method. [22] NaOH (Sigma-Aldrich 97%), H2O2 (Acros, aqueous 35%), Co(salen) (N,N′-bis(salicylidene)ethylenediaminocobalt(II) hydrate) (Aldrich, 97%), Chelex 100 (Aldrich) and oxygen (5.0, Linde) were obtained commercially.

5.4.2 Catalytic reactionsBase-catalyzed oxidation reactions were performed in a glas vial and magnetically

stirred. 0.2 g of dried lignin was dissolved in 10 mL 1 M NaOH in milliQ water after which 0.5 mL H2O2 (35% in H2O) was added. The vial was covered with aluminum foil to prevent any influence of light on the depolymerization reaction and the mixture was left to stir at room temperature overnight.

Co-catalyzed oxidation reactions were performed in a 40 mL stainless steel Parr batch autoclave. [12] The temperature was monitored with a thermocouple and stirring was performed using a magnetic driver equipped with an impellor at 750 rpm. The autoclave was loaded with 5 mg Co(salen), 10 mL 0.5 M aqueous NaOH and 500


143

mg organosolv lignin. The reactor was purged 4 times with oxygen and pressurized to 5 bar, the reaction mixture was heated to 80 ˚C and was left to stir overnight. After the reaction was stopped by cooling to room temperature and release of pressure, the reaction mixture was stirred with Chelex 100 to remove Co ions.

Solubilization experiments were performed in a 100 mL stainless steel Parr bath autoclave. The temperature was monitored with a thermocouple and stirring was performed using a magnetic driver equipped with an impellor at 750 rpm. The autoclave was loaded with 250 mg organosolv lignin, 25 mL ethanol and 25 mL water. After heating to 225 ˚C the reaction mixture was cooled to room temperature, the remaining solids were filtered off and the solution was freeze-dried to obtain the solubilized lignin.

5.4.3 AnalysisThe Mw of depolymerized lignin was determined by GPC on an alkaline SEC by

Waters Alliance. The system was equipped with a manually packed column (4.6 x 30 cm) with ethylene glycolmethacrylate copolymer TSK gel Toyopearl HW-55F according to the work of Gosselink et al. [23] Sodium polystyrene sulfonates (Mw range 891 to 976,000 Da) were used for calibration of the molar mass distribution. The oxidized lignin solutions were diluted to a concentration of 1 mg/mL lignin in 0.5 M NaOH. GPC runs were performed at 40 ˚C with 0.5 M NaOH eluent at a flow rate of 1 mL/min and UV detection at 280 nm. GC analyses were performed on a Varian GC equipped with a VF-5ms capillary column and an FID detector.

FTIR measurements were carried out at room temperature on a Bruker Tensor 27 instrument. IR data were recorded with a deuterated triglycerine sulfate (DTGS) detector. The samples are recorded in a KBr pellet in transmission mode. The optical resolution of the IR spectra was 4 cm−1 and 16 scans were accumulated for each spectrum.

For the 31P NMR studies 40 mg of dried ligin was dissolved overnight in 400 µL pyridine anhydride/CDCl3 (1.6/1, v/v). Cholesterol (200 µL, 19 mg/mL) and chromium (III) acetylacetonate (50 µL, 11.4 mg/ml) were used as internal standard and relaxation agent. 100 µL phosphitylating agent (2-chloro-4,4,5,5-tetramethyl-1,3,2-dioxaphospholane) (Aldrich, 99%) was added before the measurement. [18] 31P NMR spectra were obtained on a Varian 400 MHz NMR spectrometer using a standard phosphorus pulse program and a relaxation delay of 10 s.

Part I, Chapter 5

144

1H and HSQC NMR spectra were obtained using a Bruker Avance II 600 MHz spectrometer equipped with a 5 mm CPTCI 1H-13C/15N/2H cryogenic probe with z-gradients at 298 K using the CPMGQHSQC pulse program. [24] The lignins and lignin solutions were freeze-dried and dissolved in DMSO-d6 (40 mg/mL, 99.9%, Cambridge Isotopes Laboratories), and chemical shifts were referenced to the residual DMSO signal (2.50/39.5 ppm).

5.5 Acknowledgments

Richard Gosselink and Jacinta van der Putten from Wageningen University and Hans Wienk from Utrecht University are acknowledged for help in acquiring the GPC and NMR data, respectively. NMR experiments were performed at SONNMRLSF at the Bijvoet Institute of Biomolecular Research of Utrecht University. Peter de Peinder from Vibspec is thanked for his help in analyzing the IR spectra.

5.6 References

[1] C. Heitner, D. R. Dimmel, J. A. Schrmidt, Lignin and Lignans; Advances in Chemistry, CRC Press, 2010, Chapter 11.[2] Z. Jiang, D. S. Argyropoulos, J. Pulp Pap. Sci. 1999, 25, 25-29.[3] R. C. Francis, Y. Lai, C. W. Dence, T. C. Alexander, Tappi J. 1991, 74, 219-224.[4] A. Wuorimaa, R. Jokela, R. Aksela, Nord. Pulp Pap. Res. J. 2006, 21, 435-443.[5] J. F. Kadla, H-m Chang, H. Jameel, Holzforschung 1997, 51, 428-434.[6] J. F. Kadla, H.m Chang, H. Jameel, Holzforschung 1999, 53, 277-284.[7] S. Omori, C. W. Dence, Wood Sci. Technol. 1981, 15, 113-123.[8] L. Heuts, G. Gellerstedt, Nord. Pulp Pap. Res. J. 1998, 13, 107-111.[9] O. Legrini, E. Oliveros, A. M. Braun, Chem. Rev. 1993, 93, 671-698.[10] M. Ragnar, C. T. Lindgren, N. O. Nilvebrant, J. Wood Chem. Technol. 2000, 20, 277-305.[11] E. I. Evstigneev, Russ. J. Appl. Chem. 2011, 84, 1040-1045. [12] J. Zakzeski, A. L. Jongerius, B. M. Weckhuysen, Green Chem. 2010, 12, 1225-1236.[13] R. Hatfield, W. Vermerris, Plant Physiol. 2001, 126, 1351-1357.[14] J. Ralph, K. Lundquist, G. Brunow, F. Lu, H. Kim, P. F. Schatz, J. M. Marita, R. D. Hatfield, S. A. Ralph, J. H. Christensen, W. Boerjan, Phytochem. Rev. 2004, 3, 29-60.[15] C. Canevali, M. Orlandi, L. Zoia, R. Scotti, E. L. Tolppa, J. Sipila, F. Agnoli, F. Morazzoni, Biomacromolecules 2005, 6, 1592-1601.


145

[16] G. Socrates, Infrared Characteristic Group Frequencies, second edition, Wiley, 1994, Chapter 10. [17] C. Heitner, D. R. Dimmel, J. A. Schrmidt, Lignin and Lignans; Advances in Chemistry, CRC Press, 2010, Chapter 4.[18] D. S. Argyropoulos, H. I. Bolker, C. Heitner, Y. Archiov, J. Wood Sci. Technol. 1993, 13, 187-212.[19] J. J. Bozell, B. R. Hames, J. Org. Chem. 1995, 60, 2398-2404.[20] J. C. del Rio, J. Rencoret, G. Marques, J. Li, G. Gellerstedt, J. Jiménez-Barbero, Á. T. Martínez, A. Gutiérrez, J Agric. Food Chem. 2009, 57, 10271-10281.[21] J. Gierer, I. Pettersson, L. A. Smedman, Acta Chem. Scand. 1972, 26, 3366-3376.[22] N. T. Kleinert, 1971, US patent 3585104.[23] R. J. A. Gosselink, J. E. G. Van Dam, E. De Jong, E. L. Scott, J. P. M. Sanders, J. Li, G. Gellerstedt, Holzforschung 2010, 64, 193–200.[24] S. Heikkinen, M. M. Toikka, P. T. Karhunen, I. Kilpelinen, J. Am. Chem. Soc. 2003, 125, 4362 – 4367.

Part II

Hydrodeoxygenation of Ligin Model Compounds

Chapter 6

CoMo Sulfide-Catalyzed Hydrodeoxygenation of Lignin Model Compounds

AbstractExtensive hydrodeoxygenation (HDO) studies with a commercial sulfided CoMo/

Al2O3 catalyst were performed on a library of monomeric and dimeric lignin model compounds at 50 bar hydrogen pressure and 300 °C in dodecane, using a batch autoclave system. The catalyst was activated under hydrogen atmosphere prior to the reaction and the spent catalyst was analyzed using thermogravimetric analysis. An extended reaction network is proposed, showing that HDO, demethylation and hydrogenation reactions take place simultaneously. HDO of mono-oxygenated substrates proved to be difficult under the applied conditions. Starting from most positions in the network, phenol and cresols are therefore the main final products, suggesting the possibility of convergence on a limited number of products from a mixture of substrates. HDO of dimeric model compounds mimicking typical lignin linkages revealed that coumaran alkyl ethers and β-O-4 bonds can be broken, but 5-5 linkages not.

Based on: A. L. Jongerius, R. Jastrzebski, P. C. A. Bruijnincx, B. M. Weckhuysen, “CoMo sulfide-catalyzed hydrodeoxygenation of lignin model compounds: An extended reaction network for the conversion of monomeric and dimeric substrates“ J. Catal. 2012, 285, 315-323.

Part II, Chapter 6

150

6.1 Introduction

Pretreatment of lignocellulosic biomass, for instance in a biorefinery operation, will give rise to large process streams of lignin, for which value-added applications will have to be found. [1] The catalytic conversion of lignin into bulk chemicals or fuel components is a possible scenario for future use of this abundant lignin feedstock. To this extent, lignin or smaller fragments derived from this biopolymer (e.g. monomers or small oligomers) can be subjected to various chemical transformations. Several strategies can be followed, such as lignin bond hydrolysis [2, 3] and catalytic oxidation reactions, [4, 5] in order to obtain value-added products. One approach that carries considerable potential is the conversion by catalytic hydrodeoxygenation (HDO), as this allows removal of part of the extensive functionalities from the highly oxygenated structure resulting in the formation of bulk platform molecules, such as phenolics and benzene, toluene and xylenes (BTX). Several studies on the catalytic hydrodeoxygenation of lignin or fragments derived from lignin have been reported. [6-8]

The structural complexity and variability of lignin have prompted the use of several lignin model compounds to study lignin hydrodeoxygenation reactions. These model compounds, typically mono-aromatics, contain linkages and functional groups that resemble those found in the lignin polymer and the study of their reactivity provides insight in the reactivity of the lignin polymer itself. The models can thus be used to mimic the chemistry of typical lignin fragments and hold the additional benefit of greatly simplifying product analysis. In general, feeds derived from lignin depolymerization, as those obtained in the processes studied in Chapters 3 and 5, will likely consist of a mixture of aromatics, the constituents of this mixture can be mimicked by proper selection of a library of model compounds. [9] Moreover, reports on the hydrodeoxygenation of these lignin model compounds also bear relevance to the production and upgrading of bio-oils, as they can consist of up to 30-40% of oxygen-functionalized aromatics. Chemical processing of bio-oil, obtained for instance from a biomass pyrolysis process, is necessary to convert the bio-crude into useful transportation fuels. Stabilization of these bio-crudes can be achieved by removal of the oxygen functionalities via hydrodeoxygenation. [8, 10] The traditional hydrotreating catalysts originally developed for the removal of sulfur and nitrogen from oil feeds could also be deployed for the removal of oxygen from biomass-derived product streams. Indeed, many of the studies on HDO of bio-oil or lignin-related components employ conventional cobalt- or nickel-promoted molybdenum catalysts.


151

[6, 7] An early study by Elliott, screening different commercial catalysts for their activity in phenol HDO, already showed that a sulfided CoMo catalyst gave the best results. [11] Because of their lower intrinsic hydrogenation activity, CoMo/Al2O3

catalyst are often preferred over the nickel analogues, as the aromaticity of the feeds should preferably be kept intact. This is in contrast to deep hydrodesulfurization or hydrodenitrogenation reactions, in which the heteroatom is commonly removed after ring saturation.

Several HDO studies using CoMo/Al2O3 catalysts were performed on simple aromatic model compounds, such as phenol, cresol, anisole and guaiacol, which mimic parts of the lignin structure and are important bio-oil constituents. Products originating from deoxygenation, (de)methylation and hydrogenation reactions were reported. [12-15] Hurff and Klein, for instance, observed the formation of catechol and phenol as main products for the conversion of guaiacol over a sulfided CoMo/Al2O3 catalyst. Formation of products in time confirmed that catechol is a primary product, which further reacts to phenol. [16] Several concise reaction schemes for the conversion of guaiacol and methylguaiacol were subsequently reported without information on selectivity. [16, 17] Recently, Bui et al. reported on the promoting effect of cobalt on molybdenum sulfide catalysts and published a more comprehensive reaction scheme for guaiacol conversion also including the formation of heavy products. [18] The addition of cobalt was found to enhance the direct demethoxylation and direct deoxygenation pathways. Bredenberg et al. observed that the carbon-oxygen bond in phenol and cresols is more difficult to break than other carbon-oxygen bonds and that complete deoxygenation could therefore not be reached. [13] The role of the support material in HDO has also been investigated. The acid sites on alumina are thought to participate in the conversion of guaiacol, both in the formation of heavier products and steps that lead to ring methylated products. [19-21] The effect of support acidity on the performance of the catalyst has also been studied by addition of pyridine as selective poison. [19] Laurent et al. have studied the inhibition of CoMo/Al2O3 and NiMo/Al2O3 catalysts under HDO conditions with water, ammonia and H2S, concluding that oxygen-elimination reactions are inhibited by sulfur- and nitrogen-containing compounds. [22-24] Senol et al. reported the conversion of phenol in the presence of H2S and also concluded that both the formation of benzene and cyclohexene are inhibited. [25] Recently, a renewed interest in HDO reactions has resulted in an increasing amount of literature available on the HDO conversion of lignin-derived streams. New approaches include co-feeding aromatic lignin-derived compounds such as guaiacol with straight run gas oil for the direct addition of oxygenates to fuel

Part II, Chapter 6

152

streams [26, 27] and mixed HDO of aromatic and aliphatic oxygenates.

Furthermore, when using depolymerized lignin streams it is likely that in addition to monoaromatic compounds also various dimers and small oligomers are present in the reaction feed, necessitating the study of fragments that mimic specific structural linkages found in lignin under HDO conditions. Romero et al. [28] and Edelman et al., [29] for instance, showed that benzofuran could be hydrogenated to dihydrobenzofuran, which was further converted to ethylphenol. Petrocelli et al. researched the reactivity of lignin structural linkages by using biphenolic model compounds in their HDO experiments, which were shown to be stable under HDO conditions. [30]

Valuable insights were obtained from these studies, but direct comparison of the reported data is hampered by the varying experimental conditions under which they have been obtained. Indeed, a systematic study of the conversion of a library of lignin model compounds with one batch of catalyst under comparable reaction conditions and in the same reactor system is still lacking. Here, we report on the conversion of lignin model compounds with the aim of obtaining bulk aromatics with lower oxygen

Figure 6.1: Monomeric model compounds phenol (1), o-cresol (2), p-cresol (3), anisole (4), 4-methylanisole (5), catechol (6), guaiacol (7), 4-methylguaiacol (8), 1,3-dimethoxybenzene (9), syringol (10), and vanillin (11).


153

Figure 6.2: Dimeric model compounds mimicking the β-O-4 (12), 5-5 (13), and phenylcoumaran (14) linkages typically found in lignin.

content. Several hydroxy- and methoxy-substituted monomeric aromatic model compounds (Figure 6.1) were used to conduct catalytic studies with a sulfided CoMo/Al2O3 catalyst using a single set of reaction conditions. For the first time mono-, di- and tri-oxygenated compounds as well as both methylated and non-methylated substrates are used to provide more insight in the reaction pathways responsible for product formation. A reaction network is presented to show the major reaction pathways and reaction selectivities in more detail. In addition to conversion of mono-aromatic model compounds, the effect of the catalyst system on linkages typically present in the lignin polymer was also studied. The model compounds shown in Figure 6.2 were chosen to represent the β-O-4 (12) and phenylcoumaran (14) ether linkages, as well as the 5-5 aryl-aryl linkage (13), which all can be found in the lignin structure.


6.2.1 Conversion of monomeric model compoundsThe conversion of the monomeric lignin model compounds 1-11 by a sulfided CoMo/

Al2O3 catalyst was investigated under HDO conditions. General reaction conditions were 300 °C and 50 bar H2 pressure. The catalytic results are reported in Table 6.1 and some general observations can be made based on the presented data. Conversions are found to differ widely after 4 h of reaction as only 20% of the substrate is converted for mono-oxygenated aromatics, whereas conversions of up to 90% are observed for some of the higher functionalized molecules, e.g. methylguaiacol and syringol, with mass balances that are closed from 56% for syringol to 96% for anisole. GC analysis of the liquid reaction product revealed complex reaction mixtures in which the main

Part II, Chapter 6

154

products vary depending on the reactant. High selectivities were obtained for benzene from phenol (1), phenol from anisole (4), and phenol and catechol from guaiacol (7). Methylated substrates, such as cresols (2, 3) and methylguaiacol (8), gave rise to methylated products. Furthermore, o- and p-cresol (2, 3) were also found in reactions with non ring-methylated substrates, e.g. selectivities up to 5% were obtained in the conversion of guaiacol. Unlike in the work of Gutierrez et al. where benzene was found to be the main product obtained from guaiacol, completely deoxygenated products are not observed in large quantities. [31]

In addition to p- and o-cresol, di- and trimethylated phenolics are also formed, as well as other methylated products, such as 4-methylguaiacol and toluene. Importantly, no quantifiable amounts of m-cresol were found, not even in the conversion of 4-methylguaiacol (8). For most reactions, mono-oxygenated products, such as phenol, make up the major part of the reaction mixture. In the case of syringol (10), however, also several di- and tri-oxygenated compounds were observed. The presence of small amounts of benzene and toluene in all reaction mixtures suggests that full deoxygenation does take place, albeit with low overall selectivity. The conversion of anisole and guaiacol, for instance, only yielded minor amounts of benzene, illustrating that it is not a major product in these reactions. The formation of mainly phenolics suggests that demethylation and deoxygenation are the main reaction pathways together with ring methylation (see Scheme 6.1 in which guaiacol is shown as a typical example).

Ring hydrogenation products, such as cyclohexanol, cyclohexene and methylcyclohexene, are only detected in low quantities with the total amount adding up to no more than 5%. In comparison, Hurff and Klein already found 10% cyclohexane after 4 h in the conversion of guaiacol with a sulfided CoMo/Al2O3 catalyst in the

Scheme 6.1: Reaction pathways in the conversion of guaiacol: a) formation of phenol by HDO i) followed by methylation to o-cresol iii), b) formation of catechol by demethylation ii), followed by formation of phenol by HDO iv) and methylation to o-cresol iii).


155

presence of sulfiding agent. [16] Bredenberg et al. also observed only small amounts of around 1% hydrogenation products in the conversion of guaiacol and anisole over a sulfided NiMo/Al2O3 catalyst in a continuous flow process; [13] nevertheless, over 5% ring hydrogenation was observed when a sulfiding agent was added to the feed. [15] Bui et al. reported around 4% hydrogenation products and found that the amount of hydrogenation products increased when a non-promoted MoS2/Al2O3 catalyst was used. [18] For some reactions not all products detected by GC could be unambiguously identified with either authentic samples or by GC-MS analysis. These products added together, however, do not represent more than a total selectivity of 4%. An overview of the most common reaction products is given in Table 6.1.

6.2.2 Formation of heavy productsIntermolecular coupling between guaiacol or catechol-like molecules at higher

temperatures can cause oligomerization, leading to higher molecular weight products. [17, 18] When formed, these products cannot be detected by our GC analysis method resulting in mass balances that can be closed for 80% to 90% in most reactions. In general, we observe that substrates with more oxygen functional groups give a lower mass balance (e.g. syringol (5) 57% compared to guaiacol (7) 76%). The carbon balances of around 80% for di-oxygenated compounds such as guaiacol, around 90% for phenol and anisole and 95% for methylated phenols are comparable to those previously reported for similar substrates. [15-17, 24] The remainder being equal to the amount of heavy products observed by Bui et al. by GC-MS after gas phase HDO of guaiacol. [18, 21] In some cases only the product distribution of the isolated products is published, leaving the products that were not isolated out of the mass balance. [12, 32] Petrocelli and Klein found that the mass balance decreased with increasing reaction time, [30] also indicative of an increase in formation of heavy products that cannot be detected and identified by the typical GC methods used. In our case, ESI-MS analysis of the reaction mixture indeed revealed some peaks of higher molecular weight in the product mixtures, but these products could not be identified. As blank reactions for guaiacol and anisole, run without catalyst, show a closed mass balance, it is clear that the side reactions that cause oligomerization of the substrate are also catalyzed by the sulfided CoMo/Al2O3 catalyst. Coke formation on the catalyst to such an extent that it explains the observed weight loss, on the other hand, can be excluded. The catalyst sample shows no significant change in weight after reaction, whereas the missing weight should have increased the catalyst mass by 300%.

Thermogravimetric analysis (TGA) of the spent catalyst was carried out to

Part II, Chapter 6

156

Conv

ersi

onBa

lanc

eM

ethy

late

d Pr

oduc

ts b

Hyd

roge

nate

d Pr

oduc

ts c

X a

Phen

ol27

8637

x-

--

-4

6

Anis

ole

5896

1064

x-

-10

3-

Cate

chol

8056

136

-x

-<1

44

Guai

acol

8476

<134

311

x19

3-

1,3-

Dim

etho

xy78

60<1

32

--

21

40

benz

ene

Syri

ngol

8357

<11

<13

56

126

Conv

ersi

onBa

lanc

eM

ethy

late

d pr

oduc

tsH

ydro

gena

ted

prod

ucts

Xa

o-cr

esol

2395

65-

--

--

312

p-Cr

esol

2291

60<1

--

-1

6-

4-M

ethy

lani

sole

8585

1347

x-

-13

22

4-M

ethy

lgua

iaco

l89

912

42-

13*

x23

-4

Vani

llin

5375

-3

-9

23-

-17

Sele

ctiv

itie

s are

dis

play

ed a

s a p

erce

ntag

e of

the

conv

ersi

on in

mol

es. a

) X in

dica

tes t

he to

tal s

um o

f all

othe

r pea

ks in

the

chro

mat

ogra

m in

clud

ing:

di

met

hoxy

phen

ol, m

etho

xyca

tech

ol, r

esor

cino

l, di

met

hoxy

benz

ene,

met

hylm

etho

xyph

enol

, tri

met

hoxy

benz

ene

and

1,4-

benz

ened

iol.

b) M

ethy

late

d pr

oduc

ts in

clud

e: to

luen

e, p

- and

o-c

reso

l, di

- and

trim

ethy

late

d ph

enol

s, 3-

and

4-m

ethy

lgua

iaco

l, 4-

met

hylc

atec

hol a

nd m

ethy

lmet

hoxy

benz

ene.

For

m

ethy

late

d su

bstr

ates

onl

y th

ose

prod

ucts

wit

h m

ore

than

one

met

hyl f

unct

iona

lity

are

incl

uded

. c) H

ydro

gena

ted

prod

ucts

incl

ude:

cyc

lohe

xene

, cy

cloh

exan

ol, m

ethy

lcyc

lohe

xane

and

met

hylc

yclo

hexe

ne. *

both

4-m

ethy

lgua

iaco

l and

5-m

ethy

lgua

iaco

l are

foun

d in

a 1

:1 ra

tio.

Tabl

e 6.

1: H

DO

of d

iffer

ent l

igni

n m

odel

com

poun

ds o

ver a

sul

fided

CoM

o/Al

2O 3 cat

alys

t at 3

00 °C

and

50

bar H

2 pre

ssur

e fo

r 4 h

.


157

Conv

ersi

onBa

lanc

eM

ethy

late

d Pr

oduc

ts b

Hyd

roge

nate

d Pr

oduc

ts c

X a

Phen

ol27

8637

x-

--

-4

6

Anis

ole

5896

1064

x-

-10

3-

Cate

chol

8056

136

-x

-<1

44

Guai

acol

8476

<134

311

x19

3-

1,3-

Dim

etho

xy78

60<1

32

--

21

40

benz

ene

Syri

ngol

8357

<11

<13

56

126

Conv

ersi

onBa

lanc

eM

ethy

late

d pr

oduc

tsH

ydro

gena

ted

prod

ucts

Xa

o-cr

esol

2395

65-

--

--

312

p-Cr

esol

2291

60<1

--

-1

6-

4-M

ethy

lani

sole

8585

1347

x-

-13

22

4-M

ethy

lgua

iaco

l89

912

42-

13*

x23

-4

Vani

llin

5375

-3

-9

23-

-17

Sele

ctiv

itie

s are

dis

play

ed a

s a p

erce

ntag

e of

the

conv

ersi

on in

mol

es. a

) X in

dica

tes t

he to

tal s

um o

f all

othe

r pea

ks in

the

chro

mat

ogra

m in

clud

ing:

di

met

hoxy

phen

ol, m

etho

xyca

tech

ol, r

esor

cino

l, di

met

hoxy

benz

ene,

met

hylm

etho

xyph

enol

, tri

met

hoxy

benz

ene

and

1,4-

benz

ened

iol.

b) M

ethy

late

d pr

oduc

ts in

clud

e: to

luen

e, p

- and

o-c

reso

l, di

- and

trim

ethy

late

d ph

enol

s, 3-

and

4-m

ethy

lgua

iaco

l, 4-

met

hylc

atec

hol a

nd m

ethy

lmet

hoxy

benz

ene.

For

m

ethy

late

d su

bstr

ates

onl

y th

ose

prod

ucts

wit

h m

ore

than

one

met

hyl f

unct

iona

lity

are

incl

uded

. c) H

ydro

gena

ted

prod

ucts

incl

ude:

cyc

lohe

xene

, cy

cloh

exan

ol, m

ethy

lcyc

lohe

xane

and

met

hylc

yclo

hexe

ne. *

both

4-m

ethy

lgua

iaco

l and

5-m

ethy

lgua

iaco

l are

foun

d in

a 1

:1 ra

tio.

estimate the amount of coke deposited after the reaction with guaiacol. The TGA data show a small increase in the formation of CO2 with respect to the fresh catalyst and a total weight loss of 17%. Judging on the major weight loss at 440 °C, most of the coke formed on the catalyst is so-called hard coke, which consists of the more heavy and highly aromatic compounds. [33]

6.2.3 Catalyst deactivationThe use of a traditional HDS catalyst on a feedstock that does not contain sulfur

can lead to leaching of sulfur from the active site into solution and to loss of catalytic activity. NO adsorption on MoS2 sites is often used to determine the amount of active sites in CoMoS/Al2O3 catalysts, as the amount of NO adsorption can be quantified by IR spectroscopy and can be directly correlated to the amount of active sites. NO adsorption studies were performed on a fresh and a spent catalyst sample to determine the loss of active sites during reaction. The NO IR spectra of the freshly activated and the spent catalyst are shown in Figure 6.3. The NO IR spectrum for the fresh catalysts compares well to previously reported data. [34] The NO bands for the spent catalyst decreased significantly compared to the same bands in the freshly activated catalyst. According to literature, NO adsorption can take place either on coordinatively unsaturated cations at the edge sites of sulfide particles [35] or via a ‘push pull’-type mechanism, which releases one of the S-edge atoms in the form of H2S. [34] Either way, the NO adsorption correlates directly to the amount of HDS

150016001700180019002000

Spent Catalyst

Activated Catalyst

Wavenumber (cm )

-1

-1

Figure 6.3: IR spectra after NO adsorption on a freshly activated as well as spent CoMo/Al2O3 catalyst.

Part II, Chapter 6

158

active sites on the MoS2 phase. The decrease in NO adsorption on the spent catalyst compared to the fresh catalyst indicates that structural changes have taken place on the edges of the catalyst, sulfur atoms having been possibly replaced by oxygen ones, thus decreasing the amount of vacant sites NO can adsorb on. Catalytic experiments using the spent catalyst without further treatment for 4 consecutive runs (data not shown), however, show similar HDO activity in the 1st and 2nd run and an overall decrease in the amount of HDO products of only 40% in the 4th run. The dramatic deactivation that one would expect from the NO adsorption data is not observed. Even after the 4th run, the catalyst still exhibits, albeit reduced, HDO activity. The decrease in NO adsorption can therefore not be directly translated in a loss of HDO active sites.

6.2.4 Reaction networkThe formation of several different products in each reaction suggests that HDO and

O-demethylation reactions, indicated in Scheme 6.1, as well as hydrogenation reactions are taking place simultaneously. Depending on the substrate used, selectivity for one route may decline in favor of another. Nonetheless, the conversion of the different model compounds leads towards the formation of phenolic products via a combination of demethylation/dehydroxygenation steps. As previously observed, demethylation and methylation reactions are prone to take place on the alumina support. [19-21] Ring hydrogenation products are observed, but no oxygenated cyclohexanes are found. This implies that the demethylation and HDO reactions take place in an earlier stage of the reaction and that the reaction can be stopped at the point where HDO is completed before large amounts of ring hydrogenation products start to form.

Notable is the difference in reactivity of the compounds guaiacol (7), anisole (4) and phenol (1) compared to their ring-methylated derivatives (8, 5, 2 and 3). As depicted in Figure 6.4, conversion, mass balance and selectivity towards deoxygenation products of the methylated substrates are slightly higher than for the non-methylated model compounds. The difference in HDO selectivity between phenol and both cresols is even more than 20%. Also, less ring methylation products are found for the methylated substrates. The presence of a methyl group on the ring, therefore, most likely makes addition of a second one sterically less favorable. The relative increase in HDO activity of the methylated compounds can also probably be attributed to steric factors, as a methyl group makes the aromatic ring more difficult to approach thus limiting alternative reactions, such as oligomerization and hydrogenation. It was previously noted that in order for hydrogenation to take place, the molecule must adsorb flat on the catalyst surface whereas deoxygenation takes place by η1 coordination of the oxygen on the edge of the MoS2 phase. [15, 36, 37] Furthermore, in order for ring


159

methylation reactions to take place, the aromatic ring must also approach the catalyst surface and forced non-planar coordination might hamper ring methylation reactions.

As most aromatic rings in the original lignin structure as well as in lignin-derived depolymerization feeds carry alkyl functionalities, the increased selectivity towards HDO observed for alkyl-substituted model compounds might be useful when applied to lignin itself.

Several routes can be envisaged for the conversion of guaiacol to phenol and ultimately benzene. (i) a direct deoxygenation pathway (DDO) where phenol is formed via loss of the methoxy functionality; (ii) demethylation of the methoxy group followed by HDO. The obtained phenol is then further converted to benzene. Importantly, catechol (6) is observed as the main reaction intermediate in the conversion of guaiacol (vide infra), suggesting (ii) to be the dominant pathway. Experiments with catechol as the substrate, also show that it readily reacts further via HDO to phenol (see Table 6.1). It is not clear, however, if the reaction from guaiacol to phenol proceeds exclusively via catechol by means of consecutive demethylation and HDO (ii) or if DDO (i) also takes place. Although minor amounts of methanol are clearly observed by GC for some of the reactions, it was not possible to reliably quantify the amount of methanol formed by demethoxylation. Overall, these results thus indicate that the demethylation pathway

0102030405060708090

100

Conversion CarbonBalance

Demethylation Selectivity

HDOSelectivity

(%)

OHOMe

OHOMe

OH OH

OH OMe

OMe OHOMe

OHOMe

OH OH

OH OMe

OMe OHOMe

OHOMe

OH OH

OH OMe

OMe OHOMe

OHOMe

OH OH

OH OMe

OMe

Figure 6.4: Conversion, mass balance, demethylation and HDO selectivity for ring-methylated model compounds 2,3,5, and 8 compared to non-methylated analogues 1, 4, and 7 over a sulfided CoMo/Al2O3 catalyst.

Part II, Chapter 6

160

(ii) is the major pathway in the conversion of guaiacol, with demethoxylation (i) most likely being only a minor route. This might be expected as a result of the higher bond strength of the MeO-Ar compared to the ArO-Me bond. Indeed, the bond dissociation energy for RO-Ar bonds (422 kJ/mol) is higher than for RO-R bonds (339 kJ/mol). [7] The absence of large quantities of anisole combined with the lower reactivity of anisole compared to catechol under HDO conditions further supports that the major reaction pathway from guaiacol to phenol goes via catechol. There is no evidence, however, to fully exclude demethoxylation (ii) as one of the many reaction routes that take place simultaneously. These results are in agreement with Laurent et al. [17] and Bredenberg et al., [15] who also concluded that demethylation to form catechol is the first step in guaiacol deoxygenation. In contrast, in a general reaction network proposed by Petrocelli et al., [30] it was concluded that demethylation was only a minor pathway.

Similarly, it is difficult to distinguish between the two possible routes for catechol formation from guaiacol, as both demethylation as well as hydrogenolysis of the O-Me bond can be envisaged. The observation of various ring-methylated structures in addition to some activity observed for the support itself (vide infra) point to the former route. Nonetheless, gas phase analysis of a guaiacol conversion reaction shows some formation of methane, showing that hydrogenolysis also takes place, at least to a minor extent.

Interestingly, conversion of 4-methylguaiacol (8) gives methylated catechol (17) and p-cresol (3) as the main reaction products. As p-cresol is formed selectively and no m-cresol is observed, the HDO step must be selective for the oxygen in the 2-position. Reactions with m-cresol show no formation of p-cresol, so fast isomerization of this product can be excluded and p-cresol is the only cresol formed via HDO of methylated aromatics (Scheme 6.4). This is not in line with the results obtained by Bui et al. who reported the formation of m-cresol in the conversion of guaiacol over pure γ-alumina. [21] Huuska et al. [19] have previously reported a preference for ring methylation at the o-position in the conversion of anisole over a NiMoS/Al2O3 catalyst in a continuous flow system, claiming this as evidence for a nonplanar orientation of the phenoxide ion. However, in the conversion of non-methylated substrates we observe a slight excess in p-cresol over o-cresol, again no m-cresol is found (Scheme 6.2).

No evidence was found for ring demethylation reactions at equal or comparable rates to O-demethylation reactions. Catalytic tests on methylated model compounds,


161

4-methylguaiacol, 4-methylanisole and the cresols, show no formation of ring-demethylated products. Small amounts of phenol that are found in cresol reaction mixtures can be attributed to impurities in the starting material. Reactions with 4-methylguaiacol, however, show evidence for reversible O-methylation, as equal amounts of 4-methylguaiacol and 5-methylguaiacol are found.

The use of both methylated and non-methylated substrates provided more insight in the mechanisms responsible for the formation of the observed products. Scheme 6.3 shows a reaction network for the reductive conversion of methoxy- and hydroxy-functionalized mono-aromatics over sulfided CoMo/Al2O3. The reaction pathways presented in this network are based on the experiments discussed above and those listed in Table 6.1. For the sake of simplicity, not all minor direct demethoxylation routes are shown in the proposed reaction network. However, when the demethylated intermediate was not identified, only direct demethoxylation routes are included, for example, in going from syringol (10) to guaiacol (7).

In summary, distinction is made in the reaction network between three types of reactions that occur simultaneously. The arrow size indicates reaction selectivity for this path, but does not contain information on conversion. All tested mono-aromatic model compounds follow the same three pathways that eventually lead to the formation of phenol and benzene or cresols and toluene.

Scheme 6.2: Formation of o- and p-cresol in the sulfided CoMo/Al2O3-catalyzed conversion of ring-methylated and non-methylated monoaromatic model compounds. No m-cresol was observed for any of the substrates.

Part II, Chapter 6

162

O-demethylation is the fastest pathway; the removal of methyl groups from methoxy functionalities is more favorable than direct demethoxylation. In the conversion of anisole, for instance, phenol is observed as the main product, whereas only small amounts of benzene were formed. O-demethylation goes hand in hand with methylation of the aromatic ring. Indeed, aromatic alkylations are observed in most reaction mixtures for which the p- and o-positions are strongly favored over the m-position. It is likely that also transfer of the methyl groups between hydroxy functional groups occurs, as is indicated by the isomerization of 4-methylguaiacol to 5-methylguaiacol. Methylation and demethylation activity are often attributed to support acidity. [20, 21]

Hydrodeoxygenation is occurring at a slower rate than O-demethylation. Both methoxy and hydroxy groups can be removed via HDO although demethoxylation is much slower. Conversion of catechol has a selectivity of 36% to monooxygenated products, whereas dimethoxybenzene produces single oxygen products with a selectivity of around only 5% (note that these can also be formed by subsequent demethylation and deoxygenation). Substrates containing only one oxygen atom, such as phenol and cresols (conversion ~ 20%), appear to be much more stable than higher functionalized aromatics, such as catechol and guaiacol (conversion ~ 80%).

Hydrogenation activity has been commonly observed with HDO catalysts. Only very small amounts of hydrogenation products are observed after 4 h, however, under the conditions applied here (< 5%). Hydrogenated products mainly consist of cyclohexene and cyclohexanol, but in some cases also small amounts of methylated hydrogenation products are found. The low amounts of these products in the reaction mixture indicate that hydrogenation is the least favorable pathway for this system.

Although reactions starting from phenol and cresol show a high selectivity towards HDO products benzene and toluene, activity is rather low, as total phenol and cresol conversions are under 30%. This indicates that phenol itself is quite stable under the applied reaction conditions. From a lignin valorization point of view, phenol and cresols constitute attractive targets for this type of conversions. It is therefore important to note, that starting from every position in the reaction network, phenol and cresols can indeed be obtained as the main reaction products under these process conditions.


163

O

OHOH

Me

OOH

OOH

OHOO

OOHO

OOO

OH

OOH

O

OHOH

>35% selectivity

10-35% selectivity<10% selectivity

OHOH

O

HDO

demethylation

methylation

12/3

4

6

78 9

1011

18

20

5

17

15 16

OHOHO

23OHO

22OHHO

21 OOO

24

OOH

19

Scheme 6.3: Reaction pathways in the HDO of lignin model compounds over a sulfided CoMo/Al2O3 catalyst. Ring-hydrogenation is very limited and these pathways are not included; arrow size contains no information on conversion.

Part II, Chapter 6

164

6.2.5 Conversion of dimeric model compoundsDuring plant growth, the lignin polymer is generally believed to be formed via radical

polymerization of three phenylpropane-based monomeric units, i.e. p-coumaryl alcohol, coniferyl alcohol and sinapyl alcohol. [6] As a result of the radical nature of lignin synthesis, many different linkages can be found in the resulting biopolymer. The most common of these linkages are the so-called β-O-4, 5-5 and phenylcoumaran ones, which are present in different quantities in native lignin depending on the plant species and in processed lignin depending on the pretreatment method. Figure 6.5 shows a typical softwood lignin fragment with some linkages highlighted. The β-O-4 linkage is by far the most abundant and model compound 12 (Figure 6.2) was synthesized to investigate the reactivity of this linkage under HDO conditions. The compounds 2,2’-biphenol (13) and coumaran (14) represent some key features of the 5-5 and the phenylcoumaran-type linkages encountered in lignin and are commercially available.

Figure 6.5: Schematic representation of a softwood lignin structure, highlighting examples of the β-O-4, phenylcoumaran and 5-5 linkages.


165

The results of the HDO reaction of the β-O-4 lignin model compound 12 are shown in Table 6.2 and Scheme 6.4 1H NMR analysis of the reaction mixture showed complete conversion of the substrate. Only mono-aromatic products could be identified by GC analysis with a total amount of 33% of aromatics recovered after 4 h of reaction; the presence of these monoaromatic compounds confirms cleavage of the β-O-4 bond under HDO conditions. Phenol, guaiacol and syringol-like products are mainly obtained. No ethylbenzene derivatives could, however, be identified which indicates that all observed products likely originate from the syringol part of 12.

2,2’-Biphenol (13) was used as a simple analog of the 5-5 lignin carbon-carbon bond. The results are shown in Table 6.3 and Scheme 6.5. 2,2’-Biphenol is not soluble in the reaction mixture at room temperature, leading to some losses of the remaining starting material during workup. This can in part account for the relatively low mass balance found in this reaction. Nonetheless, all detected products still contain the

Conversion (%) 100Mass balance (%) 33Phenol (%) 9o-Cresol (%) 2p-Cresol (%) 2Dimethylphenol (%) 5Catechol (%) 1Resorcinol (%) 2Guaiacol (%) 6Methoxyphenol (%) 2Syringol (%) 4

Table 6.2: HDO conversion and selectivities in mol% of β-O-4 lignin model compound 12 over a sulfided CoMo/Al2O3 catalyst at 300 °C and 50 bar H2 pressure for 4 h.

Scheme 6.4: Conversion of the β-O-4 dimer 12.

Part II, Chapter 6

166

aryl-aryl linkage. The recalcitrance of the 5-5 bond and its insensitivity towards chemical reactions is not unexpected, as previous research on treatments for lignin-derived molecules always report the 5-5 link to remain intact. [30] In line with the observations made for catechol and other bis-oxygen functionalized substrates, only one hydroxyl group is removed by HDO activity leading to a mono-oxygenated product. Possibly, at least one oxygen atom is needed for positioning of the molecule to the catalyst surface so that the other can react. Interestingly, large amounts of dibenzofuran, are found in the reaction mixture. Dibenzofuran can be formed via an acid-catalyzed dehydration reaction of the 2,2’-biphenol in which the ring closing step is an intramolecular nucleophilic attack of one of the hydroxyl groups. [38, 39] This indicates that dibenzofuran and 2-hydroxybiphenyl are formed from 2,2’-biphenol via separate pathways and that dibenzofuran formation competes with HDO.

Dihydrobenzofuran or coumaran (14) is a common probe molecule for HDO reactions [28] and can serve as a model compound that resembles lignin β-5 linkages.

Conversion (%) 55Mass balance (%) 79Biphenyl (%) 22-Hydroxybiphenyl (%) 22Dibenzofuran (%) 34Tetrahydrodibenzofuran (%) 4

Scheme 6.5: Conversion of 2,2-biphenol (13) to dibenzofuran, 2-hydroxybiphenyl and biphenyl.

Table 6.3: HDO conversion and selectivities in mol% of 2,2-biphenol (13) over a sulfided CoMo/Al2O3 catalyst at 300 °C and 50 bar H2 pressure for 4 h.


167

After reacting over a sulfided CoMo/Al2O3 catalyst for 4 h at 300 °C and 50 bar hydrogen pressure, 18% of the substrate was converted. As can be seen in Table 6.4 and Scheme 6.6, high selectivity towards 2-ethylphenol is obtained. This is in line with the results obtained by Romero et al., [28] who found 2-ethylphenol to be the main product and ethylbenzene the main deoxygenated product after HDO over a NiMoP/Al2O3 catalyst using a fixed-bed reactor. They contributed the low conversions to an inhibitive effect of coumaran itself on the HDO activity of the catalyst. The obtained results do indicate that, albeit at low conversion, it is possible to cleave the alkyl-oxygen bond in β-5 linkages with a sulfided CoMo/Al2O3 catalyst. Cleavage of this ether bond is a promising result leading to weakening of the lignin structure, also with respect to other ether linkages present in lignin, such as β-β or dibenzodioxocin linkages.

6.3 Conclusions

The hydrodeoxygenation of mono-aromatic lignin model compounds on a sulfided CoMo/Al2O3 catalyst at 300 °C and 50 bar H2 pressure gives products with lowered oxygen content as well as demethylated and ring-methylated products. An extended reaction network shows that the reactions take place via demethylation of methoxy groups followed by deoxygenation. Mono-oxygenated compounds, such as phenol, are very stable under the applied conditions and almost no completely deoxygenated products are observed. Together with the limited hydrogenation activity, this results in the formation of phenol and cresols in high selectivities for most starting materials. Conversion of dimeric model compounds that mimic lignin linkages shows that both the β-O-4 and β-5 ether linkages can be broken under HDO conditions. The rigidity of the 5-5’ linkage prevents cleavage of the 5-5 model compound 2,2’-biphenol.

Conversion (%) 18Mass balance (%) 97Ethylbenzene (%) 3Phenol (%) 62-Ethylphenol (%) 65

Table 6.4: HDO conversion and selectivities in mol% of coumaran (14) over sulfided CoMo/Al2O3 catalyst at 300 °C and 50 bar H2 pressure for 4 h.

Scheme 6.6: Conversion of coumaran (14).

Part II, Chapter 6

168


6.4.1 Chemicals and catalystAll model compounds, additives and solvents were used as received: phenol,

o-cresol, p-cresol, anisole (Fluka, 99%), 4-methylanisole (Sigma-Aldrich, 99%), catechol (Acros, 99%), guaiacol (Sigma-Aldrich, 99%), 4-methylguaiacol (Sigma-Aldrich, 99%), 1,3-dimethylbenzene (Sigma-Aldrich, >98%), syringol (Sigma-Aldrich, 99%), vanillin (Sigma-Aldrich, 99%), coumaran (Sigma-Aldrich, 99%), 2,2’-biphenol (Sigma-Aldrich 99%), hexadecane (Sigma-Aldrich, 99%), pyridine (Acros, 99%), dodecane (Acros, 99%), diethylether (Biosolve, solvent grade), methanol (Biosolve, solvent grade). Gases (hydrogen, 5.0 and argon, 5.0) were obtained from Hoekloos.

The β-O-4 lignin model compound 12, i.e. 1-hydroxy-1-(4-hydroxy-3-methoxyphenyl)-2-(2,6-dimethoxyphenoxy)ethane, was synthesized according to a literature procedure. [40] The product was purified using a silica gel column with a hexanes and ethyl acetate mixture (1:3) as eluent, followed by repetitive crystallization. The purity of the compound was confirmed by 1H NMR.

The CoMo/Al2O3 catalyst obtained from Albemarle Catalysts was presulfided. Prior to its use, the catalyst material was crushed and sieved to particles of size between 150 and 450 μm that were activated under 60 mL/min H2 flow at 300 °C for 3 h. The activation method is similar to the industrially used activation procedure for this particular catalyst. Addition of H2S or other sulfiding agents to the H2 flow was not needed as a sulfiding agent was already present in the sample. To ensure that sulfur-loss was limited under conditions of a catalytic run, elemental analysis of the fresh and spent catalyst was performed at the Mikroanalytisches Laboratorium Kolbe. No loss of sulfur was found, but the elemental analysis data rather showed a marginal increase in sulfur content of the catalyst after reaction in absolute amounts (13% increase relative to Al). NO adsorption experiments were also performed on the fresh and spent catalyst by IR spectroscopy. The catalyst sample was mixed 1/1 with SiO2 powder and pressed to form a self-supporting wafer with a weight of 12 mg. The sample was placed in an in-situ IR transmission cell, evacuated under vacuum and dried at 300 ˚C for 1 h after which it was cooled to 50 ˚C. NO adsorption was carried out by letting 250 mbar of NO (1% in N2) in the cell untill after 30 min equilibrium was reached. Excess NO was removed by applying a vacuum and after 15 min IR spectra were obtained to compare the amounts of adsorbed NO in the fresh and spent catalyst.


169

6.4.2 Catalytic reactionsThe catalytic reactions were performed in a 100 mL stainless steel high pressure

Parr batch autoclave reactor. The temperature was monitored using a thermocouple and stirring was performed using a magnetic driver equipped with an impellor at 750 rpm. In a typical reaction, the autoclave was loaded with 150 mg of the catalyst material, 1.5 g substrate (typically guaiacol), 0.5 g hexadecane as internal standard and 30 g of the solvent dodecane. A fresh batch of catalyst was used in every run. The reactor was purged three times with argon and the reaction mixture was heated to 300 °C, unless stated otherwise, then pressurized with H2 to 50 bar and the catalytic reaction was carried out for 4 h. After the reaction was stopped by cooling and subsequent release of pressure, the reaction mixture was diluted with an equal volume of diethyl ether in order to dissolve all of the products, the solid catalyst was removed by filtration and washed with diethyl ether.

6.4.3 AnalysisThe reaction products were analyzed using a Shimadu GC-2010A gas chromatograph

equipped with a WCOT fused silica CP-WAX 57-CB column and FID detector, and a Varian 430-GC gas chromatograph equipped with a VF-5ms capillary column and a FID detector. Compounds were identified with gas chromatography coupled with a mass spectrometer using a Shimadzu QP2010 GCMS instrument equipped with a VF-5ms capillary column and compared with retention times of pure standards. Response factors relative to hexadecane were determined experimentally when authentic samples were available. The β-O-4 lignin model compound was silylated with N,O-bis(trimethylsilyl)acetamide before GC analysis. Gas phase products of were analyzed on an Interscience compact GC equipped with a molsieve 5A column and a TCD detector. 1H NMR analysis was performed on a Varian 300 Inova MHz spectrometer. Thermal gravimetric analysis (TGA) was performed with a Perkin-Elmer Pyris 1 apparatus. Typically 15 mg of impregnated silica gel was heated with a ramp of 5 °C min-1 to 600 °C in a 10 mL min-1 flow of air. Gas analysis was performed, in parallel to the TGA measurements, with a quadrupole Pfeiffer Omnistar mass spectrometer, which was connected to the outlet of the TGA apparatus. Ion currents were recorded for m/z values of 28 and 32.

Part II, Chapter 6

170

6.5 Acknowledgments

Albemarle Catalysts is kindly acknowledged for providing the CoMo/Al2O3 catalyst.Robin Jastrzebski is thanked for his contributions to the experimental work on which this chapter is based.

6.6 References

[1] R. J. A. Gosselink, E. de Jong, B. Guran, A. Abächerli, Ind. Crops Prod. 2004, 20, 121-129.[2] V. M. Roberts, V. Stein, T. Reiner, A. Lemonidou , X. Li, J. A. Lercher, Chem. Eur. J. 2011, 17, 5939-5948.[3] J. S. Shabtai, W.W . Zmierczak, E. Chornet, 1999, US Patent 5959167.[4] K. Stark, N. Taccardi, A. Bosmann, P. Wasserscheid, ChemSusChem 2010, 3, 719-723.[5] J. Zakzeski, A.L. Jongerius, B.M. Weckhuysen, Green Chem. 2010, 12, 1225-1236.[6] J. Zakzeski, P. C. A. Bruijnincx, A. L. Jongerius, B. M. Weckhuysen, Chem. Rev. 2010, 110, 3552-3599.[7] E. Furimsky, Appl. Catal. A: Gen. 2000, 199, 147-190.[8] D. C. Elliott, Energy Fuels 2007, 21, 1792-1815.[9] M. Kleinert, T. Barth, Chem. Eng. Technol. 2008, 31, 736-745.[10] F. Behrendt, Y. Neubauer, M. Oevermann, B. Wilmes, N. Zobel, Chem. Eng. Technol. 2008, 31, 667-677.[11] D. C. Elliott, Prepr. Pap. - Am. Chem. Soc. Div. Pet. Chem. 1983, 28, 667-674.[12] B. S. Gevert, J. E. Otterstedt, F. E. Massoth, Appl. Catal. 1987, 31, 119-131.[13] J. B. s. Bredenberg, M. Huuska, J. Raty, M. Korpio, J. Catal. 1982, 77, 242-247.[14] E. O. Odebunmi, D. F. Ollis, J. Catal. 1983, 80, 56-64.[15] J. B. s. Bredenberg, M. Huuska, P. Toropainen, J. Catal. 1989, 120, 401-408.[16] S. J. Hurff, M. T. Klein, Ind. Eng. Chem. Fundam. 1983, 22, 426-430.[17] E. Laurent, B. Delmon, Appl. Catal. A: Gen. 1994, 109, 77-96.[18] V. N. Bui, D. Laurenti, P. Afanasiev, C. Geantet, Appl. Catal. B: Environ. 2011, 101, 239- 245.[19] M. Huuska, J. Rintala, J. Catal. 1985, 94, 230-238.[20] A. Centeno, E. Laurent, B. Delmon, J. Catal. 1995, 154, 288-298.

[21] V.N. Bui, D. Laurenti, P. Delichère, C. Geantet, Appl. Catal. B: Environ. 2011, 101, 246- 255.[22] E. Laurent, B. Delmon, Appl. Catal. A: Gen. 1994, 109, 97-115.


171

[23] E. Laurent, B. Delmon, J. Catal. 1994, 146, 281-291.[24] E. Laurent, B. Delmon, Ind. Eng. Chem. Res. 1993, 32, 2516-2524.[25] O. I. Senol, E. M. Ryymin, T. R. Viljava, A. O. I. Krause, J. Mol. Catal. A: Chem. 2007, 277, 107-112.[26] A. Pinheiro, D. Hudebine, N. Dupassieux, C. Geantet, Energy Fuels 2009, 23, 1007-1014.[27] M. Philippe, F. Richard, D. Hudebine, S. Brunet, Appl. Catal. A: Gen. 2010, 383, 14-23. [28] Y. Romero, F. Richard, Y. Renème, S. Brunet, Appl. Catal. A: Gen. 2009, 353, 46-53.[29] M. C. Edelman, M. K. Maholland, R. M. Baldwin, S. W. Cowley, J. Catal. 1988, 111, 243- 253.[30] F.P. Petrocelli, M.T. Klein, Ind. Eng. Chem. Prod. Res. Dev. 1985, 24, 635-641.[31] A. Gutierrez, R. K. Kaila, M. L. Honkela, R. Slioor, A. O. I. Krause, Catal. Today 2009, 147, 239-246.[32] H. Y. Zhao, D. Li, P. Bui, S. T. Oyama, Appl. Catal. A: Gen. 2011, 391, 305-310.[33] S. K. Sahoo, S. S. Ray, I. D. Singh, Appl. Catal. A: Gen. 2004, 278, 83-91.[34] N.- Y. Topsøe, A. Tuxen, B. Hinnemann, J. V. Lauritsen, K. G. Knudsen, F. Besenbacher, H. Topsøe, J. Catal. 2011, 279, 337-351.[35] A. Hrabar, J. Hein, O. Y. Gutiérrez, J. A. Lercher, J. Catal. 2011, 281, 325-338.[36] Y. Romero, F. Richard, S. Brunet, Appl. Catal. B: Environ. 2010, 98, 213-223.[37] A. Popov, E. Kondratieva, L. Mariey, J. M. Goupil, J. El Fallah, J-P Gilson, A. Travert, F. Maugé, J. Catal. 2013, 297, 176-186.[38] A. Arienti, F. Bigi, R. Maggi, P. Moggi, M. Rastelli, G. Sartori, A. Trerè, Perkin Trans. 1997, 1, 391-1394.[39] T. Yamato, C. Hideshima, G. K. S. Prakash, G. A. Olah, J. Org. Chem. 1991, 56, 3192-3194.[40] S. Kawai, K. Okita, K. Sugishita, A. Tanaka, H. Ohashi, J. Wood Sci. 1999, 45, 440-443.

Chapter 7

W2C and Mo2C Carbide Catalysts for the Hydrodeoxygenation of the Lignin Model Compound Guaiacol

AbstractHydrodeoxygenation (HDO) studies over carbon nanofiber-supported (CNF) W2C

and Mo2C catalysts were performed on guaiacol, a prototypical molecule to evaluate the catalytic potential for depolymerized lignin valorization. The HDO reactions were executed at 55 bar hydrogen pressure over a temperature range of 300-375 °C for 4 h in dodecane, using a batch autoclave system. Combined selectivities of up to 87% and 69% to phenol and methylated phenolics were obtained at 375 °C for W2C/CNF and Mo2C/CNF at >99% conversion, respectively. The molybdenum carbide-based catalyst showed a higher activity and more completely deoxygenated aromatic products, such as benzene and toluene, than W2C/CNF. Catalyst recycling experiments were performed with and without regeneration of the carbide phase showing that the Mo2C/CNF catalyst was stable during recycling experiments. The most promising results were obtained with the Mo2C/CNF catalyst as it showed much higher activities and selectivities to phenolics compared to W2C/CNF.

Based on: A. L. Jongerius*, R. W. Gosselink*, J. Dijkstra, J. H. Bitter, P. C. A. Bruijnincx, B. M. Weckhuysen, “Carbon Nanofiber-supported Transition Metal Carbide Catalysts for the Hydrodeoxygenation of Guaiacol“ ChemCatChem 2013, DOI: 10.1002/cctc.201300280.* Both authors contributed equally to this work.

Part II, Chapter 7

174

7.1 Introduction

As was detailed in Chapter 2, initially, most studies on the HDO of lignin or lignin model compounds focused on the use of sulfided CoMo and NiMo on alumina catalysts. [1] In Chapter 6 we reported a systematic comparison of the HDO selectivity and activity of CoMo/Al2O3 for guaiacol and various other lignin-derived molecules, reactions which generally resulted in the formation of (alkyl)phenols as the main HDO product. However, deactivation of the catalyst if no sulfur is added to the feed and contamination of the products by leaching of sulfur are general problems associated with this class of catalysts. In addition, selective HDO over these catalysts is hampered by the acidity of the support, as isomerization reactions on Lewis acid sites lead to a broad product distribution even from a single model compound and coke formation is accelerated during the reaction. Furthermore, we have shown in Chapter 4 that at elevated reaction temperatures alumina is known to be unstable in the presence of water, which can be expected to be present in biomass-derived feedstocks. [2]

Noble metal catalysts are commonly considered as more stable alternatives for the traditional sulfided CoMo and NiMo catalysts. However, oxygen removal over these catalysts is often accompanied by complete hydrogenation of the aromatic ring, which is clearly undesirable if phenolics or BTX are targeted as end products. Indeed, cyclohexanol and cyclohexane were shown to be major products of the HDO of guaiacol over supported Pt, Rh, Pd and Ru catalysts. [3-5] The formation of aromatic HDO products was observed for the conversion of vaporized guaiacol and anisole over Pt/Al2O3. This process, however, resulted in the formation of a large amount of different products including oxygen-containing and oxygen-free hydrogenation products. [6-8]

Alternative catalyst materials for the HDO of lignin-derived feeds should preferably combine the relatively strong HDO and weak hydrogenation properties of the traditional HDS catalyst with the stability of noble metal hydrogenation catalysts. Transition metal phosphides, nitrides and particularly carbides can potentially provide such an alternative. It has been demonstrated that not only molybdenum sulfides but also carbides and nitrides of molybdenum and other transition metals show good activity in various hydrotreating reactions. [9, 10] In the field of catalytic biomass conversion, bulk tungsten carbides have already been used for lignin depolymerization [11] and for cellulose degradation when supported on activated carbon. [12] Recently, it was shown that CNF-supported tungsten carbides combine high activity in the HDO of fatty acids with low hydrogenation activity, resulting in

W2C and Mo2C Carbide Catalysts for the Hydrodeoxygenation of Guaiacol

175

the production of high amounts of linear unsaturated hydrocarbons. [13]

In addition, several bulk and supported molybdenum nitride catalysts have been reported for the HDO of model compounds that mimic lignin-derived products. Indeed, already in 1997, supported molybdenum nitrides and carbides were shown to be active in the HDO of benzofuran to ethylbenzene. [14] The HDO of guaiacol over molybdenum nitrides has been reported more recently and high selectivities towards phenol were obtained. [14-17] For molybdenum nitrides, Mo2N was reported to be the most active phase whereas catalyst prepared with other Mo/N ratios were less active. [15] In contrast to the CoMo/Al2O3-catalyzed reactions that are reported to proceed via consecutive demethylation/dehydroxylation via catechol to phenol, the HDO of guaiacol over Mo2N is thought to take place via a direct demethoxylation pathway without a catechol intermediate. Demethylation of guaiacol to catechol was only observed with Al2O3-supported molybdenum nitrides, as a result of a support-catalyzed conversion. [17] Importantly, the Mo2N catalyst generally showed a better conversion of guaiacol than the CoMo/Al2O3 catalysts used in Chapter 6 and yielded less ring-hydrogenation products, such as cyclohexene and cyclohexane.

Notably, no studies have yet been reported that use either bulk or supported molybdenum or tungsten carbides for the HDO of guaiacol. The results obtained with the fatty acids, however, clearly show that the CNF-supported transition metal carbide catalysts hold promise also for selective hydrodeoxygenation of guaiacol, as they combine high HDO with low hydrogenation activity. [13] Furthermore, carbon nanofibers (CNF) are ideally suited as support for molybdenum or tungsten carbides because of their high surface area and mesoporosity. An additional, particular advantage of using carbon supports for the carbides is that catalyst preparation does not require carburization gases. Furthermore, the CNF support is expected to be stable and inert under reaction conditions, even in the presence of water. [19, 20]

In this Chapter, we report our studies on the applicability of group 6 metal carbides for the selective HDO of guaiacol, which serves as a model for the upgrading of depolymerized lignin feeds, with the aim of obtaining phenolics. The conversion of guaiacol over CNF-supported tungsten and molybdenum carbides at different temperatures shows that high selectivities towards the mono-oxygenated products phenol and cresol can be obtained. In addition, spent catalysts are analyzed and recycled (with and without additional heat treatment) to demonstrate that in the case of molybdenum no substantial deactivation of the catalyst takes place under the

Part II, Chapter 7

176

applied reaction conditions.


7.2.1 Catalyst characterizationThe X-ray diffraction (XRD) patterns of the fresh 15 wt% W2C/CNF and 7.5 wt%

Mo2C/CNF catalysts are shown in Figure 7.1, as is the pattern for bare CNF for comparison. The graphite-like reflections of the CNF support are clearly observed at 31, 51 and 64° 2θ, corresponding to (002), (101) and (004) reflections. [21] Both the W2C and Mo2C phases show main reflections at similar angles, i.e. 40, 44 and 46° 2θ corresponding to the (100), (002) and (101) lattice planes, due to their structural resemblance. In both cases an oxide phase (WO2 or MoO2) is also present as evidenced by the shoulder at 41° 2θ. [21] This can be explained by partial oxidation of the metal carbide upon exposure to air. The observed oxide phase does not increase after prolonged exposure to air, indicating that the sample is fully passivated upon first contact with air. The observed line broadening of the carbide reflections is similar for both metal carbide catalysts, resulting in comparable average particle sizes of 4 nm, as calculated using the Scherrer equation (see also Table 7.1). [22] This is in accordance with statistical calculations on Transmission Electron Microscopy (TEM) images (Table 7.1).

To study the influence of the carbothermal reduction processes on the CNF support, N2 physisorption was performed. The BET surface area and pore volume results are depicted in Table 7.1 for bare CNF, Mo2C/CNF and W2C/CNF. A significant decrease in the BET surface area and total pore volume is observed after heat treatment, which indicates a structural change in the CNF structure. No evidence for any structural change can be seen in the TEM images or XRD diffraction patterns, however; previous

Particle size N2-physisorptionTEM (nm) XRD (nm) BET surface

area (m2/g)Pore volume

(cm3/g)

CNF - - 190 0.40Mo2C/CNF 5 4 120 0.34W2C/CNF 5 4 111 0.30

Table 7.1: Physicochemical characteristics of the CNF-supported metal carbide catalysts under study.


177

a. b.

CNF

Mo₂C/CNF

W₂C/CNF

●

●●▲

▲▲■ ■ ▲● ▲

●▲▲

▲ ■

10 30 50 70 902θ (˚)

35 40 45 50 55 602θ (˚)

Figure 7.1: Full XRD patterns (a) and zoomed in XRD patterns (b) of bare CNF, W2C/CNF and Mo2C/CNF. Reflections are given for graphitic carbon ( ), XO2 ( ) and X2C ( ), with X = W or Mo. [21]

studies with W2C/CNF also showed a decrease of the BET surface area upon an increase of the temperature of the carbothermal process, which could be attributed to, e.g., partial collapse or plugging of the pores. [13]

7.2.2 Catalytic activityA typical HDO reaction of guaiacol over W2C/CNF and Mo2C/CNF was performed

in dodecane at 350 °C with 55 bar hydrogen pressure for 4 h. The main products observed in all runs were phenol, o-cresol and p-cresol; other products were obtained in smaller amounts with a total selectivity of less than 10%. All products can be grouped into classes according to extent and type of HDO: i) phenol; ii) the cresols o- and p-cresol and dimethylphenol isomers; iii) full HDO products benzene and toluene; iv) products still containing methoxy functionalities (anisole, methylated anisole and dimethoxybenzene) and v) ring-hydrogenation products (cyclohexane, cyclohexene and cyclohexanone). In the Mo2C/CNF-catalyzed reactions, larger amounts of group iii-v products were generally formed, with a relative increase in the fractions of full HDO and ring-hydrogenation products compared to the W2C/CNF-catalyzed reactions. The exact composition of all product mixtures can be found in Table 7.2 and 7.3. It should be noted that the conversion levels over W2C/CNF vary between 66 - 82% for a typical 4 h reaction at 350 °C; the ratio between different products was always found to be constant, however.

Reactions performed in the absence of a catalyst resulted in a 40% conversion of guaiacol. A small amount was converted to phenol and cresols with a total selectivity

Part II, Chapter 7

178

of 14%, but the majority of products could not be detected by GC analysis (mass balance 68%, Table 7.2). A blank reaction with only bare CNF showed the same guaiacol conversion of 40%, and selectivities for phenol (7%) and cresol (11%). Low concentrations of methoxylated and hydrogenated products were detected with a total selectivity of 6%. Under identical conditions the reactions catalyzed with both W2C/CNF and Mo2C/CNF resulted in significantly higher conversions, selectivity towards phenolics and mass balances (Table 7.2). Combination of the product yields obtained by GC analysis resulted in mass balances of 76% for W2C/CNF and 74% for Mo2C/CNF-catalyzed reactions at 350 °C. An increase in reaction temperature to 375 °C resulted in higher mass balances for both catalysts (96% (W2C) and 91% (Mo2C)); the reactions at temperatures other than 350 °C are discussed below. The blank reactions also give low mass balances, which can only be caused, in this case, by uncatalyzed thermal reactions. The results for the catalyzed reaction point in the same direction, as recovery of the catalyst did not show any increase in weight of the catalyst and no solids were formed on the reactor wall, indicating that the 10-30% weight loss that occurred during the reaction cannot be attributed to the formation of large amounts of coke. Indeed, a common problem of the use of guaiacol as substrate is that guaiacol condensation reactions result in soluble higher molecular weight products that cannot be detected by our GC methods, thus lowering the mass balance. Guaiacol is, for instance, known to undergo a radical-induced rearrangement leading to the formation of oligomers and ultimately coke. This rearrangement is initiated by OMe bond homolysis but takes place only at temperatures above 400 °C. [23] Condensation products can nonetheless also be formed at lower temperatures, as a

Conversion (%) Product class selectivity (%) Mass balance (%)

i) ii) iii) iv) v)

No catalyst 40 6.5 8 <1 2 4 68CNF 40 7 11 <1 3 4 70W2C/CNF 66 46 12 <1 4 2 76Mo2C/CNF >99 45 13 2 8.6 2 74

i) phenol; ii) o-cresol, p-cresol and dimethylphenol isomers; iii) benzene and toluene; iv) anisole, methylated anisole and dimethoxybenzene; v) cyclohexane, cyclohexene and cyclohexanone

Table 7.2: Conversion, selectivity to phenol and cresols and mass balance for the HDO of guaiacol without catalyst, over heat-treated CNF and over Mo2C/CNF and W2C/CNF at 350 °C and ~55 bar H2 pressure after 4 h.


179

result of direct ring coupling as was shown by Bui et al. in the HDO of guaiacol over CoMo/Al2O3 and MoS2. [24] In our case, 1H NMR analysis of a reaction mixture with a 85% mass balance also shows that at least 5% more aromatic products are present in the product mixture than can be identified by GC analysis. Analysis of the reaction mixture obtained after a run without catalyst (mass balance 68%, Table 7.2) shows a new singlet at 2.2 ppm, indicating the presence of a large amount of unidentified aromatic ring-methylated products. Similar products were again reported by Bui et al. after guaiacol HDO over CoMo/Al2O3 and MoS2. [24] We therefore attribute the observed loss in mass balance to the formation of condensed, higher molecular weight products. Although the formation of such higher molecular weight products cannot be easily prevented, their contribution can be considerably reduced by speeding up guaiacol conversion, as the phenol and cresol products are much more stable against self-condensation.

Notably, no catechol was observed in any of the reaction mixtures, indicating that the conversion of guaiacol to phenol does not involve sequential demethylation (DME) and HDO steps (as was observed for sulfided CoMo/Al2O3 in Chapter 6), but rather a direct demethoxylation pathway (DMO) (Scheme 7.1). This is in accordance with results reported for bulk and carbon-supported Mo2N-catalyzed HDO of guaiacol, although small amounts of catechol were still found with the nitride catalysts. Most likely the high temperatures that were used during the carburization step resulted

Scheme 7.1: HDO of guaiacol gives phenol and cresols as the main products.

Part II, Chapter 7

180

in removal of all mildly Brønsted acidic sites on the CNF. Where Lewis acid sites on alumina-supported catalysts are responsible for DME activity, the CNFs used here can be considered inert under our reaction conditions, as the applied heat treatment removes the relevant functional groups from the support. [25] Indeed, blank reactions performed in the presence of heat-treated CNF give almost identical results as blank reactions run without any catalyst or support present (Table 7.2). The small amounts of cresols and other methylated products that are formed during the reaction probably arise from a transalkylation reaction catalyzed by the mildly acidic guaiacol and phenols. Similar to what was reported in Chapter 6 for the HDO of guaiacol over CoMo/Al2O3, phenol is much more stable under the applied conditions than guaiacol, indicating that the removal of the second oxygen to form benzene is more difficult and is not favorable when guaiacol is still present. A reaction with phenol as the substrate over the W2C/CNF catalyst at 350 °C, indeed showed a much lower conversion of 28%, with a mass balance of 82% and benzene as the major product with 30% selectivity.

To optimize the reaction conditions, reactions were performed at different reaction temperatures (300-375 °C) for 4 h at ~55 bar hydrogen pressure. Results of the temperature optimization are shown in Figure 7.2. For the W2C/CNF-catalyzed reactions, conversion and selectivity towards phenolics increased with reaction temperature. Full conversion was obtained at 375 °C after 4 h with a selectivity of 66% to phenol and 21% to cresols. The Mo2C/CNF catalyst showed a much higher activity with almost full conversion (95%) obtained after 4 h already at 325 °C. The total selectivity for phenolics (53+17% for phenol and cresols, respectively) is lower with Mo2C/CNF than with W2C/CNF, and a higher total selectivity (17%) to other products, particularly benzene (8%) and anisole (6%) is observed for the former at 375 °C after 4 h. Ring hydrogenation activity of the Mo2C/CNF catalyst also increased at higher temperatures. As mentioned previously, higher temperatures result in better mass balances for both catalysts with a maximum of 96% and 91% at 375 °C for W2C/CNF and Mo2C/CNF, respectively. Notably, the mass balances of >90% obtained for the reactions at 375 °C are among the highest reported for guaiacol HDO reactions.

The selectivities towards phenol, or more generally, phenolics are much higher than typically obtained with traditional HDS catalysts, as illustrated by the CoMo/Al2O3-catalyzed HDO of guaiacol in Chapter 6 at comparable substrate concentration and hydrogen pressure. For this system, the highest phenolics yields were obtained at 300 °C after 4 h, amounting to a selectivity to phenol of only 34% and 11% to cresols at 84% conversion. A broad range of methylated and demethylated products were also


181

formed, including catechol (11%) and methylated guaiacol and catechol (8%). The use of noble metal catalysts also generally results in lower phenolics yields, as they show much higher ring hydrogenation than HDO activity and produce mainly cyclohexanol and eventually cyclohexane. [3, 4]

A Mo2N catalyst supported on activated carbon showed a similar phenol selectivity at 300 °C in guaiacol HDO as the Mo2C/CNF catalyst presented here at 300 °C. [18] However, significantly lower conversion levels were obtained whilst all reaction parameters were comparable. [15] The total selectivity for mono-oxygenated aromatics was also lower with the activated carbon-supported Mo2N catalyst as no

a.

b.

0,0

20,0

40,0

60,0

80,0

100,0

conversion phenolselectivity

cresolselectivity

mass balance

conv

ersi

on/s

electiv

ity (%

)

0,0

20,0

40,0

60,0

80,0

100,0

conversion phenolselectivity

cresolselectivity

mass balance

conv

ersi

on/s

electiv

ity (%

)

Figure 7.2: Conversion, selectivity to phenol and cresols and mass balance for the HDO of guaiacol over a: W2C/CNF and b: Mo2C/CNF at ~55 bar H2 pressure for 4 h at different temperatures ( 300 °C, 325 °C, 350 °C and 375 °C).

Part II, Chapter 7

182

cresols or anisole were obtained. Demethylation activity, illustrated by the small quantities of catechol that were detected, was furthermore reported for these supported nitride catalysts.

The conversion of guaiacol over time at 350 °C and selectivity towards different products are shown for both catalysts in Figure 7.3. The difference in activity between the two catalysts is evident, as initial reaction rates of 0.16 and 0.64 mol/g metal/h were obtained for W2C/CNF and Mo2C/CNF, respectively. Full conversion of guaiacol over Mo2C/CNF was reached after 2 h, while guaiacol conversion over W2C/CNF is still increasing after 6 h. Phenol selectivity is found to increase initially, but leveled off when a conversion of around 60% was reached for both catalysts. In the Mo2C/CNF-catalyzed reaction the phenol selectivity decreased after 6 h in favor of full HDO and hydrogenation products (see also Table 7.3). In both reactions the mass balance decreased sharply to around 80% after the first hour after which it leveled off.

Catalyst T (˚C) t (h) Conversion (%)

Product class selectivity (%) Mass balance (%)

i) ii) iii) iv) v)

W 325 4 32 6 2 0 1 1 78W 375 4 >99 66 21 2 4 2 96W 350 1 34 7 2 <1 1 3 77

W 350 2 46 15 4 <1 1 3 76W 350 4 66 35 10 <1 3 2 84W 350 6 86 42 11 <1 4 2 73Mo 300 4 58 21 4 <1 6 2 74Mo 325 4 95 43 12 1 10 2 74Mo 375 4 >99 53 17 10 7 5 91Mo 350 1 70 35 8 1 6 2 81Mo 350 2 97 44 13 1 9 4 74Mo 350 4 >99 46 13 2 9 7 74Mo 350 6 >99 40 16 5 8 7 76W* 350 4 29 - <1 9 <1 0 83

i) phenol; ii) o-cresol, p-cresol and dimethylphenol isomers; iii) benzene and toluene; iv) anisole, methylated anisole and dimethoxybenzene; v) cyclohexane, cyclohexene and cyclohexanone; * reaction performed on phenol.

Table 7.3: HDO of guaiacol over W2C/CNF and Mo2C/CNF under different reaction conditions.


183

a.

b.

0

20

40

60

80

100

0 2 4 6

conv

ersi

on/s

electiv

ity (%

)

reaction time (h)

0

20

40

60

80

100

0 2 4 6

conv

ersi

on/s

electiv

ity (%

)

reaction time (h)

The results show that Mo2C/CNF is more active than W2C/CNF, but also shows a higher selectivity to the full HDO products benzene and toluene and the methoxylated product anisole. The selectivity to full HDO products increases with reaction time, with a maximum selectivity of 4.8% observed after 6 h reaction. The higher activity of Mo2C/CNF also resulted in a ring hydrogenation selectivity (7%) after 6 h that was higher than observed for W2C/CNF (2%). This is in line with the observation that that tungsten carbides are less active olefin hydrogenation catalysts than molybdenum carbides in the HDO of unsaturated fatty acids. [26]

Figure 7.3: Conversion ( ), selectivity to phenol ( ), cresols ( ) and mass balance ( ) for the HDO of guaiacol over: W2C/CNF and b: Mo2C/CNF at 350°C and ~55 bar H2 pressure, followed in time.

Part II, Chapter 7

184

7.2.3 Catalyst recyclingThe stability of the W2C/CNF and Mo2C/CNF catalysts was tested in two ways. A

standard reaction was first performed on a freshly prepared catalyst (run 1). After reaction, the spent catalyst was retrieved, rinsed with diethyl ether and reused directly without any reactivation step (run 2). Hereafter, the spent catalyst of run 2 was recarburized and again tested in a catalytic reaction (run 3). All runs were performed at 350 °C for 4 h and the catalysts were analyzed by XRD after every run (Figure 7.4 and 7.5). For the Mo2C/CNF catalyst, an additional recycling experiment has been performed with reactions of 70 min in order to obtain a conversion level comparable to W2C/CNF after 4 h. The recovery of each catalyst after every run was close to 100% and reaction volumes and guaiacol concentrations were in every run scaled to the amount of catalyst recovered to allow for a proper comparison of the runs. Catalytic data for both catalysts is presented in Table 7.4.

Remarkably, the 70 min reaction with the molybdenum catalyst resulted in a much higher phenol selectivity, 62% compared to 46% after 4 h, and higher mass balances, 92% compared to 76%. Differences are seen in the XRD pattern of the W2C/CNF catalyst after run 1 (Figure 7.4). A clear increase in the diffraction peak attributed to WO2 is observed for the spent W2C/CNF catalyst, whereas for Mo2C/CNF, both after the 70 min and 4 h reactions, it does not change significantly (Figure 7.5). The tungsten-based catalyst therefore seems more prone to oxidation than the molybdenum carbide. The catalytic test with the spent W2C/CNF catalyst (without any further treatment, run 2) showed an increase in guaiacol conversion from 66 to 80% and increased selectivities towards all products. Reused Mo2C/CNF showed similar conversions and selectivities in run 2 for both 4 h and 70 min reactions, as could be expected from the unchanged XRD pattern and high activity of this catalyst.

It was already noted above that the conversion levels of typical guaiacol HDO experiments catalyzed by freshly prepared batches of W2C/CNF show some variation, i.e. were found range between 66 and 82% in multiple experiments; conversion levels for Mo2C/CNF, on the other hand, showed little variation. The increased conversion observed in the second run for the tungsten catalyzed reaction could therefore simply be the result of this observed fluctuation in conversion level. A second set of W2C/CNF recycling experiments, however, again showed a higher activity in run 2 with the W2C/CNF catalyst. This indicates that this increased activity of the spent catalyst could also arise from a structural change that takes place under reaction conditions. It is very likely that the formation of an oxide phase plays an important role in these


185

a. b.

Fresh

Run 1

Run 2

Regenerated

Run 3

●

● ●▲

▲

▲

■

■ ▲● ▲

●

▲

▲

▲■

35 40 45 50 55 602θ (˚)

10 30 50 70 902θ (˚)

Figure 7.4: Full XRD patterns (a) and zoomed in XRD patterns (b) of fresh W2C/CNF (fresh), spent W2C/CNF (run 1), reused W2C/CNF (run 2), spent W2C/CNF after reactivation at 1000 °C (regenerated) and spent reactivated W2C/CNF (run 3). Reflections are given for graphitic carbon ( ), XO2 ( ) and X2C ( ), with X = W or Mo. [21]

a. b.

Fresh

Run 1

Run 2

Regenerated

Run 3

●

●●▲

▲▲■ ■ ▲● ▲

●▲▲▲ ■

10 30 50 70 902θ (˚)

35 40 45 50 55 602θ (˚)

Figure 7.5: Full XRD patterns (a) and zoomed in XRD patterns (b) of fresh Mo2C/CNF (fresh), spent Mo2C/CNF (run 1), reused Mo2C/CNF (run 2) and reactivated Mo2C/CNF after reactivation at 900 °C (regenerated) and reactivated Mo2C/CNF after reaction (run 3). Reflections are given for graphitic carbon ( ), XO2 ( ) and X2C ( ), with X = W or Mo. [21]

Part II, Chapter 7

186

reactions, however, more research is needed before the exact role and nature can be revealed. These experiments are beyond the scope of this paper.

XRD analysis of the catalyst after run 2 but before recarburization showed again an increase in the WO2 phase for the tungsten catalyst, whereas no change in the reflections for the molybdenum catalyst was observed. This unexpected difference between the stability towards oxidation of both catalysts cannot be readily explained. Although the Gibbs free energy for oxidation of a (bulk) W2C phase is more negative than for Mo2C (as calculated by HSC chemistry for Windows), the difference is not significant enough to explain the observed dissimilarity in catalyst stability. A possible explanation would be that it is caused by a kinetic factor, such as the diffusion of oxygen through the particle.

Subsequently, both catalysts were subjected to a recarburization treatment at 1000 °C and 900 °C for W2C/CNF and Mo2C/CNF, respectively. The X-ray diffractograms show that the carbide phase was restored for the tungsten-based catalyst; the thermal treatment did, however, lead to a clear increase in particle size to 9 nm, which corresponds to the size of the formed WO2 crystallites, which also were 9 nm. This increase in particle size has a clear effect on the catalytic activity, with a 20% decrease in conversion in run 3 with respect to run 1, a 10 % drop in selectivity for phenol and a sudden increase in hydrogenation selectivity to 3.4%. Interestingly, larger carbide particle sizes have been reported to be to show higher activity in hydrogenation and HDS reactions, which is attributed to a relative increase in the ratio of the (111)/(200) crystal planes. [1] However, the XRD data does not show such an increase for our catalysts. Furthermore, our results are based on 4 - 9 nm particles, which are significantly smaller than the bulk particles tested in literature. Remarkably, although no changes in the Mo2C/CNF XRD pattern could be observed after the recarburization treatment, guaiacol conversion dropped from 68% to 51% for the 70 min reaction with a concomitant drop in phenol selectivity. The characterization data nonetheless shows the carbide phase of Mo2C/CNF to remain rather stable during the reactions and the carburization treatment.

Both catalysts show activity loss after the recarburization step, which for the W2C/CNF can be attributed to an increase in particle size. The decrease in activity of the Mo2C/CNF catalyst after recarburization indicates, however, that other changes also must take place in the material during the heat treatment. For example, encapsulation of the carbide particles in the carbon support or irreversible coke formation could


187

be responsible for the observed catalyst deactivation. Notably, no loss in catalytic activity and selectivity was observed for the molybdenum catalyst after washing and reusing the catalyst without further heat treatment, though (Table 7.4, run 2, 70 min). This shows that recarburization is not required for the Mo2C/CNF catalyst, but is essential for the W2C/CNF catalyst to recover the carbide phase. Indeed, the Mo2C/CNF catalyst proved to be rather stable under reaction conditions and can be reused for several runs without difficulty.

Time Conversion (%) Product class selectivity (%) Mass balance (%)i) ii) iii) iv) v)

W2C/CNF

run 1 4 h 66 46 12 1 4 2 76run 2 81 51 15 1 7 2 80run 3 48 36 12 1 2 3 79

Mo2C/CNF

run 1 4 h >99 46 13 2 9 2 74run 2 >99 45 16 4 10 7 81run 3 >99 45 15 5 8 7 79

run 1 70 min 67 62 13 1 6 2 92run 2 68 56 12 1 10 2 87run 3 51 50 12 1 8 2 86

i) phenol; ii) o-cresol, p-cresol and dimethylphenol isomers; iii) benzene and toluene; iv) anisole, methylated anisole and dimethoxybenzene; v) cyclohexane, cyclohexene and cyclohexanone

Table 7.4: Conversion, selectivity to various products and mass balance for the HDO of guaiacol over W2C/CNF and Mo2C/CNF (run 1), the reused (run 2) and reactivated catalysts (run 3), at 350 °C and ~55 bar H2 pressure for 4 h or 70 min.

Part II, Chapter 7

188

7.3 Conclusions

High conversions and selectivities towards phenolics are obtained in the hydrodeoxygenation of guaiacol over W2C/CNF and Mo2C/CNF at 300-375 °C and ~55 bar hydrogen pressure. At higher temperatures, higher conversions and selectivities to the desired products are obtained. Mo2C/CNF is more active than W2C/CNF but also leads to the formation of more full HDO and methoxylated products. Longer reaction times and higher temperatures result in more of the full HDO products benzene and toluene. Reactions with low catalytic activity lead to the formation of higher molecular weight methylated aromatics, thus lowering the mass balance. W2C/CNF is sensitive to oxygen and the small amount of WO2 phase detected after reaction had a significant influence on catalytic activity. After recarburization the W2C phase could be restored completely, however, sintering of the particles occurred, resulting in loss of catalytic activity. In contrast, Mo2C/CNF showed excellent stability under reaction conditions and could be reused without further treatment and without loss of catalytic activity. Compared to the results reported in Chapter 6 with CoMo/Al2O3, the tungsten and molybdenum carbides show much higher selectivities towards phenolics and lower (de)methylation activity. Their low ring-hydrogenation activity makes them excellent catalysts for the production of phenolics from depolymerized lignin streams.


7.4.1 Catalyst MaterialsA 5 wt% Ni/SiO2 growth catalyst was prepared by homogeneous deposition

precipitation using 30 g silica (Aerosol 300, Degussa), 7.9 g Ni(NO3)2 (99%, Acros) and 4.85 g of urea (99%, Acros) at 90 °C for 18 h in 1.5 L demineralized water, as reported earlier. [27] 5 g Ni/SiO2 was then reduced at 700 °C (ramp 5 °C/min in N2 flow) in a H2/N2 flow for 2 h. Fishbone-type CNFs were obtained by flowing a CO/H2/N2 mixture (266/102/450 mL/min) over this catalyst at 550 °C (ramp 5 °C/min in N2/H2 flow) for 24 h. The crude product, approximately 35 g, was collected and refluxed three times for 1 h, with intermediate washing with deionized water, in 400 mL boiling 1.0 M aqueous KOH (Acros) to remove SiO2. Subsequently, the fibers were treated with 400 mL boiling, concentrated HNO3 (65%, Merck) for 1.5 h in order to remove exposed nickel and introduce oxygen-containing functional groups. The acid-treated fibers are finally washed with demineralized water until the pH of the filtrate was neutral.

The CNF were dried under vacuum at 80 °C for 4 h prior to impregnation. For


189

the tungsten-based catalysts, the support was impregnated using an ammonium metatungstate solution (66.5 wt% W, Aldrich) by pore volume (0.7 mL/g determined by water uptake) to arrive at a 15 wt% loading. After impregnation the sample was kept under vacuum at 80 °C for 24 h. Subsequently a heat treatment was applied for 3 h at 1000 °C under N2 atmosphere to yield W2C/CNF. Mo2C/CNF was also prepared by pore volume impregnation, using an ammonium molybdate (>99%, Acros) solution to arrive at a 7.5 wt% loading, followed by a heat treatment at 900 °C under N2 atmosphere. [13] Different weight loadings were used to arrive at similar molar loadings of the carbide catalysts.

7.4.2 Catalytic ReactionsThe catalytic reactions were performed in a 100 mL stainless steel high pressure

Autoclave Engineers batch autoclave reactor. The temperature was monitored using a thermocouple and stirring was performed using a magnetic driver equipped with an impellor. In a typical reaction, the autoclave was loaded with 200 mg of the catalyst material, 1.7 g guaiacol (Sigma), 0.6 g hexadecane (>99%, Sigma) as internal standard and 32 g of the solvent dodecane (99%, Acros). A fresh batch of catalyst was used for every run. The reactor was purged for 5 min with nitrogen (5.0) at a flow of 100 mL/min and 5 min with hydrogen (5.0) at a flow of 100 mL/min while stirring at 600 rpm. The reactor was then filled with hydrogen with a flow of 100 mL/min while heating to the desired temperature (usually 350 °C) with a heating ramp of 5 °C/min, typically leading to a hydrogen pressure between 50 and 60 bar during the reaction. When reaction temperature was reached, the stirring speed was increased to 1000 rpm (maximum stirring speed) and the catalytic reaction was carried out for 4 h. The reaction was stopped by cooling and release of pressure, after which the reaction mixture was diluted with an equal volume of diethyl ether in order to dissolve all products. The solid catalyst was removed by filtration and washed with diethyl ether.

7.4.3 AnalysisN2 physisorption isotherms were recorded with a Micromeritics Tristar 3000

at 77 K. The samples were dried prior to performing the measurement for at least 16 h at 473 K in a N2 flow. The surface area was determined using the Brunauer–Emmett–Teller (BET) theory. [28] The total pore volume was defined as the single-point pore volume at p/p0 = 0.95. X-ray powder diffraction (XRD) patterns were obtained on a Bruker-AXS D2 Phaser powder X-ray diffractometer using Co Kα1,2 with

Part II, Chapter 7

190

λ = 1.79026 Ǻ. Measurements were carried out between 10 and 100° 2θ using a step size of 0.09° 2θ and a scan speed of 1 s. Bright-field TEM images were obtained on an FEI Tecnai 12, operated at 120 keV. The reaction mixture was analyzed using a Varian 430-GC gas chromatograph equipped with a VF-5ms capillary column and a FID detector. Compounds were identified with gas chromatography coupled with a mass spectrometer using a Shimadzu QP2010 GCMS instrument equipped with a VF-5ms capillary column and compared with retention times of pure standards. Response factors of guaiacol and phenol relative to hexadecane were determined experimentally. 1H NMR measurements were performed on the pure reaction mixture with a few drops of deuterated chloroform on a Varian 400 MHz NMR spectrometer.

7.5 Acknowledgments

Rob Gosselink is thanked for his contributions to the manuscript on which this chapter is based. Rob Gosselink and Jelmer Dijkstra are thanked for their contributions to the experimental work.

7.6 References

[1] E. Furimsky, Appl. Catal. A: Gen. 2000, 199, 147–190.[2] Ravenelle, R. M.; Copeland, J. R.; Kim, W. G.; Crittenden, J. C.; Sievers, C. ACS Catal. 2011, 1, 552-561.[3] C. R. Lee, J. S. Yoon, Y.-W. Suh, J.-W. Choi, J.-M. Ha, D. J. Suh, Y.-K. Park, Catal. Commun. 2012, 17, 54-58.[4] A. Gutierrez, R. K. Kaila, M. L. Honkela, R. Slioor, A. O. I. Krause, Catal. Today 2009, 147, 239-246.[5] C. Zhao, J. A. Lercher, ChemCatChem 2012, 4, 64-68.[6] T. Nimmanwudipong, R. C. Runnebaum, D. E. Block, B. C. Gates, Catal. Lett. 2011, 141, 779-783.[7] R. C. Runnebaum, T. Nimmanwudipong, D. E. Block, B. C. Gates, Catal. Sci. Technol. 2012, 2, 113-118.[8] T. Nimmanwudipong, R. C. Runnebaum, D. E. Block, B. C. Gates, Energ. Fuel 2011, 25, 3417-3427.[9] G. Ertl, K. Knözinger, F. Schüth, J. Weitkamp, Handbook of Heterogeneous Catalysis, Wiley, first edition, 1996, Chapter 2.3.14, 342-356.


191

[10] E. Furimski, Appl. Catal. A: Gen. 2003, 240, 1-28.[11] S. Kotrel, M. Emmeluth, A. Benöhr, 2009, International Patent 037281.[12] N. Ji, T. Zhang, M. Zheng, A. Wang, H. Wang, X. Wang, J. G. Chen, Angew. Chem. Int. Ed. 2008, 47, 8510-8513.[13] R. W. Gosselink, D. R. Stellwagen, J. H. Bitter, Angew. Chem. Int. Ed. 2013, 52, 5089- 5092.[14] G. M. Dolce, P. E. Savage, L. T. Thompson, Energ. Fuel 1997, 11, 668-675.[15] C. Sepulveda, K. Leiva, R. Carcia, L. R. Radovic, I. T. Ghampson, W. J. DeSisto, J. L. Garcia Fierro, N. Escalona, Catal. Today 2011, 172, 232-239.[16] I. T. Ghampson, C. Sepulveda, R. Garcia, B. G. Frederick, M. C. Wheeler, N. Escalona, W. J. DeSito, Appl. Catal. A: Gen. 2012, 413-414, 78-84.[17] I. T. Ghampson, C. Sepulveda, R. Garcia, J. L. Carcia Fierro, N. Escalona, W. J. DeSisto, Appl. Catal. A: Gen. 2012, 435-436, 51-60.[18] I. T. Ghampson, C. Sepulveda, R. Garcia, L. R. Radovic, J. L. Carcia Fierro, W. J. DeSisto, N. Escalona, Appl. Catal. A: Gen. 2012, 439-440, 111-124.[19] V. Jiménez, P. Sánchez, P. Panagiotopoulou, J. L. Valverde, A. Romero, Appl. Catal A: Gen. 2010, 390, 35-44.[20] J. H. Bitter, J. Mater. Chem. 2010, 20, 7312-7321.[21] ICDD PDF-2 database package http://www.icdd.com/products/pdf2.htm.[22] A. L. Patterson, Phys. Rev. 1939, 56, 978-982.[23] M. Asmadi, H. Kawamoto, S. Saka, J. Anal. Appl. Pyr. 2011, 92, 88-98.[24] N. Bui, D. Laurenti, P. Afanasiev, C. Geantet, Appl. Catal. B: Environ. 2011, 101, 239- 245.[25] R. W. Gosselink, R. van den Berg, W. Xia, M. Muhler, K. P. de Jong, J. H. Bitter, Carbon 2012, 50, 4424-4431.[26] S. A. W. Hollak, R. W. Gosselink, J. H. Bitter, D. S. van Es, in preparation.[27] M. L. Toebes, M. K. Van der Lee, L. M. Tang, M. H. Huis in t Veld, J. H. Bitter, J. Van Dillen, K. P. De Jong, J. Phys. Chem. B, 2004, 108, 11611-11619.[28] S. Brunauer, P. H. Emmett, E. Teller, J. Am. Chem. Soc. 1938, 60, 309–319.

Part III

Combined Depolymerization and Hydrodeoxygenation of Lignin

Chapter 8

Liquid-Phase Reforming and Hydrodeoxygenation as a Two-Step Route to Aromatics from Lignin

AbstractA two-step approach to the conversion of organosolv, kraft and sugarcane bagasse

lignin to monoaromatic compounds of low oxygen content is presented. The first lignin depolymerization step involves a liquid phase reforming (LPR) reaction over a 1wt% Pt/g-Al2O3 catalyst at 225 °C in alkaline ethanol/water. The LPR step resulted in a decrease in lignin molecular weight of 32%, 57% and 27% for organosolv, kraft and bagasse lignin, respectively. GC analysis of the depolymerized lignin furthermore showed the formation of alkylated phenol, guaiacol and syringol-type products in 11%, 9% and 5% yields from organosolv, kraft and bagasse lignin, respectively. The lignin-oil that was isolated by extraction of the LPR reaction mixture solution was subjected to a subsequent hydrodeoxygenation (HDO) reaction in the second conversion step. HDO of the lignin-oil was performed in dodecane at 300 °C under 50 bar hydrogen pressure over CoMo/Al2O3 and Mo2C/CNF catalysts. The product mixture obtained after the two-step LPR-HDO process contained, amongst others, benzene, toluene, xylenes and ethylmethylbenzenes. Of the total observed monomeric products (9%), 25% of these were oxygen-free. Notably, such products cannot be obtained by direct HDO of lignin. HDO of the lignin-oil at 350 °C resulted in the conversion of all tris-oxygenated products, with 57% of the observed monomeric products now identified as mono-oxygenated phenolics.

Based on: A. L. Jongerius, P. C. A. Bruijnincx, B. M. Weckhuysen, “Liquid-Phase Reforming and Hydrodeoxygenation as a Two-Step Route to Aromatics from Lignin“ Green Chem. 2013, DOI:10.1039/C3GC41150H.

Part III, Chapter 8

196

8.1 Introduction

The valorization of the lignin component of lignocellulosic biomass would greatly aid the development of economically viable biorefinery processes. Lignin is highly aromatic and can thus be regarded as a major source for aromatics in a bio-based economy; however, selective conversion of this relatively recalcitrant resource has proven to be difficult and requires the development of new technologies. Many of the recent efforts in this direction have focused on single-step routes for the conversion of lignin into monomeric aromatic compounds. Oxygen-rich fine chemicals such as vanillin and benzoquinones are, for instance, readily obtained via oxidative routes, as was shown by Werhan et al. and Stärk et al. [1,2] and by the commercial production of vanillin from lignosulfonate lignin through the Borregaard biorefining processes.[3] Reductive one-step conversion routes, on the other hand, typically lead to the formation of mixtures of (alkylated) guaiacol and syringol-type molecules. Recently, yields of up to 50% of propyl-substituted guaiacol and syringol have been reported via a nickel-catalyzed fragmentation hydrolysis [4] or the nickel/tungsten carbide-catalyzed hydrocracking of birch wood. [5] Other reported reductive processes generally resulted in lower yields and more complex product mixtures consisting of multiple major aromatic products, [6,7] or even ring-hydrogenated monomeric products. [8] Such one-step, reductive processes generally lead to the formation of bis- or tris-oxygenated mono-aromatic molecules, however. No one-step lignin conversion processes have been reported so far for the production of oxygen-free or low-oxygen content monomeric aromatics.

Rather than performing the conversion of lignin in one step, a two-step conversion strategy, as discussed in Chapter 1, allows one to sequentially perform the depolymerization and a second step to further upgrade the product mixture of the depolymerization process in a second, independent step. In the first step of such a process, a (catalytic) depolymerization reaction is applied to yield lignin monomers, dimers and small oligomers. The reaction product of this first step can both be a depolymerized lignin solution or a lignin-oil. Such oils, for instance obtained from lignin pyrolysis reactions, commonly consist of a complex mixture of phenolic molecules including (substituted) guaiacols, catechols and syringols. [9-11] Also other thermochemical processes are available for the production of lignin-oils, of which base-catalyzed hydrogenolysis is most commonly used. [12-14] As the second conversion step can be independently chosen, there are many possibilities among which are oxidation, hydrogenation, and hydrodeoxygenation (HDO) reactions of


197

which especially the latter two are most commonly explored. Indeed, the choice of catalyst and type of conversion in the in the second step can be based on the targeted end products. Here, we focus on the use of an HDO step for the upgrading of depolymerized lignin streams. The combination of lignin depolymerization with a separate HDO step allows one to obtain products with a lower oxygen content than is customarily obtained for the one-step processes.

HDO studies aimed at the deoxygenation of lignin-oils mostly make use of industrial CoMo and NiMo catalysts or noble metal Ru, Pt and Pd-based catalysts. However, it must be noted that most of the catalytic studies that are concerned with HDO of lignin-related compounds are actually performed with model compounds, either a single molecule or a library of compounds, rather than real lignin-derived feeds, as the use of model compounds greatly simplifies product analysis and kinetic studies (see also Chapter 2). [15] Performing HDO reactions on real depolymerized lignin feeds poses considerable challenges that are absent when model compounds are used, primarily because lignin-derived bio-oils consist of a complex mixture of oxygen-rich aromatic molecules, a mixture that in addition to monomeric compounds will also contain dimers and oligomers. These higher molecular weight components are prone to repolymerization and coke formation at the relatively high temperatures (> 300 ˚C) generally needed for HDO. Of the relatively few reports that have appeared on HDO of real lignin depolymerization products, Shabtai et al. were ones of the first to report a combination of a lignin depolymerization step combined with a second HDO step. [16] The process consisted of a sequential base-catalyzed depolymerization and CoMo/Al2O3-catalyzed HDO and hydrocracking step to ultimately produce hydrocarbon gasoline. More recently, several studies on the HDO of (lignin) pyrolysis oils [17, 18] showed that higher degrees of deoxygenation could be obtained with noble metal catalysts than with the traditional CoMo and NiMo catalysts. [17] Noble metal catalysts, however, also showed a high ring-hydrogenation activity leading to the formation of cycloalkanes rather than aromatic products. [19]

Here, we report the integration of a lignin depolymerization step, i.e. a liquid phase reforming (LPR) reaction, with a second HDO step over CoMo/Al2O3 and Mo2C/CNF catalysts with the aim of obtaining aromatic product mixtures of low oxygen content (Figure 8.1). We present a rare example of actual BTX (Benzene, Toluene, Xylenes) formation from real lignin sources. The LPR reaction of organosolv, kraft and bagasse lignin is performed in ethanol/water over a Pt/Al2O3 catalyst at 225 ˚C as described in Chapter 3. The product mixture containing the depolymerized lignin

Part III, Chapter 8

198

is extracted and a lignin-oil is obtained after evaporation of the solvent. The lignin-oil is used as the feed for HDO reactions performed in dodecane at 300 ˚C under 50 bar hydrogen atmosphere over the CoMo/Al2O3 and Mo2C/CNF catalysts that were tested in Chapters 6 and 7 for the HDO of lignin model compounds. Analysis of the product mixtures after step 1 and step 2 shows a decrease in oxygen content after the HDO reaction for all different lignin sources. Oxygen-free aromatic products such as toluene and xylene are observed only after the second HDO step. This shows that depolymerized lignin-feeds obtained by LPR are suitable for further processing and that the CoMo/Al2O3 and Mo2C/CNF-catalyzed HDO reactions can be performed on real lignin feeds. Additional HDO experiments with Mo2C/CNF performed on organosolv lignin show that increasing the temperature to 350 ˚C leads to a higher degree of depolymerization. Importantly, no tris-oxygenated products were present in the product mixture anymore after HDO at this elevated temperature. Direct HDO of organosolv lignin over Mo2C/CNF is furthermore proven to be less effective than consecutive LPR and HDO.


8.2.1 Step 1: Liquid-phase reformingOrganosolv lignin, kraft lignin and sugarcane bagasse were depolymerized through

the LPR process previously described in Chapter 3. In a typical reaction, 1 g of dried lignin was dispersed in 50 mL ethanol/water (1/1 v/v) with 0.5 g of a commercial 1% Pt/γ-Al2O3 catalyst and 1 g NaOH as co-catalyst. The reaction was performed

Figure 8.1: Two-step approach for the valorization of lignin.

Step 1 LPR

Lignin-oil HDOproducts

Organosolv,

Bagasse

ethanol/water Pt/Al₂O₃225 C, 58 bar Ar, 2 h

ethanol/water Pt/Al2O3225 C, 58 bar Ar, 2 h

dodecane a. CoMo/Al₂O₃ b. Mo₂C/CNF300 C, 55 bar H₂, 4 h

˚

GPC-analysis

GC-analysis

Step 2 HDO

˚

Lignin

Kraft &


199

for 2 h under 58 bar Ar at 225 ˚C. As previously discussed, the solvent combination ethanol/water is used in the LPR process to aid lignin solubilization and to prevent (re)condensation. An acid or base is used as co-catalyst to increase monomer yields. Although sulfuric acid was shown in Chapter 3 to be the most effective co-catalyst, here we chose to perform our reactions with NaOH as co-catalyst. The solubility of lignin in basic solutions prevents coke formation at higher lignin concentrations and the use of NaOH simplifies the isolation of lignin after reactions. Compared to the reactions described in Chapter 3, the lignin loading was increased to 1 g to obtain a sufficient amount of lignin-oil for further processing in the second HDO step. To this extent, the lignin concentration in the solvent, as well as the lignin to catalyst ratio were doubled compared to the typical LPR conditions. After the reaction, any solids were removed by filtration and a lignin-oil was obtained by extracting the acidified product mixture with dichloromethane followed by solvent evaporation. Reactions performed on sugarcane bagasse contained larger amount of insoluble matter (0.2 g) than the reactions performed with organosolv and kraft lignin (less than 0.02 g). Complete evaporation of carried over ethanol proved to be difficult and small quantities were present in the lignin-oil that was used for the second step.

Almost no lignin-derived solids were obtained after the LPR reaction because of the excellent solubility of lignin in alkaline solutions; also the lignin-derived LPR products were shown to be almost completely soluble in ethanol water even after acidification to pH 1. Lignin-oil yields were generally around 0.8 g for organosolv and kraft lignin; bagasse yielded less oil (around 0.65 g) because of the insoluble ash that was removed in the filtration step.

GPC analysis of the crude reaction mixture after simple dilution of the LPR reaction mixture with alkaline H2O reveals a decrease in molecular weight for all three types of lignin. The GPC chromatograms depicted in Figure 8.2 show the formation of a shoulder at longer retention times, indicating the formation of lower molecular weight molecules. Only for the kraft lignin a shift of the peak maximum towards longer retention times was observed, which indicates that all polymer chains, including the larger ones, were shortened. These results correspond to the results reported in Chapter 3, which also showed LPR of kraft lignin to be most effective. Calculated average molecular weights (Mw) and polydispersity indices (PDI) of the lignins and depolymerized lignins are shown in Table 8.1. Despite the 30-50% decrease in average molecular weight that was observed after the LPR reaction, the observed Mw of 2200-3000 Da of the lignins after reaction indicate that a large fraction of the product

Part III, Chapter 8

200

mixture still consists of larger lignin molecules. The increase in signal intensity at the longest retention times, i.e. around 12 min, nonetheless revealed that low molecular weight monomers and dimers are formed. The retention time of phenol under these conditions is 12.2 min and any molecules eluting at around 12 min are most likely monomers.

The monomer yields and oxygen content after step 1 were estimated by GC analysis of the lignin-oil using anisole as the internal standard. Typical monoaromatic product yields were 11%, 9% and 5% for organosolv, kraft and sugarcane bagasse lignin, respectively. These yields are comparable to the results reported in Chapter 3 (12%, 13% and 2%) for the same lignin sources in the LPR reaction with NaOH co-catalyst.

A large number of mono-aromatics are obtained from the LPR reaction of organosolv and kraft lignin, including methyl-, ethyl-, propyl- and propanol-substituted guaiacol and syringol-type products as well as catechols, phenol and some ethanol-substituted aromatic products. In contrast, the bagasse-derived product mixture contained only methyl- and ethyl-substituted guaiacol, phenol and syringol. A possible explanation for the relatively simple product mixture formed after the LPR of sugarcane bagasse is its relatively low natural abundance of syringyl units. Indeed, the product distribution




a. b. c.

Figure 8.2: GPC chromatograms of a) organosolv, b) kraft and c) bagasse before (grey) and after (black) LPR. The peak maxima were normalized to the same value.

Table 8.1: Mw and PDI decrease for organosolv, kraft and bagasse lignin after LPR with NaOH as co-catalyst.

Mw before (Da)

Mw after (Da)

Decrease (%)

PDI before PDI after Decrease (%)

organosolv 3900 2700 32 4.8 5.6 -17kraft 5000 2200 57 5.9 5.5 7

bagasse 3000 2200 27 4.3 5.8 -35

Table 8.1: Mw and PDI decrease for organosolv, kraft and bagasse lignin after LPR with NaOH as co-catalyst.


201

and oxygen content of the observed monomers generally correspond to the natural abundance of syringyl, guaiacyl and p-coumaryl units in the original lignin samples. The monomeric products observed by GC in the lignin-oil originating from hardwood lignin consisted for 58% of tris-oxygenated, syringol-like molecules. In the kraft lignin LPR product, 59% was bis-oxygenated as can be expected from a softwood lignin, while the grass-type bagasse yielded a mixture of 42% bis-oxygenated and 38% mono-oxygenated phenolics. The composition of the product mixture of the organosolv and kraft lignin LPR reactions is comparable to the products obtained in the LPR of these lignins with NaOH as a co-catalyst as reported in Chapter 3. Depending on the lignin source, the same major products, i.e. guaiacol, syringol and alkyl-substituted guaiacol and syringol were observed. The reaction mixture obtained after the LPR of sugarcane bagasse showed some slight changes, however, compared to the results obtained in Chapter 3. Although higher lignin concentrations were used, higher yields of the less-substituted molecules were obtained than before. Figure 8.3 shows the monomer yields that were obtained during the first-step LPR, products are divided in groups that contain the same number of oxygen atoms.

8.2.2 Step 2: HydrodeoxygenationThe lignin-oils obtained after LPR of organosolv, kraft and sugarcane bagasse

lignin consist of highly oxygenated oligo- and mono-aromatic molecules. To lower the oxygen content of the lignin-derived product a subsequent HDO reaction was performed. Various types of catalysts have been previously reported for HDO of lignin model compounds, ranging from CoMo and NiMo on alumina catalysts to noble metal catalysts. [15, 20, 21] We choose to perform the HDO of lignin LPR-oils over CoMo/Al2O3 and Mo2C/CNF catalysts at 300 ˚C in dodecane. Reactions were performed in a 25 mL autoclave under 50 bar hydrogen atmosphere for 4 h; these conditions are comparable to those used in Chapter 6 for the HDO of lignin model compounds over CoMo/Al2O3. The CoMo/Al2O3 catalyst was activated prior to use and the Mo2C/CNF catalyst was freshly prepared; both catalysts were stored under an argon atmosphere to prevent deactivation over time. After reaction, the formation of solids was observed, probably caused by precipitation of the larger lignin molecules still present in the lignin-oil, components that are not soluble in dodecane. Individual catalyst particles were still visible and could easily be distinguished from the formed insoluble organics.

Analysis of the liquid phase of the CoMo/Al2O3-catalyzed reactions showed the formation of (alkylated) syringol, guaiacol and phenol-type molecules. The composition of the product mixtures after the LPR and HDO step is again depicted

Part III, Chapter 8

202

in Figures 8.3 and 8.4. Total observed monomer yields were 6%, 6% and 5% for organosolv, kraft and sugarcane bagasse lignin, respectively. The yields of organosolv and kraft HDO reaction products are thus lower than for the LPR lignin-oils; monomer yields for the bagasse reaction products, however, increased slightly after the HDO reaction. The presence of more monomers in the HDO product than in the LPR oil might indicate that the HDO reaction is able to contribute to further depolymerization of the lignin (albeit to a small extent), which makes it possible to use depolymerized lignin feeds that still contain dimers and small oligomers. Notably, the HDO product mixture also contained oxygen-free aromatic products such as benzene, toluene, xylenes and different ethylmethylbenzenes. The oxygen-free products comprise of 25%, 15% and 20% of the total observed products for organosolv, kraft and sugarcane bagasse lignin, respectively. None of these oxygen-free aromatics were present in the lignin-oils obtained from the LPR reaction, illustrating that the oil was at least partially deoxygenated during the HDO step. As found for the LPR reaction, the ratio between the mono-, bis- and tris-oxygenated products strongly depended on the lignin source used. Indeed, almost 50% of the observed aromatic monomers obtained from organosolv lignin are tris-oxygenated, while the kraft lignin-derived product

Figure 8.3: Yield of monomeric aromatic products after LPR and LPR + HDO reactions of different lignins: oxygen-free products, mono-oxygenated products, products containing two oxygen atoms, products with three or more oxygen atoms.

0

2

4

6

8

10

12

LPR

LPR

+ H

DO (M

o₂C/

CNF)

LPR

+ H

DO (M

o₂C/

CNF)

LPR

+ H

DO (M

o₂C/

CNF)

LPR

LPR

LP

R +

HDO

(CoM

o/Al₂O

₃)

LP

R +

HDO

(CoM

o/Al₂O

₃)

LP

R +

HDO

(CoM

o/Al₂O

₃)

organosolv kraft bagasse

Mon

omer

Yie

ld (w

t%)


203

Figure 8.4: Typical products obtained after HDO reactions of the various lignin-oils over the Mo2C/CNF and CoMo/Al2O3 catalysts.

consisted for 63% of bis-oxygenated compounds. Mono-oxygenated phenolics were the largest fraction (36%) in case of the bagasse. It should be noted that the removal of oxygen leads to a reduction of molecular weight, which can partially explain the decrease in observed monomer yields. In the organosolv lignin-derived LPR product mixture 23% of the total observed monomer weight was made up by the oxygen atoms in the various products, after HDO this was only 16%. Theoretically, this decrease of 7% results in a decrease of 0.7% in the total monomer yield.

The reaction mixtures obtained after Mo2C/CNF-catalyzed HDO of the lignin-oils generally showed the same product distribution as observed for the CoMo/Al2O3-catalyzed reactions. The composition of the product mixture is shown in Figure 8.3 and 8.4. With 9%, 7% and 6% observed total monomer yields for organosolv, kraft

Mono-oxygenated

Bis-oxygenatedTris-oxygenated

Oxygen-free

OHOO

OHOO

OHOO

OHO

OHO

OHO

OH OH

OH

Part III, Chapter 8

204

and sugarcane bagasse lignin, respectively, the yields form the Mo2C/CNF-catalyzed reactions were slightly higher than for the CoMo/Al2O3-catalyzed reactions. The 9% yield obtained after the HDO of organosolv lignin-oil is the highest obtained; the high yield is mainly caused by an increase in the amount of oxygen-free products to 36% of the total observed aromatic products, whereas in the CoMo/Al2O3-catalyzed reactions not more than 25% oxygen-free products were observed. The reduction in total mass of the products by loss of oxygen in the HDO step again accounts in part for the lower monomer yields obtained in step 2 compared to step 1. For example, the total amount of oxygen present dropped from 23 wt% in the lignin-oil to 18 wt% in the HDO product of the organosolv lignin, corresponding to a 0.5% decrease in monomer yield. The relative amounts of oxygen-free products observed in the kraft (15%) and sugarcane bagasse-derived product mixtures (16%) were comparable to the amounts obtained with the CoMo/Al2O3 catalyst. In accordance with the results discussed above for the LPR and the CoMo/Al2O3-catalyzed reactions, tris-oxygenated products (44%) constituted the largest product fraction for organosolv lignin, bis-oxygenated products (66%) for kraft lignin and mono-oxygenated phenolics (30%) for bagasse.

Somewhat surprisingly, for both the CoMo/Al2O3 and Mo2C/CNF catalysts relatively small amounts of mono-oxygenated products were observed in the HDO product mixtures of organosolv and kraft lignin-oils. In contrast, mono-oxygenated phenolics were shown to be the major products in the HDO of lignin model compounds over these catalysts in Chapter 6 and 7; full HDO to oxygen-free compounds such as benzene, toluene and xylene was only found to occur after all bis-oxygenated model compounds were converted. However, HDO of the lignin-oils led to the formation of oxygen-free products even before all bis- and tris-oxygenated products were consumed. The concentration oxygenated aromatic molecules in the second step is lower than the concentrations used for the model compound studies in Chapter 6 and 7. This lower concentration leads to a more limited availability of substrates for the HDO catalyst and possibly explains the difference in behavior in the model compounds studies compared to the conversion of the lignin-oil. In addition, the products obtained from the HDO of real lignin-oils contained larger amounts of alkylated products compared to the model compound product mixtures. The propyl side chains that are present in the original lignin structure lead to the formation of large amounts of methyl-, ethyl- and propyl-substituted aromatics. Since no alkyl side chains are present in the model compounds, no propyl- or ethyl-substituted products were formed in these reactions.

An HDO experiment with Mo2C/CNF as the catalyst performed on organosolv


205

lignin directly, i.e., without an LPR pretreatment, confirmed that small amounts of monomeric products can also be formed by HDO alone (Figure 8.5). The total amount of monomeric products observed (6%), however, was lower than the 9% observed after the HDO of the LPR-oil from the same lignin. Also, the product distribution of this “direct HDO” reaction is more similar to the product distribution of the first LPR step than to the mixture obtained in the second HDO step. Only 3% of the observed products were completely deoxygenated after direct HDO of lignin compared to 36% after LPR followed by HDO, whereas 63% still had 3 oxygen functionalities after direct HDO. This indicates that consecutive depolymerization and deoxygenation steps result in a higher deoxygenation degree than a direct HDO reaction alone.

8.2.3 Reaction optimizationAs the highest monomer yields and lowest oxygen content were obtained with

the Mo2C/CNF catalyst and the organosolv lignin-oil, additional tests to assess the influence of process parameters on the yield and oxygen content of the product mixture were performed with this lignin-catalyst combination.

0

2

4

6

8

10

12

Mon

omer

Yie

ld (w

t%)

EtOAC extr

acted

Figure 8.5: Yield in monomeric aromatic products after HDO reactions of organosolv lignin-oil over the Mo2C/CNF catalyst under different conditions; oxygen-free products, mono-oxygenated products, products containing two oxygen functionalities, products with three or more oxygen functionalities. Unless stated otherwise, the reactions were extracted with DCM and performed for 4 h at 300˚C.

Part III, Chapter 8

206

The results of the HDO of guaiacol over Mo2C/CNF described in Chapter 7 showed that by increasing reaction time and reaction temperature, higher amounts of completely deoxygenated products could be obtained. Increased reaction times possibly also help to increase the total monomer yields by additional cleavage of dimer and oligomer linkages. An HDO reaction of organosolv lignin-oil performed for 15 h rather than 4 h at 300 ˚C over Mo2C/CNF showed no increase in the yield of monomeric products, however (Figure 8.5). In fact, after a reaction performed overnight the total monomer yield (5.6%) and the fraction of oxygen-free products (16%) were significantly lower than after the 4 h reaction. This shows that higher monomer yields of this reaction cannot simply be achieved by increasing reaction time.

After performing the 4 h HDO of organosolv lignin-oil over Mo2C/CNF at 350 ˚C instead of 300 ̊ C (Figure 8.5), a total monomer yield of 6% was obtained. This is lower than for the reaction at 300 ˚C where a total yield of 9% of aromatic monomers could be identified. The oxygen content of the product mixture, however, was reduced to 14 wt%, representing a weight loss of 1%. No products with tris-oxygenated products could be observed anymore and only 24% of the total observed monomeric products consisted of bis-oxygenated molecules. In the reactions at 300 ˚C with the organosolv lignin-oil the tris-oxygenated products were always the major products. In Chapter 7, the Mo2C/CNF catalyst was shown to be more active at higher temperatures in the HDO of guaiacol. The results reported here show that also in the HDO of LPR-

10 30 50 70 902θ (˚)

35 40 45 50 55 602θ (˚)

a.

b.

b.

a.

●▲▲

▲■

Figure 8.6: X-ray diffraction patterns of a) the fresh Mo2C/CNF catalyst, b) the Mo2C/CNF catalyst after HDO of organosolv lignin-oil at 350 ˚C. Reflections are given for graphitic carbon ( ), MoO2 ( ) and Mo2C ( ). [30]


207

derived lignin-oils products with a lower oxygen content can be obtained at higher temperatures, albeit in lower overall yield. The X-ray diffraction patterns of the spent catalyst recovered after the 350 ˚C run show that the Mo2C phase was still present (Figure 8.6). This indicates that this catalyst, which was reported to be stable during the HDO of model compound guaiacol (Chapter 7), also remained stable during the HDO of lignin depolymerization oils.

From a green chemistry perspective, it is important to reduce the amounts of solvents used in a process and if solvents are required to choose, if possible, less toxic alternatives than dichloromethane. [22] The use of dichloromethane for the extraction of the lignin-oil after LPR, when also less hazardous solvents such as ethyl acetate are available, is therefore not desirable. The two-step conversion of organosolv lignin was therefore repeated, but now with extraction of the LPR reaction mixture with ethyl acetate. The HDO step was performed with Mo2C/CNF at 350 ˚C and the results are shown in Figure 8.5. The total monomer yield at the end of the second step amounts to only 4%, which is lower than the 6% previously obtained in a DCM-extracted reaction under the same conditions. The oxygen content of the observed products was comparable to the DCM-extracted reaction with no tris-oxygenated compounds and 23% oxygen-free products. It should be possible to obtain comparable yields and activities after further optimization of the extraction step for a cleaner solvent such as ethyl acetate, however.

The results show that a combination of two appropriate steps can potentially lead to a process for the production of low oxygen-content aromatic chemicals from lignin. Further process optimization is needed for both the LPR and HDO step to obtain higher yields and selectivities. Higher degrees of depolymerization in the LPR step are, for instance, expected to lead to higher monomer yields in both steps. It is very likely that the degree of deoxygenation during the HDO step can be optimized by variation of reaction conditions such as concentration, temperature and reaction time. The amount of solids that are found after the HDO step can be reduced when lignin-oils with a larger degree of depolymerization are used or when a solvent is used that is able to dissolve also the larger lignin fractions. Only the smaller molecules dissolve in the dodecane during de HDO reaction. Furthermore, precipitation of the larger molecules before extraction by changing the ethanol/water ratio of the solvent or separating the smaller from the larger molecules by destillation will probably lead to lower lignin-oil yields but cleaner HDO reaction mixtures. The flexibility of the two-step approach allows replacement of the first LPR step by other processes that produce depolymerized lignins. The product slate of lignin-oil obtained by base-

Part III, Chapter 8

208

catalyzed depolymerization in particular shows great similarity to the products obtained by LPR and has been reported to contain higher amounts of monomeric compounds. [13, 14, 23, 24]

The production of aromatics with little or no oxygen from lignin has generally proven to be very difficult. Successful methods reported for lignin depolymerization so far typically lead to the formation of oxygen-rich aromatic monomers with at least two or three oxygen functionalities. Although it has been reported before that benzene and phenolics could be formed from kraft lignin in processes such as hydrocracking, the harsh reaction conditions employed result in a loss of over 30% of the original lignin by the formation of gasses such as CO and methane. [25] Related to the work described here, the hydrodeoxygenation of pyrolysis oil for the production of second generation biofuels has been reported by several groups. Several nickel-based catalysts were used to lower the oxygen content and increase the H/C ratio. [26-28] Although no product compositions were reported, HDO of these of pyrolysis oils over nickel catalysts has been shown to result in the formation of mainly ring-hydrogenated products. [29]

For the production of BTX and phenol from lignin a two-step process is desired, however, which combines a high depolymerization activity with a HDO catalyst that does not show ring-hydrogenation activity. By combining the LPR reaction with a sequential HDO step over a CoMo/Al2O3 or Mo2C/CNF catalyst, we have been able to produce BTX without loss of aromaticity.

8.3 Conclusions

A two-step process for the catalytic conversion of lignin into bulk aromatic chemicals is presented. In the first step, the molecular weight of various sources of lignin is reduced via the LPR reaction resulting in the formation of monomers, dimers and small oligomers. The lignin-oil obtained after extraction was successfully used as substrate in a second HDO step. During the HDO of the depolymerized lignin oil, products with a lower oxygen content were obtained and yields of up to 9% monomeric aromatic products of which 24% were oxygen-free products. The oxygen content of the lignin-oil obtained after LPR and of the HDO product greatly depends on the lignin source. Product mixtures from hardwood lignins were shown to have a higher oxygen content than products obtained from softwood or grass type lignins. At increased temperatures of 350 ̊ C the total oxygen content could be lowered even more


209

with complete conversion of all molecules containing three oxygen functionalities even when starting with hardwood lignin. The majority of the products are mono-oxygenated phenolics. The two-step approach proved to be more effective for the production of oxygen-free products than the separate steps. These results show the potential of the approach for further development and optimization of a practical process for the production of aromatic chemicals from lignin.

8.4 Experimental Section

8.4.1 Chemicals and ligninsThe INDULIN AT kraft lignin (63.25% C, 6.05% H, 0.94% N, 1.64% S, 28.12% O by

difference), provided by ECN, was obtained from pine and is free of all hemicellulosic materials. The Alcell organosolv lignin (66.47% C, 5.96% H, 0.15% N, 27.43% O by difference) provided by Wageningen University, was obtained from hardwoods and isolated by the organosolv extraction method. The lignin from sugarcane bagasse (58.90% C, 4.90% H, 0.14% N, 1.53% S, 34.53% O by difference), provided by Dow Chemical, was derived from Brazilian sugarcane. Demineralized water was purified using a MilliQ system. Reagents, catalyst, solvents and gasses were purchased commercially; NaOH (Merck), HCl (Merck, 37%), ethanol (Scharlau, HPLC grade), dichloromethane (Biosolve), ethyl acetate (Biosolve), dodecane (Acros, 99%), diethylether (Biosolve), hexadecane (Sigma, 99%), anisole (Acros, 99%), Pt/Al2O3 (1% Pt, Aldrich), argon (Linde, 5.0).

8.4.2 Catalytic reactionsThe liquid phase reforming and reduction reactions were conducted in a 100

mL stainless steel high-pressure Parr batch autoclave reactor equipped with a thermocouple, a pressure transducer and gauge and a magnetic driver (750 rpm). During a typical liquid phase reforming reaction, 1 g of lignin was added to the autoclave along with 0.5 g Pt/Al2O3, 25 mL 1 M NaOH in water and 25 mL ethanol. The reactor was purged three times with argon and pressurized with argon to 58 bar. The reaction mixture was heated to 225 ˚C and depressurized regularly to maintain a pressure of around 58 bar. The reaction was stopped after 2 h by cooling rapidly to room temperature. The reaction mixture was filtered to remove all solids and subsequently acidified with HCl to pH 1. After a second filtration under reduced pressure to remove any precipitated lignin the liquid phase was extracted with dichloromethane. After evaporation of the dichloromethane and ethanol a lignin-oil was obtained.

Part III, Chapter 8

210

The HDO reactions were performed in a 25 mL stainless steel high-pressure Parr batch autoclave reactor. The temperature was monitored using a thermocouple, and stirring was performed using a magnetic driver equipped with an impellor at 750 rpm. In a typical reaction, the autoclave was loaded with the 500-800 mg of the lignin-oil, 50 mg of catalyst material, 0.35 g hexadecane as internal standard, and 7.5 g of the solvent dodecane. A fresh batch of catalyst was used in every run. The reactor was purged three times with argon, and the reaction mixture was heated to 300 ˚C, then pressurized with H2 to 50 bar, and the catalytic reaction was carried out for 4 h. The reaction was stopped by cooling and release of pressure, after which the reaction mixture was diluted with an equal volume of diethyl ether in order to dissolve all of the products, condensed lignin solids and the solid catalyst were removed by filtration and washed with diethyl ether.

8.4.3 AnalysisThe chemical composition of the isolated product mixtures was determined by

a Varian GC equipped with a VF-5ms capillary column and an FID detector. Anisole was used as an internal standard for the LPR reactions. The products were quantified using response factors determined for phenol, guaiacol and syringol. Product identification was conducted using a Shimadzu GCMS-QP2010 equipped with a VF-5ms capillary column and by comparison with pure compounds when available. The Mw of depolymerized lignin was analyzed by GPC performed on an alkaline SEC by Waters Alliance system equipped with a manually packed column (4.6 mm x 30 cm) with ethylene glycolmethacrylate copolymer TSK gel Toyopearl HW-55F, according to the work of Gosselink et al. [31] Sodium polystyrene sulfonates (Mw range 891 to 976,000 Da) were used for calibration of the molar mass distribution. The LPR lignin solutions were diluted to a concentration of 1 mg/mL lignin in 0.5 M NaOH. GPC runs were performed at 40 ˚C with 0.5 M NaOH eluent at a flow rate of 1 mL/min and UV detection at 280 nm.

X-ray powder diffraction (XRD) patterns were obtained on a Bruker-AXS D2 Phaser powder X-ray diffractometer using Co Kα1,2 with λ = 1.79026 Å. Measurements were carried out between 10 and 100° 2θ using a step size of 0.09° 2θ and a scan speed of 1 s.


211

8.5 Acknowledgments

We would like to thank Jaap van Hal (ECN), Richard Gosselink (Wageningen University) and Matthijs Ruitenbeek (The Dow Chemical Company) for supplying the kraft, organosolv and sugarcane lignin, respectively. Rob Gosselink is thanked for the preparation of the Mo2C/CNF catalyst. Albemarle Catalysts is kindly acknowledged for providing the CoMo/Al2O3 catalyst.

8.6 References

[1] H. Werhan, J. M. Mir, T. Voitl, P. R. von Rohr, Holzforschung 2011, 65, 703-709.[2] K. Stärk, N. Taccardi, A. Bösmann, P. Wasserscheid, ChemSusChem 2010, 3, 719-723.[3] www.borregaard.com.[4] Q. Song, F. Wang, J. Cai, Y. Wang, J. Zhang, W. Yu, J. Xu, Energy Environ. Sci. 2013, 6, 994-1007.[5] C. Li, M. Zheng, A. Wang, T. Zhang, Energy Environ. Sci. 2012, 5, 6383-6390.[6] W. Xu, S. J. Miller, P. K. Agrawal, C. W. Jones, ChemSusChem 2012, 5, 667-675.[7] P. T. Patil, U. Armbruster, M. Richter, A. Martin, Energy Fuels 2011, 25, 4713-4722.[8] K. Barta, T. D. Matson, M. L. Fettig, S. L. Scott, A. V. Iretskii, P. C. Ford, Green Chem. 2010, 12, 1640-1647.[9] R. Y. Nsimba, C. A. Mullen, N. M. West, A. A. Boateng, ACS Sustainable Chem. Eng. 2013, 1, 260-267.[10] P. R. Patwardhan, R. C. Brown, B. H. Shanks, ChemSusChem 2011, 4, 1629-1636.[11] H. Ben, A. J. Ragauskas, ACS Sustainable Chem. Eng. 2013, 1, 316-324.[12] M. P. Pandey, C. S. Kim, Chem. Eng. Technol. 2011, 34, 29-41.[13] V. M. Roberts, V. Stein, T. Reiner, A. Lemonidou, Chem. Eur. J. 2011, 17, 5939-5948.[14] J. M. Lavoie, W. Baré, M. Bilodeau, Bioresource Technol. 2011, 102, 4917-4920.[15] E. Furimsky, Appl. Catal. A: Gen. 2000, 199, 147–190.[16] J. S. Shabtai, W. W. Zmierczak, E.Chornet, 1999, WO Patent 9910450.[17] Q. Bu, H. Lei, A. H. Zacher, L. Wang, S. Ren, J. Liang, Y Wie, Y. Liu, J. Tang, Q. Zhang, R. Ruan, Bioresource Technol. 2012, 124, 470-477.[18] J. C. Hicks, J. Phys. Chem. Lett. 2011, 2, 2280-2287.[19 H. Ben, W. Mu, Y. Deng, A. J. Ragauskas, Fuel 2013, 103, 1148-1153.[20] C. R. Lee, J. S. Yoon, Y.-W. Suh, J.-W. Choi, J.-M. Ha, D. J. Suh, Y.-K. Park, Catal. Commun. 2012, 17, 54-58.[21] Y. Want, T. He, K. Liu, J. Wu, Y. Fang, Bioresource Technol. 2012, 108, 280-284.

Part III, Chapter 8

212

[22] T. Laird, Org. Process. Res. Dev. 2012, 16, 1-2.[23] A. Vigneault, D. K. Johnson, E. Chornet, Can. J. Chem. Eng. 2007, 85, 906-916.[24] A. Toledano, L. Serrano, J. Labidi, J. Chem. Technol. Biotechnol. 2012, 87, 1593-1599.[25] D. T. A. Huibers, H. Jonson, 1983, DE Patent 3228897.[26] A. R. Ardiyanti, S. A. Khromova, R. H. Venderbosch, V. A. Yakovlev, I. V. Melián- Cabrera, H. J. Heeres, Appl. Catal. A: Gen. 2012, 449, 121-130.[27] N. Joshi, A. Lawal, Chem. Eng. Sci. 2012, 74, 1-8.[28] I. Graça, J. M. Lopes, H. S. Cerqueira, M. F. Ribeiro, Ind. Eng. Chem. Res. 2013, 52, 275- 287.[29] C. Zhao, J. A. Lercher, Angew. Chem. Int. Ed. 2012, 51, 5935-5940.[30] ICDD PDF-2 database package http://www.icdd.com/products/pdf2.htm.[31] R. J. A. Gosselink, J. E. G. Van Dam, E. De Jong, E. L. Scott, J. P. M. Sanders, J. Li, G. Gellerstedt, Holzforschung 2010, 64, 193–200.

Chapter 9a

Summary and Concluding Remarks

Chapter 9a

214

Summary

The chemical industry currently depends heavily on fossil resources for the production of its essential building blocks. With the depletion of fossil fuels and increasing environmental awareness, there is much interest in biomass as an alternative to fossil feedstocks for the more sustainable production of renewable fuels and chemicals. Moreover, as a result of the recent ‘shale gas revolution’, the production of lighter fossil fuel fractions is growing compared to the heavier crude oil, a shift in feedstock that has major consequences for downstream refining operations. Indeed, the availability of those commodity chemicals produced by steam cracking of these heavier fractions will decrease. New, dedicated routes for the on-purpose production of particularly butadiene and BTX, preferably from renewable resources, need to be developed.

The major and most sustainable source of biomass for the industrial production of renewable fuels and chemicals is non-edible, lignocellulosic material that is obtained, for instance, from purposefully grown crops or agricultural waste. Lignocellulosic biomass consists mainly of three components, the carbohydrate polymers cellulose and hemicellulose and an aromatic polymeric fraction called lignin. The conversion of lignocellulose to fuels and chemicals takes place in a so-called biorefinery. In analogy to the petrochemical refinery, all components of lignocellulosic biomass need to be valorized in order to limit the generation of waste and to make the biorefinery economically viable. While commercial processes have already been developed for the cellulose and hemicellulose fractions, the lignin fraction has received much less attention and only a few examples of successful lignin valorisation are available. Chemically, lignin is completely different from the carbohydrate fraction of lignocellulose and its chemical recalcitrance actually makes it a challenging feedstock for the production of anything of added value. Lignin is an amorphous polymer, containing many different chemical functionalities in a structure that varies with the plant species it is extracted from and the pre-treatment method used for this extraction. It can be found in the plant cell walls where it gives strength and rigidity to the plant and protects the cells from microbial attack, as a result of which lignin is difficult to break down and recalcitrant towards most chemical transformations. In addition to this structural heterogeneity and general recalcitrance, the low solubility of lignin in most common solvents also makes it a difficult substrate to work with. The structural intricacies of lignin and challenges associated with its conversion are detailed in the literature review presented in Chapter 2, which both covers the


215

properties of lignin, such as structure, type of linkages and solubility, as well as its reactivity in catalytic reactions. Although the older literature on lignin and lignin conversion is somewhat scarce and scattered, much has been done over the years and a recent increased interest has led to a large growth in the number of publications on the topic.

Many of the studies on catalytic lignin conversion involve the use of catalysts that were originally developed and later optimized for the conversion of fossil feedstocks. However, for the conversion of a ‘new’ oxygen-rich substrate such as lignin, also new catalysts showing high catalyst activity, selectivity and stability under typical lignin conversion conditions need to be developed. An overview of the literature also shows that many of the research efforts regarding catalytic lignin conversion are concerned with the use of model compounds mimicking different parts of the lignin structure rather than real lignin feeds to reduce the complexity of the process and to simplify product analysis. Although a lot can be learned from model compound studies, the structural complexity of the lignin macromolecule cannot be caught in a simple model compound. One can conclude that, although much work has already been done, efficient processes for the conversion of real lignins are still lacking and fundamental knowledge of the required properties of suitable catalysts is limited.

The aim of the work described in this thesis was 1) to develop a two-step conversion route for the production of bulk chemicals such as phenolics or BTX from lignin and 2) to gain more insight into the properties required of a good catalyst in terms of activity, selectivity and stability in the lignin conversion processes. The two-step route for the production of bulk aromatic chemicals from lignin studied here comprises a catalytic depolymerisation reaction as the first step, resulting in a mixture of oxygen-rich aromatic monomers, followed by a hydrodeoxygenation step to reduce the oxygen content of the mixture and limit the amount of final products. With this aim, different reactions were developed and tested on real lignin samples obtained from various sources.

Part I of this thesis is concerned with studies of two different routes for the catalytic depolymerization of lignin. A graphical depiction of the different routes is shown in Figure 9.1. Chapter 3 describes the solubilization, liquid phase reforming and reductive depolymerization of organosolv, kraft and sugarcane bagasse lignin. It was found that all different types of lignin readily dissolve in a 1/1 ethanol/water mixture when heated to 225 ˚C making this a generic method that can be applied

Chapter 9a

216

to all kinds of lignin. GPC and quantitative HSQC NMR analyses of the solubilized lignins showed that solubilization takes place by cleavage of the lignin ether linkages, primarily the predominant β-O-4 linkage, resulting in a decrease in lignin molecular weight. The solubilized lignin could now be subjected to various catalytic conversions. In a variation on the well-known aqueous phase reforming process, Pt/Al2O3 in combination with sulphuric acid, phosphotungstic acid or NaOH as a co-catalyst is used as the catalyst for the depolymerization of lignin to monomeric aromatic molecules in a process coined liquid phase reforming (LPR). Higher yields and very low formation of solids were observed compared to the aqueous-phase reforming of lignin, which takes place in water alone. This was attributed to ethanol acting as a capping agent to prevent recondensation reactions from taking place. An acid or base co-catalyst aided in lignin hydrolysis, with the best choice of co-catalyst depending on the particular lignin feedstock. Maximum yields of 17% monomeric aromatic products were obtained from kraft lignin using sulphuric acid as a co-catalyst. Reduction of solubilized lignin over Pt/Al2O3, Ru/C, Pd/C and Ni/SiO2 catalysts yielded up to 6% monomeric aromatic products. For both reactions the product slate and yields depended heavily on the type of lignin and the catalyst used. Generally the organosolv lignin that was derived from hardwood produced a large quantity of syringol-like products, whereas the kraft lignin from softwood mainly yielded guaiacol and the use of grass-type bagasse resulted in a substantial amount of phenol-type products.

The stability of the Pt/γ-Al2O3 catalyst under the reaction conditions used in Chapter 3 for the liquid phase reforming reaction is the topic of Chapter 4. The stability of the support and the metal phase were studied, as well as the formation of carbon-rich deposits on the catalyst surface. A multi-technique approach was taken to study the changes that took place after treating the catalyst in ethanol/water at LPR reaction temperatures and to determine the influence of lignin and model compounds representing components of a typical lignin depolymerization product on this process. Analysis of the catalyst after treatment with ethanol/water with XRD, 27Al MAS NMR and N2 physisorption showed that the γ-Al2O3 support is hydrated over time and loses surface area as it is transformed into boehmite. Some sintering was also observed, as the Pt particles were found to increase in size from 2 to 3 nm during this treatment. The addition of mono-aromatic compounds to the solution slowed down the formation of boehmite. This stabilization of the support is believed to take place via interaction with the support hydroxyl groups, thereby preventing water from accessing these sites. Monomers containing more oxygen atoms had a larger influence on support stability. Sintering of the Pt particles, on the other hand, increased when the phase


217

Figure 9.1: Two approaches to the first lignin depolymerization step described in Part I.

transformation of γ-Al2O3 to boehmite was slowed down more. H2 chemisorption measurements furthermore indicated that the Pt particles got encapsulated in the support during the stability tests. Remarkably, the addition of lignin to the reaction mixture completely stopped the conversion of γ-Al2O3 to boehmite and prevented sintering of the Pt particles. IR and TGA analysis showed that a lignin layer formed on the catalyst surface, protecting the catalyst by preventing the hydration of the support and sintering of the metal particles.

In Chapter 5 a bleaching method that was originally developed for the removal of residual lignin from cellulose pulps is applied to the oxidative depolymerization of lignin itself. The alkaline hydrogen peroxide treatment reduced the molecular weight of organosolv lignin by 25% overnight. While the addition of NaOH aided in the solubilization of lignin and the deprotonation of the peroxide, no influence of its concentration on the extent of depolymerization could be observed. Increasing the amount of peroxide, on the other hand, led to an increased rate of depolymerization.

Chapter 9a

218

Notably, the efficiency of the depolymerization reaction heavily depended on the type of lignin used. This was attributed to the different impurities in the various samples, and an increased concentration of free phenolic OH groups in some of the alkaline-treated lignins. Transition metal impurities can, for instance, influence the reaction by catalyzing the decomposition of hydrogen peroxide. Bleaching of residual lignin in cellulose pulps is thought to take place via a Dakin or Dakin-like reaction pathway, which involves the addition of a hydroperoxy anion to a lignin fragment containing a free phenolic OH group and a ketone on the α-carbon atom. It was shown in Chapter 3 that free phenolic OH groups are formed by cleavage of lignin ether linkages during the solubilization of lignin in ethanol/water. A lignin pretreatment by solubilization in ethanol/water before alkaline oxidation indeed resulted in a higher decrease of molecular weight to 32%. An attempt to increase the amount of α-ketones by applying a Co-catalyzed oxidation prior to alkaline oxidation, did not result in increased depolymerization. In contrast, the molecular weight of the lignin was found to increase during the Co-catalyzed oxidation reaction. A Dakin or Dakin-like mechanism should lead to the introduction of carboxylic acid groups on the lignin and the formation of monomers. Indeed, IR and NMR studies showed that large amounts of carboxylic acid groups were present in lignin after alkaline oxidation. The lignin that was treated with a Co-catalyzed oxidation did not only show formation of ketones, but also carboxylic acid groups. Furthermore, 31P NMR revealed that no free phenolic OH groups could be detected anymore after Co-catalyzed oxidation.

Part II is dedicated to the second step of the two-step conversion route of lignin to aromatics, i.e. a study of two hydrodeoxygenation (HDO) catalysts with lignin model compounds. Figure 9.2 shows the two different classes of catalysts that are studied in these chapters. In Chapter 6 the HDO of a library of model compounds over a commercial sulfided CoMo/Al2O3 catalyst is described. A complex reaction network was proposed based on these reactions. It was shown that the HDO of lignin-derived aromatic monomers such as guaiacol takes place via consecutive demethylation and deoxygenation reactions. The catalyst showed a very low ring hydrogenation activity and as a result led to the formation of predominantly aromatic products. Removal of the final phenolic –OH group was found to be most difficult with phenol and cresols as a result being relatively stable under reaction conditions. These mono-oxygenated aromatic molecules were therefore found to be the main products, regardless of the starting point in the reaction network. The reaction of three different lignin model dimers showed that both β-O-4 and β-5 ether linkages could be broken under HDO reaction conditions; the 5-5 linkage, however, was very stable and survived the


219

HDO conditions. Analysis of the spent catalyst after reaction with guaiacol showed a decrease in sulfur content and a loss of active sites, pointing at catalyst deactivation under reaction conditions.

In Chapter 7 the HDO of guaiacol over carbon nanofiber-supported tungsten and molybdenum carbides is reported. Carbide-based catalysts provide an attractive alternative to the use of traditional sulfur-containing catalysts such as CoMo/Al2O3 and supported noble metal catalysts, because they combine the low ring hydrogenation activity of the CoMo/Al2O3 catalyst with the high activity and stability of the supported noble metals. Higher conversions and yields to phenolics were obtained with these catalysts than with the sulfided CoMo/Al2O3 catalyst reported in the previous chapter. The selectivity to phenolics and mass balances were the highest reported to date. It was shown that the conversion of guaiacol to phenol takes place via a direct demethoxylation pathway rather than by sequential demethylation and hydrodeoxygenation steps as is the case for the CoMo/Al2O3-catalyzed reactions. The Mo2C/CNF catalyst proved to be much more active than the W2C/CNF catalyst with

Figure 9.2: Two different classes of catalyst have been studied for the HDO of lignin model compounds in Part II.

Chapter 9a

220

the faster conversion of guaiacol leading to a higher selectivity for the completely deoxygenated products benzene and toluene. In addition, higher mass balances were obtained as side reactions had less of a chance to take place. Recycling experiments showed that the Mo2C/CNF catalyst could be re-used without loss of activity and selectivity. Analysis of the catalyst with X-ray diffraction showed that the Mo2C phase is stable under reaction conditions, whereas the W2C phase was more prone to oxidation.

In Part III of this thesis the depolymerization reaction is combined with a hydrodeoxygenation step in a two-step lignin conversion process. A graphical summary of this two-step process is presented in Figure 9.3. Chapter 8 presents the liquid phase reforming of organosolv, kraft and sugarcane bagasse lignin over Pt/Al2O3 with NaOH as a co-catalyst. The reaction product was isolated as a lignin-oil by extraction and subsequently hydrodeoxygenated with the CoMo/Al2O3 and Mo2C/CNF catalysts that were studied in Part II. GC analyses of the lignin-oils obtained after LPR and before HDO showed that the highest yield (11%) of monomeric aromatic products was obtained from organosolv lignin. GPC analysis of the LPR reaction products, however, showed that a higher degree of depolymerization was achieved with the kraft lignin. Depending on the lignin that was used the main monomeric aromatic products that were detected were (alkylated) syringol, guaiacol and phenol-type molecules, reflecting the original monolignol composition of the various lignins. The oxygen content of the lignin-oils was reduced successfully by the hydrodeoxygenation reaction with both the CoMo/Al2O3 and Mo2C/CNF catalysts. After HDO reactions at 300˚C, the oxygen-free aromatic compounds toluene, xylene and ethylmethylbenzene were obtained. Importantly, direct hydrodeoxygenation of organosolv lignin without performing first the liquid phase reforming step, did not lead to the formation of oxygen-free products, indicating that the consecutive depolymerization and hydrodeoxygenation steps are a successful way to obtain low oxygen-content bulk aromatic chemicals from lignin. For all three lignins, the total product yields obtained with the Mo2C/CNF catalyst were higher than for the CoMo/Al2O3; the total amounts of oxygen-free products were comparable. The highest yield in monomeric aromatic products and oxygen-free aromatic products were obtained after the HDO of the organosolv lignin-oil over the Mo2C/CNF catalyst. Increasing the temperature of the reactions led to higher degrees of deoxygenation and after hydrodeoxygenation of organosolv lignin-oil at 350˚C, all tris-oxygenated monomeric compounds were completely converted and the majority of the obtained products were oxygen-free and mono-oxygenated aromatic monomers.


221

Figure 9.3: The two-step conversion of lignin via consecutive LPR and HDO reactions as described in Chapter 8, exemplified with organosolv lignin.

Concluding Remarks

The results presented in this thesis contribute to a better understanding of the various factors influencing the production of bulk aromatic chemicals from lignin. Valuable knowledge on catalyst activity and stability, optimal conditions for the solubilization of lignin that prevent recondensation and analytical procedures for the characterization of starting materials and products was obtained. Based on the results and insights gained, some general recommendations can be formulated.

Lignin is a recalcitrant biopolymer that is difficult to handle and difficult to analyze. As a result, it is essential to use combinations of analytical techniques that complement each other, for example by combining GC and GPC results that give information on the changes in molecular weight of the polymer and monomer formation with IR and NMR data that illustrate the type of chemical transformations that take place. In addition, it is essential to present data in a way that gives the most valuable information. Reporting, for example, lignin conversion in depolymerization reactions is ineffectual, because it is difficult to distinguish between simple solubilization and actual reactions that take

Chapter 9a

222

place; reporting yields of products or changes in molecular weight instead is much more valuable. Standardization of how to report data of catalytic lignin conversions, similar to a standardization of lignin characterization techniques as advocated by the International Lignin Institute is therefore desired. Furthermore, although the use of model compounds mimicking important lignin functionalities can provide valuable information, it is important to know the restrictions of this approach. The complex structure of the lignin polymer limits the use of molecules mimicking lignin linkages as model compounds for depolymerization reactions. As chemical transformations that are tested on pure models do not necessarily perform similarly in the real polymer, using these models to gain mechanistic insight in a reaction that has been shown to work on lignin might therefore be more meaningful.

Well-established catalyst systems are of course the obvious starting point for studies aimed at the development of catalytic routes for biomass valorization, but one has to keep in mind that these catalyst systems have originally been developed and optimized for the conversion of apolar hydrocarbons of fossil origin. These substrates contain limited amounts of oxygen and water and their conversion reactions are generally run under gas phase conditions. In contrast, the catalytic conversion of biomass-derived feeds involves highly polar and oxygenated substrates in processes that are typically run in the liquid phase in the presence of water. Indeed, several investigations on the stability of these catalysts during biomass conversion reactions, as well as the results reported in Chapter 4 and 6 of this thesis, have shown that the high water content can lead to leaching, sintering and transformations of the support. Ultimately, new types of catalysts will have to be developed and optimized specifically for the demanding conditions required for the conversion of biomass. These should have support materials that are stable in the presence of water, even in acidic or basic environment, as well as contain metal particles that are stable under oxidizing conditions. On the other hand, catalysts that are not stable under general reaction conditions might still have advantages in specific cases. The sulfided CoMo/Al2O3 catalyst, for example, leaches sulfur and deactivates during HDO reactions when no sulfur is added to the feed, as described in Chapter 6. Lignins isolated using a kraft pulping method, however, contain up to 2-3 wt% sulfur in their structure. Although the presence of sulfur in lignin can lead to poisoning of other catalysts, it can prevent deactivating of the CoMo catalyst. The use of a sulfided CoMo catalyst can therefore be a solution for these types of feeds.

The complicated mixture of lignin and lignin products that is formed during the


223

reaction makes it difficult to see what reactions and side-reactions are taking place. Often the exact role of the solvent, catalyst and reactants is not clear and more mechanistic insight in the reaction is needed to optimize reaction parameters and catalyst choice. Since the reactions that take place on model compounds cannot be translated directly to the reactivity of real lignin, other solutions will have to be found to visualize the changes that take place in lignin. Options such as the use of isotope-labeled solvents or lignin have hardly been explored and online monitoring tools will have to be developed.

As the petrochemical industry has had decades to develop and optimize the technologies currently used for the conversion of fossil resources to fuels and chemicals, it is promising to be able to note that the recent, strongly renewed interest in the conversion of lignin has already led to interesting developments. The necessity of biorefineries to utilize all components of lignocellulosic biomass in order to compete with the traditional petrochemical refineries in combination with the need to produce chemicals that cannot be easily obtained from fossil fuel sources such as shale gas will further drive the research efforts aimed at the valorisation of lignin. Continuous efforts and interest in the replacement of fossil fuel-based chemicals will eventually lead to the development of new sustainable lignin-based processes.

Chapter 9b

Nederlandse Samenvatting

Chapter 9b

226

Samenvatting

Op dit moment is de chemische industrie erg afhankelijk van fossiele brandstoffen voor de productie van essentiële bouwstenen voor chemicaliën en materialen. Nu deze fossiele grondstoffen uitgeput beginnen te raken, is er in toenemende mate aandacht voor het produceren van deze bouwstenen op een meer duurzame wijze. De interesse in het gebruik van biomassa als een alternatief voor fossiele brandstoffen voor de productie van brandstoffen en chemicaliën neemt als gevolg hiervan dan ook sterk toe. Recentelijk heeft de productie van schaliegas als bron van energie in de Verenigde Staten een enorme vlucht genomen, een ontwikkeling die ten koste zal gaan van het relatieve belang van de (zwaardere) ruwe olie fracties voor de productie van chemicaliën en materialen. Omdat een aantal essentiële chemicaliën nu juist worden geproduceerd uit het kraken van deze zwaardere fracties, heeft deze ontwikkeling een negatieve invloed op de beschikbaarheid van veel belangrijke bouwstenen zoals butadieen en de aromaten benzeen, tolueen en xyleen (BTX). Om dit tekort op te vangen, zullen er nieuwe routes voor de productie van deze stoffen moeten worden ontwikkeld, dit biedt gelegenheid om te kiezen voor meer duurzame processen.

De belangrijkste en meest duurzame bron van biomassa voor de productie van hernieuwbare brandstoffen en chemicaliën is lignocellulosische biomassa. Lignocellulose is niet eetbaar en kan bijvoorbeeld worden verkregen uit afval van de land- of bosbouw of via cultivering van specifieke gewassen. Lignocellulose bestaat voornamelijk uit drie componenten: de suikerpolymeren cellulose en hemicellulose en een aromatisch polymeer, lignine. Analoog aan het omzetten van fossiele grondstoffen in een petrochemische raffinaderij, zal het omzetten van lignocellulose naar brandstoffen en chemicaliën plaatsvinden in een bioraffinaderij. Om te kunnen concurreren met producten die in een petroleumraffinaderij worden gemaakt, zal de bioraffinaderij alle bestanddelen van lignocellulose moeten gebruiken voor het maken van waardevolle producten en daarbij zo min mogelijk afval moeten produceren. Hoewel er recent al commerciële processen zijn ontwikkeld voor het gebruik van de cellulose- en hemicellulosefracties, is er nog relatief weinig aandacht besteed aan het omzetten van lignine en zijn er slechts een handvol voorbeelden beschikbaar van het succesvol verwaarden van lignine. Het ontwikkelen van de ligninetak van een geïntegreerd, efficiënt bioraffinageprocess is dan ook een grote uitdaging. Lignineverwaarding is moeizaam omdat lignine chemisch gezien erg verschilt van de suikerfracties in lignocellulose en bovendien erg moeilijk chemische reacties aangaat.


227

Lignine is een amorf polymeer met veel verschillende chemische functionaliteiten en een structuur die afhangt van de plant waaruit het geïsoleerd is en de methode die gebruikt is voor de isolatie. Het bevindt zich in de celwanden van de plant waar het zorgt voor stevigheid en bescherming tegen micro-bacteriële aanvallen. Lignine is dan ook een sterke verbinding, die zowel via natuurlijke processen als met chemische behandeling moeilijk is af te breken. Daar komt bij dat lignine nauwelijks oplost in gangbare oplosmiddelen, waardoor het een moeilijk substraat is om mee te werken. De structuur van lignine en de uitdagingen van lignineconversie zijn in detail beschreven in een literatuuroverzicht in Hoofdstuk 2. Hier worden zowel de chemische eigenschappen van lignine, zoals de structuur, de aanwezige type bindingen en de oplosbaarheid, als de reactiviteit van lignine in katalytische reacties behandeld. Hoewel de oudere literatuur over lignine en de omzettingen van lignine relatief schaars is en verspreid over veel verschillende vakgebieden zeker in vergelijking met het werk rond andere biopolymeren, is er over de jaren wel veel kennis opgedaan over lignine en heeft de toegenomen interesse van vooral de laatste jaren gezorgd voor een enorme toename in het aantal publicaties op dit onderwerp.

Het literatuuroverzicht laat zien dat een groot deel van het onderzoek naar methoden voor de katalytische omzetting van lignine gebruik maakt van katalysatoren die oorspronkelijk ontwikkeld en geoptimaliseerd zijn voor de omzetting van fossiele grondstoffen. Voor een nieuw, zuurstofrijk substraat als lignine zouden echter ook nieuwe katalysatoren moeten worden ontwikkeld die een hoge katalytische activiteit, selectiviteit en stabiliteit hebben onder de specifieke omstandigheden waarbij lignine typisch wordt omgezet. Het overzicht laat ook zien dat in veel onderzoek naar katalytische lignineconversie gebruik gemaakt wordt van modelverbindingen die verschillende onderdelen van de ligninestructuur proberen na te bootsen. Hoewel studies met modelverbindingen ons veel kunnen leren en de analyse van deze reacties veel minder ingewikkeld is dan wanneer echte ligninemengsels worden gebruikt, wordt de structurele complexiteit van het ligninemacromolecuul toch niet goed nagebootst. Hoewel er al veel onderzoek is gedaan, kan worden geconcludeerd dat een efficiënt proces voor het omzetten van echte lignine nog steeds niet volledig is ontwikkeld en dat de beschikbare fundamentele kennis over de benodigde eigenschappen van geschikte katalysatoren beperkt is.

Het doel van het werk beschreven in dit proefschrift is 1) het ontwikkelen van een twee-staps reactie voor de omzetting van lignine naar bulkchemicaliën zoals fenolen en BTX en 2) het verkrijgen van meer inzicht in eigenschappen als activiteit,

Chapter 9b

228

selectiviteit en stabiliteit van goede katalysatoren voor deze reacties. De eerste stap van het proces voor het omzetten van lignine dat hier wordt beschreven, bestaat uit een katalytische depolymerisatie, waarin een mengsel van zuurstofrijke, aromatische monomeren wordt gevormd. In de tweede stap wordt m.b.v. een hydrodeoxygenatiereactie het zuurstofgehalte van het mengsel omlaag gebracht om uit te komen op een beperkt aantal aromatische eindproducten. Met dit als doel zijn er verschillende reacties ontwikkeld en getest op ligninemonsters verkregen uit uiteenlopende plantenbronnen.

Deel I van dit proefschrift beschrijft twee verschillende routes voor de katalytische depolymerisatie van lignine, deze routes zijn weergegeven in Figuur 9.1. Hoofdstuk 3 beschrijft het oplossen, de vloeistoffasereforming en de reductieve depolymerisatie van organosolv-, kraft- en suikerrietbagasselignines. Het bleek mogelijk om al deze types lignine op te lossen in een 1:1 mengsel van ethanol en water door dit te verwarmen naar 225 ̊ C. GPC en kwantitatieve HSQC NMR analyse van de opgeloste lignines lieten zien dat tijdens dit proces de etherbindingen in de lignines, voornamelijk de β-O-4 bindingen, worden gebroken, waardoor het molecuulgewicht van de ligninepolymeren werd gereduceerd. De opgeloste lignines werden vervolgens gebruikt voor verschillende katalytische reacties. Geïnspireerd door het zogenaamde ‘aqueous-phase reforming’ (LPR) proces dat plaatsvindt in water en waarbij zuurstofrijke verbindingen worden omgezet, is een Pt/γ-Al2O3 katalysator in combinatie met zwavelzuur, fosforwolfraamzuur of natronloog als co-katalysator gebruikt voor het depolymeriseren van lignine naar monomere, aromatische chemicaliën in een proces dat ‘liquid-phase reforming’ is gedoopt. Vergeleken met de reactie waarbij alleen water als oplosmiddel wordt gebruikt, worden er bij de reactie in ethanol/water hogere opbrengsten behaald en minder onoplosbare vaste stoffen gevormd. Dit is omdat ethanol niet alleen een rol speelt als oplosmiddel maar ook als reagens en de net-gevormde ligninefragmenten stabiliseert en daarmee recondensatiereacties van de fragmenten bemoeilijkt. Een base of zuur was nodig als co-katalysator om ligninebindingen te hydrolyseren, afhankelijk van het type lignine dat werd gebruikt werden de beste resultaten behaald met verschillende soorten co-katalysatoren. De hoogste opbrengst van aromatische monomeren was 17% en werd behaald door zwavelzuur te gebruiken als co-katalysator bij het LPR van kraftlignine. Reductie van de opgeloste lignines met Pt/γ-Al2O3, Ru/C, Pd/C en Ni/SiO2 katalysatoren resulteerde in maximale opbrengsten van 6% aromatische monomeren. Voor zowel het LPR als de reductiereactie gold dat de samenstelling van het productenmengsel en de maximale opbrengsten afhankelijk waren van het type lignine en de katalysator die


229

werden gebruikt. Over het algemeen kan worden gezegd dat de organosolvlignine die geïsoleerd is uit hardhout voornamelijk syringol-achtige producten opleverde, terwijl de kraftlignine die uit dennenhout afkomstig was, leidde tot guaiacol als belangrijkste product en dat het gebruik van de bagasse, afkomstig uit het grasachtige suikerriet, voornamelijk leidde tot de productie van fenolen.

Het onderwerp van Hoofdstuk 4 is de stabiliteit van de Pt/γ-Al2O3 katalysator onder de reactieomstandigheden die zijn gebruikt in Hoofdstuk 3. Zowel de stabiliteit van het dragermateriaal en de metaaldeeltjes als de vorming van koolstofrijke afzettingen op het oppervlak van de katalysator zijn onderzocht. Meerdere analytische technieken zijn gebruikt om veranderingen in de katalysator, die plaatsvonden na behandeling in ethanol/water onder reactiecondities, te bestuderen en om te bepalen wat de invloed hierop is van lignine zelf en ligninemodelverbindingen die de typische producten van het LPR proces vertegenwoordigen. Analyse van de katalysator na behandeling in ethanol/water met röntgendiffractie, aluminium NMR en N2

fysisorptie liet duidelijk zien dat de γ-Al2O3 drager oppervlakte verloor, als gevolg van

Figuur 9.1: Twee routes voor de eerste ligninedepolymerisatiestap zoals beschreven in deel I.

Chapter 9b

230

hydratering en de faseverandering naar boehmite. Tegelijkertijd vond sintering van de Pt-deeltjes plaats waardoor deze in grootte toenamen van 2 naar 3 nm. Wanneer er mono-aromatische verbindingen aan de oplossing werden toegevoegd nam de snelheid waarmee boehmite gevormd werd aanzienlijk af. Dit komt waarschijnlijk doordat deze moleculen coördineren aan de hydroxylgroepen op de aluminadrager en daarmee de interactie van de drager met watermoleculen bemoeilijken. Tevens werd gevonden dat monomeren met meerdere zuurstofgroepen een grotere invloed hadden op de stabiliteit van de drager. Wanneer de snelheid waarmee de γ-Al2O3-drager naar boehmite werd omgezet afnam, nam het sinteren van de Pt-deeltjes toe. Uit H2 chemisorptiemetingen bleek verder dat de Pt-deeltjes ingekapseld werden in het dragermateriaal tijdens de behandeling. Het toevoegen van lignine aan het reactiemengsel leidde tot een opmerkelijke toename in stabiliteit. De transformatie van γ-Al2O3 naar boehmite en sintering van de Pt-deeltjes stopten volledig. Uit IR en TGA analyse bleek dat er lignine was afgezet op het oppervlak van de katalysator waardoor hydratie van het dragermateriaal en sintering van de metaaldeeltjes kon worden voorkomen.

In Hoofstuk 5 wordt een methode die oorspronkelijk is ontwikkeld in de papierindustrie voor het bleken van cellulosevezels door het verwijderen van achtergebleven lignine, toegepast op het oxidatief depolymeriseren van lignine zelf. Na deze alkalische waterstofperoxidebehandeling nam het molecuulgewicht van lignine af met 25%. De hoeveelheid toegevoegde natronloog, nodig voor zowel het oplossen van de lignine als het deprotoneren van de peroxide, had geen invloed op de depolymerisatie zelf. Toename van de hoeveelheid waterstofperoxide in het reactiemengsel leidde wel tot een toename in de depolymerisatiesnelheid. Omdat onzuiverheden in de lignine een grote invloed hebben op de alkalische depolymerisatie, varieerde de efficiëntie van de depolymerisatie aanzienlijk voor de verschillende ligninemonsters. Overgangsmetalen kunnen bijvoorbeeld de afbraak van waterstofperoxide katalyseren waardoor het oxidant niet meer beschikbaar is voor de depolymerisatiereactie. Het bleken van achtergebleven lignine in cellulosepulp vind plaats via een Dakin, of Dakin-achtig, reactiemechanisme. Tijdens deze reacties vindt de additie plaats van een hydroperoxy anion op een ligninefragment met een vrije, fenolische OH-groep en een keton op het α-koolstofatoom. In Hoofdstuk 3 werd al aangetoond dat deze vrije, fenolische OH-groepen kunnen worden gevormd door het breken van lignine-etherbindingen tijdens het oplossen van lignine in ethanol/water. Voorbehandeling van de lignine door het op te lossen in ethanol/water leidde er dan ook toe dat de reductie in molecuulgewicht na de alkalische oxidatiereactie


231

toenam tot 32%. Een poging om het aantal α-ketonen te doen toenemen door middel van een kobalt-gekatalyseerde oxidatie resulteerde niet in een verbetering van de efficiëntie van de depolymerisatiereactie. In tegendeel, het molecuulgewicht van de lignine ging omhoog na deze voorbehandeling. IR en NMR analyse lieten zien dat een grote hoeveelheid carbonzuren werd gevormd in de alkalische oxidatie van lignine. Dit komt overeen met het Dakin(-achtige) reactiemechanisme dat de vorming van deze functionele groepen voorspelt. Na de kobalt-gekatalyseerde oxidatie van lignine waren er niet alleen ketonen gevormd maar ook een groot aantal carbonzuren. Fosfor NMR liet verder zien dat er geen vrije fenolische OH groepen meer aanwezig waren na deze oxidatie.

In Deel II wordt de tweede stap van de tweestaps omzetting van lignine naar aromaten beschreven, namelijk de hydrodeoxygenatie (HDO) van modelverbindingen met twee verschillende type katalysatoren. In Figuur 9.2 staan de twee verschillende klassen katalysatoren die worden gebruikt weergegeven. Hoofdstuk 6 beschrijft de HDO van een groot aantal modelverbindingen over een commercieel verkrijgbare,

Figure 9.2: De twee verschillende soorten katalysatoren die zijn bestudeerd in de HDO van ligninemodelverbindingen in deel II.

Chapter 9b

232

ingezwavelde CoMo/Al2O3 katalysator. Op basis van deze reactie wordt een uitgebreid reactienetwerk beschreven, dat laat zien dat de HDO van ligninemodelverbindingen plaatsvindt via opeenvolgende demethylerings- en deoxygenatiestappen. Omdat de katalysator een erg lage ringhydrogeneringsactiviteit heeft worden er voornamelijk aromatische producten gevormd. Het verwijderen van de laatste fenolische OH groep bleek moeilijk te gaan. Fenolen en cresolen zijn als gevolg hiervan, relatief stabiel tijdens deze reacties. Aromatische moleculen met één zuurstoffunctionaliteit waren daarom de belangrijkste producten, onafhankelijk van het molecuul waarmee de reactie was begonnen. De reacties met drie verschillende modeldimeren lieten zien dat zowel de β-O-4 en de β-5 etherbindingen in lignine kunnen worden gebroken tijdens de HDO reactie terwijl de 5-5 binding zo stabiel was dat deze niet kon worden gebroken. Uit analyse van de katalysator bleek dat het zwavelgehalte na de reactie met guaiacol was gereduceerd. Dit wijst op een afname van het aantal katalytisch actieve sites afnam en deactivering van de katalysator tijdens de reactie.

In Hoofdstuk 7 wordt de HDO van guaiacol over wolfraam- en molybdeencarbides op koolstofnanovezels beschreven. Deze carbidekatalysatoren zijn een goed alternatief voor de traditionele zwavelhoudende katalysatoren zoals de CoMo/Al2O3 en de overgangsmetaalkatalysatoren. De carbides combineren de lage ringhydrogeneringsactiviteit van de CoMo/Al2O3 met de hoge activiteit en stabiliteit van de gedragen overgangsmetaalkatalysatoren. Met de carbides werden hogere omzettingen en fenolopbrengsten behaald dan met de ingezwavelde CoMo/Al2O3

katalysator beschreven in het vorige hoofdstuk. De selectiviteit naar fenolen en de massabalansen die werden gehaald zijn hoger dan de resultaten die tot nu toe zijn gerapporteerd voor deze reactie. De reactie van guaiacol naar fenol over de carbides vindt plaats via een directe demethoxyleringsstap en niet zoals met de CoMo/Al2O3

katalysator via opeenvolgende demethylerings- en hydrodeoxygenatiestappen. De Mo2C/CNF katalysator was veel actiever dan de W2C/CNF katalysator. De snellere reactie van guaiacol over de molybdeenkatalysator leidde tot een hogere selectiviteit voor volledig zuurstofvrije producten zoals benzeen en tolueen. Verder was de massabalans beter, omdat de reactie zo snel ging dat nevenreacties niet konden plaatsvinden. De Mo2C/CNF katalysator kon worden hergebruikt zonder verlies van activiteit en selectiviteit. Röntgendiffractieanalyse van de katalysator na de reactie liet zien dat de Mo2C fase stabiel was onder reactieomstandigheden terwijl de W2C fase geoxideerd werd.

In Deel III van dit proefschrift worden de eerste depolymerisatiestap en de


233

tweede hydrodeoxygenatiestap gecombineerd, zoals weergegeven in Figuur 9.3. Hoofdstuk 8 beschrijft de ’liquid-phase reforming’-reactie van organosolv-, kraft- en suikerrietbagasselignine over een Pt/γ-Al2O3 katalysator met natronloog als co-katalysator. Na de reactie werd een lignine-olie geïsoleerd die vervolgens verder werd omgezet in een hydrodeoxygenatiereactie over de CoMo/Al2O3 en Mo2C/CNF katalysatoren die beschreven werden in Deel II. De organosolvlignine-olie bevatte voor de HDO reactie de grootste hoeveelheid aromatische monomeren, GPC analyse van deze olies liet echter zien dat de kraftlignine het meest gedepolymeriseerd was. Afhankelijk van de gebruikte lignine waren de meest voorkomende aromatische monomeren (gealkyleerde) syringol-, guaiacol- en fenolmoleculen. Het zuurstofgehalte van de lignine-olie kon worden gereduceerd d.m.w. hydrodeoxygenatie met zowel de CoMo/Al2O3 als de Mo2C/CNF katalysator. Na reactie bij 300 ˚C werden ook de volledig zuurstofvrije producten tolueen, xyleen en ethylmethylbenzeen gevonden. Dat na HDO reactie direct op en zonder voorafgaande depolymerisatiestap geen zuurstofvrije producten werden gevormd, laat zien dat opeenvolgende depolymerisatie- en hydrodeoxygenatiestappen een goede manier zijn om aromatische chemicaliën met een laag zuurstofgehalte te maken uit lignine. De totale opbrengst van aromatische monomeren uit alle drie de lignines was hoger met de Mo2C/CNF katalysator dan

Figure 9.3: De tweestaps reactie van lignine via een LPR en HDO stap zoals beschreven in hoofdstuk 8.

Chapter 9b

234

met de CoMo/Al2O3 katalysator, de totale hoeveelheden zuurstofvrije producten waren vergelijkbaar. De hoogste opbrengst werd gehaald met organosolvlignine-olie en de Mo2C/CNF katalysator. Reacties die werden gedaan bij hogere temperaturen resulteerden in olies met een lager zuurstofgehalte. Na reactie van organosolvlignine-olie bij 350 ˚C over de Mo2C/CNF katalysator waren zelfs alle aromatische monomeren met drie zuurstofgroepen omgezet, het grootste gedeelte van de gevonden producten had geen of slechts één zuurstofhoudende functionele groep.

Slotbeschouwing

De resultaten die worden beschreven in dit proefschrift dragen bij aan een beter begrip van de diverse factoren die van invloed zijn op de productie van aromatische bulkchemicaliën uit lignine. Waardevolle kennis is opgedaan over de katalysatoractiviteit en stabiliteit, de optimale condities voor het oplossen van lignine om condensatiereacties te voorkomen en analytische methoden voor de karakterisering van begin- en eindproducten. Gebaseerd op deze resultaten en inzichten kunnen er nu enkele algemene aanbevelingen worden gedaan.

Lignine is een recalcitrant biopolymeer dat moeilijk te hanteren en te analyseren is. Juist daarom is het van belang om combinaties van verschillende, complementaire analytische technieken te gebruiken voor de analyse van de complexe reactiemengsels. Bijvoorbeeld door GC en GPC resultaten, die informatie geven over respectievelijk veranderingen in molecuulgewicht van het polymeer en de type en hoeveelheid gevormde monomeren, te combineren met IR en NMR data, die beiden inzicht geven in de chemische veranderingen die hebben plaatsgevonden tijdens de reactie. Verder is het van groot belang dat de verkregen data wordt weergegeven op een manier die de meest belangrijke informatie laat zien. Het is bijvoorbeeld niet zinvol om lignineconversies te rapporteren in depolymerisatiereacties omdat het lastig is een onderscheid te maken tussen lignine die enkel is opgelost en daadwerkelijke conversie; opbrengsten van producten en veranderingen in molecuulgewicht zijn hier veel belangrijker. Standaardmethodes voor het rapporteren van data in katalytische lignineomzettingen, vergelijkbaar met de standaardmethodes voor ligninekarakterisering die zijn opgezet door het International Lignin Institute moeten daarom worden ontwikkeld.

Hoewel het gebruik van modelverbindingen die veelvoorkomende


235

ligninefunctionaliteiten nabootsen waardevolle informatie kan opleveren, is het belangrijk om je bewust te zijn van de beperkingen van deze aanpak. De complexiteit van het ligninepolymeer zorgt ervoor dat het nut van moleculen die specifieke bindingen in lignine nabootsen voor depolymerisatiereacties beperkt is. Reacties die worden getest op pure modelverbindingen geven niet per se dezelfde resultaten wanneer ze worden toegepast op het echte polymeer. Het is daarom veel interessanter om deze modelverbindingen te gebruiken om meer inzicht te krijgen in het reactiemechanisme van reacties waarvan al is bewezen dat ze werken op lignine zelf.

Het gebruik van bestaande katalysatorsystemen is een voor de hand liggend startpunt in onderzoek gericht op het ontwikkelen van katalytische routes voor de valorisatie van biomassa. Het is belangrijk dat men zich realiseert dat deze katalysatoren oorspronkelijk zijn ontwikkeld en geoptimaliseerd voor het gebruik op de apolaire koolwaterstoffen die worden verkregen uit fossiele grondstoffen. Deze substraten bevatten relatief weinig zuurstof en water en de reacties vinden vaak plaats in de gasfase. Daar staat tegenover dat de substraten die afkomstig zijn uit biomassa vaak erg polair zijn, een hoog zuurstofgehalte hebben en dat katalytische omzettingen van deze verbindingen vaak plaatsvinden in de vloeistoffase in aanwezigheid van veel water. Diverse studies naar de stabiliteit van deze katalysatoren tijdens de omzetting van substraten afkomstig uit biomassa hebben, net als de resultaten beschreven in Hoofdstuk 4 en 6 van dit proefschrift, laten zien deze hoge concentraties water kunnen leiden tot het verlies van actieve sites, sintering van metaaldeeltjes en transformatie van het dragermateriaal. Uiteindelijk zullen nieuwe soorten katalysatoren moeten worden ontwikkeld, die specifiek zijn ontwikkeld en geoptimaliseerd voor de veeleisende omstandigheden van biomassaomzettingen. Deze katalysatoren moeten bestaan uit dragermaterialen die stabiel zijn in de aanwezigheid van water, zelfs onder zure of basische condities, en metaaldeeltjes bevatten die stabiel zijn onder oxiderende omstandigheden. Verder zijn er katalysatoren die niet stabiel lijken te zijn onder algemene reactieomstandigheden maar in specifieke gevallen grote voordelen kunnen hebben. Zoals beschreven in Hoofdstuk 6 verliest de ingezwavelde CoMo/Al2O3 katalysator bijvoorbeeld zwavel in HDO reacties waar geen zwavel in het reactiemengsel zit. Dit leidt tot deactivatering van de katalysator en vervuiling van de producten. Lignines die geïsoleerd zijn via de kraftmethode bevatten over het algemeen echter 2-3% zwavel. Waar deze aanwezige zwavel bij andere katalysatoren kan leiden tot katalysatorvergiftiging, kan het bij de CoMo katalysator juist deactivering voorkomen. Het gebruik van ingezwavelde CoMo/Al2O3 katalysatoren zou daarom een oplossing kunnen zijn voor dit soort grondstoffen.

Chapter 9b

236

Het reactiemengsel dat wordt verkregen na reacties met lignine bestaat uit een complex mengsel van lignine en kleinere aromaten. Dit maakt het lastig om te zien welke reacties en nevenreacties er allemaal plaatsvinden. De rol van het oplosmiddel, de katalysator en de reactanten is vaak niet precies duidelijk en meer informatie over het mechanisme is nodig om de reactieparameters en katalysator te kunnen optimaliseren. Aangezien reacties met modelverbindingen niet direct vertaald kunnen worden naar de conversie van echte ligninestromen, zullen er andere oplossingen moeten worden gevonden om de veranderingen die plaatsvinden in de lignine te visualiseren. Het gebruik van bijvoorbeeld isotoop-gelabelde oplosmiddelen en lignines en de ontwikkeling van nieuwe online analysemethodes bieden hier mogelijkheden.

De petrochemische industrie heeft decennia de tijd gehad om de technologieën te ontwikkelen en te optimaliseren die nu gebruikt worden voor het omzetten van fossiele grondstoffen. Het is dan ook veelbelovend om te zien dat de recentelijk toegenomen interesse in het gebruik van lignine al heeft geleid tot enkele interessante ontwikkelingen. Om te kunnen concurreren met traditionele op petroleum gebaseerde raffinaderijen, is het echter noodzakelijk dat bioraffinaderijen alle componenten van lignocellulose kunnen gebruiken voor de productie van brandstoffen en chemicaliën. Dit, in combinatie met de noodzaak om chemicaliën te produceren die niet of moeilijk te produceren uit fossiele grondstoffen zoals schaliegas, zal de drijvende kracht zijn achter voortzetting van het onderzoek naar de valorisatie van lignine. Aanhoudende inspanningen en interesse in het vervangen van de huidige, op fossiele grondstoffen gebaseerde chemicaliën zullen uiteindelijk leiden tot het ontwikkelen van nieuwe, op lignine gebaseerde, duurzame processen.

237

List of Publications

This thesis was based on the following publications:

Chapter 2: J. Zakzeski, P. C. A. Bruijnincx, A. L. Jongerius and B. M. Weckhuysen, “The Catalytic Valorization of Lignin for the Production of Renewable Chemicals”, Chem. Rev. 2010, 110, 3552–3599.

Chapter 3:J. Zakzeski*, A. L. Jongerius*, P. C. A. Bruijnincx and B. M. Weckhuysen, “Catalytic Lignin Valorization Process for the Production of Aromatic Chemicals and Hydrogen”, ChemSusChem 2012, 5, 1602-1609.

Chapter 4:A. L. Jongerius, J. R. Copeland, G. S. Foo, J. P. Hofmann, P. C. A. Bruijnincx, C. Sievers and B. M. Weckhuysen, “Stability of Pt/γ-Al2O3 Catalysts in Lignin and Lignin Model Compound Solutions under Liquid Phase Reforming Reaction Conditions”, ACS Catal. 2013, 3, 464-473.

Chapter 6:A. L. Jongerius, R. Jastrzebski, P. C. A. Bruijnincx and B. M. Weckhuysen, “CoMo sulphide-Catalyzed Hydrodeoxygenation of Lignin Model Compounds: An Extended Reaction Network for the Conversion of Monomeric and Dimeric Substrates”, J. Catal. 2012, 285, 315-323.

Chapter 7:A. L. Jongerius*, R. W. Gosselink*, J. Dijkstra, J. H. Bitter, P. C. A. Bruijnincx and B. M. Weckhuysen, “Carbon Nanofiber-supported Transition Metal Carbide Catalysts for the Hydrodeoxygenation of Guaiacol”, ChemCatChem 2013, DOI: 10.1002/cctc.201300280.

Chapter 8:A. L. Jongerius, P. C. A. Bruijnincx and B. M. Weckhuysen, “Liquid-Phase Reforming and Hydrodeoxygenation as a Two-Step Route to Aromatics from Lignin“ Green Chem. 2013, DOI:10.1039/C3GC41150H.

* Authors contributed equally to the work.

238

Other publications by the author:

N. Franssen, B. de Bruin, E. Jellema, A. L. Jongerius, “Functionalized Materials by Catalyzed Carbene Polymerization”, International patent: WO 2011/157444 A1.

J. Zakzeski, A. L. Jongerius and B. M. Weckhuysen, “Transition Metal Catalyzed Oxidation of Alcell Lignin, Soda Lignin, and Lignin Model Compounds in Ionic Liquids”, Green Chem. 2010, 12, 1225-1236.

E. Jellema, A. L. Jongerius, A. J. C. Walters, J. M. M. Smits, J. N. H. Reek and B. de Bruin, “Ligand Design in Rh(diene)-Mediated ‘Carbene’ Polymerization; Efficient Synthesis of High-Mass, Highly Stereoregular, and Fully Functionalized Carbon-Chain Polymers”, Organometallics 2010, 12, 2823-2829.

E. Jellema, A. L. Jongerius, J. N. H. Reek and B. de Bruin, “C1 Polymerization and Related C-C Bond Forming ‘Carbene Insertion’ Reactions”, Chem. Soc. Rev. 2010, 39, 1706-1723. E. Jellema, A. L. Jongerius, G. Alberda van Ekenstein, S. D. Mookhoek, T. J. Dingemans, E. M. Reingruber, A. Chojnacka, P. J. Schoenmakers, R. Sprenkels, E. R. H. van Eck, J. N. H. Reek and B. de Bruin, “Rhodium-Mediated Stereospecific Carbene Polymerization: From Homopolymers to Random and Block Copolymers”, Macromolecules 2010, 43, 8892–8903.

239

Dankwoord

Onderzoek doe je nooit alleen. Daarom wil ik nu aan het eind van het boekje nog even terugblikken en iedereen die mij de afgelopen vier jaar heeft geholpen en aangemoedigd bedanken voor de ideeën, de samenwerkingen en de goede tijd die ik tijdens mijn promotie heb gehad.

Als eerste denk ik daarbij natuurlijk aan Bert mijn promotor die mij de kans heeft geboden mijn promotieonderzoek te doen in deze geweldige groep. Als coach heeft hij eerst de lijnen uitgezet maar daarna kreeg ik alle vrijheid om mijn eigen “kabouters” te tekenen. Het vertrouwen in mij dat je altijd uitsprak heb ik erg gewaardeerd. En met Pieter als co-promotor kon het eigenlijk allemaal niet meer mis gaan. Met jou als mentor heb ik de afgelopen vier jaar heel veel geleerd, je was daarbij ook meer dan eens een soort wetenschappelijk geweten dat het altijd door had als ik me ergens te makkelijk van af probeerde te maken. Altijd kon ik bij jou binnenlopen, soms enthousiast met nieuwe resultaten maar ook om te klagen als iets niet werkte, al was dat dan vaak steeds hetzelfde apparaat. Aan de wieg van het lignineonderzoek staat ook Joe Z met wie ik de eerste twee jaar van mijn onderzoek veel heb samengewerkt. Hoofdstuk 2 en 3 zijn hier direct het resultaat van, daarnaast heeft onze samenwerking nog zeker zijn sporen nagelaten in de rest van dit proefschrift en ik denk dat we veel van elkaar hebben geleerd.

Voor het draaiende houden van alle zichtbare en onzichtbare radartjes in het lab ben ik net als elke andere AIO het technisch personeel heel dankbaar. AdM, AdE, Fouad, Vincent, Marjan en Rien, als er ergens een schroefje aangedraaid moest worden (of kwijt was), metingen gedaan moesten worden en ingewikkelde bestellingen afgehandeld, jullie waren er altijd. Daarbij kan ik natuurlijk Dymph & Monique niet vergeten zonder wie de groep niet op volle snelheid zou kunnen functioneren.

Zelfs met een laboratorium dat vol staat met apparatuur is het toch eens in de zoveel tijd nodig om uit te wijken naar een ander lab waar dan de dingen staan die je wel nodig hebt. Richard en Jacinta, heel erg bedankt dat ik zo vaak naar Wageningen kon komen voor de GPC metingen en voor alle tijd die jullie hebben genomen om hierbij te helpen. Hans ook heel erg bedankt voor alle tijd die je samen met mij bij de

240

NMR hebt doorgebracht. Dit waren zeker enkele van de gezelligste metingen die ik ooit heb gedaan, waarbij je toch altijd stiekem het apparaat probeerde in te stellen terwijl je maar bleef zeggen dat ik dat de volgende keer echt zelf moest doen. Ook bij organische chemie, onze nieuwe bovenburen en mijn nieuwe collega’s, heb ik veel hulp gekregen. Bert, Johan en Henk, bedankt voor het beschikbaar stellen van een zuurkast, destijds nog in het Kruyt-gebouw en voor het gebruik van jullie NMR die bij biomassa onderzoek bijna onmisbaar is. Behalve met de bovenburen waren er ook goede connecties met de onderburen en daarom wil ik ook Mies dan nog bedanken voor alle hulp met de vriesdrogers en natuurlijk die gezellige muziek die uit zijn kantoor schalde (mits het niet te lang duurde).

Furthermore I would like to thank the CatchBio program and specifically the industrial partners who have always shown a lot of interest in my research over the course of the project. I very much appreciated the helpful discussions and valuable input that we received during the biannual User Committee meetings.

The fourth chapter of this thesis could not have been made without the cooperation with the Georgia Institute of Technology. Carsten, John and Guo, thanks for the TEM and 27Al MAS NMR measurements and all the input we have received via e-mail and Skype meetings. Additionally, I would like to thank Jan Philipp who also contributed to this chapter.

Het onderzoek waarop hoofdstuk 7 is gebaseerd is het resultaat van een samenwerking binnen de Anorganische Chemie & Katalyse onderzoeksgroep. Rob en Harry, ik wil jullie hartelijk bedanken voor jullie bijdrage hieraan, ik denk dat we goed werk hebben geleverd dat het resultaat is van echt teamwork.

Het uitwerken van IR spectra van lignine en materialen die in contact zijn geweest met lignine is niet altijd heel makkelijk, daarom wil ik Peter de P en Joop van harte bedanken voor hun input en suggesties op het gebied van de IR-spectroscopie.

Special thanks to everybody from the biomass lab, together we made a sanctuary for people working with brown and black sticky stuff. For sure our colleagues realize by now: Biomass is there to stay!

De afgelopen jaren heb ik het genoegen gehad samen te werken met 5 studenten die allemaal op hun eigen manier hebben bijgedragen aan het onderzoek. Robin was

241

de eerste die begon toen ik net een half jaar bezig was, en hoewel dat al best lang geleden is werk je nog steeds aan lignine maar nu als collega, en doen we nog steeds op vrijdagmiddag af en toe een dansje in het lab. Sophia was my second masters student who came all the way from Taiwan to Utrecht, it was a lot of fun working with you and I will never forget the Taiwanese candy and tea you always brought back from me. Daarnaast hebben maar liefst drie bachelorstudenten interesse getoond in het lignine onderzoek, Gerben, Wouter en Jelmer ook met jullie heb ik met veel plezier samengewerkt.

Met veel plezier kan ik terugdenken aan een tijd met geweldige collega’s waar ik zowel binnen als buiten de universiteit veel tijd mee heb doorgebracht. Een groot deel hiervan ging op aan Rob, mijn promotiemaatje. Helaas kunnen we niet tegelijk ons proefschrift verdedigen want hoewel we op dezelfde dag zijn begonnen zal jij je diploma ongeveer 21 uur en 45 minuten eerder ontvangen, toch voelt het alsof we samen klaar zijn. Many good memories are also shared with Clare, although sometimes blurred by our shared love, G&T, which helped us through many evenings, late nights and difficult thesis times. Ilona heeft mijn goede voorbeeld gevolgd en is vanuit Amsterdam naar Utrecht gekomen, niemand had van tevoren bedacht dat het een slecht idee was om ons op kantoor bij elkaar te zetten maar dat hebben ze geweten. Tegelijk met het begin van mijn promotieonderzoek ben ik ook met een nieuwe sport begonnen. Vele uren heb ik met Peter M in de klimhal doorgebracht wat vaak resulteerde in het niet openkrijgen van flessen met oplosmiddelen de volgende ochtend. Na al die uren Nederlandse les vind ik ook dat Dilek dit gewoon in het Nederlands moet kunnen lezen. Je pogingen om mij Turks te leren hadden vaak hilarische resultaten toch was het altijd çok leuk en hebben we ons nooit sıkıldı.

Many hours spend in the office were always made enjoyable by great office mates therefore thanks to all my “oude kantoorgenoten”: Tamara, Peter H en Ilona, but also the new “open office people on the awesome 4th floor north side balcony”: Ilona (again), Dilek, Upakul, Davide, Luis, Piter, Peter N, Jinbao, Hirsa, and later: Zafer, Robin, Frank, Ramon, Rogier, Sam, Peter B, Gang, Nazila and Jeroen.

Everybody who’s name I still haven’t mentioned, you are not forgotten, I will always remember coffee/lunch breaks nights out, (proton) parties, and general awesome good times: Carlo, Wenhao, Joe S, Eduardo, Andrei, Elena, Luis, Upakul, Arjan, Daniel, Tomas, Fernando, Evelien, Dimitrije, Jesper, Chris, Bart, Emiel, Inge, Korneel, Hirsa, Roy, Gavi, Gonzalo, Qingwing, Harry, Elena, Fouad, Gareth, Marleen, Mikael, Fiona, anyone I might have forgotten en iedereen die mij onderdak

242

heeft geboden als ik midden in de nacht niet meer terug naar Amsterdam kon/wilde. I should not forget the upstairs neighbours who are always willing to share their beers as well as their labs: Peter S, Ties, Eric, Manuel, Suresh, Emma, Vital, en later Stefan.

En met het gevaar dat het bijna lijkt alsof zo’n promotieonderzoek alleen maar een groot feest was wil ik toch nog mensen bedanken in de volgende drie categorien: Alle laatblijvers uit de CatchBio-bar: Rob, Vlien, Zea, Ilona, Daniel, Carlo, Tomas, Wenhao, Ties, Stefan, Frits, Daan, Harry. Motivatie om te gaan sporten: Peter M, Clare, Guido, Bart en Vlien (klimmen), Joe Z, Amelie, Rob, Joe S, Dimitrije and (even) Clare (zwemmen) and all the NIOK and Debye volleyball teams. Reisgenoten tijdens de jaarlijkse citytrips: Clare, Dilek, Ingeborg en Joe Z.

Natuurlijk heb ik niet de volledige 4 jaar doorgebracht in Utrecht, al leek het wel bijna zo. Guido en Vien (mijn trouwe paranimfjes), Ruben (altijd tot diep in de nacht over onzin ouwehoeren op de bank) en de rest van de achterblijvers in Amsterdam. Nooit heb ik spijt gehad dat ik hier ben blijven wonen.

Als laatste, mijn ouders, ik vind het fijn dat jullie mij altijd hebben aangemoedigd om te doen wat ik leuk vind en altijd interesse tonen in waar ik mee bezig ben; hier ligt nu het resultaat van al dat harde werk.

Annelie

243

Curriculum Vitae

Annelie Jongerius was born on the 1st of October 1985 in Zaandam, the Netherlands. After graduating from secundary school (St. Michael College in Zaandijk) she started her undergraduate studies in chemistry at the University of Amsterdam in 2003. In 2006 she obtained her Bachelor of Science degree with the thesis entitled: “Synthesis of amphiphilic ligands for biphasic catalysis” under the supervision of Prof. Dr. C.J. Elsevier. In that same year she started the master Chemistry in the master track Molecular Design, Synthesis and Catalysis at the University of Amsterdam. She performed her master’s thesis research entitled: “Rhodium Mediated (Co)Polymerization of Carbenes” in the group of Prof. Dr. J.N.H. Reek and Prof. Dr. B. de Bruin. She spend a 4 months internship at Albemarle Catalysts in Amsterdam in the Alternative Fuels Technology group and obtained her Master of Science degree (cum laude) late 2008. In 2009 she started her PhD research project funded by the CatchBio program at Utrecht University under the supervision of Prof. Dr. B.M. Weckhuysen and Dr. P.C.A. Bruijnincx, results of which are described in this thesis. Parts of the work were presented at several national and international conferences, namely NCCC 2011 and 2012 (Noordwijkerhout), ACS Spring Meeting 2011 (Anaheim) and ICC 2012 (Munich).

Date post:	14-Feb-2017
Category:	Documents
Upload:	vukhanh
View:	224 times
Download:	6 times

Catalytic Conversion of Lignin for the Production of Aromatics

Documents