University of Groningen Oxidative dehydrogenation of ethylbenzene under industrially relevant conditions Zarubina, Valeriya IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below. Document Version Publisher's PDF, also known as Version of record Publication date: 2015 Link to publication in University of Groningen/UMCG research database Citation for published version (APA): Zarubina, V. (2015). Oxidative dehydrogenation of ethylbenzene under industrially relevant conditions: on the role of catalyst structure and texture on selectivity and stability. [S.l.]: [S.n.]. Copyright Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons). Take-down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum. Download date: 03-10-2020
Transcript
University of Groningen
Oxidative dehydrogenation of ethylbenzene under industrially relevant conditionsZarubina, Valeriya
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite fromit. Please check the document version below.
Document VersionPublisher's PDF, also known as Version of record
Publication date:2015
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):Zarubina, V. (2015). Oxidative dehydrogenation of ethylbenzene under industrially relevant conditions: onthe role of catalyst structure and texture on selectivity and stability. [S.l.]: [S.n.].
CopyrightOther than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of theauthor(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons thenumber of authors shown on this cover page is limited to 10 maximum.
This work was financially supported by the Dutch Technology Foundation (STW), which is the applied science division of NWO, and the Technology Program of the Ministry of Economic Affairs, Agriculture and Innovation (Green and Smart Process Technologies, GSPT), under the project No. 07983 entitled “CO2 plus O2 Oxidative Dehydrogenation of Hydrocarbons to Olefins and Styrene Production”. CB&I is acknowledged for financial support. ISBN: 978-90-367-7691-2 ISBN E-book: 978-90-367-7690-5 This thesis is also available in electronic format at: http://dissertations.rug.nl/ Cover design: Stepan Belyakov, www.styopa.ru Printed by: Gildeprint – www.gildeprint.nl.gildeprint.nl
Oxidative dehydrogenation of
ethylbenzene under industrially
relevant conditions
On the role of catalyst structure and texture on
selectivity and stability
Proefschrift
ter verkrijging van de graad van doctor aan de Rijksuniversiteit Groningen
op gezag van de rector magnificus, prof. dr. E. Sterken
en volgens besluit van het College voor Promoties.
De openbare verdediging zal plaatsvinden op
vrijdag, 27 maart 2015 om 12.45 uur
door
Valeriya Zarubina
geboren op 20 december 1986 te Omsk, Sovjet Unie
Promotor Prof. dr. ir. H.J. Heeres Co-promotor Dr. I.V. Melián Cabrera Beoordelingscommissie Prof. dr. A.A. Broekhuis Prof. dr. ir. K. Seshan Prof. dr. E.J.M. Hensen
To my beloved mother, who never stops believing in
Chapter 3 ON THE STABILITY OF CONVENTIONAL AND NANO-STRUCTURED CARBON-BASED CATALYSTS IN THE OXIDATIVE DEHYDROGENATION OF ETHYLBENZENE UNDER INDUSTRIALLY RELEVANT CONDITIONS . . . . . . . . . .
Styrene (ST) is industrially produced by direct dehydrogenation of ethylbenzene (EB) using steam
at 580-630 °C. The process suffers from high energy consumption due to low conversion per pass
because of equilibrium limitations and the high temperatures required for the endothermic
reaction. However, many research groups and companies have investigated alternative styrene
production processes. Oxidative dehydrogenation process is one of the most important ones. The
big advantage of oxidative dehydrogenation is that the process can be operated at lower
temperatures. There is no need for the co-feeding of superheated steam, and it is free of
thermodynamic limitations regarding the conversion of ethylbenzene. Thus, high conversion per
pass can be achieved. However, various pitfalls in the oxidative dehydrogenation of ethylbenzene
to styrene still exist. These aspects are discussed as well as alternative dehydrogenation processes,
economic and environmental aspects of styrene production and the thermodynamics of the styrene
chemistry.
Chapter 1
10
1.1 Economic and environmental scope of styrene product
Styrene is one of the most important monomer for the polymer industry. Commercial production started in the 1930th on small scale. In 2010 the total annual production of styrene made 26.4 million metric tonnes [1], that makes the industrial
dehydrogenation of ethylbenzene one of the most important industrial processes [2-4]. The expected consumption of styrene in 2020 is estimated to be increased to 41 million metric tonnes worldwide [5]. The market price of styrene in Western Europe
is about 1550 $/tonne based on the data of December 2012 [6]. It makes it evident that the total market size of styrene worldwide is immense ($30-$50 billion).
An estimated energy consumptions of 6.3 GJ/tonne styrene [7]. The
worldwide energy consumption of the styrene production process by dehydrogenation with an annual production of 26.4 million tonnes can be estimated at 1.7×1017 J/year (170 PJ/year). This means that an energy efficient production
process would be developed, the total worldwide energy consumption and, therefore, the emmission of greenhouse gasses could be considerably reduced. This means that both from an economical and environmental point of view, a reduction of the energy
consumption of the styrene production process is of great interest.
1.2 Styrene
Styrene is a colorless oily liquid with a sweet smell. It is an aromatic olefin (Figure 1) which is easy can be polymerized due to the presence of carbon double bond.
Styrene is named after „styrax‟, the resin from the oriental sweet gum tree, native in the eastern Mediterranean region. An overview of some physical properties of styrene is given in the Table 1.
Figure 1. Structure of styrene molecule
Table 1. Physical properties of styrene [8].
Molecular weight [g/mol] 104.152 Density [kg/m3] 903
Melting point [°C] -30.6 Boiling point [°C] 145.2 Critical temperature [°C] 373
According to the “Styrene Producers Association” [9] the main purpose of the styrene is the production of polystyrene (62%) and acylonitrile-butadiene-styrene (ABS) resins (14%). All main applications are given in Figure 2.
Figure 2. Global demand for styrene monomer derivatives in 2004 [9]. (Reproduced with permission) [9]
Styrene is a main monomer block for the polystyrene production. Polystyrene
is widely used because it is relatively inexpensive to produce and easy to polymerize
and co-polymerize [10]. The main uses of polystyrene are for disposable cups, trays and bowls, packaging, household appliances, consumer goods, and as building and construction material. For products which need more stiffness, ABS resign is often
used. Other smaller uses are as a co-polymer in several synthetic rubbers and resins as shown in Figure 2.
1.3 Styrene production chemistry and thermodynamics
Most of the commercial styrene is produced by direct dehydrogenation of
ethylbenzene (85-90%). The remaining part (10-15%) is obtained as a by-product in the production process of propylene oxide [10]. Ethylbenzene is dehydrogenated according to the following reaction:
ΔHr0= 117.6 kJ/mol This equilibrium gas-phase catalytic reaction is highly endothermic (ΔrH
0298 = 117.6
kJ/mol [11,12]) and it performs in the presence of steam. The equilibrium constant is defined by:
Due to the reaction stoichiometry and the fact that the reaction takes place in
the gas phase, a high pressure drives the equilibrium towards EB (Le Chatelier‟s
principle [13]). It means that at low pressure, the system adjusts the position of the equilibrium towards the side of the balance with the larger number of reactants in order to resist the effect of the pressure. Therefore, lower pressures favour the
conversion to styrene. The low pressure as 0.4 bar are often applied to the system to increase the styrene yield [10].
This makes it clear that lowering the pressure initiate a larger driving force for
the reaction, to the side of styrene and hydrogen. High temperatures also lead the equilibrium to be on the side of styrene. That
is why a low pressure and high temperatures are used in the industrial practice of
styrene production by direct dehydrogenation. The effect of low pressure and high temperatures on the ethylbenzene equilibrium conversion is shown in the Figure 3.
Figure 3. The effect of temperature and pressure on the ethylbenzene equilibrium conversion [14]. (Reproduced with permission)
An excess of superheated steam of 720°C is added with steam:EB molar ratios
of 6-13:1 for the styrene production before entering the dehydrogenation reactor. Main reasons are listed below:
Energy in the form of steam is needed to supply the heat for the reaction [11]. High temperatures of 550-700°C [2,10,11,15] are needed because the
equilibrium constant increases with temperature [10].
The equilibrium is shifted to higher conversion of ethylbenzene by diluting the reaction system with steam, in order to lower PEB.
It reduces the formation of unwanted coke deposition on the catalyst particles
[2-11]. The use of conventional styrene production by steam dehydrogenation has
also several disadvantages: High energy consumption due to the use of superheated steam.
Introduction
13
The reaction is equilibrium- and thermodynamically limited at 50- 65%, which
requires a large reactant recycle [11,15] Separation of EB and ST is difficult due to a similar boiling point of respectively
136 °C and 145°C
Consumption of feedstock and product by side reactions forming syngas (CO + H2) [11].
1.4 Styrene production technologies
1.4.1 Lummus/UOP classic styrene technology
The Lummus/UOP Classic SMTM is a major technology for the styrene production. Approximately 43 plants worldwide operate using this technology, with a cumulative
production of 8.3 million tonne annually [16]. In the Figure 4 is shown, that for this process the outlet flow of the dehydrogenation section is cooled down, and then it is distilled to separate the different products (styrene, benzene, toluene, and tar); the
non-reacted ethylbenzene is recycled. In the dehydrogenation section, ethylbenzene is dehydrogenated over
potassium promoted iron catalyst in an adiabatic fixed bed reactor. More than one
reactor is used since the temperatures drop in a theoretical adiabatic reactor under 100% ethylbenzene (theoretical) conversion is ~330°C [14] (Figure 4). A
temperature drop is undesirable for a good performance of the dehydrogenation reaction.
Lummus/UOP developed a more efficient process for styrene production called
the SMART process, which is implemented in several plants worldwide with an annual cumulative production of 1.4 million tonnes. This process is based on the classic styrene monomer process with a difference in the dehydrogenation section (Figure
5). The number of plants using the SMART technology is limited due to the safety risks involving a high temperature mixture of oxygen and hydrogen which presents in the reactor.
Figure 4. PFD of the Lummus/UOP classic SM process [10]. (Reproduced with permission)
Chapter 1
14
600°C
650°C650°C
600°C
EBSteam
Crude ST
EBSteam
Steam
Figure 5. Conventional reactor configuration for the dehydrogenation section of ST production (Steam/EB = 12- 17 mol/mol) [14]. (Reproduced with permission)
The dehydrogenation section of the SMART process contains an extra reactor between the existing dehydrogenation reactors. This extra reactor contains both an ethylbenzene dehydrogenation and a hydrogen oxidation catalyst, as shown in Figure
6. The additional conversion of hydrogen causes the equilibrium to shift towards ethylbenzene, resulting in a higher conversion per pass of up to 75% [16]. The energy release by oxidation is used to decrease the amount of steam used and
Figure 6. Dehydrogenation section Lummus/UOP SMARTTM process [16]. (Reproduced with permission)
1.4.2 Badger/ATOFINA styrene technology
The Badger/ATOFINA process is another major technology for the styrene production, with 47 plants licenced worldwide with a cumulative annual production of
9 million tonnes [16]. This process uses potassium promoted iron catalyst as well [18].
The main difference with the Lummus/UOP process is the distillation section.
In the Badger/ATOFINA process benzene and toluene are separated from styrene in the first distillation column downstream of the settling drum (Figure 7, a). In the next column, ethylbenzene and styrene are separated and the remaining
Introduction
15
ethylbenzene is mixed with fresh ethylbenzene and it is fed back to the first
dehydrogenation reactor. Finally, in the last column styrene is separated from residues. All columns are designed to operate below atmospheric pressures to minimize the operating temperature and to prevent polymer formation [10].
Figure 7. PFD of the Badger/ATOFINA styrene process [10]. (Reproduced with permission)
The dehydrogenation section of this technology is partly different from the Lummus/UOP process. In the Badger/ATOFINA process the dehydrogenation section
also includes two packed bed columns with interstage heating of the reaction mixture. It helps to cope with the temperature decrease due to the endothermic nature of the reaction. However, the exit stream of the first reactor in the
Badger/ATOFINA process is not injected with steam directly, but is reheated by a heat exchanger (Figure 8).
Figure 8. PFD of the Badger/ATOFINA styrene process [10]. (Reproduced with permission)
1.4.3 SNOW process
Since the main purpose of ethylbenzene is the production of styrene, it is produced in the most of cases on site of a styrene production plant by alkylation of benzene with ethylene. Thus, the raw material price is costs of benzene (66%) and ethylene
(34%) [10]. In order to reduce the risk of ethylene prices fluctuations, Snamprogetti and Dow (hence SNOW) developed a process that can run on both ethylene and ethane. Furthermore, ethane is often a cheap by-product of petrochemical streams
[17], which makes it possible to integrate a styrene plant into a petrochemical plant
Chapter 1
16
without the use of a steam cracking unit to produce ethylene. Moreover, the
integration of ethylene- and ethylbenzene production with styrene production possibly can generate a great reduction in capital expenses for the total styrene production process.
A plant running on SNOW technology is fed with benzene and ethane, the latter being dehydrogenated in the same reactor as ethylbenzene, to produce the stoichiometric amount of ethylene for the alkylation of benzene (Figure 9, top).
Alternatively, a plant with SNOW technology can run on benzene and ethylene as feedstock, working similarly as the conventional styrene technology described earlier (Figure 9, bottom).
Figure 9. Flow scheme for SNOW technology with the ethane option (top) and the conventional ethylene option (bottom). (Reproduced with permission)
The reactor section of the SNOW process is considerably different to the other direct dehydrogenation processes due to the simultaneous dehydrogenation of ethylbenzene and ethane. The reactor section consists of a riser type reactor in
which the gas inlet stream is mixed in co-current with fresh catalyst and moves upwards under gas velocities of 4-20 m/s (Figure 10). The catalytic reactions are performed rapidly (approximately 1-5 seconds) in the riser [17]. The temperature
ranges among 590-700°C; it does not run below atmospheric pressures to shift the equilibrium and increase the selectivity in comparison with the more conventional dehydrogenation process. The temperature is supplied by the heat capacity of the
catalyst particles [18]. The regeneration of the spent catalysts takes place in a bubbling fluidized bed
under air to burn of possible coke formation. Then regenerated catalyst is fed back in
the bottom of the riser. The reactor outlet stream is separated and processed using conventional separation technology [17].
Introduction
17
Figure 10. Reactor section of the SNOW process [21]. (Reproduced with permission)
The production process of styrene by direct dehydrogenation is developed to a
high degree of maturity, and there is not much can be improved [17]. Also, the price of feedstock greatly determines the profit margin of styrene production, as 80% of the production costs comes from raw material feedstock [10]. The development of
the SNOW process responds to this and decreases the raw material cost with approximately 14% [17]. Moreover, it decreases the energy consumption due to the absence of superheated steam using for dilution of the reaction mixture.
1.5 Oxidative dehydrogenation of ethylbenzene
In the oxidative dehydrogenation of ethylbenzene to styrene EB is feed simultaneously with oxygen for the styrene formation according to the following
reaction:
In contrast to direct dehydrogenation this reaction is oxidative and, therefore, exothermic (ΔHr0=-124.3 kJ/mol [11,12]). The big advantage of oxidative
dehydrogenation is that the process can be operated at lower temperatures. There is no need in the co-feeding of superheated steam, and it is free of thermodynamic limitations regarding the conversion of ethylbenzene [19]. Thus, high conversion per
pass can be achieved without using a vacuum.
1.5.1 Nature of the active coke
The ODH of EB has been studied for four decades. In 1973 Alkhazov et. al. [20] first proposed that actual catalyst for the ODH is the layer of carbonaceous deposits
formed on acidic catalysts as alumina during the first hours of the reaction. This was later confirmed by many authors who studied this phenomenon and became the general conclusion [21-29].
Chapter 1
18
The layer of active coke consists of carbon, oxygen, and hydrogen. The ratio
of these species in the molecules of the coke layer varies with the time on stream. It was shown that the C/H ratio increases with the reaction time and varies between 0.5-4 in steady state [23,27,28]. Active coke contents between 5.0-33.7 wt.% have
been reported [22-25,28,30-33]. The active coke in the ODH of ethylbenzene is ascribed to redox couples
formed on a polycyclic aromatic basis, being the coke molecules. A reaction
mechanism based on experiments with zirconium phosphate as a support material was proposed by Emig and Hofmann (Figure 11).
According to this model, coke is formed from the condensation of styrene on
the catalyst support material. It has to be noted, that styrene can be present without the availability of active coke because direct dehydrogenation can occur to a small
extent [26]. However, later Lisovskii and Aharoni [32] showed that the reactivity‟s of styrene and ethylbenzene are very similar. The mechanism proposed by Emig and Hofmann is very similar to the reaction mechanism introduced by Iwasawa et al. [21]
for the ODH of ethylbenzene over polynaphtoquinone.
Figure 11. Mechanism for the ODH of EB proposed by Emig and Hofmann [23]. (Reproduced with permission)
In this proposed reaction scheme, styrene is condensed to a system of
polycyclic aromatic rings on the catalyst support surface. Afterwards, these rings are
oxidized and form the polyquinone structure, which has a name of “coke” in Figure 11. This polyquinone structure oxidizes EB to styrene and reacts to a polyhydroquinone intermediate. Thereafter, the polyhydroquinone structure is
oxidized by half molecule of O2 to polyquinone one more time. Several research groups confirmed that the mechanism demonstrated in Figure 11 is the most probable reaction mechanism [12,25,28,29,32]. Recently, it was shown that
carbonyl/quinone groups indeed act as active sites for the ODH reaction. Hence, the activity of the catalyst is directly related to the concentration of the carbonyl groups in the coke layer on the catalyst [34].
Moreover, Lisovskii and Aharoni [32] showed that in the case of interruption of ethylbenzene supply with the constant space velocity, the production of styrene stops immediately. This implies that styrene is not formed out of a carbonaceous
intermediate but directly from ethylbenzene, which makes the proposed reaction mechanism in Figure 11 more feasible.
Introduction
19
Vrieland showed that the active coke is not the major source of COx [23]. It
appears that styrene and ethylbenzene react more readily with oxygen than the deposited carbon does. There are indications that the active coke actually catalyses the burning of styrene and ethylbenzene, as the COx formation increases with
increasing carbon coverage of the support [20,23].
1.5.2 Supports properties for the active coke formation
1.5.2.1 Surface acidity of the support
For the formation of an active coke layer, the support must have some acidity; basic
supports as magnesia and titania are almost completely inactive [30]. Coke formation is accelerated by acidic centers [23], but a narrow distribution of acidity is required for obtaining an active and selective coke layer. In general, acidic sites with
moderate to low acid strength, give the largest contribution to the formation of catalytically active coke for the styrene production [22,24,25]. In several publications has been stated that the supports with the highest total acidity have the
greatest active coke formation [28,36], while other researchers state that very strong acidic site are either ineffective or promote cracking and other side products [22,23]. Although some authors discuss the total acidity of the support (Brønsted
and Lewis), moderate Lewis acid strength is considered necessary for the formation of proper coke [29].
1.5.2.2 Textural properties of the support
Textural properties of the support are important factors to achieve good performance. For this reason it is interesting to know the effect of coking on a support as a high rate of coke formation can block the pore mouth of the micropores
and sometimes mesopores. Figure 12 illustrates the effect of slow and rapid coking. The former results in an equal distributed layer of coke in the micropores (a), the latter results in pore mouth plugging by the coke (b).
a) b)
Figure 12. Schematic representation of the coke formation rate [31]. (Reproduced with permission)
Olefins are known to have a large rate of coke formation compared to other
hydrocarbons [31]. This corresponds with the reports that catalyst particles with
meso- and macro pores show better results in the ODH of ethylbenzene by active carbons than microporous materials, because the micropores are quickly and almost completely blocked by the formed coke [35,36]. These studies used carbon as
catalyst, but it shows the influence of the coke on the textural properties. This results of the BET surface area approaching the area of the meso- and macropores also known as „external surface area‟. Furthermore, catalytic behaviour cannot be
Chapter 1
20
directly related to the surface area, which makes it clear that this is not the only
essential parameter for the reaction [37]. Hence, the previous implies that the surface molecular structure should also play an important role in the catalysis of ODH [38].
1.5.2.3 Textural properties of the support
The phosphorous modified catalysts were found active in the ODH of EB to ST [2,19,22-23,35,38]. The catalyst preparation method in the Chapter 5 of this research is based on a solid state reaction between the support structure and the
impregnated phosphorous solution. The reaction between the phosphate and the support, which is already extensively investigated [19,39-41], can yield several structures. An overview of the different surface groups resulting from the reaction
during calcination are given in Figure 13. For all the different phosphorous containing surface active groups, the oxygen
atom of the phosphate molecule has bonded with a silica atom on the surface of the
silica support. From Figure 13, it becomes clear that the phosphate ion can either bond with one, two, or three silica atoms. Furthermore, oxygen bridges can be formed between two phosphate ions. For supports containing alumina, the aluminium
atom can coordinate the phosphate group.
a) b) c)
d) e)
Figure 13. Possible surface active groups resulting from the impregnation with H3PO4-: a)
[19,42], b) [42,43], c) [43], d) [19], e) [44].
1.6 Process related pitfalls in scientific research
Constraints given by the process are not frequently accounted for in scientific
research. However, they are decisive in whether or not a new process route can be economically feasible. In this section some important process related aspects regarding the ODH reaction experiments are discussed for the ODH of ethylbenzene
in relation with industrial application.
1.6.1 Selectivity
In contrast to direct dehydrogenation process, currently commercially used to produce styrene, the ODH process consumes oxygen. This means that in the case
Introduction
21
the selectivity to styrene is not 100%, COx can be formed in combination with mainly
toluene and benzene, which are also being formed in direct dehydrogenation. Benzene and toluene can be separated from the styrene product and sold as by-products, although it increases the operational and investment costs. However, when
COx is produced, EB feedstock is simply combusted. Moreover, when ethylbenzene is converted to COx, eight times more oxygen is
consumed compared to ODH to styrene due to the stoichiometry of the reaction. This
has an influence on both the oxygen availability for the oxidative dehydrogenation reaction, influencing the maximal conversion, as well as the temperature control of the process. This has effect on the temperature control because the burning of
ethylbenzene to COx is highly exothermic, especially compared to the ODH. The high exothermicity is due to stoichiometry, which is a problem particularly in a fixed bed
reactor which is known for their poor heat transfer [45]. Limited selectivity to styrene is one of the main issues in the oxidative
dehydrogenation of ethylbenzene. Activated carbons [33-38,48,52-59], carbon
nanofibers [60-67], onion-like carbons [68-70], diamonds [56,58,70], nanofilaments [60], graphites [37,58,60,64], multiwall carbon nanotubes (MWCNTs) [56,58,66,69-79], and other type of carbon materials or mixtures of the above mentioned [80-84]
were studied for this reaction widely. It is generally found that these materials are readily active and selective; the reported selectivities are moderate lying between 55-85 %. Only in some cases the reported selectivity is exceptionally high, in the
range of 90-97% [58,76,77]. There are other two types of catalysts based on phosphorous such as metal
pyrophosphates [23,26,85] and phosphates [2,26,39,49,85-91], or P-supported
silica [19,26] that have been reported to be active and selective for EB ODH. However, the catalyst‟s stability under industrially relevant conditions is unknown and more insight in this direction is needed to prove its commercial viability.
The styrene selectivity obtained using Lewis acid-based -alumina is relatively
low for commercialization, up to 70%, compared to the commercial process of steam dehydrogenation (ST selectivity>95%). Few examples indicate that the acidity
enhancement of -Al2O3 by H3PO4 [26] or HBO3 [92,93] has a positive impact on the
styrene yield. To commercialize the EB ODH process, it is needed to develop a catalyst which
will have higher selectivity and stability than traditional catalyst for direct
dehydrogenation. Since the K-promoted Fe2O3 catalyst for the conventional dehydrogenation process is highly selective to styrene, typically >97%, and stable [16], developing a selective and stable catalyst for oxidative dehydrogenation
process is a rather ambitious target.
1.6.2 O2:EB ratio
The O2:EB ratio has an influence on the safe operation of a ST production plant in the case of ODH. It is preferred that the reaction mixture always stays outside of the
flammability limits of ethylbenzene. At 30°C the lower flammability limit (LFL) of ethylbenzene in air is 1%, and the upper flammability limit (UFL) is 6.7%. The choice of 10 vol.% oxygen with 10 vol.% ethylbenzene gives a safe operation in all parts of
the plant [11]. In addition, 100% oxygen conversion is desired, to prevent flammable mixtures in purge streams and on the distillation trays.
1.6.3 Stability
After selectivity, probably one of the most important aspects of industrial catalyst development is stability and hence the catalyst lifetime. Since a new catalyst bed is
Chapter 1
22
often a very large capital expenditure, the stability of a catalyst can determine
whether a catalytic process is economically feasible or not [46]. Under the reported reaction conditions, and time on stream, most of the
carbon-based materials are stable with the exception of the activated carbons that
are steadily decomposed [36,38,56,73]; the rate of gasification/burning is faster than that of the coke build-up. Some of the most stable systems are the carbon nanotubes and ordered mesoporous carbons, though they show a pronounced initial
deactivation in 5 h [56]. Su et al. reported a decay from 90 to 70% EB conversion in a time frame of 5 h as well [79]. A similar initial deactivation was observed for furfuryl alcohol-based CMK-3 type carbons [56,79].
The deactivation of the catalysts based on the inorganic supports due to excessive coking is still a major concern [94] as well as enhancing the selectivity;
the conventional process achieves extremely high selectivity to ST. Two types of instabilities have been found for the metal pyrophosphates, phosphates, or P-supported silica catalysts in the EB ODH. In time on stream having a maximum in
the conversion curve after that it drops [23]. Other type of instability observed after each in-situ regeneration (~2-4% selectivity and conversion) [95]. The main source of deactivation can come from the support itself under the reaction conditions that
have a steam concentration up to 10 vol. %. In general, catalytic tests of only 5-10 hours are reported in many
publications and conclusions are drawn about the performance of the catalyst
[19,22,25,34,37,38,40,47]. However, this performance can seriously deteriorate with longer time on stream (TOS). Therefore, this research focuses on longer TOS to see whether a catalyst is interesting from the industrial point of view.
1.6.4 Space velocity
In order to have a laboratory scale experiment which is comparable with an industrial scale, it is important to choose a comparable space velocity. Since industrial space velocities are in the range of 2000-20000 h-1 (GHSV) [46], having a
much lower space velocity deteriorate the industrial relevance of the experiment.
1.7 Concluding remarks on the ODH process
Despite over forty years of research, ODH for styrene production has never come further than the experimental state for several reasons. Looking at Table 2 the
following aspects stand out as infeasible for industrial scale-up of the discussed catalysts:
Many publications report a selectivity that is unsatisfactory, especially since
COx is formed, which drastically deteriorate the process economics.
Most experiments only show the first few hours of the reaction, ignoring the deactivation of the catalyst on a longer timescale.
O2:EB ratios are often not stated or too high for industrial purposes. If not all
oxygen in converted during the reaction, a high temperature mixture of hydrocarbons and oxygen is flowing downstream of the reactor. It causes explosion risks in the reaction and distillation section of the process.
Gas hourly space velocities are often not reported or too low compared to industrial processes.
Thus, from the comparison it is evident that a reliable evaluation is needed in terms of activity, selectivity, and stability under industrially relevant conditions. Note that no one has even been able to reproduce the impressive results with PNQ and
PPAN with the selectivity nearing 100% (Table 2).
Introduction
23
Table 2. ODH reaction results of various catalyst systems from literature
The general aim of the research described in this thesis is to develop improved heterogeneous catalysts based on commercial supports such as aluminas, silicas, alumina-silicas, zeolites, and carbon-based materials for the oxidative
dehydrogenation (ODH) of ethylbenzene (EB) to styrene under industrially relevant conditions. The main objectives are to improve styrene selectivity and catalyst stability, and to establish structure-performance relations. Regarding selectivity, the
catalyst should show at least comparable selectivity to the direct dehydrogenation catalysts (i.e. >95%). This is especially relevant when COx is formed during the reaction, which is highly undesirable regarding process economics and environmental
aspects. When considering conversion, a conversion higher than the conventional process (60-65%) is aimed for and it is preferentially at least similar to the SMARTTM
process (i.e. 80%). To achieve these goals, high throughput catalyst screening
studies have been performed involving catalysts based on bare commercial carriers, metal-based counterparts, carbon-based materials (commercial and tailor-made),
and P-promoted catalysts. In the Chapter 1 (Introduction) an overview of styrene production processes
is presented, and the oxidative dehydrogenation process is discussed in detail.
Various process-related aspects (i.e. selectivity, O2:EB ratio, stability, space velocity) for the ODH process are described and evaluated.
In Chapter 2 the positive impact of the thermal activation of a silica-
stabilized -alumina on the oxidative dehydrogenation of ethylbenzene to styrene is
Chapter 1
24
discussed. A systematic study was performed in a 6-flow reactor set-up. Catalysts
were characterized in detail. In Chapter 3, a systematic study on the use of carbon-based materials,
home-made carbon-silica hybrids, commercial activated carbon, and nanostructured
multi-walled carbon nanotubes (MWCNT) for the oxidative dehydrogenation of ethylbenzene is reported. Special attention was given to the reaction conditions. Relatively concentrated EB feeds (10 vol. % EB), a limited excess of O2 (O2:EB=0.6),
and lower temperatures (425-475 oC) in comparison with the commercial steam dehydrogenation process were applied.
In Chapter 4 a strategy to enhance the thermal stability of home-made
carbon-silica hybrids is proposed. It involves P-addition before the pyrolysis. In this study, the effects of P addition on a furfuryl alcohol based silica hybrid were
investigated. The performance of the P-modified hybrid catalytic materials was compared to state of-the-art P/SiO2 and MWCNT. In addition, the catalyst stability under the ODH reaction conditions was evaluated from the apparent activation
energies of the combustion reaction. In Chapter 5, the feasibility to regenerate MWCNT under mild conditions is
discussed. The regeneration method is described in detail, and the effect of the
regeneration time on the pore volume and surface area was investigated. In Chapter 6 the effect of phosphorous addition to the various inorganic
supports for ODH is described. The performance of various bare supports (silicas,
alumino-silicate, zeolites, and zeolites with low alumina content) and the corresponding phosphorous-based catalysts is presented. The fresh, spent and regenerated catalysts were analysed with various techniques and the results are
[2] Bautista F, Campelo J, Luna D, Marinas J, Quirós R, Romero A. Screening of amorphous metal–phosphate catalysts for the oxidative dehydrogenation of ethylbenzene to styrene. Appl Catal B-Environ 2007;70(1-4):611-20.
[3] Vislovskiy VP, Chang JS, Park MS, Park SE. Ethylbenzene into styrene with carbon dioxide over modified vanadia–alumina catalysts. Catal Commun 2002;3(6):227-31.
[4] James D. Ullmann. Encyclopedia of Industrial Chemistry. Wiley-VCH. 1994:329. [5] PRLog, http://www.prlog.org/11727607-styrene-global-markets-to-2020-
substitution-of-polystyrene-by-polypropylene. Accessed on Sept. 19, 2014.
[6] DeWitt & Company incorporated, http://www.dewittworld.com/portal/Default.aspx?ProductID=2. Accessed on December 5, 2013.
[7] Mimura N, Saito. Dehydrogenation of ethylbenzene to styrene over Fe2O3/Al2O3 catalysts in the presence of carbon dioxide. Catal Today 2000;55(1-2):173-8.
[8] Perry RH, Green DW, Maloney JO. Perry's chemical engineers' handbook. New
[12] Maciá-Agulló JA, Cazorla-Amorós D, Linares-Solano A, Wild U, Su DS, Schlögl R.
Oxygen functional groups involved in the styrene production reaction detected by quasi in situ XPS. Catal Today 2005;102–103:248-53.
[13] Jenkins HDB. Chemical Thermodynamics at a Glance. UK:Wiley-Blackwell.
2008:160-4. [14] Moulijn JA, Makkee M, van Diepen A. Chemical process technology. Chichester:
John Wiley & Sons. 2001.
[15] Badstube T, Papp H, Dziembaj R, Kustrowski P. Screening of catalysts in the oxidative dehydrogenation of ethylbenzene with carbon dioxide. Appl Catal A-Gen 2000;204(1):153-65.
[16] Meyers RA. Handbook of petrochemicals production processes. New York NY: McGraw-Hill. 2005: 11.3-11.34.
[17] Sanfilippo D, Capone G, Cipelli A, Pierce R, Clark H, Pretz M. SNOW: Styrene from Ethane and Benzene. Stud Surf Sci Catal 2007;167:505-10.
[18] Buonomo F, Donati G, Micheli E, Tagliabue L. Process for the production of
styrene. EP0905112 A2, 1999. [19] Tagiyev DB, Gasimov GO, Rustamov MI. Carbon deposits on the surface of
CaO/SiO2 as active catalysts for the oxidative dehydrogenation of ethylbenzene.
dehydrogenation of alkyl aromatic hydrocarbons on aluminum oxide catalysts.
III. Kinetics and mechanism of the oxidative dehydrogenation of ethylbenzene on aluminum oxide. Kinet Catal+ 1973;14(5):1182–8.
[21] Iwasawa Y, Nobe H, Ogasawara S. Reaction mechanism for styrene synthesis
over polynaphthoquinone. J. Catal 1973;31(3):444-9. [22] Fiedorow R, Przystajko W, Sopa M, Dalla Lana IG. The nature and catalytic
influence of coke formed on alumina: Oxidative dehydrogenation of
ethylbenzene J Catal 1981;68:33-41. [23] Emig G, Hofmann H. Action of zirconium phosphate as a catalyst for the
oxydehydrogenation of ethylbenzene to styrene. J Catal 1983;84(1):15-26.
[24] Cadus LE, Arrua LA, Gorriz OF, Rivarola JB. Action of activated coke as a catalyst: Oxydehydrogenation of ethylbenzene to styrene. Ind Eng Chem Res 1988;27:2241-6.
[25] Schraut A, Emig G, Hofmann H. Kinetic investigations of the oxydehydrogenation of ethylbenzene. J Catal 1988;112(1):221-8.
[26] Vrieland GE. Oxydehydrogenation of ethylbenzene to styrene over metal
pyrophosphates. 1. Catalyst composition and reaction variables. J Catal 1988;111(1):1-13.
[27] Vrieland GE. Oxydehydrogenation of ethylbenzene to styrene over metal
pyrophosphates. 2. Microbalance studies of carbon deposition and burnoff. J Catal 1988;111(1):14-22.
[28] Cadus LE, Gorriz OF, Rivarola JB. Nature of active coke in the
oxydehydrogenation of ethylbenzene to styrene. Ind Eng Chem Res 1990;29:1143-6.
[29] Vrieland GE, Menon PG. Nature of the catalytically active carbonaceous sites for
the oxydehydrogenation of ethylbenzene to styrene: A brief review Applied Catalysis 1991;77(1):1-8.
[30] Alkhazov TG, Lisovskii AE. Role of condensation products in oxidative
dehydrogenation process of ethylbenzene on aluminium-oxide catalyst. Kinet Catal 1976;17(2):375-9.
[31] Menon P. Coke on catalysts-harmful, harmless, invisible and beneficial types. J
Mol Catal 1990;59(2):207-20.
Chapter 1
26
[32] Lisovskii AE, Aharoni C. Carbonaceous deposits as catalysts for
oxydehydrogenation of alkylbenzenes. Catal. Rev. Sci. Eng. 1994;36(1):25-74, and references therein.
[33] Pereira MFR, Orfão JJM, Figueiredo JL. Oxidative dehydrogenation of
[37] Guerrero-Ruiz A, Rodriguez-Reinoso F. Oxydehydrogenation of ethylbenzene to styrene catalyzed by graphites and activated carbons. Carbon 1994;32(1):23-9.
[38] Pereira MFR, Orfão JJM, Figueiredo JL. Influence of the textural properties of an activated carbon catalyst on the oxidative dehydrogenation of ethylbenzene. Colloid Surface A 2004;241:165-71.
[39] Murakami Y, Iwayama K, Uchida H, Hattori T, Tagawa T. Study of the oxidative dehydrogenation of ethylbenzene.1. Catalytic behavior of SNO2-P2O5. J Catal 1981;71(2):257-69.
[40] Kim J, Weller S. Oxidative dehydrogenation of ethylbenzene over lanthanide oxide-promoted catalysts. Appl Catal 1987;33(1):15.
[41] Echigoya E, Sano H, Tanaka M. Eighth International Congress on Catalysis,
Berlin, 1984;5:623-633. [42] Maki Y, Sato K, Isobe A, Iwasa N, Fujita S, Shimokawabe M, Takezawa N.
Structures of H3PO4/SiO2 catalysts and catalytic performance in the hydration of
ethane. Appl Catal A-Gen 1998;170(2):269-75. [43] G. Busca, G. Ramis, V. Lorenzelli, P.F. Rossi, A. La Ginestra, P. Patrono.
Phosphoric acid on oxide carriers. 1. Characterization of silica, alumina, and
titania impregnated by phosphoric acid. Langmuir 1989;5(4):911-6. [44] Vinek H, Rumplmayr G, Lercher JA. Catalytic properties of post synthesis
[46] Hagen J. Industrial catalysis: a practical approach. Weinheim: Wiley-VCH.
2006. [47] Degannes PN, Ruthven DM. The oxidative dehydrogenation of ethylbenzene to
styrene. Can J Chem Eng 1979;57(5):627.
[48] Grunewald GC, Drago RS. Oxidative dehydrogenation of ethylbenzene to styrene over carbon-based catalysts. J Mol Catal 1990;58(2):227-33.
[49] G. Bagnasco, P. Ciambelli, M. Turco, A. La Ginestra, P. Patrono. Layered
zirconium-tin phosphates: II. Catalytic properties in the oxydehydrogenation of ethylbenzene to styrene. Appl Catal 1991;68(1):69-79.
[50] Wang L, Delgado JJ, Frank B, Zhang Z, Shan Z, Su DS, et al. Resin-derived
hierarchical porous carbon spheres with high catalytic performance in the oxidative dehydrogenation of ethylbenzene. ChemSusChem 2012;5(4):687-93.
[51] Zarubina V, Nederlof C, Van der Linden B, Kapteijn F, Heeres HJ, Makkee M,
Melián Cabrera I. Making coke a more efficient catalyst in the oxidative dehydrogenation of ethylbenzene using wide-pore transitional aluminas. J Mol
Catal A: Chem 2014;381:179-87.
Introduction
27
[52] Alkhazov TG, Lisovskii AE, Guiakhmedova TKh. Oxidative dehydrogenation of
ethylbenzene over a charcoal catalyst. React Kinet Catal Lett 1979;12(2):189-93.
[53] Drago RS, Jurczyk K. Oxidative dehydrogenation of ethylbenzene to styrene
over carbonaceous catalysts. Appl Catal A-Gen 1994;112(2):117-24. [54] Badstube T, Papp H, Kustrowski P, Dziembaj R. Oxidative dehydrogenation of
ethylbenzene with carbon dioxide on alkali-promoted Fe active carbon catalysts.
Catal Lett 1998;55(3-4):169-72. [55] Ikenaga N, Tsuruda T, Senma K, Yamaguchi T, Sakurai Y, Suzuki T.
Dehydrogenation of ethylbenzene with carbon dioxide using activated carbon-
supported catalysts Ind Eng Chem Res 2000;39(5):1228-34. [56] Zhang J, Su D, Zhang A, Wang D, Schlögl R, Hebert C. Nanocarbon as robust
catalyst: Mechanistic insight into carbon-mediated catalysis. Angew Chem Int Ed 2007;46:7319-23.
[57] De Oliveira SB, Barbosa DP, De Melo Monteiro AP, Rabelo D, Do Carmo Rangel
M. Evaluation of copper supported on polymeric spherical activated carbon in the ethylbenzene dehydrogenation. Catal Today 2008;133-135:92-8.
[58] Zhang J, Su DS, Blume R, Schlögl R, Wang R, Yang X, et al. Surface chemistry
and catalytic reactivity of a nanodiamond in the steam-free dehydrogenation of ethylbenzene. Angew Chem Int Ed 2010;49(46):8640-4.
[59] Malaika A, Rechnia P, Krzyzynska B, Kozlowski M, The influence of texture of
activated carbons on their catalytic activity in the process of ethylbenzene dehydrogenation coupled with nitrobenzene hydrogenation, Micropor Mesopor Mater 2012;163:300-6
[60] Mestl G, Maksimova NI, Keller N, Roddatis VV, Schlögl R. Carbon nanofilaments in heterogeneous catalysis: An industrial application for new carbon materials? Angew Chem Int Ed 2001;40(11):2066-8.
[61] Pereira MFR, Orfão JJM, Figueiredo JL. Oxidative dehydrogenation of ethylbenzene on activated carbon fibers. Carbon 2002;40(13):2393-401.
[62] Delgado JJ, Vieira R, Rebmann G, Su DS, Keller N, Ledoux MJ, et al. Supported
carbon nanofibers for the fixed-bed synthesis of styrene. Carbon 2006;44(4):809-12.
[63] Delgado JJ, Su DS, Rebmann G, Keller N, Gajović A, Schlögl R. Immobilized
carbon nanofibers as industrial catalyst for ODH reactions. J Catal 2006;244(1):126-9.
[64] Li P, Li T, Zhou JH, Sui ZJ, Dai YC, Yuan WK, et al. Synthesis of carbon
nanofiber/graphite-felt composite as a catalyst. Micropor Mesopor Mat 2006;95(1-3):1-7.
[65] Zhao TJ, Sun WZ, Gu XY, Rønning M, Chen D, Dai YC, et al. Rational design of
the carbon nanofiber catalysts for oxidative dehydrogenation of ethylbenzene. Appl Catal A-Gen 2007;323:135-46.
[66] Su DS, Chen X, Liu X, Delgado JJ, Schlögl R, Gajović A. Mount-etna-lava-
supported nanocarbons for oxidative dehydrogenation reactions. Adv Mater 2008;20(19):3597.
[67] Delgado JJ, Chen XW, Frank B, Su DS, Schlögl R. Activation processes of highly
ordered carbon nanofibers in the oxidative dehydrogenation of ethylbenzene. Catal Today 2012;186(1):93-8.
[68] Keller N, Maksimova NI, Roddatis VV, Schur M, Mestl G, Butenko YV, et al. The
catalytic use of onion-like carbon materials for styrene synthesis by oxidative dehydrogenation of ethylbenzene. Angew Chem Int Ed 2002;41(11):1885.
[69] Su DS, Maksimova N, Delgado JJ, Keller N, Mestl G, Ledoux MJ, et al.
Nanocarbons in selective oxidative dehydrogenation reaction. Catal Today 2005;102-103:110-4.
Chapter 1
28
[70] Su D, Maksimova NI, Mestl G, Kuznetsov VL, Keller V, Schlögl R, et al.
Oxidative dehydrogenation of ethylbenzene to styrene over ultra-dispersed diamond and onion-like carbon. Carbon 2007;45(11):2145-51.
of carbon nanotubes in the oxidative dehydrogenation of ethylbenzene. Carbon 2004;42(14):2807.
[72] Nigrovski B, Zavyalova U, Scholz P, Pollok K, Müller M, Ondruschka B.
Microwave-assisted catalytic oxidative dehydrogenation of ethylbenzene on iron oxide loaded carbon nanotubes. Carbon 2008;46(13):1678-86.
[73] Rinaldi A, Zhang J, Mizera J, Girgsdies F, Wang N, Hamid SBA, et al. Facile
synthesis of carbon nanotube/natural bentonite composites as a stable catalyst for styrene synthesis. Chem Commun 2008;48:6528-30.
[74] Nigrovski B, Scholz P, Krech T, Qui NV, Pollok K, Keller T, et al. The influence of microwave heating on the texture and catalytic properties of oxidized multi-walled carbon nanotubes. Catal Commun 2009;10(11):1473-7.
[75] Frank B, Zhang J, Blume R, Schlögl R, Su DS. Heteroatoms increase the selectivity in oxidative dehydrogenation reactions on nanocarbons. Angew Chem Int Ed 2009;48(37):6913-7.
[76] Delgado JJ, Chen X, Tessonnier JP, Schuster ME, Del Rio E, Schlögl R, et al. Influence of the microstructure of carbon nanotubes on the oxidative dehydrogenation of ethylbenzene to styrene. Catal Today 2010;150(1-2):49-
54. [77] Qui NV, Scholz P, Krech T, Keller TF, Pollok K, Ondruschka B. Multiwalled
carbon nanotubes oxidized by UV/H2O2 as catalyst for oxidative
dehydrogenation of ethylbenzene. Catal Commun 2011;12(6):464-9. [78] Qui N, Scholz P, Keller T, Pollok K, Ondruschka B, Ozonated Multiwalled Carbon
Nanotubes as Highly Active and Selective Catalyst in the Oxidative
Dehydrogenation of Ethyl Benzene to Styrene, Chem Eng Technol 2013;36(2): 300-6.
[79] Qi W, Liu W, Zhang B, Gu X, Guo X, Su D, Oxidative Dehydrogenation on
Nanocarbon: Identification and Quantification of Active Sites by Chemical Titration, Angew Chem Int Ed 2013; 52:14224-8.
[80] Du Y, Li J, Ya X. Polyaniline as nonmetal catalyst for styrene synthesis by
oxidative dehydrogenation of ethylbenzene. Catal Commun 2008;9(14):2331-3. [81] Su DS, Delgado JJ, Liu X, Wang D, Schlögl R, Wang L, et al. Highly ordered
mesoporous carbon as catalyst for oxidative dehydrogenation of ethylbenzene
to styrene. Chem Asian J 2009;4(7):1108-13. [82] Wang L, Delgado JJ, Frank B, Zhang Z, Shan Z, Su DS, et al. Resin-derived
hierarchical porous carbon spheres with high catalytic performance in the
oxidative dehydrogenation of ethylbenzene. ChemSusChem 2012;5(4):687-93. [83] Xiao N, Zhou Y, Ling Z, Zhao Z, Qiu J, Carbon foams made of in situ produced
carbon nanocapsules and the use as a catalyst for oxidative dehydrogenation of
Albero AM, Rodrıguez-Reinoso F, A simplified route to the synthesis of CMK-3
replica based on precipitation polycondensation of furfuryl alcohol in SBA-15 pore system, Carbon 2013;64: 252-61.
[85] Vrieland GE, Friedli HR. US patent 3933932, 1976.
[86] Kurakami Y, Iwayama K, Uchida H, Hattori T, Tagawa T. Screening of catalysts for the oxidative dehydrogenation of ethylbenzene. Appl Catal 1982;2(1-2): 67-74.
[87] Schraut A, Emig G, Sockel H-G. Composition and structure of active coke in the oxydehydrogenation of ethylbenzene. Appl Catal 1987;29(2):311.
Introduction
29
[88] Arrúa LA, Ardissone DE, Quiroga OD, Rivarola JB. Oxidehydrogenation of
ethylbenzene on P−O−Ni catalyst. React Kinet Catal Lett 1995;56(2):383-389. [89] Dziewiecki Z, Jagiello M, Makowski A. Investigation of polymer organic deposit
formed on nickel phosphate in oxidative dehydrogenation of ethylbenzene.
Making Coke a more Efficient Catalyst in the Oxidative Dehydrogenation of Ethylbenzene using wide-pore Transitional Aluminas
The thermal activation of a silica-stabilized -alumina impacts positively on the oxidative
dehydrogenation of ethylbenzene (EB) to styrene (ST). A systematic thermal study reveals that the
transition from -alumina into transitional phases at 1050 oC leads to an optimal enhancement of
both conversion and selectivity under pseudo-steady state conditions; where active and selective
coke has been deposited. The effect is observed in the reaction temperature range of 450-475 oC
at given operation conditions resulting in the highest ST yield, while at 425 oC this effect is lost due
to incomplete O2 conversion. The conversion increase is ascribed to the ST selectivity improvement
that makes more O2 available for the main ODH reaction. The fresh aluminas and catalytically
active carbon deposits on the spent catalysts were characterized by gas adsorption (N2 and Argon),
acidity evaluation by NH3-TPD and pyridine adsorption monitored by FTIR, thermal and elemental
analyses, solubility in CH2Cl2 and MALDI-TOF to correlate the properties of both phases with the ST
selectivity enhancement. Such an increase in selectivity was interpreted by the lower reactivity of
the carbon deposits that diminished the COx formation. The site requirements of the optimal
catalyst to create the more selective coke is related to the higher density of Lewis sites per surface
area, no mixed Si-Al Brønsted sites are formed, while the acid strength of the formed Lewis sites is
relatively weaker than those of the bare alumina.
Chapter 2
32
2.1 Introduction
Styrene (ST) is industrially produced by direct dehydrogenation of ethylbenzene (EB) using steam at 580-630 °C [1]. The process suffers from high energy consumption due to low conversion per pass because of equilibrium limitations, and the high
temperatures required for the endothermic reaction. The operation cost mainly depends on oil prices and utilities; decreasing the reaction temperature can have a remarkable market impact. Oxidative dehydrogenation (ODH) can address these
drawbacks [2], but it is not commercialized among others mainly due to the limited catalyst stability and the lower selectivity to styrene.
A distinctive catalyst support for this reaction is -alumina due to its mild
acidity as compared to e.g. zeolites. The coke deposits generated on the alumina
under these oxidative reaction conditions do not produce deactivation, but promote the activity and selectivity [3-6]. There is a general consensus that these carbon
deposits species are partially oxygenated and act as active and selective catalytic sites [3-17]. This has been confirmed by employing pre-coked catalysts [18] and by measuring the intrinsic activity of the carbon deposits separated from the alumina.
Those carbon deposits were superior in respect to the alumina [6,15]. Nederlof et al. [17] demonstrated that this also holds for CO2-asisisted EB ODH; during the first 15 h time on stream the alumina shows an increase in ethylbenzene conversion from 15
to 60% and styrene selectivity from 60 to 92%. Characterizations of the Al2O3 samples after different times on stream show a clear correlation between the formed coke and the activity and selectivity. This is also shown in this work for the O2-based
EB ODH, where the activation period can be observed at reaction temperatures lower than 475 oC; at temperatures > 475 oC the coke build-up is fast.
The alumina by itself is not an inactive support but it provides Lewis centres
and those generate „good coke‟ [7,14,15]; the Brønsted acidity does not play a key role in forming selective coke for this reaction. These findings on active coke established a trend setting by using synthetic carbons that mimic the in-situ
produced deposits during EB ODH [18-41]. Typically, their conversion levels are acceptable, 50-70%, while the styrene selectivity is relatively low for commercialization, up to 70%, compared to the commercial process of steam
dehydrogenation (ST selectivity > 95%). In the framework of the EB ODH catalyst development, less attention has
been given on the use of inorganic supports, in particular about strategies to improve
either conversion, selectivity or stability of their carbonaceous sites or combinations thereof. Few examples indicate that the acidity enhancement of -Al2O3 by H3PO4
[11] or HBO3 [42,43] has a positive impact on the styrene yield. The effect of the
texture of the alumina phases at high temperature, to the best of our knowledge, has not been investigated systematically. In this study, we use thermally treated SiO2 promoted -Al2O3 at high temperature; the pore size is expanded by coarsening
the nanoparticles to investigate the effect of pore size on the reaction performance.
We have observed that this treatment also modifies the nature of the carbon deposit formed; making it less reactive and consequently more selective to ST. The relationships between EB ODH performance and the alumina and coke properties are
discussed in this paper and rationalize previous observations [44] where the EB conversion and ST selectivity were optimal for a specific phase composition at high temperature.
Making Coke a more Efficient Catalyst in the Oxidative Dehydrogenation of Ethylbenzene using wide-pore Transitional Aluminas
33
2.2 Experimental
2.2.1 Materials
In this study a low SiO2 stabilized γ-Al2O3 extrudates (Albemarle Catalysts BV) were employed. The extrudates were crushed and sieved into a 212-425 μm fraction used
for the catalytic tests and characterization. The material was thermally activated at temperatures ranging from 500 to 1200 oC in ceramic crucibles using a box furnace in static air at a rate of 4 oC/min and held for 8 h. The samples were labeled as
AluX00, where X00 corresponds to the treatment temperature in degree Celsius. Other inorganic commercial supports and catalysts were tested as reference: TiO2 (Degussa, P25), SiO2 (Fuji Silysia, G-6 5 microns), CeO2 nanopowder (Sigma-Aldrich,
700290), mesoporous Al-MSU-F type (Sigma Aldrich, 643629) and Merck ultrapure alumina (1.01095.1000).
2.2.2 Characterization of bare and spent catalysts
The coke content and stability of the spent catalysts (after 60 h time on stream)
were determined by thermogravimetric analysis in a Mettler-Toledo analyzer (TGA/SDTA851e) using a flow of synthetic air of 100 ml/min (STP). The temperature was increased from 30 to 900 °C at 10 °C/min. Blank curve subtraction using an
empty crucible was employed. CHN elemental analyses were carried out in a EuroVector 3000 CHNS
analyzer. Approximately 2 mg of sample was accurately weighed in a 6-digit analytic
balance (Mettler Toledo). The samples were burnt at 1800 oC in the presence of an oxidation catalyst and decomposed into CO2, H2O, and N2. These gases are then separated in a Porapak QS column at 80 oC and quantified with a TCD detector.
Acetonitrile (99.9%) was used as an external standard. Textural analysis of the fresh and spent catalysts was carried out by N2 and Ar
physisorption at -196 oC and -186 oC, respectively, in a Micromeritics ASAP 2420
analyzer. Prior to the measurements all samples were outgassed under vacuum at 350 oC for 10 h for the fresh materials, while for the spent catalysts a mild degassing was applied, 130 oC for 24 h. The surface area was calculated by BET method [45],
SBET. The single point pore volume (VT) was estimated from the amount adsorbed at a relative pressure of 0.98 in the desorption branch. The pore size distribution was derived from the BJH model [45]. For the fresh materials, the adsorption pore sizes
are used. In case of spent catalysts, the desorption branch is favored to evidence the presence of closed pores.
NH3-TPD experiments were carried out in a Micromeritics AutoChem II system
equipped with a thermal conductivity detector (TCD). The sample (ca. 30 mg) was pre-treated by heating it up to 500 °C in He at 10 °C/min. The sample was cooled to 120 °C at a similar cooling rate, and then exposed to 1 vol% NH3/He (25 ml/min) for
30 min. Subsequently, a flow of He (25 ml/min) was passed through the reactor for 60 min to remove weakly adsorbed NH3 from the samples‟ surface. After this
baseline stabilization, the desorption of NH3 was monitored in the range of 120-1000 °C using a heating rate of 10 °C/min.
The acidity was also evaluated by pyridine adsorption monitored by IR. FTIR
spectra were recorded on a Thermo Scientific Nicolet 6700 FTIR spectrometer equipped with a MCT-B detector and a quartz home-made IR cell having CaF2 windows connected to a vacuum system that allows the controlled dosing of pyridine.
Around 50 mg of sample were pressed in a 1.767 cm2 self-supporting wafer and mounted into the quartz cell. Before dosing the pyridine, the sample was preheated under vacuum (5.10-6 mm Hg) as follows: 120 oC for 2 h at 1 oC/min, then at 400 oC
for 2h, 1 oC/min followed by cooling to 150 oC. Pyridine gas was dosed into the
Chapter 2
34
sample wafer until full saturation at 150 oC (i.e. the pyridine bands become stable).
Afterwards, the sample wafer was evacuated and cooled down to room temperature. Spectra were measured by accumulating 256 scans at a resolution 4 cm−1; the spectrum of the sample after evacuation was subtracted to each spectrum. The
strength of the acid sites was evaluated by recording the spectra after evacuating the sample at increasing temperatures, with 50 oC increments. The spectra were taken once the pressure remains stable. The density of acid sites was calculated
according to Tamura et al. [46]: W
SACW
where CW is the acid site density (mol/g); is the molar extinction coefficient (1.71
cm.mol-1 for Lewis sites on alumina [46]); S is the surface of the sample disc area
(cm2), W the sample weight (mg), and A is the peak area (cm-1). The skeletal density was obtained by He pycnometry at room temperature
after evacuating the sample chamber (1 ml) 5 times and measuring in 10 cycles; the standard deviation for each analysis is given.
XRD analysis was carried out in a Philips PW1840 using a CuK radiation. The
phase identification was done using the JCPDS database.
2.2.3 Characterization of the active coke
The coke nature and composition were analyzed after separating it from the inorganic support. The spent catalyst was treated in dichloromethane (CH2Cl2) under
reflux for 6 h, which removes the soluble coke compounds retained on the sample [47]. The percentage of insoluble coke (IC) in total was determined by the difference in the TGA weight loss between the spent catalyst and the catalyst after refluxing in
CH2Cl2. The carbonaceous insoluble compounds were obtained by treating the alumina matrix in a hydrofluoric acid solution (40 wt.% HF) at room temperature overnight, washing of the residue by centrifugation and subsequently drying at 90 °C
for 10 h. This produces brown-to-black particles that were subjected to CHN analysis to quantify the empirical CxHyOz formula. For the high-temperature spent samples, the alumina matrix could not be fully dissolved due to its inertness; the inorganic
residue was taken into account for the calculations of the corrected C and H concentrations in the coke while O was calculated by difference. The presence of O2, a temperature of 1800 oC and an oxidation catalyst in the CHN analysis can ensure
that the formation of Al-oxycarbide is avoided. The insoluble carbon material was mildly analyzed by MALDI time of flight
(TOF) spectrometry [48-51]. These IC particles were suspended in tetrahydrofuran
(THF) by sonication and mixed with dithranol as a matrix. MALDI spectra were recorded on an Applied Biosystems, Inc. Voyager-DE PRO spectrometer by pulsed ion extraction using a solid state laser with a wavelength of 355 nm. Laser desorption-
ionization (LDI)-TOF was also performed in the same instrument. Samples were suspended in THF, 1 L was spotted directly on a stainless steel target plate. In both
modes, the measurements were performed in linear mode using positive ionization.
Spectra were accumulated between 100 or 300 and 3000 or 10000 Da (depending on spectrum) and calibrated externally.
2.3 Catalytic tests
The catalytic tests were carried out in a parallel flow micro reactor using a fixed volume of catalyst (0.8 ml) using down-flow 4 mm quartz reactors. The reactors are
typically loaded as follows: quartz wool, 10 cm glass pearls (0.5 mm diameter), the catalyst, 10 cm glass pearls (0.5 mm diameter), and a second quartz wool plug. In this way, it is assured that the catalyst bed is located in the isothermal operational
zone of the furnace. In all cases the same volume of catalyst bed was used,
Making Coke a more Efficient Catalyst in the Oxidative Dehydrogenation of Ethylbenzene using wide-pore Transitional Aluminas
35
corresponding to 65 mm of bed length; so the sample mass increases with increasing packing density of the catalyst. The inertness of the glass beads was verified as they showed less than 3% EB conversion under all applied conditions. The reactor gas
feed is a mixture that can consist of CO2, N2, and air that counts for a gas-flow rate of 36 ml (NTP)/min; the liquid EB-feed flow rate is 1 g/h (3.54 ml (NTP)/min vapour) that is evaporated upstream each reactor in a -Al2O3 column, resulting in a 1:9
molar ratio of ethylbenzene to gas (10 vol. % EB). This corresponds with operation at a GHSV of 3000 l/l/h. The total pressure was 1.2-1.3 105 Pa. The pseudo steady-state conditions refer at the reaction time where both the conversion and selectivity
achieve the highest values, which are typically achieved after 2 to 6 h time on stream; the reported conversion data are given at 6-10 h time on stream or longer. The conversion and selectivity were averaged in the isothermal period after reaching
pseudo steady-state conditions. We state „pseudo steady state conditions‟ because after the conversion and selectivity achieve a maximum, they start both to decline slowly; most probably due to excessive coke build-up. The standard deviations were
below 2% (see example Table S-1 in appendix) for all cases. The reactor exhaust gas was analyzed by gas chromatography using a combination of columns (0.3m Hayesep Q 80-100 mesh with back-flush, 25m × 0.53mm Porabond Q, 15 m ×
0.53mm molsieve 5A and RTX-1 with 30m × 0.53mm), and TCD and FID detectors. This configuration allows the quantification of permanent gases such as CO2, H2, N2, O2, CO as well as hydrocarbons (typically, methane, ethane, ethene, benzene,
toluene, ethylbenzene, styrene, and heavy aromatics). The catalytic tests were carried out under practical conditions of 20 % excess
O2 with respect to the ODH reaction, a concentrated EB feed of 10% and constant
bed volume. Unless stated otherwise, for all EB conversion data the oxygen conversion is complete. The oxygen is used to generate coke, to convert ethylbenzene mainly into styrene, and to convert deposited coke into COx. Since the
selectivities to side products are small and almost constant as a function of the reaction conditions, the selectivity to styrene (ST) is in principle directly coupled to COx. These two combined selectivities (COx and ST) are on average responsible for
96% of the converted ethylbenzene. All physical characterizations for the spent catalysts and derived coke were done after the complete testing cycle of 60 h. Due to the O2 gradient in the reactor, the coke will never be a steady-state material.
Because of that, we took the complete sample, mix it and used it for characterization as an average sample of the reactions conditions applied, location in the bed, and
time on stream.
2.4 Results and discussion
2.4.1 Pseudo steady-state ODH performance
Under pseudo steady-state conditions, typically after 5 h on-stream, the bare -Al2O3
outperforms other commercial supports such as CeO2, TiO2, and SiO2 (data not
shown). The bare alumina (SBET = 272 m2/g and 0.389 ml/g pore volume) is an optimized industrial catalyst support that is stabilized with SiO2; it has an extraordinarily high surface area and pore volume. The thermal activation of the
bare -alumina was carried out to expand the pore size by coarsening the
nanoparticles and investigate its effect on the reaction performance. Remarkably, both conversion of EB and selectivity to ST steadily increase with pretreatment temperature (Figure 1), showing an optimal temperature at 1050 oC (Alu1050)
above which both quantities dropped significantly.
Chapter 2
36
Figure 1. Ethylbenzene conversion and styrene selectivity at 475oC, 1g EB/h and O2:EB=0.6 with a total flow of 36 mL STP/min measured at 105 Pa with 0.8 ml catalyst after 10 h time on stream; GHSV of 3000 h-1.
In all cases the oxygen was completely converted. The maximum ST yield of
35% was found for the Alu1050, which corresponds to ca. 22% relative increase to
the Alu500 at the given GHSV of 3000 l/l/h. Figure 2-a represents the ST yield versus time on stream at various temperatures and EB:O2 ratios for Alu500 and Alu1050. The positive effect of the calcination was observed at 450 and 475 oC
reaction temperature at which full O2 conversion was observed. At 425 oC the O2 conversion was not complete for the Alu1050 and the catalyst performance trend was reversed (Alu500>Alu1050). It should be noted that the O2 conversion was not
always complete for some catalysts at 425 oC; this means that the O2 concentration profiles decrease variably over the reactor bed for the Alu series (i.e. smoother or sharper). Hence, the „active catalyst reactor volume‟ where oxygen is present is
variable as a function of reactor length. This makes the comparison of the intrinsic activity complex at 425 oC. Therefore, the discussion in this paper refers to reaction temperatures ≥450 oC, where full oxygen conversion was achieved. At these
temperatures, the complete catalyst bed was taken as reference for the interpretation.
Figure 2-a also evidences that the catalysts are deactivating as a function of
time on stream, most probably due to ongoing coke deposition on the coked catalysts. Figure 2-b represents the time dependency at 450 oC for Alu500 (O2:EB=0.6); it shows the activity and selectivity increase during the initial coke
formation before achieving a pseudo stationary state. This is in agreement with Nederlof et al. [17] for the CO2-assisted EB ODH, it is attributed to the formation and deposition of active coke on the alumina surface. From a practical standpoint it is
beneficial starting the catalyst performance tests at the highest temperature, i.e. 475 oC, because the coke build-up is faster than that at 425-450 oC; in this way the pseudo-steady state is achieved at shorter time. This can be seen by comparing the
initial performance (<5 h) for Alu500 in Figure 2-a (475 oC) and Figure 2-b (450 oC): at 450 oC the pseudo stationary conditions are achieved after 6 h, while this is
reduced to 3 h at 475 oC. From Figure 2-a, increasing the O2:EB ratio may look favorable in terms of the
overall ST yield but this will have a penalty in the selectivity to ST; a high selectivity
is one of the major targets in the O2 EB ODH to compete with the existing steam-based process. It must be emphasized that the catalytic performance in Figure 1
400 600 800 1000 1200
10
20
30
40
50
60
EB
con
vers
ion
/ %
Pretreatment temperature / oC
50
60
70
80
90
100
ST
sele
ctivity / %
SELECTIVITY CONVERSION
Making Coke a more Efficient Catalyst in the Oxidative Dehydrogenation of Ethylbenzene using wide-pore Transitional Aluminas
37
corresponds to pseudo-steady state conditions between 5-10 h time on stream with significant coke build up. Assuming the density of graphene (0.76 mg/m2) the number of coke layers is nearly 2.4 for Alu500 and it increases up to 3.4 for
Alu1050. Therefore, we assume this increased performance corresponds to the coke deposits; no interaction between the reactants with the alumina surface or the interface alumina-coke is expected.
Figure 2. a) Time on stream performance at various temperatures of the optimal Alu1050 and
Alu500 catalysts at various temperatures (475, 450, 425, and 450 oC) and O2:EB= 0.6, 0.4 and 0.2 (vol). GHSV of 3000 l/l/h; b) time on stream dependency at 450 oC and O2:EB= 0.6 (Alu500).
We first attempt to correlate this promoting effect to the total amount of carbon deposited, as it is considered the dominating active phase once the deposits are formed under pseudo-steady state conditions. The coke contents are displayed in
Figure 3 as gravimetric quantity (i.e. g coke/g fresh alumina) as well as the volumetric coke (g coke/ml reactor bed); the latter can be more appropriately related to the product yield as it represents the actual amount of coke per reactor
volume, taking into account that the performance is measured at fixed catalyst volume. From this representation, it can be unambiguously seen that the amount of volumetric coke progressively decreases with calcination temperature with a
pronounced reduction above 1050 oC. Considering the increased ST product yield, this suggests that the coke becomes apparently more active (different species) or more selective for the ODH reaction. Calculation of the coke per surface area
indicates that the coke on the low-temperature aluminas is more dispersed (160 mol C/m2 for Alu500) than the optimal Alu1050 (205 mol C/m2).
0 10 20 30 40 50 60 700
5
10
15
20
25
30
35
40
45
50 Alu1050
Alu500
Yie
ld / %
Time / h
425 450 450
0.6 0.4 0.6 0.6 0.4 0.6 0.4
475 oC
O2/EB
0 25 50 75 1000
10
20
30
40
ST
se
lectivity
/ %
EB
co
nve
rsio
n
/ %
Time / h
70
80
90
100
a)
b)
Chapter 2
38
Figure 3. Coke on catalyst (gravimetric and volumetric) determined by TGA as a function of the pre-treatment temperature of the bare alumina after 60 hours, time on stream.
Regarding the selectivity effect, less of the available oxygen is used for the
combustion-gasification of the deposited carbon; for Alu500 around 35% of the
converted oxygen is used for ODH reaction, whereas this increases to 50% for the Alu1050 catalyst. The conversion increase cannot be justified by an additional pathway. Because we run at full O2 conversion, part of the bed will be O2 free. We
did not observe changes in the heavy by-products composition indicating the stability of the by-products in the absence of O2 (2% benzene/toluene and 2% oxygenated and heavy aromatics were observed in all the cases). The reaction temperatures
(425-475 oC) are relatively low to justify the direct EB dehydrogenation; the conventional steam dehydrogenation is operated above 600 oC. If direct dehydrogenation would occur in the absence of O2, a decrease in the EB conversion
is expected due to the low equilibrium conversion at these temperatures (e.g. 20% at 450 oC, EB partial pressure of 0.1 bar [52]). The key aspect to explain the conversion increase is otherwise the improved selectivity to ST; a lower COx
formation will make more O2 available for the main ODH reaction, and this is responsible for the conversion increase. 2.4.2 Textural properties of the alumina upon thermal treatment
The alumina surface has limited intrinsic activity with a low EB selectivity in the order of approximately 75% (Fig. 2-b), also observed in [15]. It is the coke deposited on
the alumina surface that makes the catalyst more selective and active. However, the initial surface area and pore volume of the fresh alumina (and acidity) are important as these will determine how much and well dispersed the coke is. In addition,
accessibility is highly important in heterogeneous catalysis, as diffusion limitations may completely deteriorate selectivity and reduce activity, especially for the much larger ethylbenzene with respect to oxygen. The texture and accessibility of the fresh
aluminas were evaluated by textural properties as derived from N2 physisorption. The textural parameters are plotted in Figure 4-a as a function of the calcination temperature. The marked reduction of surface area and pore volume is associated to
the thermal coarsening of the alumina nanoparticles because the pore size increases. The BJH pore size distributions (PSD) shift towards larger pores (Figure 4-b); this is consistent with the reduction of the textural parameters. Figure 4-b additionally
Pretreatment temperature / oC
Co
ke o
n c
atal
yst
g/g
or
g/m
L)
Making Coke a more Efficient Catalyst in the Oxidative Dehydrogenation of Ethylbenzene using wide-pore Transitional Aluminas
39
shows a broadening of the pore size distribution that is more evident for the Alu1000 upwards while full collapse occurs at 1150 oC. At 1050 oC, this corresponds to the optimal catalyst performance, the pore size distribution is broad, but still well-
defined with a surface area of 101 m2/g. Analysis of an ultrapure gamma-alumina (without silica stabilization) calcined at 1050 oC shows 11 m2/g (Table S-3) and no PSD (Figure 4-c) due to structural collapse. Therefore, the preservation of the
texture, or lack of pore collapse for the Alu1050 is associated to the promotion effect of silica [53]. The changes in texture were accompanied by a densification of the material with an increase of the skeletal density from 3.035 to 4.067 g/ml (Table S-2
in supplementary information). The changes in texture of two relevant spent catalysts containing the coke
were investigated as well and compared to the fresh counterparts. The isotherm of
the fresh Alu500 is of the type IV with hysteresis H1 [54]; representing solids with cylindrical pore geometry with relatively high pore size uniformity and facile pore connectivity. However, the hysteresis changes into type H2 for the spent catalyst
with a closure point at 0.45 relative pressure. Hysteresis H2 is believed to occur in solids where the pores have narrow necks and wide bodies, also called ink-bottle type pores, or when the porous material has interconnected pores. As the original
material has no interconnectivity around 0.45 relative pressure, the spent Alu500 possesses pore neck restrictions attributed to the excessive coke build-up. This is
consistent with the observed TSE effect at p/po = 0.45 in the desorption branch, supposedly showing a uniform pore at 3.9 nm (inset in Figure 5-top). This is not due to a real pore but a physical phenomenon associated to the nature of the adsorbate;
it is known that pore restrictions below 3.9 nm cannot be detected by N2 because of this effect [55]. Therefore, we can only conclude that the TSE points out to pore restrictions smaller than 3.9 nm; it is uncertain whether these restrictions are
relevant for this reaction in terms of mass-transfer limitations. The isotherm for the spent Alu1050 (Figure 5-bottom) maintains the H1 hysteresis (as for the fresh counterpart) with no detection of the TSE effect in the desorption PSD. This indicates
that the pore expansion by high temperature calcination and limited coke build-up keep the pores accessible during the reaction.
Chapter 2
40
Figure 4. a) Specific surface area and total pore volume as a function of the calcination temperature of the alumina; b) BJH pore size distribution using the adsorption branch of the isotherm; both derived from N2 physisorption at -196 oC; c) Comparison of the PSD between Alu1050 and an ultrapure alumina (Merck) calcined at 1050 oC (C1050).
20 100 10000.0
0.5
1.0
Alu1050
C1050
BJH
Ad
so
rptio
n d
V/d
log
(w)
Po
re V
olu
me
pore size /
100 10000.0
0.5
1.0
1.5
2.0
Alu500
Alu600
Alu700
Alu800
Alu900
Alu1000
Alu1050
Alu1100
Alu1150
Alu1200
BJH
Ad
so
rptio
n d
V/d
log
(w)
Po
re V
olu
me
pore size / Å
a)
b) 400 600 800 1000 12000
50
100
150
200
250
300
Pretreatment temperature / oC
SBET (m
2/g
)
0.0
0.2
0.4
0.6
0.8
1.0
VT (c
m3/g
)
c)
Å
Making Coke a more Efficient Catalyst in the Oxidative Dehydrogenation of Ethylbenzene using wide-pore Transitional Aluminas
41
Figure 5. N2 physisorption isotherms and corresponding desorption BJH PSD (inset) for Alu500 and Alu1050: a) fresh and b) spent catalysts. The indicated pore sizes correspond to the desorption branch.
2.4.3 Acidic properties of the alumina upon thermal treatment
The acidity properties were evaluated by temperature programmed desorption of NH3 in order to understand the reasons why the wide-pore SiO2 promoted -alumina
phase becomes a better selective coke-forming support with the calcination
temperature. The TPD profiles (Figure S-1) are formed by two contributions centered at 250 and 600 oC that do not change in position in the Alu series, representing weak and stronger sites. The quantification of the total acidity per catalyst volume displays
a threshold value of 385 to 420 mol/ml-cat, which drops significantly after
calcination of the alumina at 1100 oC to 245 mol/ml-cat. A commercial ultra-pure -
Al2O3 was thermally treated at 1000 oC and its acidity was quantified by NH3-TPD (Figure S-1 and Table S-2). It clearly shows a less intense desorption profile with a
very low acidity (90 mol/g), while the corresponding Alu1000 has 436 mol/g;
hence the high acidity is attributed to the silica promotion. The acidity of relevant catalysts of the series (Alu500 and Alu1050) was
further assessed by pyridine adsorption, monitored by FTIR; this technique is unique in providing information about the nature and strength of the acid sites. The spectrum of the bare Alu500 (Figure 6-left and a) shows a distinctive adsorption of
pyridine on Lewis Al sites at 1449 cm-1; no Brønsted sites were detected that
Chapter 2
42
typically appears at 1545 cm-1 [46]. This consequently implies that there are no
aluminosilicate domains; silica and alumina are present as separate phases. This is in agreement with results of Daniell et al. [56] that observed independent phases up to 5 wt.% silica in alumina. Therefore, the 1 wt. % silica present in bare alumina is a
textural promoter that avoids thermal coarsening of the alumina crystallites and prevents pore collapse, as discussed in 2.4.2. At room temperature the Lewis sites density of Alu500 is 126 mol/g (Table 1) that is in agreement with reported values
[46,57]. It is remarkable that the density obtained from NH3 TPD is a factor 5 higher than pyridine. This is likely due to the fact that NH3 is less selective and accounts for the adsorption on crystallographic defects. The density of acid sites per surface area
was further quantified using Argon physisorption (Table S-2) instead of N2. The quadrupole moment of the nitrogen molecule on highly hydroxylated surfaces causes an orientation effect of the adsorbed nitrogen molecules [58]; this can be solved
using Argon physisoprtion at 87 K. The acid site surface density results in 0.33 sites/nm2 for the Alu500 (Table 1). The strength of the Lewis sites was evaluated after desorbing pyridine at increasingly higher temperatures up to 400 oC (Figure 6-
left, spectra b-g). Quantification of the acid sites density at each evacuation temperature indicates a drop from 126 to 33.9 mol/g at 400 oC (Table 1).
Alu1050 shows Lewis sites at 1449 cm-1 (Figure 6-right and a) with a density
of 138.9 mol/g (Table 1). Brønsted sites were not detected, thus the thermal
treatment at 1050 oC does not produce the mixing between silica and alumina and remain as separate phases. The acid site surface density using Ar physisorption is of 1.39 sites/nm2, that is relatively higher than that of the Alu500 (0.33 sites/nm2). The
acid strength of the Lewis sites was distinctively inferior to the Alu500. At 250 oC Alu1050 has 5 times less sites than that of Alu500, and pyridine was fully desorbed
at 350 oC, while Alu500 still has 49.7 mol/g at this temperature of 350 °C.
It is concluded that Alu1050 has a high acid site density per surface area, higher surface area than non-silica promoted alumina calcined at 1050 oC; both effects are ascribed to the silica promotion, while the Lewis sites are relatively
weaker than that of Alu500.
Making Coke a more Efficient Catalyst in the Oxidative Dehydrogenation of Ethylbenzene using wide-pore Transitional Aluminas
43
Figure 6. IR spectra of adsorbed pyridine after evacuation at various temperatures for Alu500 (left) and Alu1050 (right): a) room temperature; b) 150; c) 200; d) 250; e) 300; f) 350 and g) 400oC.
Table 1. Lewis acid sites densities determined by FTIR pyridine adsorption.
Alu500 Alu1050
Temperature (oC) a mol/g sites/nm2 b mol/g sites/nm2 b
25 126.2 0.33 138.9 1.39
150 140.7 0.37 57.5 0.58
200 119.2 0.31 31.6 0.32
250 85.3 0.23 16.4 0.17
300 66.6 0.18 4.7 0.05
350 49.7 0.13 0 0
400 33.9 0.09 0 0
a. Evacuation temperature; b. Using SBET values derived from Ar physisorption at -186 oC.
2.4.4 Coke nature and reactivity
In this section the nature of the coke will be discussed based on solubility in CH2Cl2,
thermal analysis (TPO), chemical analysis (CHN), and MALDI information; the latter two techniques were applied on carbonaceous residues after a process of dissolution of the alumina.
The importance of the insoluble coke comes from the fact that the fraction of coke that is less active in styrene production, burns more easily under oxidative
1650 1600 1550 1500 1450 1400
Absorb
ance
Wavenumbers / cm-1
1650 1600 1550 1500 1450 1400
Absorb
ance
Wavenumbers / cm-1
a
b
c
d
e
f
g
a
b
c
d
e
f
pyH+ py on Al3+ 0.2 1545 cm-1 1456 cm-1
0.2 pyH+ py on Al3+ 1545 cm-1 1456 cm-1
Chapter 2
44
conditions, and its nature corresponds to soluble coke; therefore insoluble coke (IC)
determines the product yield (conversion times selectivity) [9,12,15]. The proportion of insoluble coke after refluxing the spent catalysts in CH2Cl2 was quantified (Table 2). The differences among the aluminas are undetectable, always IC > 97%, and
cannot explain the differences in selectivity discussed in Figure 1. The chemical composition of the insoluble carbon deposits was further characterized after dissolving in HF the alumina from the spent catalysts. It was found that the high
temperature aluminas were substantially inert and hard to dissolve completely, showing an ash residue of ca. 8-9 wt. %. Because of that, the elemental composition was tentatively corrected using the TGA residue. The chemical composition of
representative samples of the low- and high-temperature treatments is given in Table 2 and plotted in a Van Krevelen-type plot (Figure 7-a) together with several
reference catalysts. The C:H molar ratio stays in the typical range (1.8-2.5) reported for other untreated -Al2O3 and modified counterparts [7,15], that is typical of
polycondensed/aromatic coke. The high temperature aluminas produced a priori a substantial amount of oxygen-rich coke with O:C molar ratios of 0.20 (0.19 duplo)
and 0.29 (0.27 duplo). In particular, the Alu1050 has a higher concentration of oxygenates and less hydrogen, C:H=2.21; indicating less condensation apparently.
Table 2. Chemical characteristics of the coke for representative low and high temperature spent catalysts after HF treatment.
b) Alumina residue due to incomplete dissolution in 30 wt. % HF. c) Pore diameter (PD) for the corresponding alumina at the pore size distribution maxima
(Figure 4). d) Cylindrical geometric value calculated as 4(VT)/SBET as having a substantially more
asymmetric pore size distribution.
Making Coke a more Efficient Catalyst in the Oxidative Dehydrogenation of Ethylbenzene using wide-pore Transitional Aluminas
45
Figure 7. a) Modified Van Krevelen plots representing the composition of the carbon deposits of various Alu samples (Table 2) as well as data adapted from prior work [,7] and [,15]; b)
oxidation rate in air (TPO) using the TGA derivative patterns of the spent Alu catalysts (>60 h time on stream).
It is remarked that the variability in the ash content within a sample for the
high temperature alumina introduces uncertainty in the representativeness of the samples used in the TGA and CHN analyses. As the oxygen content is very sensitive to the ash residue, we believe that the relatively high oxygen content for Alu1050
derived coke can be due to an overestimation. This is supported by the coke oxidation patterns (discussed later in Figure 7-b); no promotion to lower temperature was observed that would be expected for such an oxygen-rich coke.
The insoluble coke was analyzed by MALDI and LDI time-of-flight mass spectrometry. These techniques use a mild ionization and reflect directly the molecular distribution of the carbon deposits as the energy input of the laser is low.
LDI is particularly selective to oxidized products and was chosen for these ODH carbon deposits, as the matrix addition made the spectra interpretation more complex. The spectrum of the Alu500 derived insoluble coke shows a wide molecular
distribution of masses centered on ca. 1500 Da (Figure 8-a). The spectrum can be assigned to large platelets of polyaromatic hydrocarbons up to 4500 Da, as it shows long series with 24 Da of mass increase revealed upon high resolution magnification
(insets in Figure 8). This is due to the addition of –CH=CH– entity to build a new aromatic ring, with the abstraction of two hydrogen atoms from the main aromatic-polycondensed structure. The interpretation of this spectrum, is in agreement with
the Iwasawa-Ogasawara quinone-based active site model postulated for the EB ODH reaction: condensed aromatic rings with conjugated double bonds containing quinone/hydroxyquinone C=O functional groups [5,59-61]. However, the O groups
must be embedded in the lighter units (<500 m/z) as no mass increase involving O was observed. The spectrum of the Alu1050 (Figure 8-c) shows also a 24 Da C2H2 increase, in agreement with the enlargement of the condensed rings. The emission
intensity of a LDI spectrum gives a proof of the coke nature; the intensity is a measure of the coke reactivity (i.e. oxidation ability) as the emission is directly coupled to ionization. The comparison of the emission intensities in Figure 8
evidences the low reactivity of the Alu1000 and Alu1050 derived cokes as compared to the Alu500, in a factor of ca. 2.5 to 10.
a) b)
Chapter 2
46
Figure 8. LDI time-of-flight spectra of the thermally treated alumina: a) 500 (300 shoots, top); b) 1000 oC (400 shoots, middle) and c) 1050 oC (400 shoots, bottom).
The thermal stability of the carbon deposits is related to the mechanism of
total oxidation in EB ODH and can explain the selectivity trends; COx in EB ODH results from the combustion of the carbon deposits rather than from the direct combustion of ethylbenzene [15]. Looking at the oxidation rate patterns of the spent
catalysts (Figure 7-b), a trend exists with a shift of ca. 20 K towards higher temperatures in the maxima with increasing the calcination temperature of the alumina. This enhanced oxidation stability can explain the lower formation of COx
and, consequently the increase of the ST selectivity. The TPO shift in the carbon oxidation rate is consistent with the lower ionization intensity of the MALDI pattern for the optimal Alu1050. Hence, the increased oxidation stability indicates the lower
reactivity of the Alu1050-derived coke and can explain the selectivity increase to ST. The increased oxidation stability of the coke can be ultimately attributed to the acidity and textural changes upon thermal treatment. Alu1050 has higher density of
Lewis acid sites per surface area, it does not collapse texturally keeping 101 m2/g (compared to a non-silica promoted alumina calcined at 1050 oC with 11 m2/g); both effects are ascribed to the silica promotion. Si-Al Brønsted acid sites are not
involved, while the acid strength of the Lewis sites is relatively weaker (the “good” Lewis acidity) than the low-temperature Alu500. Other phenomenon that changes the surface chemistry of the alumina upon thermal treatment cannot be ruled out to
play a role too; for instance, the surface segregation of impurities upon thermal treatment [62].
2.5 Conclusions
Modification of a SiO2-stabilized -alumina by thermal treatment into a transitional
phase having wider pores and a high density of Lewis sites per surface area enhance the conversion and selectivity for the oxidative dehydrogenation of ethylbenzene to
styrene; the effect is observed when the O2 conversion is complete. The EB conversion increase comes from the ST selectivity enhancement that makes more O2 available for the main ODH reaction. The ST selectivity improvement is attributed to
the coke nature that is less reactive and consequently less prone to COx formation. The site requirements of the optimal catalyst to create the more selective coke seem to be related to the higher density of Lewis sites per surface area, no mixed Si-Al
1000 2000 3000 4000 50000
150
300
450
600
750
900
1050
1200
1350
1500
Absolu
te e
mis
sio
n in
tensity
mass (m/z)
cb
c
a
a
Making Coke a more Efficient Catalyst in the Oxidative Dehydrogenation of Ethylbenzene using wide-pore Transitional Aluminas
47
Brønsted sites are involved, while the acid strength of the Lewis sites is relatively weaker.
2.6 References
[1] Meyers RA. Handbook of petrochemicals production processes. New York NY:
McGraw-Hill. 2005: 11.3-11.34.
[2] Cavani F, Trifiro F. Alternative processes for the production of styrene. Appl
Catal A-Gen 1995;133(2):219-39.
[3] Lisovskii AE, Alkhazov TG, Dadasheva AM, Feizullaeva SA. Oxidative
dehydrogenation of alkylaromatic hydrocarbons on alumina catalysts. 4. Effect
of alkalis on acid properties of alumina and its catalytic activity in oxidative
dehydrogenation of ethyl benzene, Kinet Catal 1975;16(2):385-9.
[56] Daniell W, Schubert U, Glöckler R, Meyer A, Noweckb K, Knözinger H.
Enhanced surface acidity in mixed alumina-silicas: a low-temparature FTIR
study. Appl Catal A-Gen 2000;196(2):247-60.
[57] Carre S, Tapin B, Gnep NS, Revel R, Magnoux P. Model reactions as probe of
the acid-base properties of aluminas: nature and strength of active sites.
Correlation with physicochemical characterization. Appl Catal A-Gen
2010;372(1):26-33.
[58] Thommes M. Physical adsorption characterization of nanoporous materials.
Chem Ing Tech 2010;82(7):1059-73.
[59] Iwasawa Y, Ogasawara S. Control of selectivity and increase of catalytic
activity of polynaphthoquinone by various Lewis acids. J Catal
1975;37(1):148-57.
[60] Iwasawa Y, Fujitsu H, Onishi T, Tamaru K. Various reactions catalyzed by
electron donor-acceptor complex of polynaphthoquinone with potassium. J.
Chem. Soc. Faraday T 1 1974;70(2):202-7.
[61] Iwasawa Y, Mori H, Ogasawara S. Catalytic hydrogen transfer reactions
between hydroaromatics and nitrobenzene over polynaphthoquinone. J. Catal
1980;61(2):366-73.
[62] Yang BL, Lee SB, Cheng DS, Chang WS. Surface enrichment of impurities and
its effect on catalytic properties of iron oxide. J Catal 1993;141:161-70.
Making Coke a more Efficient Catalyst in the Oxidative Dehydrogenation of Ethylbenzene using wide-pore Transitional Aluminas
51
Chapter 2
52
Figure S-1. NH3 temperature programmed desorption (TPD) profiles: a) Alu series and b) comparison with a commercial Alumina (Merck 1.01095.1000) calcined at 1000 oC.
Figure S-2. Nitrogen sorption isotherms at -196 oC K for the fresh Alu series.
200 300 400 500 600 700 800 900 1000
0.000
0.004
0.008
0.012
Alu bare
Alu900
Alu1000
Alu1050
Alu1100
Alu1200
TC
D S
ign
al (a
.u.)
Temperature, °C
200 300 400 500 600 700 800 900 1000
0.000
0.004
0.008
0.012 Alu1000
gamma alumina Merk (1000)
TC
D S
ign
al (a
.u.)
Temperature, °C
a)
b)
0,0 0,2 0,4 0,6 0,8 1,0
0
100
200
300
400
500 Alu500
Alu600
Alu700
Alu800
Alu900
Alu1000
Alu1050
Alu1100
Alu1150
Alu1200
Qu
an
tity
Adsorb
ed (
cm
³/g
ST
P)
Relative Pressure (p/p°)
Making Coke a more Efficient Catalyst in the Oxidative Dehydrogenation of Ethylbenzene using wide-pore Transitional Aluminas
53
Table S-1. Example of the accuracy of the conversion and selectivity quantities
Alu500
Data points Conversion (%) Selectivity (%)
1 35.71 81.36
2 36.18 82.71
3 35.62 84.13
4 36.08 82.64
5 34.99 82.94
6 34.98 82.26
7 34.54 82.79
Average 35.44 82.69
(%) 1.6 0.9
Chapter 2
54
Tab
le S
-2.
Str
uctu
ral, t
extu
ral and a
cid
ic p
ropert
ies o
f th
e fre
sh t
herm
ally t
reate
d a
lum
inas.a
Sam
ple
P
hase
VT
(cm
3/
g)
VT’
(cm
3/
ml
bed
)b
SB
ET
(m
2/g
cat.
)
S’ B
ET
b
(m
2/
ml
bed
)
D B
JH
ad
s
(Å
)
Den
sit
yc
(g
/cm
3)
Acid
ity
(m
ol/
g)
Acid
ity
b
(m
ol/
ml
bed
)
Bare
e
0.6
39
(0.6
28)
0.3
89
(0.3
82)
272
(227)
165
(138)
84
3.0
35
637
388
Alu
500 e
0.6
49
(0.6
61)
0.4
31
(0.4
39)
271
(228)
180
(152)
86
3.1
08
Alu
600
0.6
44
0.4
46
255
177
93
3.0
84
Alu
700
0.6
35
0.4
02
239
151
101
3.0
03
Alu
800
0.6
36
0.4
36
214
147
118
3.0
99
Alu
900 e
0.6
08
(0.5
93)
0.4
73
(0.4
61)
179
(138)
139
(108)
138
3.2
95
540
420
Alu
1000
0.4
92
0.4
35
119
105
141
3.3
16
436
385
Alu
1050 e
0.4
58
(0.3
82)
0.3
27
(0.2
73)
101 (
60)
72 (
43)
230
3.3
76
398
284
Alu
1100
0.3
54
0.3
56
54
54
294
3.6
58
244
245
Alu
1150
0.1
65
0.1
57
20
19
330 d
4.0
09
Alu
1200
0.1
17
0.1
43
16
20
293 d
4.0
67
20
25
C1050 f
0.0
51
11
185 d
a)
N2 (
-196 o
C)
isoth
erm
s a
re g
iven in F
igure
S-2
and N
H3-T
PD
in F
ig.
S-1
. b)
Quantity
per
reacto
r volu
me.
c)
Skele
tal density.
d)
Geom
etr
ical pore
siz
e a
s t
here
is n
o m
axim
um
in t
he B
JH p
ore
siz
e d
istr
ibution.
e)
Valu
es b
etw
een b
rackets
are
deri
ved fro
m A
rgon p
hysis
orp
tion a
t -1
86 o
C.
f)
Ultra
pure
alu
min
a (
com
merc
ial:
Merc
k 1
.01095.1
000)
therm
ally t
reate
d a
t 1050 o
C.
On the stability of conventional and nanostructured carbon-based catalysts in the oxidative dehydrogenation of ethylbenzene under industrially relevant conditions
and nanostructured multi-walled carbon nanotubes (MWCNT) were tested in the oxidative
dehydrogenation of ethylbenzene (EB). Special attention was given to the reaction conditions,
using a relatively concentrated EB feed (10 vol. % EB), and limited excess of O2 (O2:EB=0.6) in
order to work at full oxygen conversion and consequently avoid O2 in the downstream processing
and recycle streams. The temperature was varied between 425-475 oC, that is about 150-200 oC
lower than that of the commercial steam dehydrogenation process. The stability was evaluated
from runs of 60 h time on stream. Under the applied reactions conditions, all the carbon-based
materials are apparently stable in the first 15 h time on stream. The effect of the
gasification/burning was significantly visible only after this period where most of them fully
decompose. The carbon of the hybrids decomposes completely rendering the silica matrix and the
activated carbon bed is fully consumed. Nano structured MWCNT is the most stable; the structure
resists the demanding reaction conditions showing an EB conversion of 30% (but deactivating)
with a steady selectivity of 80%. The catalyst stability under the ODH reaction conditions is
predicted from the combustion apparent activation energies.
Chapter 3
56
3.1 Introduction
Styrene (ST) is one of the major platform chemicals. It is the monomer for
polystyrene formulations and an essential additive for tires, and many more small-scale uses. The global production of styrene in 2010 was 25 million metric tons and it is forecast to grow at an average of 3.6% per year [1]. It is industrially produced for
about 85 % by steam dehydrogenation (also called direct dehydrogenation) of ethylbenzene (EB) in an excess of steam over a K-promoted Fe2O3 catalyst [2,3], eq. 1.
EB ⇄ ST + H2 (1)
The catalyst is highly selective to styrene, typically >97%, and stable [2]. One
of the aspects for improvement is reducing the huge amount of steam needed to carry out the dehydrogenation chemistry, which is an endothermic process. Thermodynamic limitation of direct dehydrogenation is also a reason why new routes
for the styrene production are still investigated. The use of oxidants, such as O2 [4, 5], is very promising because the reaction is shifted by producing H2O, eq. 2. The equilibrium dehydrogenation boundary is broken (with O2) and this allows operating
at lower temperature as long as an active, selective, and stable catalyst is available.
EB + ½ O2 → ST + H2O (2)
This oxidative pathway was initially studied over inorganic supports, such as
alumina [6]. It was demonstrated that the coke deposited on the oxide‟s surface having acid Lewis sites, generated during the early stage of the reaction is the actual “coke” catalyst [6-9]; this is supported by a recent study [10]. The coke was
characterized as polyaromatic having a relatively high oxygen content (O:C=0.10-0.15 at.); the proposed active sites are surface ketonic groups that act as redox sites for hydrogen abstraction from EB [11-14].
Carbon based materials have been investigated broadly on this reaction, trying to reduce or eliminate the „activation period‟ (i.e. formation of the active/selective carbonaceous species) that occurs over the oxides. Activated
carbons [15-29], carbon nanofibers [30-37], onion-like carbons [38-40], diamonds [26,28,40], nanofilaments [30], graphites [17,28,30,34], multiwall carbon nanotubes (MWCNTs) [26,28,36,39-49], and other type of carbon materials or
mixtures of the above mentioned [50-54] have been studied for this reaction. It is generally found that these materials are readily active and selective; the reported selectivities are moderate lying between 55-85 %. In some cases the reported
selectivity is exceptionally high, in the range of 90-97% [28,46,47]. Under the reported reaction conditions and time frame most of the carbon-based materials are stable with the exception of the activated carbons that are steadily decomposed
[20,21,26,43]; the rate of gasification/burning is faster than that of the coke build-up. Some of the most stable systems are the carbon nanotubes and ordered mesoporous carbons, though they show a pronounced initial deactivation in 5 h;
from 35 to 20 mmol.g-1h-1[26] while Su et al. reported a decay from 90 to 70% EB conversion in a time frame of 5 h as well [51]. A similar initial deactivation was observed for furfuryl alcohol-based CMK-3 type carbons [54].
Comparing the reaction performance and catalyst stability among the reported studies is difficult, as the reaction conditions vary enormously. In addition to that,
the used EB concentration is mostly relatively low, ranging 2-4 vol. % and the O2:EB ratio is far above the stoichiometric, 1-5. From a practical stand point, it is desirable to operate at higher EB concentration, for higher product yield, and at complete O2
conversion. The latter aspect is very important when looking at the overall process in order to avoid O2 in the downstream processing and recycle streams. The thermal
On the stability of conventional and nano-structured carbon-based catalysts in the oxidative dehydrogenation of ethylbenzene under industrially relevant conditions.
57
stability is typically studied at low reaction temperatures, where the rate of gasification and combustion are limited, and likely not seen in the reported cycle. It should be noted that for a catalytic fixed bed operation a temperature rise of more
than 100 °C will occur with only 25 % EB conversion with 100 % selectivity towards ST. Hence, it is essential to evaluate the catalyst stability at higher temperatures.
The objective of this study is to investigate the catalytic performance and
stability of various types of carbon-based materials (tailor made with controlled carbon loading, nature and commercial ones) in the oxidative dehydrogenation of EB under identical and industrially relevant conditions; using a concentrated EB feed of
10 vol.%, a limited excess of O2 (20 % excess O2, equivalent to O2:EB=0.6) in order to have full oxygen conversion. The proposed temperature range is between 425-475 oC, which is nearly 150-200 oC lower than that of the conventional steam
dehydrogenation process. The high initial reaction temperature at 475 oC intends to assess the gasification and combustion under demanding conditions to see changes in a reasonable time frame (<60 h). However, this range also mimics what would
happen if the reactor suffers an undesired temperature increase due to a moderate hot-spot (typically carbons are evaluated in EB ODH at 400-425 oC) or a morderate adiabatic temperature increase over a fixed catalyst bed reactor operation. Physico-
chemical characterization of the fresh and spent catalysts was carried out to understand the catalyst life behaviour; the apparent activation energies of the
combustion were particularly useful to rationalize the results.
3.2 Experimental
3.2.1 Materials
Commercial amorphous SiO2 (denoted as AS, entry 1 in Table 1) used for the hybrid catalysts preparation was supplied by Saint-Gobain NorPro (SS61138). Activated carbons (AC, entry 2 in Table 1) were supplied by Norit (ROX 0.8), and purified
MWCNT (C>96 wt.%) was obtained from Hyperion (CS-02C-063-XD). All materials were supplied in pellets, which were crushed and sieved in the fraction 212-425 m.
Furfurylalcohol (Acros Organics, 98%), D(+)-glucose (anhydrous for biochemistry,
Merck 1.08337.1000), H2SO4 (1.00731.1000, Merck, 95-97% for analysis), and oxalic acid (Acros Organics, >99%) were employed without further purification for the preparation of the carbon-silica hybrids.
Chapter 3
58
Table 1. Summary of sample codes and their preparative conditions.
Entry Sample code
Method of preparation
1 AS Commercial amorphous silica, Saint-Gobain Norpro SS61138. 2 AC Commercial activated carbon, Norit ROX 0.8.
5 AS-900 AS material calcined at 900 oC. 6 AS-1100 AS material calcined at 1100 oC. 7 G/AS700 Polymerization of glucose and pyrolysis at 700 oC over AS
support. 8 G/AS900 Polymerization of glucose and pyrolysis at 900 oC over AS
support.
9 FA/AS700 Polycondensation of furfuryl alcohol and pyrolysis at 700 oC over AS support.
10 FA/AS900 Polycondensation of furfuryl alcohol and pyrolysis at 900 oC
over AS support. 11 FA/AS1100 Polycondensation of furfuryl alcohol and pyrolysis at 1100 oC
over AS support.
For the impregnations, the liquid pore volume (VLPV) of the bare silica was
used [55], which was experimentally determined (VLPV=1.05 cm3.g-1) as follows: water was added to ca. 10 g material that is repeatedly shaken after each water addition until the material turned shiny. The VLPV (1.05 cm3.g-1) is larger than the gas
adsorption VT value (0.842 cm3.g-1, AS in Table 2) because water fills pores larger than 100 nm, which is the upper limit in gas adsorption.
Table 2. Textural parameters of the AS modified materials derived from N2 adsorption at 196 oC.
a. Values in parenthesis are the average geometrical pore size calculated by 4.103VT/SBET.
3.2.2 Thermal stability of the amorphous silica support
The thermal stability of the AS support was investigated by analysing the textural and structural properties of the material that was subjected to heat treatment at
700, 900, and 1100 oC under dried nitrogen. These treatments were carried out in LT9/11 Nabertherm box furnace. The obtained samples correspond to entries 4-6 in Table 1.
3.2.3 Preparation of hybrid-silica hybrids
3.2.3.1 Furfuryl alcohol based SiO2 hybrid materials
The carbon was added by incipient wetness impregnation of the carbon precursor
using 5% extra liquid volume regarding the liquid pore volume, VLPV. Furfurylalcohol
On the stability of conventional and nano-structured carbon-based catalysts in the oxidative dehydrogenation of ethylbenzene under industrially relevant conditions.
59
(FA) was used as carbon source and oxalic acid (OA) as polymerization catalyst. Three grams of support was degassed under vacuum at 150 oC during 4 h. An aqueous solution containing the carbon precursor was prepared by mixing FA and OA
in a molar ratio of 250 based on previous studies [56]. Then 3.15 mL of the FA/OA aqueous solution was added to the support to obtain a carbon loading of 15 wt.%, in the final pyrolyzed material, assuming that only carbon is present in the final
material. The wet material was shaken (VWR, digital DVX-2500) during 4 min at 2500 rpm at room temperature to distribute the solution evenly. Afterwards, the FA polymerization was induced by heating the sample at 160 oC in an atmospheric oven
during 8 h. The samples prepared under this protocol are summarized in Table 1, entries 9-11.
3.2.3.2 Glucose based carbon-SiO2 hybrids
Glucose (G) was also investigated as carbon source and H2SO4 as a catalyst in molar ratio of G/H2SO4=0.5 [57] for the hybrid catalysts preparation. The protocol is
identical to above mentioned for FA, using a similar carbon loading in the final hybrid, ca. 15 wt.%. The samples prepared under this protocol are summarized in
Table 1, entries 7 and 8.
3.2.4 Carbonization
The carbonization of the organic precursor was carried out by pyrolysis in a quartz-tube housed tubular oven (Nabertherm RT 50/250-11). The sample was loaded in a flat quartz crucible that was placed horizontally in the centre of the furnace heating
zone. After closing and purging the tube for 30 min with a flow of dry N2 (150 mL.min-1 NTP), the sample was heated at 1 °C.min-1 from room temperature until 700, 900, 1100 oC and kept for 3 h; the sample was kept under N2 flow during the
cooling down until room temperature.
3.2.5 Catalysts characterization
The organic content of the fresh and spent catalysts was quantified by thermogravimetric analysis (TGA) on a Mettler-Toledo analyzer (TGA/SDTA851e) using a flow of synthetic air of 100 mL.min-1 STP. The temperature was increased
from 30 to 900 °C at 10 °C.min-1. Blank curve subtraction using an empty crucible was taken into account. The oxidation rate patterns (TPO) were obtained in the same instrument using the derivative of the TGA patterns.
Nitrogen physisorption analyses (196.2 oC) were carried out in a
Micromeritics ASAP 2020. The samples were degassed in vacuum at 200 oC for 10 h. The surface area was calculated using the standard BET method (SBET) [58]. The
single point gas adsorption pore volume (VT) was calculated from the amount of gas adsorbed at a relative pressure of 0.98 in the desorption branch. The pore size distributions (PSD) were obtained from the BJH method [59] using the adsorption
branch of the isotherms. The mean pore size (BJH) is given by the position of the
PSD maximum. The t-plot method [60] was employed to quantify the micropore volume (Vm).
High resolution transmission electron microscopy (HR-TEM) images were acquired using a TEM/STEM JEOL 2100F operated at 200 kV. Samples were prepared by grinding, dispersion in ethanol and evaporative deposition onto holey carbon films
on 300 mesh Cu grids.
Chapter 3
60
Powder X-ray diffraction (XRD) measurements were done on a Bruker D8
powder X-ray diffractometer using CuK radiation, =0.154056 nm. The spectra
were recorded with a step size of 0.02° and 3 seconds of accumulation time, in the 2θ angle range of 10-60°.
Raman spectra were obtained with 785 nm excitation line, 30 mW on the Perkin Elmer Ramanstation 400 spectrometer.
The apparent activation energy of the carbon combustion was calculated by
Ozawa method [61] using a correlation between peak temperature for a given conversion and the heating rate for four thermal analysis derivative curves (DTA).
The DTA curves were obtained on a Mettler-Toledo analyzer (TGA/SDTA851e) using
a flow of synthetic air of 100 mL.min-1 NTP. The temperature was increased from 30 to 900 °C with a heating rate of 1, 3, 5, 10 °C.min-1. Blank curve subtraction using an empty crucible was taken into account. The apparent activation energy was
calculated for 20 and 80% conversion levels.
3.2.6 Catalyst performance protocol
The catalytic tests were done in a 6-flow micro reactor using a fixed volume of catalyst (0.8 mL, corresponding to 65 mm of bed length) in 4 mm quartz reactors in down-flow mode. To guarantee that the catalyst bed is located in the isothermal
zone of the furnace, the reactors were loaded with quartz wool, 10 cm glass pearls (0.5 mm diameter), the catalyst, 10 cm glass pearls (0.5 mm diameter), and a second quartz wool plug. The glass beads have limited conversion; less than 3% EB
conversion under all applied conditions. The reactor gas feed is a mixture that can consist of CO2, N2, and air that counts for a gas-flow rate of 36 ml (NTP).min-1; the liquid EB-feed flow rate is 1 g.h-1 (3.54 mL STP.min-1 vapor) that is evaporated
upstream each reactor in a Al2O3 column, resulting in a 1:9 molar ratio of
ethylbenzene to gas (10 vol. % EB). This corresponds with operation at a GHSV of 3000 L/L/h. The total pressure was 1.2-1.3∙105 Pa. The reactor exhaust gas was
analyzed by gas chromatography using a combination of columns (0.3m Hayesep Q 80-100 mesh with back-flush, 25m×0.53mm Porabond Q, 15 m × 0.53mm molsieve 5A and RTX-1 with 30m×0.53mm) and TCD and FID detectors. This configuration
allows quantifying permanent gases such as CO2, H2, N2, O2, CO as well as hydrocarbons (methane, ethane, ethene, benzene, toluene, ethylbenzene, styrene, and heavy aromatics). The catalytic test was carried out at various temperatures
(475, 450, 425, and 450 oC) and O2/EB = 0.6 and 0.2 (vol.) to access the effect of temperature and O2:EB on conversion and selectivity. It is noted that for low EB conversion (<15%) there was a considerable error in the ST/COx selectivities leading
to substantial noise. For the sake of clarity, some graphs are, therefore, plotted until 30 h TOS. For all EB conversion data the oxygen conversion is 100%, unless otherwise is stated. All physical characterizations for the spent catalysts were done
after the complete testing cycle of 60 h. The applied temperature program, starting at 475 oC for 15 h, was chosen to accelerate the gasification/combustion processes under the applied reaction conditions.
3.3 Results and discussion
3.3.1 Thermal stability of the bare silica support
The support used in this study is commercial precipitated silica, whose thermal
stability was studied by gas adsorption and XRD analysis of the thermally treated materials between 700-1100 oC. Characterization by gas adsorption of the bare silica shows well-defined mesopores centered at 19.1 nm (AS in Fig. 1). The isotherm is
type IV with hysteresis H1 [62,63]; representing solids with cylindrical pore
On the stability of conventional and nano-structured carbon-based catalysts in the oxidative dehydrogenation of ethylbenzene under industrially relevant conditions.
61
geometry with relatively high pore size uniformity and facile pore connectivity. This is consistent with the TEM micrographs that show pores defined by the interspace of nanoparticles ranging 5-30 nm (Fig. 2). The average particle size was calculated
assuming the particles to be spherical using the SBET and the skeletal density for the silica (2.3103 kg.m-3, experimentally determined by helium picnometry). The particle
size was found to be around 12 nm on average which is consistent with TEM
micrographs (Fig. 2-b). XRD analysis evidences that the bare AS silica is structurally formed by amorphous particles (AS in Fig. 3).
Figure 1. Nitrogen sorption isotherms at 196 oC for the bare amorphous silica (AS) and
calcined AS-700, AS-900 and AS-1100. Inset: BJH pore size distributions.
0.0 0.2 0.4 0.6 0.8 1.0
0
100
200
300
400
500
600
20 40 60 80 100
17.8
20.8 AS
AS-700
AS-900
AS-1100
Pore width / nm
AS
AS-700
AS-900
AS-1100
Volu
me a
dsorb
ed /
cm
³/g
ST
P
Relative Pressure (p/p°)
Chapter 3
62
Figure 2. a) TEM micrograph and b) HR-TEM micrograph of the bare silica support indicating the dimension of some particles. Scale bar = 20 nm
Figure 3. Powder XRD patterns: bare silica, AS-700, AS-900, and AS-1100. Phase identification: C (cristobalite) and T (tridymite) [64].
At 700 oC the silica remains almost unchanged; the XRD reflection at 22o is
still broad (AS-700 in Fig. 3) and the N2 isotherm is almost identical to the fresh AS material (Fig. 1). Consequently the surface area, pore volume, and pore size remains almost identical (Table 2), with differences in the physical values within 5% higher
values. The higher values are attributed to the cleaning of the sample surface upon thermal treatment. When the temperature is increased to 900 oC, the material suffers from a sintering process. Although the XRD reflection is rather similar (AS-
a
b
T T
C
C C
C C
C C
C
C
C
On the stability of conventional and nano-structured carbon-based catalysts in the oxidative dehydrogenation of ethylbenzene under industrially relevant conditions.
63
900 in Fig. 3), the material adsorbs significantly less N2 gas (Fig. 1); the surface area and pore volume are reduced to ca. half of the initial values (Table 2). The pore size is also diminished with the pore maxima shifted to 17.8 nm while the average
geometrical pore size is reduced with 11 %. These are all indications for thermal sintering. Severe changes occurred at 1100 oC, where the amorphous silica is transformed into a mixture of cristobalite and tridymate polymorphs [64] (AS-1100
in Fig. 3) and both those transformations have almost no surface area nor pore volume (Table 2).
The textural properties of thermally treated silicas were compared to the
furfuryl alcohol-silica hybrids that after impregnating the carbon source, and acid catalyzed polymerization, were pyrolyzed between 700-1100 oC; denoted as FA/AS700, FA/AS900 and FA/AS1100 (see Table 1). The textural comparison can be
found in Figure 4. The three hybrids show typical isotherms for a micro- and mesoporous material, type IV with hysteresis H1. Comparison of the isotherms for FA/AS700 and the AS-700 counterpart (Fig. 4-a) points out that the hybrid adsorbs
more gas in the low range of the isotherm (0-0.3 p/po), and, consequently, it has a higher surface area (cf. values reported in Table 2 and 3). This is attributed to the enhanced micropore volume, that is nearly 3 times higher than that of the silica AS-
700 (cf. values given in Table S-1 for AS-700 and FA/AS700), while the total pore volume of the hybrid is inferior (Table 3).
For FA/AS900, FA/AS1100 and corresponding silica counterparts, the effect is
very different. The isotherms of the silicas show inferior adsorption quantities, which is reflected in much lower surface areas than those of the hybrids; 281 versus 110
m2.g-1 and 215 versus 2 m2.g-1. The same holds for the pore volume. This is an indication of the enhanced stability against sintering of the silica in the hybrids.
These observations undoubtedly indicate that the thermal stability of the pure
silica is very different from silica where a carbon coating is present; the carbon acts as a protective means against sintering. From this comparative study, it is concluded that hybrids can be prepared by pyrolysis up to 1100 oC retaining the mesoscopic
texture. Therefore, 3 temperatures were considered for pyrolysis, ranging 700-1100 oC.
Table 3. Summary of characterization data derived from N2 physisorption, TGA, and Raman for the fresh and spent catalysts.
Sample SBET
m2g-1 VT
cm3g-1
SBET
m2g-1
VT
cm3g-1
Carbon a wt. %
ID/IG ratio
fresh spent fresh spent fresh
AC 959 0.609 FDb FDb 90.0 FDb 1.4
MWCNT 406 1.100 110 0.291 96.2 96.1 1.8
G/AS700 287 0.634 221 0.841 14.5 1.4c 1.3
G/AS900 286 0.630 224 0.831 14.9 2.0c 1.9
FA/AS700 297 0.640 222 0.853 19.1 2.3c 1.6
FA/AS900 281 0.597 223 0.835 20.3 2.4c 1.5
FA/AS1100 215 0.549 207 0.744 17.0 2.3c 1.5
AS 213 0.842 NMd NMd 1.6c 0.9c NAe
a. Determined by TGA (TGA patterns are compiled in Figs. S-1 to S-8); b. FD = fully decomposed
during the catalytic run; c. Weight loss due to silica dehydroxylation; d. NM = not measured (as it
shows very low activity); e. NA = not applicable.
Chapter 3
64
Figure 4. Nitrogen sorption isotherms at -196 oC for hybrid FA/AS materials (in red) including
the corresponding calcined support (in black): a) 700 oC; b) 900 oC, and c) 1100 oC. Inserts
corresponds to the BJH pore size distributions.
a
b
c
0.0 0.2 0.4 0.6 0.8 1.00
100
200
300
400
500
600
20 40 60 80 100
22.1
20.8
Pore width / nm
AS-700
FA/AS700
Volu
me a
dsorb
ed /
cm
³/g
ST
P
Relative Pressure (p/p°)
0.0 0.2 0.4 0.6 0.8 1.00
100
200
300
400
500
600
20 40 60 80 100
21.6
17.8
Pore width / nm
AS-900
FA/AS900
Volu
me a
dsorb
ed /
cm
³/g
ST
P
Relative Pressure (p/p°)
0.0 0.2 0.4 0.6 0.8 1.0
0
100
200
300
400
500
600
20 40 60 80 100
21.0
Pore width / nm
AS-1100
FA/AS1100
Vo
lum
e a
dso
rbe
d /
cm
³/g
ST
P
Relative Pressure (p/p°)
On the stability of conventional and nano-structured carbon-based catalysts in the oxidative dehydrogenation of ethylbenzene under industrially relevant conditions.
65
3.3.2 Structural and textural properties of the C/SiO2 hybrids, MWCNT, and AC materials
The carbon content of the hybrid materials was determined by TGA (Figs. S-3 to S-7, values in Table 3). The TGA patterns indicate that the total amount of organic content is between 14.5 and 20.3 wt.%, that is close to the nominal value of 15
wt.% (on carbon basis). The position of the maximum decomposition rate ranged between 575 to 625 oC, which shifts to higher temperatures with the applied pyrolysis temperature. MWCNT are almost purely formed by a carbon material with
an inorganic residue of 3.8 wt. % (Fig. S-2, Table 3). The activated carbon (AC) shows a relatively high residue of 10 wt. % (Fig. S-1, Table 3), attributed to the binder. The maximum of the decomposition temperature of AC and MWCNT materials
is relatively high, located at around 625 oC, at the applied heating rate of 10 °C/min. The nature of the carbon species of the fresh materials was evaluated by
Raman spectroscopy. The Raman patterns in Figure 5 reveal two broad absorptions
centred at 1600 and 1310 cm-1, which are characteristic of amorphous carbon (G and D bands, respectively) [65,66]. Quantification of the intensity ratio (ID/IG), Table 3, reveals that the defective type of carbon is dominant with ratios ranging 1.3 to 1.9.
The isotherms of the furfuryl alcohol-based C/SiO2 hybrids were preliminary discussed in section 3.1 (Fig. 4); there is a reduction of the surface area with the
pyrolysis temperature (297, 281 to 215 m2.g-1); similarly the pore volume is reduced. Even at 1100 oC the hybrids contain a mesoscopic texture. A substantial fraction of micropores was detected (Table S-1) that explains the higher surface area
of the FA/AS700 compared to AS-700, and in general that FA/AS700> FA/AS900>FA/AS1100>AS in terms of surface area. The glucose-based C/SiO2 hybrids have a similar texture, with an isotherm shape type IV with hysteresis H1,
Fig. S-9. There is also a substantial fraction of micropores (Table S-1). For glucose, the pyrolysis temperature was screened at 700 and 900 oC (1100 oC pyrolysis temperature was considered to be of no interest for the glucose-based hybrids, as
will be discussed later); no significant differences in the texture were noticed among these two pyrolysis temperatures. The presence of micropores is characteristic in all hybrid materials, while the thermally treated bare silica has negligible micropore
volume (<0.006 cm3.g-1, in Table S-1). Therefore, the micropore volume was ascribed to be located in the carbon deposited layer; the micropores explain the higher surface areas of the hybrids compared to the bare silicas. All hybrid materials
have a relatively narrow pore size distribution with the peak maxima located between 17.8 and 21.0 nm.
Chapter 3
66
Figure 5. Raman spectra for the fresh carbon-based materials of the carbon-
based materials under this study.
The texture of the structured MWCNT shows an isotherm type IV with
hysteresis H1 as well (Fig. S-10). The texture is purely mesoporous as the micropore volume can be negligible (Table S-1). The pore size distribution (Fig. S-10, inset) has a main contribution of pores centered at 31.5 nm that is substantially higher than
the hybrids; the high surface are is explained by the presence of small mesopores (3.0 nm) that are located in the inner space of opened tubes that contribute substantially to the surface area. The total pore volume is remarkably high with 1.1
cm3.g-1 compared to all other materials; this can also be deduced from the comparative over layer in Fig. S-11.
The texture of the activated carbon (Fig. S-10) is typical of a micro- and
mesoporous material; isotherm type I with mesoporous hysteresis H2, displaying pore restrictions. The high surface area (959 m2.g-1) comes from the relatively high fraction of micropores that represents around 50% of the total pore volume (cf.
values in Table 3 and S-1), while the pore volume (0.609 cm3.g-1) is comparable to the hybrids. These trends in surface area and pore volume can be deduced from Fig. S-11.
The TPO patterns (Fig. 6) can provide information about the stability under the harsh oxidative conditions of this reaction, eq. 2. From Fig. 6 it can be deduced that the thermal oxidative stability for the C/SiO2 hybrids depends on the applied
carbonization temperature and carbon source used. With a higher applied pyrolysis temperature, the maxima of the oxidation
pattern shift to higher temperatures, 582 up to 627 oC (FA samples) and 570 up to
613 oC (G samples). This is reasonable since the carbon becomes more graphitic with the increasing pyrolysis temperature, though this is not directly seen by Raman due
to the heterogeneity of the materials displaying broad bands. All hybrids materials start to burn at higher temperatures than those of AC and MWCNT, at ca. 450-475 oC, while MWCNT and AC begin below 450 oC. From the TPO maxima, it seems that
all carbon material could be stable, at least for short reaction times, under the oxidative conditions since the highest applied reaction temperature is 475 oC. The effect of time-on-stream (TOS) stability will be further discussed in the next section.
2000 1800 1600 1400 1200 1000 800
AC
MWCNT
G/AS700
G/AS900
FA/AS700
FA/AS900
FA/AS1100
Inte
nsity /
a.u
.
Wavenumber / cm-1
On the stability of conventional and nano-structured carbon-based catalysts in the oxidative dehydrogenation of ethylbenzene under industrially relevant conditions.
67
3.3.3 Catalytic performance, characterization of the spent catalysts and
oxidation stability
The ODH catalyst performance is given in Figure 7, where the EB conversion, ST and
COx selectivity as well as the overall ST yield are represented as a function of the time on stream. All the carbon-based materials are readily active and selective. They are all stable up to 15 h TOS. In this period, the hybrid catalysts are more active
than AC and MWCNT. Among the hybrids the following trend in EB conversion was found: FA/AS700<G/AS900<FA/AS1100<FA/AS900<G/AS700. The selectivity of the
hybrids varied considerably; the glucose-based hybrid renders very selective catalysts, higher than MWCNT and AC. The FA-based hybrids yield low ST selective catalysts; where the best values were found for FA/AS700, with selectivities in the
range of the AC.
Figure 6. Oxidation rate patterns (TPO) for the carbon-based materials under
this study. = (WoW)/Wo; where Wo is the initial weight. Conditions:
synthetic air, 100 ml.min-1, heating rate of 10 °C. min-1.
450 500 550 600 650 7000.0
0.2
0.4
450 500 550 600 650 7000.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
G/AS700
G/AS900
FA/AS700
FA/AS900
FA/AS1100
Temperature / oC
- d
/dT
/
(%
. oC
-1)
AC
MWCNT
Temperature / oC
- d/d
T
/
(
%.
oC
-1)
Chapter 3
68
The selectivity to COx is inversely coupled to ST (Fig. 7-c). This is because the selectivity to benzene/toluene and heavy condensates is much lower than COx/ST
and independent of the reaction conditions applied. It is noted that the selectivity values suffer from accuracy for low conversion; for the AS catalyst, the selectivity to ST and COx summed up 80%, for EB conversions <5%, and no other by-products
were detected (besides ~2% benzene and toluene). After 20 h TOS the effect of the catalyst stability, against gasification/burning
under the reaction conditions, becomes visible. The conversion and selectivity of all hybrid material severely decline until reaching a marginal conversion. For the glucose hybrids and FA/AS700 the EB conversion level down to the value found for the bare
amorphous silica (AS in Fig. 7-a). The materials FA/AS900 and FA/AS1100 are relatively more stable with a steady EB conversion slightly above 10% after 60 h TOS. The selectivity also drops after 20 h TOS (Fig. 7-b) for all hybrids. It was
observed that for low EB conversions (<15 %) the error is significant and the ST and
Figure 7. Time on stream EB conversion (a), selectivity to ST (b), selectivity to COx (c), ST yield,
and (d) at various temperatures (475, 450, 425, and 450 oC) and O2/EB= 0.6 and 0.2 (vol.); GHSV
of 3000 l/l/h; 10 vol. % EB. The top headings correspond to the O2:EB (vol.) ratio and reaction
temperature in oC. It is noted that for low EB conversion (<15%) there were considerable errors in
the ST/COx selectivities leading to substantial noise; for the sake of clarity, some graphs are,
therefore, plotted until 30 h TOS.
0 10 20 30 40 50 600
10
20
30
40
50
60
70
80
90
100
AC FA/AS700
MWCNT FA/AS900
G/AS700 FA/AS1100
G/AS900 AS
Sele
ctivity t
o S
T /
%
Time / h0 10 20 30 40 50 60
0
5
10
15
20
25
30
35
40
45
50
55 AC FA/AS700
MWCNT FA/AS900
G/AS700 FA/AS1100
G/AS900 AS
Convers
ion / %
Time / h
425 450450
0.6 0.2 0.6 0.60.20.6 0.2
475
0 10 20 30 40 50 600
10
20
30
40
50
60
70
80
90
100 AC FA/AS700
MWCNT FA/AS900
G/AS700 FA/AS1100
G/AS900 AS
Se
lectivity t
o C
Ox /
%
Time / h
0 10 20 30 40 50 600
10
20
30
40
50
60 AC FA/AS700
MWCNT FA/AS900
G/AS700 FA/AS1100
G/AS900 AS
S
T Y
ield
/ %
Time / h
a b
dc
0.6 0.2 0.60.20.6 0.2
425 450475 450450
0.6 0.2 0.60.20.6
475
0.6
450
0.6 0.2 0.60.20.6 0.2
425 450475
0.6
450
0.2
425
0.6
On the stability of conventional and nano-structured carbon-based catalysts in the oxidative dehydrogenation of ethylbenzene under industrially relevant conditions.
69
COx selectivity scatter severely; because of that and for the sake of clarity, some graphs are only plotted until 30 h TOS.
Characterization of the spent hybrid catalysts after 60 TOS by TGA (Fig. S-3 to
S-7; weight loss values are given in Table 3) shows that there is no carbon phase on the spent materials after the catalytic run. The materials became fully white and the weight loss detected by TGA (1.4-2.4 wt. %, 200-900 oC, Table 3) is attributed to
surface Si-OH dehydroxylation. The texture of the spent samples is consistent with the carbon removal by gasification/burning under reaction. The textural parameters of all the spent hybrids are very similar, 221-224 m2.g-1, and comparable to the bare
AS silica, 213 m2.g-1. Note that the porosity of the spent hybrids cannot be directly compared to the thermally treated AS materials (Table 2), because it was shown that the carbon stabilizes the silica against sintering. Performance and characterization
clearly indicate that under limited O2 (O2:EB=0.6) the effect of the gasification/burning is only visible after 20 h TOS.
The activated carbon (AC) is readily active and selective, with moderate EB
conversion (27-30%) and ST selectivity (70-80%). This material is also gasified/burnt under reaction. The conversion goes to zero after 60 h TOS, though the selectivity remains high. Inspection of the catalyst bed after the run showed that
the catalyst was completely consumed. Nanostructured MWCNT starts with a relatively low EB conversion (35 %) and
moderate selectivity (80 %); it is remarkably stable during the complete run. After
55 h TOS the conversion at 450 oC remained at ca. 33 %, the ST selectivity at 80 % and almost 20 % COx selectivity. The high selectivity to COx is a major handicap because it is waste product; reducing this parameter by making the catalyst more ST
selective, or other useful by-products, such as benzene, toluene, can make MWCNT potentially interesting for EB ODH. The pronounced initial deactivation reported for nanotubes and ordered CMK-3 [26,51,54] was not observed for this Hyperion
MWCNT, even starting at 475 oC. The textural properties of the spent catalyst are considerably reduced, and this is attributed to coke deposition or MWCNT modification, or both. The TGA derivative on the spent catalyst (Fig. S-2) shows two
oxidation steps, one peak located at ca. 625 oC due to the oxidation of the original MWCNT and one centred at 550 oC associated to the oxygen-containing polyaromatic coke coming from the ODH reaction, or oxidized original MWCNT; both can be
considered as „reaction coke‟. Roughly 60% of the coke is associated to the reaction coke. It was found that the catalyst bed after the run was reduced in 20% in volume. The selectivity remains very stable while the EB conversion decreased steadily as a
function of TOS; consequently the ST yield decreased steadily as well (Fig. 7-d). It was tried to correlate the stability for all the catalysts with the TPO
patterns. Neither the light-off temperature nor the peak maxima agrees with the
observed reaction stability. The most stable catalyst under the reaction conditions, MWCNT, begins its oxidation at 450 oC, which is the lowest temperature in the series of materials. On the other hand, the temperature of the TPO maxima for AC and
MWCNT nearly coincide, while the AC decomposes under the reaction conditions and MWCNT only partially. A plausible explanation could be that the carbon materials are formed by a heterogeneous mixture of amorphous and more crystalline domains. For
instance, it is known that MWCNT contains amorphous domains at the outer surface or filling the inner cavity of the tubes; these amorphous domains burn at lower temperature. Such a lower stability has been used to purify nanotubes; air
calcination purifies double-walled carbon nanotubes by burning the amorphous part [67]. Therefore, the TPO interpretation is troublesome since the amorphous domains are characteristic of the lower-temperature decomposition steps and can contribute
to the peak maxima as well. In the case of the investigated MWCNT, it is reasonable to assume that the amorphous domains burn under the reaction conditions and can
be attributed to the observed 20% volume reduction of the catalyst bed. The same
Chapter 3
70
sample heterogeneity holds for the activated carbon and hybrid materials, based on
the broad Raman spectra. The stability was assessed from the apparent activation energies (Ea) of the
air-assisted combustion process. This is because the apparent activation energy is
related to the rate determining step, which by definition comes from the most stable domains of the carbon that resists more to the ODH conditions for a long run (i.e. in this case TOS>50h). Thus, this approach excludes the easily burnable coke
associated to the light-off temperatures; the latter, a parameter that did not correlate to the observed catalyst life time, as discussed in the previous section.
The Ea values were plotted for the different systems at 20 and 80%
conversion level (Fig. 8). The Ea increased with the pyrolysis temperature for the FA-based samples from 102(116) to 123(125) kJ.mol-1, associated to the formation of
more graphitic and stable carbon. The highest values were found for the AC and MWCNTs with 128(132) and 133(144) kJ.mol-1; the values for the MWCNT are consistent with Frank et al., reporting an Ea of 138 kJ.mol-1 for unmodified CNTs
[68]. The major differences in Ea were found at 80% conversion, which is representative of the more stable carbon; it was found that the trend in the Ea values (MWCNT > AC > FA/AS1100 > FA/AS900 > FA/AS700) fully agrees with the
observed stability under the ODH reaction conditions (Fig. 7), which involve gasification and oxidation reactions. The Ea values do not only give a quantification of the stability but also indicate about the low reactivity of the carbon species, which is
in turn related to the low selectivity to COx. Since the fractions of benzene/toluene and heavy condensates are marginal (<2%) and independent of the reaction conditions, the low selectivity to COx implies a higher selectivity to ST for the most
stable carbons. This was observed in fact in this study for MWCNT compared to AC or hybrids (cf. profiles in Fig. 7-b). This ST selectivity and coke reactivity relationship indicates that COx is formed through the deposited coke, instead of the complete
oxidation of EB or ST to COx. These trends are consistent with previous work on aluminas [69,70]; where the most ST selective alumina showed a less reactive coke.
Overall, among the investigated carbon-based materials, nanostructured
MWCNT has potential for EB ODH due to its higher oxidative stability when the deactivation or regeneration is remediated and the selectivity to COx is reduced by
enhancing further the oxidative stability.
Figure 8. Ozawa-derived apparent activation energies (Ea, kJ/mol) of the combustion at 20
and 80% conversions. Values are given in Table S-2.
0
20
40
60
80
100
120
140
160
FA/AS700 FA/AS900 FA/AS1100 AC MWCNT
Ea at 20% conversion Ea at 80% conversion
Ea, k
J/m
ol -1
On the stability of conventional and nano-structured carbon-based catalysts in the oxidative dehydrogenation of ethylbenzene under industrially relevant conditions.
71
It is noted that an isothermal stability test at low temperature, e.g. 425 oC will only delay the gasification and full consumption of the catalytic material in view of the apparent activation energies, which clearly showed that the MWCNT is the most
stable structure. A distributed oxygen feed to lower the local O2 concentration, as recently reported by Nederlof et al. [71], would help in enhancing both ST selectivity and oxidation stability.
3.4 Conclusions Carbon/SiO2 hybrids, activated carbon and nanostructured MWCNT are readily active and selective in the oxidative dehydrogenation of ethylbenzene (EB), under
industrially relevant conditions, implying concentrated EB feed, limited O2 and a demanding reaction temperature range, 425-475 oC to assess gasification and combustion. All these materials are stable up to 15 h TOS. After this period, the
gasification and burning phenomena take place; the hybrid materials decompose into their silica matrix, regardless of the applied pyrolysis temperature (700-1100 oC). A similar behaviour was observed for the activated carbon, whose bed is fully
consumed after 60 h TOS. MWCNT are exceptionally stable, no initial conversion decay was observed
even at 475 oC, with an EB conversion of 33% and ST selectivity at 80 % at the end
of the run. The selectivity does not decline during the run, though the EB conversion decreases steadily. This is attributed to ODH-derived coke deposition with a reduction of the surface area and pore volume. The 20% reduction of the catalyst
bed volume is attributed to the gasification/burning of the more amorphous domains. Interpretation of the reaction stability with light-off temperatures of the TPO
patterns was not suitable due to the heterogeneous nature of the carbon materials,
formed by amorphous and more crystalline domains. The apparent activation energies, however, correlate well to the catalyst stability.
In terms of the hybrid materials, it was learnt that the stability against
thermal sintering of the amorphous silica is remarkably enhanced by the presence of a carbon coating, by suppressing the particle growth and phase transitions even at 1100 oC.
3.5 References
[1] US DOE,http://www1.eere.energy.gov/office_eere/pdfs/exelus_case_study.pdf, Accessed on Aug. 19, 2013.
[2] Meyers RA. Handbook of petrochemicals production processes. New York NY:
McGraw-Hill. 2005: 11.3-11.34. [3] Cavani F, Trifiro F. Alternative processes for the production of styrene. Appl
Catal A-Gen 1995;133(2):219-39.
[4] Bogdanova OK, Belomestnykh IP, Voikina NV, Balandin AA. The oxidative dehydrogenation of ethylbenzene to styrene. Petrol Chem 1967;7(3):186-90.
[5] Raitmeier RE, Meyfield FD. Catalytic dehydrogenation process and
compositions. US patent 3437703, 1969. [6] Lisovskii AE, Aharoni C. Carbonaceous deposits as catalysts for
oxydehydrogenation of alkylbenzenes. Catal. Rev. Sci. Eng. 1994;36(1):25-74, and references therein.
[7] Lisovskii AE, Alkhazov TG, Dadasheva AM, Feizullaeva SA. Oxidative
dehydrogenation of alkylaromatic hydrocarbons on alumina catalysts. 4. Effect of alkalis on acid properties of alumina and its catalytic activity in oxidative dehydrogenation of ethyl benzene, Kinet Catal 1975;16(2):385-9.
ethylbenzene over a charcoal catalyst. React Kinet Catal Lett 1979;12(2):189-
93. [16] Grunewald GC, Drago RS. Oxidative dehydrogenation of ethylbenzene to
styrene over carbon-based catalysts. J Mol Catal 1990;58(2):227-33.
[17] Guerrero-Ruiz A, Rodriguez-Reinoso F. Oxydehydrogenation of ethylbenzene to styrene catalyzed by graphites and activated carbons. Carbon 1994;32(1):23-9.
[18] Pereira MFR, Orfão JJM, Figueiredo JL. Oxidative dehydrogenation of ethylbenzene on activated carbon catalysts I. Influence of surface chemical groups. Appl Catal A-Gen 1999;184(1):153-60.
[21] Pereira MFR, Orfão JJM, Figueiredo JL. Influence of the textural properties of an activated carbon catalyst on the oxidative dehydrogenation of ethylbenzene. Colloid Surface A 2004;241:165-71.
[22] Drago RS, Jurczyk K. Oxidative dehydrogenation of ethylbenzene to styrene over carbonaceous catalysts. Appl Catal A-Gen 1994;112(2):117-24.
[23] Kane MS, Kao LC, Mariwala RK, Hilscher DF, Foley H. Effect of porosity of
carbogenic molecular sieve catalysts on ethylbenzene oxidative dehydrogenation. Ind Eng Chem Res 1996;35(10):3319-31.
[24] Badstube T, Papp H, Kustrowski P, Dziembaj R. Oxidative dehydrogenation of
ethylbenzene with carbon dioxide on alkali-promoted Fe active carbon catalysts. Catal Lett 1998;55(3-4):169-72.
[25] Ikenaga N, Tsuruda T, Senma K, Yamaguchi T, Sakurai Y, Suzuki T.
Dehydrogenation of ethylbenzene with carbon dioxide using activated carbon-supported catalysts Ind Eng Chem Res 2000;39(5):1228-34.
[26] Zhang J, Su D, Zhang A, Wang D, Schlögl R, Hebert C. Nanocarbon as robust
catalyst: Mechanistic insight into carbon-mediated catalysis. Angew Chem Int Ed 2007;46:7319-23.
On the stability of conventional and nano-structured carbon-based catalysts in the oxidative dehydrogenation of ethylbenzene under industrially relevant conditions.
73
[27] De Oliveira SB, Barbosa DP, De Melo Monteiro AP, Rabelo D, Do Carmo Rangel M. Evaluation of copper supported on polymeric spherical activated carbon in the ethylbenzene dehydrogenation. Catal Today 2008;133-135:92-8.
[28] Zhang J, Su DS, Blume R, Schlögl R, Wang R, Yang X, et al. Surface chemistry and catalytic reactivity of a nanodiamond in the steam-free dehydrogenation of ethylbenzene. Angew Chem Int Ed 2010;49(46):8640-4.
[29] Malaika A, Rechnia P, Krzyzynska B, Kozlowski M, The influence of texture of activated carbons on their catalytic activity in the process of ethylbenzene dehydrogenation coupled with nitrobenzene hydrogenation, Micropor Mesopor
Mater 2012;163:300-6. [30] Mestl G, Maksimova NI, Keller N, Roddatis VV, Schlögl R. Carbon nanofilaments
in heterogeneous catalysis: An industrial application for new carbon materials?
Angew Chem Int Ed 2001;40(11):2066-8. [31] Pereira MFR, Orfão JJM, Figueiredo JL. Oxidative dehydrogenation of
ethylbenzene on activated carbon fibers. Carbon 2002;40(13):2393-401.
[32] Delgado JJ, Vieira R, Rebmann G, Su DS, Keller N, Ledoux MJ, et al. Supported carbon nanofibers for the fixed-bed synthesis of styrene. Carbon 2006;44(4):809-12.
[33] Delgado JJ, Su DS, Rebmann G, Keller N, Gajović A, Schlögl R. Immobilized carbon nanofibers as industrial catalyst for ODH reactions. J Catal
2006;244(1):126-9. [34] Li P, Li T, Zhou JH, Sui ZJ, Dai YC, Yuan WK, et al. Synthesis of carbon
nanofiber/graphite-felt composite as a catalyst. Micropor Mesopor Mat
2006;95(1-3):1-7. [35] Zhao TJ, Sun WZ, Gu XY, Rønning M, Chen D, Dai YC, et al. Rational design of
the carbon nanofiber catalysts for oxidative dehydrogenation of ethylbenzene.
Appl Catal A-Gen 2007;323:135-46. [36] Su DS, Chen X, Liu X, Delgado JJ, Schlögl R, Gajović A. Mount-etna-lava-
supported nanocarbons for oxidative dehydrogenation reactions. Adv Mater
2008;20(19):3597. [37] Delgado JJ, Chen XW, Frank B, Su DS, Schlögl R. Activation processes of highly
ordered carbon nanofibers in the oxidative dehydrogenation of ethylbenzene.
Catal Today 2012;186(1):93-8. [38] Keller N, Maksimova NI, Roddatis VV, Schur M, Mestl G, Butenko YV, et al. The
catalytic use of onion-like carbon materials for styrene synthesis by oxidative
dehydrogenation of ethylbenzene. Angew Chem Int Ed 2002;41(11):1885. [39] Su DS, Maksimova N, Delgado JJ, Keller N, Mestl G, Ledoux MJ, et al.
Nanocarbons in selective oxidative dehydrogenation reaction. Catal Today
2005;102-103:110-4. [40] Su D, Maksimova NI, Mestl G, Kuznetsov VL, Keller V, Schlögl R, et al.
Oxidative dehydrogenation of ethylbenzene to styrene over ultra-dispersed
[45] Frank B, Zhang J, Blume R, Schlögl R, Su DS. Heteroatoms increase the
selectivity in oxidative dehydrogenation reactions on nanocarbons. Angew Chem Int Ed 2009;48(37):6913-7.
[46] Delgado JJ, Chen X, Tessonnier JP, Schuster ME, Del Rio E, Schlögl R, et al.
Influence of the microstructure of carbon nanotubes on the oxidative dehydrogenation of ethylbenzene to styrene. Catal Today 2010;150(1-2):49-54.
[47] Qui NV, Scholz P, Krech T, Keller TF, Pollok K, Ondruschka B. Multiwalled carbon nanotubes oxidized by UV/H2O2 as catalyst for oxidative dehydrogenation of ethylbenzene. Catal Commun 2011;12(6):464-9.
[48] Qui N, Scholz P, Keller T, Pollok K, Ondruschka B, Ozonated Multiwalled Carbon Nanotubes as Highly Active and Selective Catalyst in the Oxidative
Dehydrogenation of Ethyl Benzene to Styrene, Chem Eng Technol 2013;36(2): 300-6.
[49] Qi W, Liu W, Zhang B, Gu X, Guo X, Su D, Oxidative Dehydrogenation on
Nanocarbon: Identification and Quantification of Active Sites by Chemical Titration, Angew Chem Int Ed 2013; 52:14224-8.
[50] Du Y, Li J, Ya X. Polyaniline as nonmetal catalyst for styrene synthesis by
oxidative dehydrogenation of ethylbenzene. Catal Commun 2008;9(14):2331-3. [51] Su DS, Delgado JJ, Liu X, Wang D, Schlögl R, Wang L, et al. Highly ordered
mesoporous carbon as catalyst for oxidative dehydrogenation of ethylbenzene
to styrene. Chem Asian J 2009;4(7):1108-13. [52] Wang L, Delgado JJ, Frank B, Zhang Z, Shan Z, Su DS, et al. Resin-derived
hierarchical porous carbon spheres with high catalytic performance in the
oxidative dehydrogenation of ethylbenzene. ChemSusChem 2012;5(4):687-93. [53] Xiao N, Zhou Y, Ling Z, Zhao Z, Qiu J, Carbon foams made of in situ produced
carbon nanocapsules and the use as a catalyst for oxidative dehydrogenation of
carbons as highly active catalysts for hydrogen production by CH4
decomposition. Chem Commun 2008;48:6585-7. [57] Jun S, Joo SH, Ryoo R, Kruk M. Synthesis of new, nanoporous carbon with
hexagonally ordered mesostructure. J Am Chem Soc 2000;122(43):10712-3.
[58] Brunauer S, Emmett PH, Teller E. Adsorption of gases in multimolecular layers. J Am Chem Soc 1938;60(2):309-19.
[59] Barrett EP, Joyner LG, Halenda PH. The determination of pore volume and area
distributions in porous substances. 1. Computations from nitrogen isotherms. J Am Chem Soc 1951;73(1):373-80.
[60] Lippens BC, De Boer JH. Studies on pore systems in catalysis. V. T method. J
Catal 1965;4(3):319. [61] Hatakeyama T, Zhenhai L, Handbook of Thermal Analysis. Chichester: John
Wiley & Sons Ltd. 1998: 47-61.
[62] Sing KSW, Everett DH, Haul RAW, Moscou L, Pierotti R, Rouquerol J, et al. Reporting physisorption data for gas solid systems with special reference to the determination of surface area and porosity (recommendations 1984). Pure Appl
Chem 1985;57(4):603-19. [63] Kruk M, Jaroniec M. Gas adsorption characterization of ordered organic-
inorganic nanocomposite materials. Chem Mater 2001;13(10):3169-83.
On the stability of conventional and nano-structured carbon-based catalysts in the oxidative dehydrogenation of ethylbenzene under industrially relevant conditions.
75
[64] Elzea JM, Rice SB. TEM and X-ray diffraction evidence for cristobalite and tridymite stacking sequences in opal. Clay Clay Miner 1996;44(4):492-500.
[65] Wang Z, Lu Z, Huang X, Xue R, Chen L. Chemical and crystalline structure
characterizations of polyfurfuryl alcohol pyrolyzed at 600 degrees C. Carbon 1998;36(1-2):51-9.
[66] Bertarione S, Bonino F, Cesano F, Damin A, Scarano D, Zecchina A. Furfuryl
alcohol polymerization in H-Y confined spaces: Reaction mechanism and structure of carbocationic intermediates. J Phys Chem B 2008;112(9):2580-9.
[67] Liu Q, Ren W, Li F, Cong H, Cheng HM. Synthesis and high thermal stability of
double-walled carbon nanotubes using nickel formate dihydrate as catalyst precursor. J Phys Chem C 2007;111(13):5006-13.
[68] Frank B, Rinaldi A, Blume R, Schlögl, Su DS. Oxidation Stability of Multiwalled
Carbon Nanotubes for Catalytic Applications, Chem Mater 2010;22(15):4462-70.
[69] Nederlof C, Zarubina V, Melián-Cabrera IV, Heeres HJ, Kapteijn F, Makkee M.
Oxidative dehydrogenation of ethylbenzene to styrene over alumina: effect of calcination. Catal Sci Technol 2013; 3(2):519-26.
[70] Zarubina V, Nederlof C, Van der Linden B, Kapteijn F, Heeres HJ, Makkee M,
Melián-Cabrera I. Making coke a more efficient catalyst in the oxidative dehydrogenation of ethylbenzene using wide-pore transitional aluminas. J Mol
Catal A: Chem 2014;381:179-87. [71] Nederlof C, Zarubina V, Melián-Cabrera I, Heeres HJ, Kapteijn F, Makkee M,
Application of staged O2 feeding in the oxidative dehydrogenation of
ethylbenzene to styrene over Al2O3 and P2O5/SiO2 catalysts, Appl Catal A: General. 2014;476:204-14.
Chapter 3
76
On the stability of conventional and nano-structured carbon-based catalysts in the oxidative dehydrogenation of ethylbenzene under industrially relevant conditions.
77
Figure S-1. TGA pattern of the fresh AC catalyst. The material fully decomposes after the catalytic
run and, therefore, no characterization could be carried out.
Figure S-2. TGA pattern of the fresh (in black) and spent (in red) MWCNT catalyst.
200 400 600 800 10000
20
40
60
80
100
-2.0
-1.5
-1.0
-0.5
0.0
0.5
fresh
rela
tive
we
igh
t
/
%
Temperature / oC
de
riva
tive
/
%(o
C)-1
derivative
200 400 600 800 10000
20
40
60
80
100
-1.0
-0.5
0.0
0.5
fresh
spent
rela
tive
we
igh
t
/
%
Temperature / oC
de
riva
tive
/
%(o
C)-1derivative
Chapter 3
78
Figure S-3. TGA pattern of the fresh (in black) and spent (in red) G/AS700 catalyst.
Figure S-4. TGA pattern of the fresh (in black) and spent (in red) G/AS900 catalyst.
200 400 600 800 100070
80
90
100
-0.3
-0.2
-0.1
0.0
0.1
0.2
0.3
rela
tive
we
igh
t
/
%
Temperature / oC
de
riva
tive
/
%(o
C)-1
derivative
200 400 600 800 100070
80
90
100
-0.3
-0.2
-0.1
0.0
0.1
0.2
0.3
fresh
spent
rela
tive
we
igh
t
/
%
Temperature / oC
de
riva
tive
/
%(o
C)-1
derivative
On the stability of conventional and nano-structured carbon-based catalysts in the oxidative dehydrogenation of ethylbenzene under industrially relevant conditions.
79
Figure S-5. TGA pattern of the fresh (in black) and spent (in red) FA/AS700 catalyst.
Figure S-6. TGA pattern of the fresh (in black) and spent (in red) FA/AS900 catalyst.
200 400 600 800 100070
80
90
100
-0,3
-0,2
-0,1
0,0
0,1
0,2
0,3
fresh
spent
rela
tive w
eig
ht
/ %
Temperature / oC
deri
vative /
%(o
C)-1
derivative
200 400 600 800 100070
80
90
100
-0.3
-0.2
-0.1
0.0
0.1
0.2
0.3
fresh
spent
rela
tive w
eig
ht /
%
Temperature / oC
deri
vative
/ %
(oC
)-1derivative
Chapter 3
80
Figure S-7. TGA pattern of the fresh (in black) and spent (in red) FA/AS1100 catalyst.
Figure S-8. TGA pattern of the fresh (in black) and spent (in red) AS catalyst.
200 400 600 800 100070
80
90
100
-0.3
-0.2
-0.1
0.0
0.1
0.2
0.3
fresh
spent
rela
tive w
eig
ht
/ %
Temperature / oC
deri
vative /
%(o
C)-1
derivative
200 400 600 800 100060
70
80
90
100
-0.2
-0.1
0.0
0.1
0.2
fresh
spent
rela
tive w
eig
ht
/ %
Temperature / oC
deri
vative /
%(o
C)-1
derivative
On the stability of conventional and nano-structured carbon-based catalysts in the oxidative dehydrogenation of ethylbenzene under industrially relevant conditions.
81
Figure S-9. Nitrogen sorption isotherms at -196 oC for the fresh glucose-based
silica hybrid materials (G/AS). Bare silica is added for comparison. Inset includes
the BJH pore size distributions. Isotherms for G/AS700 and G/AS900 are identical.
0.0 0.2 0.4 0.6 0.8 1.00
100
200
300
400
500
600
10 20 30 40 50 60 70
20.118.4
20.8
Pore width / nm
G/AS700
G/AS900
AS
Vo
lum
e a
dso
rbe
d /
cm
³/g
ST
P
Relative Pressure (p/p°)
Table S-1. Micropore volume for fresh and spent catalysts.
Sample Vm (cm3g-1) fresh Vm (cm3g-1) spent
AC 0.316 FDa
MWCNT <0.001 0.004
G/AS700 0.019 0.001
G/AS900 0.018 0.001
FA/AS700 0.041 0.001
FA/AS900 0.038 0.001
FA/AS1100 0.019 0.003
AS 0.003 NMb
AS-700 0.006 NAc
AS-900 <0.001 NAc
AS-1100 <0.001 NAc
a. FD = fully decomposed after the catalytic run; b. NM = not measured; c. NA =
not applicable.
Chapter 3
82
Figure S-10. Nitrogen sorption isotherms at -196 oC for the fresh MWCNT
and AC materials. Inset includes the BJH pore size distributions.
Figure S-11. General overview of the nitrogen sorption isotherms at -196 oC for all materials (fresh).
0.0 0.2 0.4 0.6 0.8 1.00
100
200
300
400
500
600
700
800
0 20 40 60 80 100 120 140
3.0
31.5
Pore width / nm
MWCNT
AC
Volu
me a
dsorb
ed /
cm
³/g
ST
P
Relative Pressure (p/p°)
0.0 0.2 0.4 0.6 0.8 1.00
100
200
300
400
500
600
700
800
MWCNT
AC
G/AS700
G/AS900
FA/AS700
FA/AS900
FA/AS1100
AS
Vo
lum
e a
dso
rbe
d /
cm
³/g
ST
P
Relative Pressure (p/p°)
On the stability of conventional and nano-structured carbon-based catalysts in the oxidative dehydrogenation of ethylbenzene under industrially relevant conditions.
83
Table S-2. The apparent activation energies of the combustion of fresh catalysts
Catalyst Ea at 20% conversion
kJ.mol-1 Ea at 80% conversion
kJ.mol-1
AC 128 132
MWCNT 133 144 FA/AS700 102 116 FA/AS900 122 124
FA/AS1100 123 125
Chapter 3
84
Phosphorous-induced thermal stabilization for carbon-supported SiO2 catalysts in the oxidative dehydrogenation of ethylbenzene to styrene
A strategy to enhance the thermal stability of C/SiO2 hybrids for the O2-based oxidative
dehydrogenation of ethylbenzene by P addition is proposed. The preparation consists of the
polymerization of furfuryl alcohol (FA) on a mesoporous precipitated SiO2. The polymerization is
catalysed by oxalic acid (OA) at 160 oC (FA:OA=250). Phosphorous was added as H3PO4 after the
polymerization and before the pyrolysis that was carried out at 700 oC and will extend the overall
activation procedure. The carbon and P loadings slightly alters the textural parameters; the pores
retain the cylindrical shape with a fraction of micropores, resembling a coating. The catalytic tests
were carried out under limited O2, high EB concentration, and temperature in the range of 425-475
oC. The tests show that the P/C/SiO2 hybrids are readily active, selective, and stable in the applied
reactions conditions for 60 h time on stream. The comparison with MWCNT reveals that the
P/C/SiO2 hybrids are more active and selective at high temperatures (450-475 oC), while the
difference becomes negligible at lower temperature. Characterization of the spent catalysts shows
that P/C/SiO2 hybrids build up substantial coke and due to the overcoking, having pore network
restrictions, pore volume, and surface area depletion.
Chapter 4
86
4.1 Introduction
In the petrochemical industry styrene (ST) production is considered one of the major
processes; styrene is used principally as a monomer for polystyrene with different grades and as a component in the synthesis of styrene-butadiene co-polymer for automobile tires. ST is industrially over 85 % produced by direct dehydrogenation of
ethylbenzene (EB) over a K-promoted Fe2O3 catalyst at 580-630 °C using an excess of steam [1]. The major feature of this process is the extremely high selectivity (>96 %), that makes the downstream processing relatively simpler, as compared to other
catalytic processes. The (small) amounts of byproducts (benzene, toluene, and hydrogen) have also commercial value. The process suffers, however, from high steam consumption, moderate conversion per pass due to the equilibrium limitations,
and high temperatures are required for the endothermic reaction [2]. Remarkable efforts now and in the past have been put in to overcome the equilibrium recycle, reducing the operation temperature, reducing the steam to EB ratio, and replacing
steam by oxidants, such as O2 or CO2, or combinations thereof. The use of oxidants in this process (ODH route), can in theory and will in practice reduce the reaction temperature as the equilibrium limitation disappears for the case of O2 or it is
improved for the CO2-based ODH. It is nevertheless not an easy task finding a highly selective catalyst in the presence of an oxidant. This is because EB is also gasified and/or fully oxidized into CO/CO2 that has no economic value. Besides the selectivity,
the ODH process is not commercialized yet due to the limited catalyst stability. Two types of catalyst families have been proven to be active and selective for
EB ODH. Inorganic-based materials such as aluminas [3-8], metal pyrophosphates
[9-11], and phosphates [10-20], or P-supported silica [10,21] have been reported as a first class of catalysts. There is a general consensus that the coke deposit generated on the acid sites under these oxidative reaction conditions is more active
and selective for the production of ST than the inorganic material itself [3-6,8,22]. In fact, the formation of coke initially improves the activity and selectivity, however, in the long term an excess of carbon deposit leads to catalyst deactivation. Depending
on the reactions conditions there is an activation period to achieve pseudo-stationary conditions, i.e. a period of time necessary to achieve full coverage of active coke on the surface, where the conversion and selectivity to styrene achieve both a
maximum. We recently showed that for alumina it can be significantly shortened by working at 475 °C instead of 450 °C [8].
The second catalyst family includes carbon based materials, such as activated
carbons [23-29], carbon nano-fibers [30-35], onion-like carbon [36,37], diamonds [37,38], and multi-walled carbon nanotubes (MWCNT) [38-42]. These materials are
readily active and selective; the reported selectivities lie within 70-90 % that is low to moderate, as compared to the commercial steam dehydrogenation process (>96%). In some cases, the reported selectivity is significantly high, ranging 90-
97%, however, in an excess of oxygen [42-44]. The performance and stability of these carbon-based catalytic materials is in general difficult to compare due to large differences in the applied operation conditions. Principally temperature range, EB
concentration, space velocity, and O2:EB vary significantly among the reported literature. We have recently compared, under identical and relevant ODH conditions, various types of carbon-based catalysts, ranging from conventional to nano-
structured carbons [45]. It was found that nano-structured MWCNT is the most stable material; the structure resists the reaction conditions showing an EB conversion of ca. 30% (but deactivating) with a steady selectivity of ca. 80%. On the
other hand, the low-cost carbon-SiO2 (C-SiO2) hybrids prepared by furfuryl alcohol polymerization and pyrolysis were stable only for 15h; after that, the carbon of the hybrids decomposes completely rendering the almost inactive and non-selective
silica matrix.
Phosphorous-induced thermal stabilization for carbon-supported SiO2 catalysts in the oxidative dehydrogenation of ethylbenzene to styrene.
87
In this work we propose a strategy to enhance the thermal stability of these C/SiO2 hybrids for the ODH reaction. It is well-known that phosphorous has an inhibiting effect on the carbon combustion [46-50]; in this study we investigated the
effect of the P addition (on two loadings) on a furfuryl alcohol based silica hybrid. The performance of these P-modified hybrid catalytic materials is compared to state-of-the-art P/SiO2 and MWCNT. The catalyst stability under the ODH reaction
conditions will be evaluated from the combustion apparent activation energies.
4.2 Experimental methods
4.2.1 Materials
Bare precipitated amorphous silica extrudates (61138, denoted here as AS) was kindly supplied by Saint-Gobain NorPro, a Division of Saint-Gobain Ceramic Materials GmbH. Low SiO2 stabilized -Al2O3 extrudates were kindly provided by Albemarle
Catalysts B.V. Purified MWCNT (C>96 wt.% experimentally determined) was kindly
supplied by Hyperion (CS-02C-063-XD). Activated carbons (denoted as AC) were supplied by Norit (ROX 0.8). For the impregnations, we employed the liquid pore volume (VLPV) that was experimentally determined (VLPV=1.05 cm3.g-1): water was
added to ca. 10 g material that is repeatedly shaken after each water addition until the material turned shiny. Note that this method is different to the reported one by
Innes [51] based on the caking of the catalyst particles that agrees with the gas adsorption pore volume. Note that VLPV (1.05 cm3.g-1) is larger than that of the gas adsorption value (0.842 cm3.g-1, AS in Table 1) because water fills pores larger than
100 nm. A similar VLPV was used for the (Furfuryl Alcohol) FA/SiO2 composites during the P addition.
4.2.2. Catalyst preparation
4.2.2.1 Synthesis of the C/SiO2 catalysts
Among the various carbon sources, we considered using the furfuryl alcohol polymerization [52-61] route since it has given successful examples of stable carbon nanoreplicas [62-64]. Commercial amorphous precipitated SiO2 (AS) was used as
support. The solid was pre-sieved into the 212-425 µm fraction that was obtained by crushing the commercial extrudates. The carbon precursor was added by incipient wetness impregnation of the carbon precursor using 5% extra liquid volume
regarding the solid pore volume. Furfurylalcohol (FA, Acros Organics, 98%) was used as carbon precursor and oxalic acid (OA, Acros Organics, >99%) as catalyst.
The support (typically 3 g) was first degassed under vacuum at 150 oC during
4 h. A carbon-precursor solution was prepared by mixing FA and OA in a molar ratio of 250 based on previous polymerization studies [64] and water. Then 3.15 mL of the FA/OA aqueous solution was added to the support to obtain a carbon loading of
15 wt.%, in the final pyrolyzed hybrids; assuming that only carbon is present in the final composite material. The wet material was shaken (VWR, digital DVX-2500)
during 4 minutes at 2500 rpm at room temperature to distribute the solution evenly. Afterwards, the FA polymerization of the samples was induced at 160 oC in an atmospheric oven during 8 h; the sample is named 15C/AS.
Chapter 4
88
4.2.2.2 Preparation of the P/C/SiO2 hybrid catalysts
The phosphorus was added onto the dried 15C/AS material before pyrolysis as diluted orthophosphoric acid (H3PO4); a solution (10 mL) was prepared by mixing
0.651 mL (or 1.345 ml) of H3PO4 (Merck, 85%) with water; then 3.15 mL of the solution was (incipient wetness) impregnated into the 15C/AS dried material (3 g). This corresponds to a phosphorous loading of ca. 3 or 6 wt.% (as P). Afterwards, the
samples were shaken during 4 min at 2500 rpm at room temperature and dried at 70 oC overnight in an atmospheric oven. The catalysts are denoted as 3P/15C/AS and 6P/15C/AS.
4.2.2.3 Carbonization protocol
The carbonization of the C/SiO2 and P/C/SiO2 samples was carried out by pyrolysis in a quartz-tube housed tubular oven (Nabertherm RT 50/250-11). The sample (~3
gram) was loaded in a flat quartz crucible and placed horizontally in the centre of the heating zone of the furnace. After closing and purging the tube for 30 min with N2, the sample was heated at 1 °C.min-1 from room temperature until 700 oC for 3 h and
subsequent cooling down all in a nitrogen flow (150 mL.min-1 NTP). 4.2.2.4 Preparation of the P/SiO2 reference catalysts
Reference catalysts based on P on silica (3 and 6 wt.%) were prepared for comparison. The P addition protocol is nearly identical to that described above for
the P/C/SiO2 hybrid; using 212-425 m pre-sieved AS material that was obtained by
crushing the commercial extrudates. The samples were then shaken during 4 min at 2500 rpm at room temperature and dried at 70 oC overnight in an atmospheric oven. The samples were calcined in an air box furnace (Nabertherm LT9/11) from room
temperature at 4 °C.min-1 until at 500 oC and kept for 8 h. The catalysts are denoted as 3P/AS and 6P/AS.
4.2.3 Characterization
The total organic content of the fresh and spent catalysts was quantified by
thermogravimetric analysis (TGA) on a Mettler-Toledo analyzer (TGA/SDTA851e) using a flow of synthetic air of 100 mL.min-1 NTP. The temperature was increased from 30 to 900 °C at 10 °C.min-1. Blank curve subtraction using an empty crucible
was taken into account. The oxidation rate patterns (TPO) were obtained in the same instrument using the derivative of the TGA patterns. Elemental carbon, hydrogen, and nitrogen in the hybrids were analysed in a EuroVector 3000 CHNS analyser, after
dissolving the silica and the complete removal of “POx” (40 wt.% HF, 2 days, at ambient temperatures). Approximately 2 mg of the organic derived material was accurately weighed in a 6-digit analytic balance (Mettler Toledo). The samples were
completely (no ash deposits after the TGA experiment) burnt at 1800 oC in the presence of an oxidation catalyst and decomposed into CO2, H2O, and N2. These gases were then separated in a Porapak QS column at 80 oC and quantified with a
TCD detector. Acetonitrile (99.9 % purity) was used as an external standard. Oxygen content was calculated by difference.
Nitrogen physisorption analyses (196 oC) were carried out in a Micromeritics
ASAP 2020. The samples were degassed in vacuum at 250 oC for 10 h. The surface area was calculated using the standard BET method (SBET) [65]. The single point gas adsorption (or internal) pore volume (VT), below 100 nm, was estimated from the
amount of gas adsorbed at a relative pressure of 0.98 in the desorption branch. The pore size distributions (PSD) were obtained from the BJH method [66] using the
Phosphorous-induced thermal stabilization for carbon-supported SiO2 catalysts in the oxidative dehydrogenation of ethylbenzene to styrene.
89
adsorption branch of the isotherms; the mean pore size (BJH) is given by the
position of the PSD maximum, while the t-plot method [67] was employed to quantify the micropore volume (V) and area (S). It is noted that the samples do
not show microporosity and this was excluded from the discussion.
The apparent activation energy (AAE) of the carbon combustion was calculated by Ozawa method [68] using a correlation between peak temperature for a given conversion and the heating rate for four thermal analysis derivative curves
(TGA). The TGA curves were obtained on a Mettler-Toledo analyzer (TGA/SDTA851e) using a flow of synthetic air of 100 mL.min-1 NTP. The temperature was increased from 30 to 900 °C with a heating rate of 1, 3, 5, 10 °C.min-1. Blank curve
subtraction using an empty crucible was taken into account. The apparent activation energy was calculated for 20 and 80% conversion levels.
4.2.4 Catalytic tests
The catalytic tests were carried out in a 6-flow micro reactor using a fixed volume of
catalyst (0.8 ml, corresponding to 65 mm of bed length) in down-flow 4 mm quartz reactors. To assure that the catalyst bed is located in the isothermal zone of the furnace, the reactors were loaded with quartz wool, 10 cm glass pearls (0.5 mm
diameter), the catalyst, 10 cm glass pearls (0.5 mm diameter), and a second quartz wool plug. The glass pearls have limited conversion; less than 3% EB conversion under all applied conditions. The reactor gas feed is a mixture that can consist of
CO2, N2, and air that counts for a gas-flow rate of 36 mL (NTP).min-1; the liquid EB-feed flow rate is 1 g/h (3.54 ml (NTP).min-1 vapor) that is evaporated upstream each reactor in a -Al2O3 column, resulting in a 1:9 molar ratio of ethylbenzene to gas (10
vol. % EB). This corresponds with operation at a GHSV of 3000 l/l/h. The total
pressure was 1.2-1.3 105 Pa. The reactor exhaust gas was analyzed by gas chromatography using a combination of columns (0.3m Hayesep Q 80-100 mesh with back-flush, 25m × 0.53mm Porabond Q, 15 m×0.53mm molsieve 5A, and RTX-
1 with 30m×0.53mm) with TCD and FID detectors. This configuration allows quantifying permanent gases such as CO2, H2, N2, O2, CO as well as hydrocarbons (methane, ethane, ethene, benzene, toluene, ethylbenzene, styrene, and heavy
aromatics). The catalytic tests were carried out under practical conditions of 20 % excess O2 with respect to the ODH reaction and a concentrated EB feed of 10%. If
not stated otherwise, for all EB conversion data the oxygen conversion is complete. All physical characterizations for the spent catalysts were done after the complete testing cycle of 60 h. A more complete description of the testing protocol can be
found elsewhere [69].
4.3 Results
4.3.1 Characterization of the fresh hybrid catalysts
The thermal stability and amount of organic material of the hybrids was determined by TGA (Fig. S-1). The hybrid carbon starts to burn in air at around 450 oC with a maximum rate at 575 oC for both hybrids. The organic content of the P-hybrids is
17.2 and 19.1 wt. %. The oxidation stability in air was compared to the MWCNTs and a P-free counterpart (15C/AS) using the oxidation rate profiles (TPO, Fig. 1). Most of the materials start to burn in air above 450 oC, but the maxima vary among the
samples; the most stable are MWCNT with a maximum at 625 oC. The maxima of the P-promoted hybrids virtually coincide with the 15C/AS material, so P has apparently no effect on inhibiting the combustion of the deposited carbon. However, the
Chapter 4
90
determination of the apparent activation energies will give a more thorough
interpretation of the thermal stability, as it was shown in our previous study [45].
Figure 1. Oxidation rate patterns (TPO) in air (10 oC.min-1) for MWCNT, 3P/15C/AS,
ml.min-1, heating rate of 10 °C. min-1. Typical sample amount ranged 5-8 mg.
Figure 2. Ozawa-derived apparent activation energies (AAE, kJ.mol-1) for the combustion at
80% conversion.
300 400 500 600 700 800
a
b
c
d
e
f
Temperature / oC
Oxid
atio
n r
ate
/
a
.u.
MWCNT
3P/15C/AS
6P/15C/AS
15C/AS
3P/15C/AS spent
6P/15C/AS spent
116
131 130
144
60
80
100
120
140
160
15C/AS 3P/15C/AS 6P/15C/AS MWCNT
Apparent activation energy at 80% conversion (kJ.mol-1)
Phosphorous-induced thermal stabilization for carbon-supported SiO2 catalysts in the oxidative dehydrogenation of ethylbenzene to styrene.
91
Figure 2 shows the apparent activation energies (AAE) as determined by the Ozawa method. A clear trend is seen where the P-based hybrids have an increased AAE of around 13% higher than that of the 15C/AS, and lower than that of MWCNT (24%
higher than 15C/AS). Therefore, the AAE predict a better thermal stability attributed to the P effect.
High oxygen content of the deposited carbon has been one of the crucial
compositional features for the selective and active sites for the reaction-produced active coke in the EB ODH reaction [22]. The elemental composition of both 3P/15C/AS and 6P/15C/AS hybrids was determined by CHN (Table S-1) and
compared to reported values for ODH cokes (Fig. S-2) [22,70]. The O:C ratios (0.10-0.15 at.) lie within the typical reported limits for ODH coke. Therefore, it can be expected that the FA-based hybrids will be readily active/selective for the ODH
reaction. Studies about FA-based carbons, reports low O:C ratios (0.02-0.12) [54,56,57,59]. Burket et al. [59] demonstrated that the O removal takes place during the collapse of the mesopores. Thus the high O:C ratio in the 3P/15C/AS and
6P/15C/AS can be then attributed to the stable mesopores (in both cases 96% of the total pore volumes are mesoporous) due to the presence of the mesoporous silica support underneath that prevents the carbon collapse.
Other important features are the porosity of the hybrid material in comparison to the fresh silica and the type of coke formed. The Raman spectra of both materials
(3P/15C/AS and 6P/15C/AS, Fig. S-3) reveal two broad superimposing adsorptions centered at 1610 and 1375 cm-1, which are characteristic of aromatic and amorphous carbon (G and D peaks, respectively) [71,72].
Comparison of the gas adsorption N2 isotherms was done in Figure 3. The shape of the isotherms does not change upon deposition of carbon and phosphorous; isotherms of type IV with hysteresis HI were observed as for the bare AS silica;
representing solids with cylindrical highly uniform pores [73,74]. No pore network effects were observed, indicating that the pores are well connected after the carbon/phosphorous addition. There is, however, a substantial reduction of the pore
volume of 37% (3P/15C/AS) and 42% 6P/15C/AS); while the surface area only changes 8 and 18%, respectively. The t-plot analysis evidence the formation of micropores that contribute in 22-23% of the total surface area (Table S-2).
Therefore, there are two opposing effects on the texture; significant reduction of mesopore volume and formation of micropore volume will overall result in a slight reduction of the total surface area. This happens because the created micropores
contribute significantly to the surface area. The BJH pore size distribution in Figure 3 (inset) shows a decrease of the intensity, associated to the filling effect of the carbon. Calculation of the average geometrical pore size (Table 1) reveals a
reduction of pore size from 15.8 nm (AS) to 10.9 nm (3P/15C/AS) and 11.3 nm (6P/15C/AS). Therefore, the overall interpretation of the textural data is that a P-containing carbon coating has been created with the shrinking of the existing
mesopores and the formation of new micropores. Both will contribute in maintaining a relatively high surface area.
Chapter 4
92
Figure 3. Nitrogen sorption isotherms at196 oC for AS-700, 3P/15C/AS, and 6P/15C/AS.
Inset: BJH pore size distribution.
Table 1. Textural parameters derived from N2 adsorption at 196 oC.
Material SBET (m2.g-1) VT (cm3.g-1)a,c BJH
(nm)d
AS bare 213 0.842 20.8 (15.8)
3P/15C/AS 195 (8)b 0.532 (37)b 22.1 (10.9)
3P/15C/AS spent 58 (70)a 0.128 (76)a broad (8.8)
6P/15C/AS 174 (18)b 0.492 (42)b 23.3 (11.3)
6P/15C/AS spent 42 (76)a 0.101 (79)a broad (9.6)
MWCNT 406 1.100 3, 30.8
3P/AS 165 0.765 21.9 (18.5)
3P/AS spent 61 (63)a 0.155 (80)a 22.5 (10.1)
6P/AS 74 0.682 140 (36.9)
6P/AS spent 56 (24)a 0.392 (43)a broad (28.0)
AC 959 0.609 broad a. Value in parenthesis is the % of reduction with respect to the fresh material; b. % of reduction
with respect to the AS bare; c. gas adsorption or internal pore volume; d. values in parenthesis are the average geometrical pore size as 4.103VT/SBET.
4.3.2 Pseudo-steady state EB ODH catalyst performance
The catalyst performances are summarized in Figure 4; EB conversion, selectivity to COx, selectivity to ST, and ST yield are represented as a function of time-on-stream
(TOS) at various temperatures and O2:EB ratios. The selectivities to side products (not shown) are small and almost constant as a function of the reaction conditions. Benzene/toluene are formed in low percent concentrations along with trace amounts
of oxygenates, the selectivities to these byproducts do not substantially vary. The sum of ST and COx selectivities (Fig. 4-b and -c) is on average above 96% of the converted ethylbenzene.
0.0 0.2 0.4 0.6 0.8 1.00
100
200
300
400
500
600
2 10 100
AS-700
3P/15C/AS
6P/15C/AS
Volu
me a
dsorb
ed
/ c
m3/g
ST
P
Relative Pressure / P/Po
23.3
22.1
Pore width / nm
20.9
Phosphorous-induced thermal stabilization for carbon-supported SiO2 catalysts in the oxidative dehydrogenation of ethylbenzene to styrene.
93
Figure 4 compares the P-hybrids with various types of relevant EB ODH catalysts under similar conditions. It was found that the P-hybrids are stable. Although there is a slight gasification/burning as reflected in the COx profiles, the P-
hybrids were stable for 60 h. In our previous work [45] we reported that the P-free hybrid (15C/AS), starts to decompose after 20 h with the complete disappearance of the introduced carbon and no conversion after 35 h on the bare silica. Therefore, the
addition of P has a remarkable effect on the stability. This is consistent with the calculated apparent activation energies that showed values for the P-hybrids close to the MWCNT.
The performance of the P-hybrids is compared to state-of-the-art P/SiO2, MWCNT and prior work on C/SiO2 hybrids. The selectivity to ST ranges 85-90%, which is high, but slightly lower when compared to the P/SiO2 counterparts and
much better than C/SiO2 [46] that ranged between 50 and 80%. A more competitive carbon-based catalyst would be the MWCNT. When the ST selectivity or EB conversion of these P-based hybrids are compared to the MWCNT, it can be seen
that the P-hybrids behave much better at high temperatures (450-475 oC) both, in conversion and selectivity, while the difference becomes negligible at lower temperature (425 oC). The trends in selectivity were opposing; while the selectivity
of the MWCNT was best at lower temperature (in agreement with literature that often reports a range between 350-400 oC operation temperatures), higher
temperature were most favourable for the P-hybrids. Thus, the P-hybrids seem to be a high-temperature catalyst for this application with higher selectivity and conversion than that of MWCNT.
An additional remarkable feature is that the P-hybrids are readily active/selective, while the P-based silica catalysts (3P/AS and 6P/AS) require substantial time to reach the pseudo-stationary conversion (Fig. 4-a); 8 and 15 h
respectively, under the applied conditions. This is thought to be due to the slow formation of active coke on the catalyst surface, induced by the acidity of the phosphates groups [10,21]. This deposited coke is readily selective since the
selectivity profiles do not have such activation period (Fig. 4-b). The ST yield of these P/silica catalysts (Fig. 4-d) is determined by a complex activation-deactivation phenomena of the conversion, related to the unlimited coke build up process.
When comparing the performance at 450 oC between the second and last sequence in the catalytic test, it can be observed that the P-based hybrids drop in conversion; the conversion levels do not come back to the values recorded in the
second cycle but decline. This can be associated to the on-going coking, though other deactivating phenomena associated to the silica itself cannot be ruled out. Despite this, it is noteworthy that such a drop is much lower than those for MWCNT
and P/SiO2 catalysts.
Chapter 4
94
Figure 4. Time on stream EB conversion (a), selectivity to ST (b), selectivity to COx (c), and ST yield (d) at various temperatures (475, 450, 425, and 450 oC) and O2/EB= 0.6 and 0.2 (vol); GHSV of 3000 l/l/h. Note that 3P/15C/AS and 6P/15C/AS overlap for TOS > 20 h.
In conclusion, the P-based hybrids show remarkable features. They lack of the
stabilization period, i.e. readily active/selective, and the selectivity and conversion at
high temperature (450-475 oC) are superior when compared to the MWCNT. The difference in selectivity and conversion becomes negligible when working at 425 oC. They seem to be thermally stable, with no apparent gasification/burning for 60 h
under the applied harsh reaction conditions; the on-going deactivation is less extensive than the other reference catalysts tested.
4.3.3 Spent catalysts and deactivation
The coke content in the spent catalysts was determined by TGA (Fig. S-4) and the
coke build-up was calculated by the difference with the starting organic loading. Figure 5 shows that the P-hybrids build up coke during the reaction to ca. 23 wt. %.
Thus, P can stabilize the FA-based coating but apparently also builds up additional coke. The P/AS samples also build coke in a much wider range, from 14% (6P/AS) to 43 % (3P/AS), which is related to the differences in the surface areas of the fresh
catalysts. It is not strictly proportional, but also to the fact the 6P/AS is still in the activation process, so not all the surface has been covered by coke.
0 10 20 30 40 50 600
10
20
30
40
50
60 3P/15C/AS
6P/15C/AS
3P/AS
6P/AS
MWCNT
ST
Yie
ld /
%
Time / h
0 10 20 30 40 50 600
10
20
30
40
50
60
70
80
90
100
3P/15C/AS
6P/15C/AS
3P/AS
6P/AS
MWCNT
Sele
ctivity to
ST
/ %
Time / h
2
0 10 20 30 40 50 600
10
20
30
40
50
60
70
80
90
100 3P/15C/AS
6P/15C/AS
3P/AS
6P/AS
MWCNT
Sele
ctivity to C
Ox /
%
Time / h
425 450450
0.6 0.2 0.6 0.60.20.6 0.2
475
450450
0.6 0.2 0.60.20.6
475
0.2
425
0.6
0 10 20 30 40 50 600
10
20
30
40
50
60 3P/15C/AS
6P/15C/AS
3P/AS
6P/AS
MWCNTs
C
onvers
ion / %
Time / h
0.6 0.2 0.60.20.6 0.2
425 450475
0.6
450
0.6 0.2 0.60.20.6 0.2
425 450475
0.6
450
a b
dc
Phosphorous-induced thermal stabilization for carbon-supported SiO2 catalysts in the oxidative dehydrogenation of ethylbenzene to styrene.
95
Figure 5. Organic contents for the fresh and spent catalysts after reaction. Coke build-up is
defined as the difference (spent – fresh). Raw data are given in Table S-5 and TGA patterns
in Fig. S-4.
Chemical analysis proves higher oxygen content of the spent hybrid catalysts
than that of the fresh counterparts (Table S-1); it comes from the additional coke formed during the reaction. The higher O content is consistent with a more facile oxidation in air, evidenced by a shift towards lower temperatures in the TGA maxima
(Fig. 1). Representation of the composition in a modified van Krevelen plot (Fig. S-2) shows significant changes. The C:H varies from a relatively high ratio, characteristic of FA-based carbons having low H content [54,56,57,59], to lower ratios, near the
values often reported for the ODH coke (Fig. S-2). It is obvious, that the fresh hybrid catalysts can not be hydrogenated under the ODH conditions. Hence this
compositional effect is ascribed to the deposition of new ODH coke that is richer in H during the reaction.
Raman spectra of the spent hybrid catalysts indicate the presence of a more
ordered structure that has been formed during the reaction. The D/G peak ratio becomes smaller (Fig. S-3 and Table S-6) and this can be explained by two effects: 1) removal of the more amorphous carbon domains by gasification/burning and 2)
deposition of sp2-rich carbon materials, or 3) both effects. The latter is due to the buildup of new polyaromatic ODH coke.
Therefore, TGA/TPO, CHN, and Raman reveal the formation of a coke with
different nature than the starting hybrid. The porosity of the spent hybrids evidences that reduction of surface area and
pore blockage is the main source of catalyst modification during the applied testing
protocol. The specific surface area and pore volume are reduced around 70-80% for both parameters (Table 1) resulting in a less active catalytic system. The isotherms change the shape having a closure point around 0.45 in the relative pressure (Fig. S-
5) indicating the presence of pore network restrictions associated to the overcoking from the reaction. The spent 3P/AS and 6P/AS also manifested a substantial reduction of surface area (63 and 24%) and pore volume (80 and 43%) (Table 1,
Fig. S-6). The 3P/AS has pore network effects, while 6P/AS does not because it has much larger pores. The 6P/AS is less affected possibly because the surface is not yet fully covered in coke. The conversion keeps increasing after 55 h which is an
indication of on-going coke build-up. Thus, overcoking is identified as responsible for the decrease of texture for
both, P/silica and P/C/silica hybrids during reaction.
3P/AS 3P/15C/AS 6P/AS 6P/15C/AS
0
19.1
0
17.2
42.9 42.8
14.0
40.5 42.9
23.7
14.0
23.3
FRESH (wt.%) SPENT (wt.%) COKE BUILD-UP (wt.%)
Chapter 4
96
4.4 Conclusions
A composite material based on the polymerization and pyrolysis of furfuryl alcohol on a commercial silica, containing P is stable under relevant EB ODH conditions up to 60 h TOS. This composite material was readily active and highly selective. It is shown
that P has a combustion inhibiting effect as well as it contributes to the selectivity greatly. The P-hybrid materials outperform at high temperatures (450-475 oC) the hybrid catalysts, while the differences with respect to MWCNT become negligible at
lower temperature (425 oC), with a lower conversion. Both, conversion and selectivity were optimal at high temperature, which makes this hybrid a better alternative than MWCNT. Overcoking with a reduction of pore volume and surface
area takes place during the reaction, while the composition of the carbon gets richer in O due to the deposited coke. The apparent activation energies predicted the thermal stability under the ODH reaction conditions.
4.5 References
[1] Meyers RA. Handbook of petrochemicals production processes. New York NY:
McGraw-Hill. 2005: 11.3-11.34. [2] Cavani F, Trifiro F. Alternative processes for the production of styrene. Appl
of oxidational dehydrogenation of ethylbenzene on an aluminum oxide catalyst.
Rus Chem Bul 1977;26(3):477-80. [4] Iwasawa Y, Nobe H, Ogasawara S. Reaction mechanism for styrene synthesis
over polynaphthoquinone. J Catal 1973;31(3):444-9.
[5] Alkhazov TG, Lisovskii AE. Role of condensation products in oxidative dehydrogenation process of ethylbenzene on aluminium-oxide catalyst. Kinet Catal 1976;17(2):375-9.
[6] Nederlof C, Kapteijn F, Makkee M. Catalysed ethylbenzene dehydrogenation in CO2 or N2 - Carbon deposits as the active phase. Appl Catal A-Gen 2012;417:163-73.
[7] Nederlof C, Zarubina V, Melián-Cabrera IV, Heeres HJ, Kapteijn F, Makkee M. Oxidative dehydrogenation of ethylbenzene to styrene over alumina: effect of calcination. Catal Sci Technol 2013; 3(2):519-26.
[8] Zarubina V, Nederlof C, Van der Linden B, Kapteijn F, Heeres HJ, Makkee M, et al. Making coke a more efficient catalyst in the oxidative dehydrogenation of ethylbenzene using wide-pore transitional aluminas. J Mol Catal A: Chem
2014;381:179-87. [9] Emig G, Hofmann H. Action of zirconium-phosphate as a catalyst for the
oxidative dehydrogenation of ethylbenzene to styrene. J Catal 1983;84(1):15-
26. [10] Vrieland GE. Oxydehydrogenation of ethylbenzene to styrene over metal
pyrophosphates. 1. Catalyst composition and reaction variables. J Catal
1988;111(1):1-13. [11] Vrieland GE, Friedli HR. Method of oxydehydrogenation of ethyl benzene. US
patent 3933932, 1976.
[12] Bautista FM, Campelo JM, Luna D, Marinas JM, Quirós RA, Romero AA. Screening of amorphous metal-phosphate catalysts for the oxidative dehydrogenation of ethylbenzene to styrene. Appl Catal B-Environ 2007;70(1-
4):611-20.
Phosphorous-induced thermal stabilization for carbon-supported SiO2 catalysts in the oxidative dehydrogenation of ethylbenzene to styrene.
97
[13] Murakami Y, Iwayama K, Uchida H, Hattori T, Tagawa T. Study of the oxidative dehydrogenation of ethylbenzene.1. Catalytic behavior of SNO2-P2O5. J Catal 1981;71(2):257-69.
[14] Murakami Y, Iwayama K, Uchida H, Hattori T, Tagawa T. Screening of catalysis for the oxidative dehydrogenation of ethylbenzene. Appl Catal 1982;2(1-2):67-74.
[15] Schraut A, Emig G, Sockel HG. Composition and structure of active coke in the oxydehydrogenation of ethylbenzene. Appl Catal 1987;29(2):311-26.
[16] Bagnasco G, Ciambelli P, Turco M, La Ginestra A, Patron P. Layered zirconium-
tin phosphates. 2. Catalytic properties in the oxydehydrogenation of ethylbenzene to styrene. Appl Catal 1991;68(1-2):69-79.
[17] Arrúa LA, Ardissone DE, Quiroga OD, Rivarola JB. Oxidehydrogenation of
ethylbenzene on P-O-Ni catalyst. React Kinet Catal Lett 1995;56(2):383-9. [18] Dziewiecki Z, Jagiello M, Makowski A. Investigation of polymer organic deposit
formed on nickel phosphate in oxidative dehydrogenation of ethylbenzene.
React Funct Polym 1997;33(2-3):185-91. [19] Vrieland GE, Friedli HR. Oxydehydrogenation of ethyl benzene. US patent
3923916, 1975.
[20] Hofmann H, Emig G, Ruppert W. Process for preparing styrene, and an appropriate agent. US patent 4400568, 1983.
[21] Tagiyev DB, Gasimov GO, Rustamov MI. Carbon deposits on the surface of CaO/SiO2 as active catalysts for the oxidative dehydrogenation of ethylbenzene. Catal Today 2005;102-103:197-202.
[22] Lisovskii AE, Aharoni C. Carbonaceous deposits as catalysts for oxydehydrogenation of alkylbenzenes. Catal Rev Sci Eng 1994;36(1):25-74, and references therein.
[23] Alkhazov TG, Lisovskii AE, Guiakhmedova TKh. Oxidative dehydrogenation of ethylbenzene over a charcoal catalyst. React Kinet Catal Lett 1979;12(2):189-93.
[24] Grunewald GC, Drago RS. Oxidative dehydrogenation of ethylbenzene to styrene over carbon-based catalysts. J Mol Catal 1990;58(2):227-33.
[25] Pereira MFR, Orfão JJM, Figueiredo JL. Oxidative dehydrogenation of
ethylbenzene on activated carbon catalysts I. Influence of surface chemical groups. Appl Catal A-Gen 1999;184(1):153-60.
[26] Pereira MFR, Orfão JJM, Figueiredo JL. Oxidative dehydrogenation of
[28] Pereira MFR, Orfão JJM, Figueiredo JL. Influence of the textural properties of an
activated carbon catalyst on the oxidative dehydrogenation of ethylbenzene. Colloid Surface A 2004;241:165-71.
[29] Guerrero-Ruiz A, Rodriguez-Reinoso F. Oxydehydrogenation of ethylbenzene to
styrene catalyzed by graphites and activated carbons. Carbon 1994;32(1):23-9.
[30] Mestl G, Maksimova NI, Keller N, Roddatis VV, Schlögl R. Carbon nanofilaments
in heterogeneous catalysis: An industrial application for new carbon materials? Angew Chem Int Ed 2001;40(11):2066-8.
[31] Pereira MFR, Orfão JJM, Figueiredo JL. Oxidative dehydrogenation of
ethylbenzene on activated carbon fibers. Carbon 2002;40(13):2393-401. [32] Delgado JJ, Vieira R, Rebmann G, Su DS, Keller N, Ledoux MJ, et al. Supported
carbon nanofibers for the fixed-bed synthesis of styrene. Carbon
2006;44(4):809-12.
Chapter 4
98
[33] Delgado JJ, Su DS, Rebmann G, Keller N, Gajović A, Schlögl R. Immobilized
carbon nanofibers as industrial catalyst for ODH reactions. J Catal 2006;244(1):126-9.
[34] Li P, Li T, Zhou JH, Sui ZJ, Dai YC, Yuan WK, et al. Synthesis of carbon
nanofiber/graphite-felt composite as a catalyst. Micropor Mesopor Mat 2006;95(1-3):1-7.
[35] Zhao TJ, Sun WZ, Gu XY, Rønning M, Chen D, Dai YC, et al. Rational design of
the carbon nanofiber catalysts for oxidative dehydrogenation of ethylbenzene. Appl Catal A-Gen 2007;323:135-46.
[36] Keller N, Maksimova NI, Roddatis VV, Schur M, Mestl G, Butenko YV, et al. The
catalytic use of onion-like carbon materials for styrene synthesis by oxidative dehydrogenation of ethylbenzene. Angew Chem Int Ed 2002;41(11):1885.
[37] Su D, Maksimova NI, Mestl G, Kuznetsov VL, Keller V, Schlögl R, et al. Oxidative dehydrogenation of ethylbenzene to styrene over ultra-dispersed diamond and onion-like carbon. Carbon 2007;45(11):2145-51.
[38] Pereira MFR, Figueiredo JL, Orfão JJM, Serp P, Kalck P, Kihn Y. Catalytic activity of carbon nanotubes in the oxidative dehydrogenation of ethylbenzene. Carbon 2004;42(14):2807-13.
[39] Su DS, Maksimova N, Delgado JJ, Keller N, Mestl G, Ledoux MJ, et al. Nanocarbons in selective oxidative dehydrogenation reaction. Catal Today 2005;102-103:110-4.
[40] Zhang J, Su D, Zhang A, Wang D, Schlögl R, Hebert C. Nanocarbon as robust catalyst: Mechanistic insight into carbon-mediated catalysis. Angew Chem Int Ed 2007;46:7319-23.
[41] Nigrovski B, Scholz P, Krech T, Qui NV, Pollok K, Keller T, et al. The influence of microwave heating on the texture and catalytic properties of oxidized multi-walled carbon nanotubes. Catal Commun 2009;10(11):1473-7.
[42] Qui NV, Scholz P, Krech T, Keller TF, Pollok K, Ondruschka B. Multiwalled carbon nanotubes oxidized by UV/H2O2 as catalyst for oxidative dehydrogenation of ethylbenzene. Catal Commun 2011;12(6):464-9.
[43] Zhang J, Su DS, Blume R, Schlögl R, Wang R, Yang X, et al. Surface chemistry and catalytic reactivity of a nanodiamond in the steam-free dehydrogenation of ethylbenzene. Angew Chem Int Ed 2010;49(46):8640-4.
[44] Delgado JJ, Chen X, Tessonnier JP, Schuster ME, Del Rio E, Schlögl R, et al. Influence of the microstructure of carbon nanotubes on the oxidative dehydrogenation of ethylbenzene to styrene. Catal Today 2010;150(1-2):49-
54. [45] Zarubina V, Talebi H, Nederlof C, Kapteijn F, Makkee M, Melián-Cabrera I, On
the stability of conventional and nano-structured carbon-based catalysts in the
oxidative dehydrogenation of ethylbenzene under industrially relevant conditions. Carbon 2014, 77 329-40 or 10.1016/j.carbon.2014.05.036.
[46] McKee DW. Effect of adsorbed phosphorus oxychloride on oxidation behavior of
graphite. Carbon 1972;10:491-7. [47] McKee DW, Spiro CL, Lamby EJ. The inhibition of graphite oxidation by
phosphorous additives. Carbon 1984;22(3):285-90.
[48] Lee YJ, Radovic LR. Oxidation inhibition effects of phosphorus and boron in different carbon fabrics. Carbon 2003;41(10):1987-97.
[49] Frank B, Rinaldi A, Blume R, Schlögl R, Su DS. Oxidation stability of multiwalled
carbon nanotubes for catalytic applications. Chem Mater 2010;22(15):4462-70. [50] Frank B, Zhang J, Blume R, Schlögl R, Su DS. Heteroatoms increase the
selectivity in oxidative dehydrogenation reactions on nanocarbons. Angew
Chem Int Ed 2009;48(37):6913-7. [51] Innes WB. Total porosity and particle density of fluid catalysts by liquid
titration. Anal Chem 1956;28(3):332-4.
Phosphorous-induced thermal stabilization for carbon-supported SiO2 catalysts in the oxidative dehydrogenation of ethylbenzene to styrene.
99
[52] Eckert H, Levendis YA, Flagant RC. Glassy carbons form poly (furfuryl alcohol) copolymers: structural studies by high-resolution solid state NMR techniques. J Phys Chem 1988;92(17):5011-9.
[53] Lafyatis DS, Tung J, Foley HC. Poly (furfuryl alcohol)-derived carbon molecular sieves: dependence of adsorptive properties on carbonization temperature, time, and poly (ethylene glycol) additives. Ind Eng Chem Res 1991;30(5):865-
73. [54] Shindo A, Izumino K. Structural variation during pyrolysis of furfuryl alcohol and
[60] Wang H, Yao J. Use of poly (furfuryl alcohol) in the fabrication of nanostructured carbons and nanocomposites. Ind Eng Chem Res 2006;45(19):6393-404.
Preparation of microporous carbon fibers through carbonization of Al-based porous coordination polymer (Al-PCP) with furfuryl alcohol. Chem Mater 2011;23(5):1225-31.
[62] Kruk M, Jaroniec M, Ryoo R, Joo SH. Characterization of ordered mesoporous carbons synthesized using MCM-48 silicas as templates. J Phys Chem B 2000;104(33):7960-8.
[63] Kruk M, Jaroniec M, Kim TW, Ryoo R. Synthesis and characterization of hexagonally ordered carbon nanopipes. Chem Mater 2003;15(14):2815-23.
[70] Fiedorow R, Przystajko W, Sopa M, Dalla Lana IG. The nature and catalytic influence of coke formed on alumina: oxydative dehydrogenation of ethylbenzene. J Catal 1981;68(1):33-41.
[71] Wang Z, Lu Z, Huang X, Xue R, Chen L. Chemical and crystalline structure characte-rizations of polyfurfuryl alcohol pyrolyzed at 600 degrees C. Carbon 1998;36(1-2):51-9.
[72] Bertarione S, Bonino F, Cesano F, Damin A, Scarano D, Zecchina A. Furfuryl
alcohol polymerization in H-Y confined spaces: Reaction mechanism and structure of carbocationic intermediates. J Phys Chem B 2008;112(9):2580-9.
[73] Sing KSW, Everett DH, Haul RAW, Moscou L, Pierotti R, Rouquerol J, et al.
Reporting physisorption data for gas solid systems with special reference to the determination of surface area and porosity (recommendations 1984). Pure Appl Chem 1985;57(4):603-19.
[74] Kruk M, Jaroniec M. Gas adsorption characterization of ordered organic-inorganic nanocomposite materials. Chem Mater 2001;13(10):3169-83.
Phosphorous-induced thermal stabilization for carbon-supported SiO2 catalysts in the oxidative dehydrogenation of ethylbenzene to styrene.
101
Chapter 4
102
Figure S-1. TGA patterns of the fresh hybrids: 3P/15C/AS and 6P/15C/AS. Conditions: 100 ml.min-1, synthetic air, heating rate of 10 °C.min-1.
Table S-1. Chemical composition of the organic material after dissolution of the
silica matrix and “POx”-removal by HF treatment.
Sample C
(wt.%)
H
(wt.%)
O
(wt.%)
Compositional
formula (at.)
C:H
(at.)
O:C
(at.)
3P/15C/AS 83.1 1.6 15.3 C7.23H1.62O 4.57 0.14
3P/15C/AS spent
78.6 2.1 19.4 C5.39H1.68O 3.22 0.19
6P/15C/AS 79.9 1.4 18.6 C5.73H1.21O 4.73 0.17
6P/15C/AS spent
77.6 2.1 20.3 C5.10H1.63O 3.13 0.20
200 400 600 800 100070
75
80
85
90
95
100
-0.15
-0.10
-0.05
0.00
0.05
0.10
3P/15C/AS
6P/15C/AS
rela
tive w
eig
ht / %
Temperature / oC
derivative / %
(oC
)-1
derivative
19.1% 17.2%
Phosphorous-induced thermal stabilization for carbon-supported SiO2 catalysts in the oxidative dehydrogenation of ethylbenzene to styrene.
103
Figure S-2. Modified Van Krevelen plots representing the composition of the hybrids ( ): (3P/15C/AS and 6P/15C/AS), fresh and spent, in comparison with prior work, adapted from [,70] and [,22]. Raw data are given in Table S-1.
0.0 0.1 0.2 0.3 0.4 0.50.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
6P/15C/AS spent3P/15C/AS spent
6P/15C/AS
3P/15C/AS
atomic O:C
ato
mic
C:H
1.0Na
0.5Na
0.3Na
0.1Na
0.02K
0.01K0.005K0.0025K
Al2O
3
Al2O
3
0.05B
0.1B
Figure S-3. Raman spectra for the fresh and spent 3P/15C/AS and 6P/15C/AS catalysts.
2000 1800 1600 1400 1200 1000 800
D
3P/15C/AS fresh 3P/15C/AS spent
6P/15C/AS fresh 6P/15C/AS spent
Inte
nsity
/ a
.u.
Wavenumber / cm-1
G
Chapter 4
104
Table S-2. Micropore area and volume derived from the t-plot method from N2 adsorption volumetric data at 196 oC.
a. Values in parenthesis are the total BET surface area and (b) total pore volume. c. BD: the micropore area is not reported because either the micropore volume is negative or below the detection limit (<0.001 cm3.g-1).
Table S-3. The apparent activation energies of the combustion of fresh catalysts.
Catalyst Ea at 80% conversion, kJ.mol-1
15C/AS 116 3P/15C/AS 131 6P/15C/AS 129
MWCNT 144
Phosphorous-induced thermal stabilization for carbon-supported SiO2 catalysts in the oxidative dehydrogenation of ethylbenzene to styrene.
105
Figure S-4. TGA patterns of the spent catalysts: a) 3P/15C/AS; b) 6P/15C/AS; c) 3P/AS, and d) 6P/AS. Conditions: 100 ml.min-1, synthetic air, heating rate of 10 °C.min-1.
Table S-5. TGA weight loss at 200-900 oC from the patterns of
the spent catalysts from Fig. S-4.
Sample TGA weight loss 200-900 oC (wt.%)
3P/15C/AS 42.8 6P/15C/AS 40.5 3P/AS 42.9
6P/AS 13.7
Table S-6. Raman characterization of the fresh and spent catalysts.
Sample ID/IG
3P/15C/AS 1.58 3P/15C/AS
spent 0.99 (-37%, respect to fresh)
6P/15C/AS 1.62 6P/15C/AS
spent 1.18 (-27%, respect to fresh)
200 400 600 800 100050
60
70
80
90
100
-0.8
-0.4
0.0
0.4
0.8 a
b
c
d
rela
tive
we
igh
t
/
%
Temperature / oC
de
riva
tive
/
%(o
C)-1
Chapter 4
106
Figure S-5. Top) Nitrogen sorption isotherms at 196 oC of the spent and fresh catalysts
3P/15C/AS and 6P/15C/AS. Inset: BJH pore size distribution. Bottom) Zoom-in of the spent catalysts’ isotherms. The arrow indicates the low-pressure closure point at ca. 0.45 p/po in the
Phosphorous-induced thermal stabilization for carbon-supported SiO2 catalysts in the oxidative dehydrogenation of ethylbenzene to styrene.
107
Figure S-6. Top) Nitrogen sorption isotherms at 196 oC of the spent and fresh catalysts 3P/AS
and 6P/AS. Inset: BJH pore size distribution. Bottom) Zoom-in of the spent catalysts’ isotherms. The arrow indicates the low-pressure closure point at ca. 0.45 p/po in the desorption step. Solid
Oxidative dehydrogenation of ethylbenzene to styrene over MWCNT. Regeneration: myth or reality?
Selectivity and activity are two important aspects in the oxidative dehydrogenation (ODH) of
ethylbenzene (EB) to styrene. Besides those attributes, it is desirable that a catalyst is stable, or at
least a regenerable, that can be used for a number of runs. Inorganic-based catalysts are easier to
be regenerated by coking as the thermal stability is superior to carbon-based counterparts, though
changes in texture and active sites can occur. These aspects imply that the material will have a
limited life time, after that it must be recycled by a fresh batch. Despite the practical importance of
stability, it is hardly found any study on regeneration of EB ODH-based catalysts.
In this work, two representative ODH-based catalysts (-Al2O3 and MWCNTs) have been studied
regarding their regeneration towards coke removal by calcination. Before regeneration, the
catalysts were exposed to EB ODH reaction conditions. Various regeneration strategies were
investigated and the derived materials’ properties were characterized by physico-chemical
methods, and those properties are compared to the fresh counterparts.
Chapter 5
110
5.1 Introduction
Direct dehydrogenation of ethylbenzene (EB) in the presence of steam is the main way to produce styrene in industry. About 85% of styrene (ST) is produced by this way over a K-promoted Fe2O3 catalyst which is highly selective to styrene and stable
[1,2]. The process takes place at temperatures as high as 580-630 °C [1], due to the endothermic thermodynamics and therefore suffers from high energy consumption. The conversion per pass is low due to the equilibrium limitations,
involving a large recycle, and this has been the major driver to develop alternative ST production routes.
One of the aspects which can be improved is reducing the large amount of
steam in the direct dehydrogenation reaction. Many efforts have been put forward to improve the process by making it steam-free and applying lower temperatures. The use of oxidants such as O2 [3,4] is very promising as it helps to shift the reaction by
H2O formation. Therefore the oxidative dehydrogenations (ODH) have been thoroughly investigated, but it is not commercialized yet due to the limited catalyst stability and low selectivity to styrene.
The O2-based oxidative pathway was originally studied on oxide-based carriers such as alumina [5]. It was shown that the coke deposited on the alumina’s surface under the oxidative conditions promotes the activity and selectivity [6-9] due to
presence of Lewis acid sites. It was further proposed [10-13], and recently supported [14-16], that the formed coke acts as a selective and active site. Such type of coke has relatively high oxygen content (O:C=0.10-0.15 at.) and is considered to be
polyaromatic. Another family of catalysts that drew significant attention for the O2-based EB
oxidative dehydrogenation to styrene are carbon-based materials. Many types of
carbon-based structures have been claimed to be active and selective, including activated carbons [17-30], onion-like carbons [31-33], carbon nanofibers [34-41], carbon nanofilaments [34], diamonds [28,30,33], graphite’s [19,34,38,42], MWCNT
[28,30,40,32,33,42-49] , and other types of carbons [50-52]. It is known that some of those materials, e.g. activated carbons, gasify/decompose under the reaction conditions [22,23,28,45]. The most stable ones are MWCNT. In a previous study, it
was shown that it has a slow catalyst deactivation and good thermal stability under harsh conditions with the deposition of reaction-based coke [53]. Thus, MWCNT is a promising catalyst for the alternative ODH process.
Besides of activity and selectivity, it is desirable that a catalyst is stable, or at least a regenerable, that can be used for a number of runs. Inorganic-based
catalysts are easier to be regenerated by coking as the thermal stability is high, though changes in texture and active sites can occur. These aspects imply that the material will have a life time, after that it must be recycled by a fresh batch. Despite
the practical importance of stability, it is hardly found any study on regeneration of EB ODH-based catalysts.
In this work, two representative spent ODH-based catalysts (-Al2O3 and
MWCNTs) have been studied regarding their regeneration towards coke removal by
calcination. Before regeneration, the catalysts were tested under EB ODH conditions. Various regeneration strategies were investigated and the derived materials were characterized by physico-chemical methods and the properties are compared to the
fresh counterparts.
Oxidative dehydrogenation of ethylbenzene to styrene over MWCNT. Regeneration: myth or reality?
111
5.2 Experimental methods
5.2.1 Materials
In this study a low SiO2-stabilized -Al2O3 extrudates (Albemarle Catalysts BV,
denoted as GA-F) and purified MWCNT (Hyperion CS-02C-063-XD, denoted as MW-F) were employed. The extrudates were crushed and sieved into a 212-425 μm fraction
used for the catalytic tests and characterization
5.2.2 Regeneration
Two regeneration protocols were carried out on the spent materials, mild and conventional. The mild regeneration (denoted as MR) was carried out in a quartz-
tube housed tubular oven (Nabertherm RT 50/250-11). The sample was loaded in a flat quartz-based crucible that was placed horizontally in the centre of the furnace’s isothermal heating zone. The spent MWCNT (denoted as MW-S) and spent -alumina
(GA-S) were regenerated at temperature at 450 °C in 1% vol. O2/Ar at a heating rate of 3 °C.min-1 and held for 2, 5, 8, and 24 h for MWCNT. In the case of -alumina
5 and 24 h were applied. The conventional regeneration (denoted as R) was applied
on the spent -alumina; this was done in the in LT9/11 Nabertherm box furnace at
450 °C in air at a heating rate of 3 °C.min-1 and held for 5 h. Fresh, spent and regenerated materials with their codes and treatments are given in Table 1.
Table 1. Materials and their treatments
Sample code a Material Treatment
GA-F -Al2O3 as-received
GA-S -Al2O3 spent after reaction test, 60 h TOS
GA-R5 -Al2O3 spent regeneration in air, 5 h
GA-MR5 -Al2O3 spent mild regeneration in 1 %O2/Ar, 5 h
GA-MR24 -Al2O3 spent mild regeneration in 1 %O2/Ar, 24 h
MW-F MWCNT as-received
MW-S MWCNT spent after reaction test, 60h TOS MW-MR2 MWCNT spent mild regeneration in 1%O2/Ar, 2 h
MW-MR5 MWCNT spent mild regeneration in 1%O2/Ar, 5 h MW-MR8 MWCNT spent mild regeneration in 1%O2/Ar, 8 h MW-MR24 MWCNT spent mild regeneration in 1%O2/Ar, 24 h
a. The suffix is related to the treatment: F (fresh), S (spent after the reaction cycle); R means conventional regeneration and MR mild regeneration; the number indicates the number of hours after reaching 450 °C.
5.2.3 Catalysts characterization
The organic content of the fresh and spent catalysts was quantified by thermogravimetric analysis (TGA) on a Mettler-Toledo analyzer (TGA/SDTA851e).
The weight loss was monitored for a temperature program from 30 to 1000 °C at a heating rate of 10 °C/min using a flow of synthetic air of 100 mL.min-1 NTP. Blank curve subtraction using an empty crucible was taken into account. The oxidation rate
patterns (TPO) were obtained in the same instrument making use of the TGA derivative patterns.
Chapter 5
112
The textural properties of the catalysts and the bare supports were analyzed
by N2-physisorption at -196°C using a Micromeritics ASAP 2420. Fresh and regenerated -alumina samples were degassed at 300 oC for 10 h under vacuum.
Spent samples were degassed at 200 °C for 10 h to ensure that the coke deposited
on -alumina and carbon of MWCNT are not altered during the degassing. The
surface area (SBET) was calculated with the standard BET method [54] in the relative pressure range 0-0.25. The pore volume (VT) was calculated using the single point
total desorption pore volume at the relative pressure 0.98. Pore size distributions were calculated using the BJH-model [55].
Raman spectra were obtained with 785 nm excitation line, 30 mW on the
Perkin Elmer Ramanstation 400 spectrometer.
5.2.4 Catalytic tests
The catalytic tests were performed in a setup with six parallel quartz fixed bed reactors (inner diameter 4 mm) in down-flow operation. The reactors were loaded from top to bottom with a quartz wool plug, 10 cm glass beads (0.5 mm diameter),
and 65 mm catalyst bed (0.80 ml) to ensure that the catalyst bed is located in the isothermal zone of the furnace. The glass beads have limited conversion which is less than 3% EB conversion under all applied conditions.
Each reactor gas feed has a flow of 36 ml/min (NTP) and consist of a mixture of nitrogen, oxygen, and ethylbenzene. A liquid ethylbenzene flow of 1 g.h-1 evaporates (3.6 ml.min-1 vapour at NTP) resulting in the 1:10 volume ratio of
ethylbenzene and gas (10 vol.% EB) with a GHSV of 3000 l/l/h. The EB liquid evaporates in a -Al2O3 filled tube in a synchronized flow with the gas feed. Pressure
in the reactor system was 1.2-1.3 bars and an atmospheric outlet pressure drop was
typically 0.2-0.3 bars. The reactor outlet flows were analyzed using an online two channel gas
chromatograph with a TCD (columns: 0.3m Hayesep Q 80-100 mesh with back-flush,
25m×0.53mm Porabond Q, 15 m × 0.53mm molsieve 5Å) for permanent gasses analysis (CO2, H2, N2, O2, CO) and a FID column (30 m × 0.53 mm, Df = 3 mm, RTX-1) for hydrocarbon analysis (methane, ethane, ethene, benzene, toluene,
ethylbenzene, styrene, and heavy aromatics). The catalytic test was carried out at various temperatures (475, 450, 425, and 450 °C) and O2/EB = 0.6 and 0.2 (vol.). For all EB conversion data the oxygen conversion is 100%, unless otherwise is
stated. All characterizations for the spent catalysts were done after the complete testing cycle of 70 h.
5.2.5 Regeneration efficiency of the carbon-based materials
The regeneration efficiency of the carbon-based materials was defined via the
textural parameters (
, eq. 1) as the ratio between the final surface area,
corrected by the weight loss during regeneration, to the initial surface area, which
means the fraction of surface area that is recovered after regeneration or fraction that is lost, in case that this concept is applied to the spent catalyst:
[
]
(1)
where
is the BET surface area of the regenerated material,
is the
corresponding BET of the fresh sample and is the weight loss during
regeneration.
Oxidative dehydrogenation of ethylbenzene to styrene over MWCNT. Regeneration: myth or reality?
113
The total textural efficiency ( ) contains the weight loss during the
reaction:
[
] [
]
(2)
is the weight loss or gain during the ODH reaction. For MWCNT in the applied
catalyst run .
5.2.6 Regeneration efficiency of the alumina-based materials
This is defined as the corrected weight loss, relative to the bare alumina, from the TGA patterns as:
(
) (3)
where superscript ‘x’ refers to any material (i.e. spent, regenerated, fresh). The
recovery of the texture can be defined as:
(4)
5.3 Results
5.3.1 Catalytic tests
Figure 1 shows the catalyst performance of MWCNT (MW-F) and -alumina (GA-F)
catalysts under relevant conditions, meaning a high EB concentration and limited O2
with 20% excess compared to the stoichiometric amount. EB conversion, ST, COx selectivity, and ST yield are plotted as a function of the time on stream (TOS). MWCNT is a readily active and selective catalyst at 475 oC, but -alumina requires
approximately five hours to be active. However the conversion and selectivity of -
alumina are higher. At the second temperature range at 450 °C during 20-40 h TOS, both catalysts have identical performance. The lowest temperature of the cycle (425 °C) shows higher activity and selectivity of MWCNT in comparison with -alumina,
which makes it attractive for low-temperature ODH of EB to styrene in comparison with -alumina. At the end of the catalytic cycle, at 450 °C, conversion and
selectivity are identical, and in agreement with step 2, though the conversion and selectivity values are lower due to deactivation. So the comparison of two steps at
450 °C shows that both catalysts deactivate in 2% for MWCNT and 5% for -alumina.
Both catalysts demonstrate moderate stability, and MWCNT still have some advantages at low reaction temperature; higher activity and selectivity than alumina-
based catalyst. The selectivity to COx is inversely coupled to ST (Fig. 1-c). This is because the selectivity to benzene/toluene and heavy condensates are much lower than COx/ST and independent of the applied reaction conditions.
Chapter 5
114
Figure 1. Time on stream EB conversion (a), selectivity to ST (b), selectivity to COx (c), and ST yield (d) at various temperatures (475, 450, 425, and 450 oC) and O2/EB= 0.6 and 0.2
(vol.); GHSV of 3000 l/l/h; 10 vol. % EB.
5.3.2 Spent catalysts
Temperature programmed oxidation was performed on the fresh and spent -alumina
and MWCNT materials for the evaluation of the coke burning profiles (Fig. 2). The
TPO profile of fresh MWCNT shows the presence of two types of coke which start to decompose at 400 °C and ends at 635 °C, with a maximum at 560 °C. It indicates a high thermal stability of the fresh MWCNT (Figure 2-b). TPO for the spent -alumina
starts to decompose earlier at 400 °C, and it ends at 520 °C with a maximum at 460
°C (Figure 2-a) in a single step. Spent MWCNT has two contributions (I and II), which includes the multi-walled backbone (peak II), while the low-temperature peak
(I) comes from a new type of carbon-based material that is deposited during the reaction (Figure 2-b), and it is ascribed to ODH-based polyaromatic coke.
425 450450
0.6 0.2 0.6 0.60.20.6 0.2
475
0 10 20 30 40 50 60 700
10
20
30
40
50
60
GA-F
MW-F
EB
Co
nve
rsio
n /
%
Time / h
0 10 20 30 40 50 60 7010
20
30
40
50
60
70
80
90
100
GA-F
MW-F
Se
lectivity t
o S
T /
%
Time / h
425 450450
0.6 0.2 0.6 0.60.20.6 0.2
475
425 450450
0.6 0.2 0.6 0.60.20.6 0.2
475
0 10 20 30 40 50 60 700
10
20
30
40
50
60
70
80
90
100
GA-F
MW-F
Se
lectivity t
o C
Ox / %
Time / h
425 450450
0.6 0.2 0.6 0.60.20.6 0.2
475
0 10 20 30 40 50 60 700
10
20
30
40
50
60
GA-F
MW-FS
T Y
ield
/ %
Time / h
a b
c d
Oxidative dehydrogenation of ethylbenzene to styrene over MWCNT. Regeneration: myth or reality?
115
The coke nature was investigated by Raman spectroscopy. The comparison of Raman spectra between fresh and spent MWCNT are shown in the Figure 3. The spectrum of the raw MWCNT material reveals two broad absorptions centered at
1600 and 1310 cm-1, which are indicative of amorphous carbon (G and D peaks). Raman spectrum of the spent catalyst indicates the presence of a more ordered structure that has been formed during the reaction. The D/G peak intensity becomes
smaller and this can be explained by two possible effects: a) removal of the more amorphous carbon domains by gasification/burning and b) deposition of sp2-rich carbon materials, or both effects; the latter is due to the buildup of new polyaromatic
ODH coke.
Figure 2. Oxidation rate patterns (TPO) for the fresh and spent: a) -alumina and b)
MWCNTs. = (Wo-W)/Wo; where Wo is the initial weight. Conditions: synthetic air, 100
ml.min-1, heating rate of 3 °C.min-1.
300 400 500 600 700 8000,0
0,2
0,4
0,6
a GA-F
GA-S1
Temperature / oC
- d
/dT
/ (
%.
oC
-1)
300 400 500 600 700 8000,0
0,2
0,4
0,6
0,8
1,0
1,2
1,4
1,6
b MW-F
MW-S1
Temperature / oC
- d
/dT
/
(%
. oC
-1)
a
b
I
II
Chapter 5
116
Figure 3. Raman spectra for the fresh (bottom) and spent (top) MWCNTs.
The textural properties of the fresh and spent -alumina (Fig. 4) and MWCNT
(Fig. 5) samples were investigated by N2-physisorption analysis. The isotherm of fresh -alumina has a type IV with hysteresis H1 [56]. This type represents
cylindrical pore geometry of solid particles with pore size equability and superficial
pore connectivity. The hysteresis changes to H2 type for the spent -alumina (Fig. 4)
with a closure point at 0.45 relative pressure. Hysteresis H2 [56] occurs in solids, where the pores have narrow necks and wide bodies or when the porous material has interconnected pores. The original material has no interconnectivity at 0.45
relative pressure, thus the pore neck restrictions of the spent sample is due to the excessive coke build-up. Pore size distribution curves of spent samples shift slightly
towards low pore sizes that also indicates pore blockage. The specific surface area is depleted from 272 (GA-F) to 152-154 m2.g-1 (spent, Table 2) which is ~44% reduction. The spent sample was measured twice to ensure that sample is
homogeneous; this was in agreement with TGA in terms of identical DTGA peak positions (Fig. 6-a, b) and residual weight losses (Table 2).
2000 1800 1600 1400 1200 1000 800
x5
D MWCNT fresh
MWCNT spent
Inte
nsity / a
.u.
Wavenumber / cm-1
G
Oxidative dehydrogenation of ethylbenzene to styrene over MWCNT. Regeneration: myth or reality?
117
Figure 4. Nitrogen sorption isotherms at -196 oC for the fresh and spent Alumina samples. Inset: BJH pore size distribution.
0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1,00
50
100
150
200
250
300
350
400
450
10 20 30 400,0
0,5
1,0
1,5
2,0
Pore width / nm
GA-F
GA-S1
GA-S2
Qua
ntity
Adsorb
ed (
cm
³/g
ST
P)
Relative Pressure (p/p°)
Figure 5. Nitrogen sorption isotherms at -196 oC for the fresh and spent MWCNT materials.
Inset: BJH pore size distribution.
0,5 0,6 0,7 0,8 0,9 1,00
200
400
600
800
0 20 40 60 80 100 120 140 1600,0
0,5
1,0
1,5
Pore width / nm
MW-F
MW-S1
MW-S2
Qua
ntity
Adsorb
ed (
cm
³/g
ST
P)
Relative Pressure (p/p°)
Chapter 5
118
Figure 6. TGA curves of two spent -alumina (top) and two spent MWCNT (bottom). Conditions:
synthetic air, 100 ml.min-1, heating rate of 3 °C.min-1. A) GA-S1; b) GA-S2; c) MW-S1; d) MW-S2. Peak I: ODH-based coke; peak II: MWCNT backbone.
Table 2. Texture properties of fresh, spent, and regenerated -alumina samples.
GA-MR5 5.8 0.93 b 160 (59) 0.341 (53) 0.59 GA-MR24 3.3 1.02 b 256 (94) 0.650 (102) 0.94
a. Determined by TGA weight loss between 200-800 oC; b. Spent GA-S1 was employed for the
regeneration study; c. between parentheses the percentage of recovery is given, relative to the fresh counterpart.
Figure 5 shows the isotherms and pore size distribution of fresh and spent MWCNT. Fresh MWCNT has an isotherm type IV with hysteresis H1, which characterizes cylindrical pore geometry and relatively high pore size uniformity. Pore
size distribution of the fresh MWCNT shows the presence of two peaks (Fig. 5, inset). The broad peak with a maximum of 30 nm is related to inter-rod mesopores and the small peak at 5 nm that can be related to smaller mesopores, likely from internal
open tubes. Spent samples show much less adsorption capacity, due to the excessive coke build-up. In contrast to -alumina, the shape of the isotherm and hysteresis
remain identical, of type IV H1, while the small peak at 5 nm pore size disappears. It
200 400 600 8000
20
40
60
80
100
-0,4
-0,2
0,0
0,2
0,4
re
lative w
eig
ht / %
Temperature / oC
derivative
/ %
(oC
)-1
derivative
200 400 600 8000
20
40
60
80
100
-0,4
-0,2
0,0
0,2
0,4
rela
tive w
eig
ht / %
Temperature / oC
derivative
/ %
(oC
)-1
derivative
200 400 600 8000
20
40
60
80
100
-1,0
-0,5
0,0
0,5
rela
tive
we
igh
t
/
%
Temperature / oC
de
riva
tive
/
%(o
C)-1
derivative
200 400 600 8000
20
40
60
80
100
-1,0
-0,5
0,0
0,5
rela
tive w
eig
ht / %
Temperature / oC
derivative
/ %
(oC
)-1derivative
a b
dc
I II I II
Oxidative dehydrogenation of ethylbenzene to styrene over MWCNT. Regeneration: myth or reality?
119
all indicates that the mesopores are blocked by the formed ODH coke during the reaction; however there is no indication of pore restrictions, due to the absence of low closure point. A duplo of the spent MWCNT was measured to account for sample
homogeneity. The homogeneity of the spent MWCNT is lower than for -alumina; the
BET varied in 7% while the pore volume in 16% (Table 3). This can be related to differences in ODH coke deposition along the reactor. Sample MW-S1 was employed
for the regeneration studies. TGA patterns (Fig. 6) show two combustion steps; the low-temperature process (type I) is due to the ODH-coke, while the higher temperature step originates from the burning of the MWCNT backbone, as discussed
in Fig. 2-b. The comparison of the DTGA curves indicates that MW-S2 has a slightly higher fraction of the ODH coke than MW-S1, which is consistent with the lower BET and pore volume (Table 3).
Table 3. Texture properties of fresh, spent, and regenerated MWCNT samples.
Material SBET (m2.g-1)a VT (cm3.g-1)a Wloss (%) (-)b
MW-MR24 502 (124) 1.239 (112) 58 0.52 1.24 a. Between parentheses the percentage of recovery is given, relative to the fresh counterpart; b. In parenthesis, the regeneration efficiency of the pore volume is given.
5.3.3 Regeneration studies
For the regeneration of MWCNT, different regeneration times under mild conditions
were applied to evaluate the recovery of the textural properties similar to the original material. Figure 7 shows a comparison between various regenerated MWCNT together with the fresh and spent materials. It is visible from the graph that MW-
MR2 sample has higher adsorption capacity, a trend that continues with MR5 and MR8. Thus, increasing the regeneration time at the applied oxygen partial pressure, gives possibilities to reach the textural properties of the original material.
Remarkably, 24 h of regeneration (MW-MR24) produces a material with better textural properties than the fresh MWCNT. This is also seen with a more intense PSD
than fresh MWCNT, while the PSD maximum shifts towards higher sizes. In general, the pore size distribution after regeneration increases with increasing the regeneration time and the small peak at 5 nm appears back for the regenerated
samples, except of MW-MR2. So, it demonstrates that 2 h is not enough to perform a good regeneration of the MWCNT. The derived textural properties are presented in the Table 3. It can be seen from the table that the spent MWCNT samples have 0.3
times its original surface area and ca. 0.4 times of the original pore volume. The BET of the regenerated increases notably and at 5 h (MR5) the BET surface area surpasses that for the raw material up to a factor 1.24 for the 24 h-regenerated
sample. The pore volume is also recovered, but it does not arrive to the original MWCNT’s quantity, with the exception of the 24 h that is a factor 1.12. The situation with the MR24 can be attributed to the fact that the regeneration can also clean the
MWCNT backbone by removing amorphous material; therefore the final material is more porous. Thus, from this study, it is concluded that both, BET and pore volume can be recovered after a mild regeneration. To make the comparison complete, the
weight loss due to the regeneration itself must be considered. This is considered in
Chapter 5
120
equation 1, where the regeneration efficiency based on the BET surface area and
weight loss is defined. When using this parameter, an optimal value is found. The regeneration efficiency increases from 0.77 (MR2) with a maximum of 0.88 at 5 h regeneration, and then decreases for 8 and 24 h. This information indicates that
because of the weight loss, the surface area cannot come back to the original value due to the weight loss; it also concludes that the optimal conditions in terms of BET surface area are 5 h. A second factor that affects the overall regeneration efficiency
is the weight loss associated to the ODH reaction itself; the total regeneration efficiency is defined in eq. 2 that accounts for the reaction weight loss. The consequence of this is the lower regeneration efficiency with a maximum of 0.70 for
MR5. This means that under the optimal conditions of regeneration, 30% of the total surface area is lost due to the reaction plus regeneration.
Figure 8 shows the oxidation rate patterns for the fresh, spent, and mild-regenerated samples for the MWCNT. It is clear from the figure that the peak of the spent catalyst at ca. 520 oC disappears with the increasing regeneration time. The
first peak of the spent catalyst disappears completely after regeneration for 24 h. The maximum of the regenerated samples’ peak shifts to 580 oC. It demonstrates that the coke which was formed on the MWCNT is removed upon regeneration, while
the MWCNT structure remains unmodified with improved thermal stability. The regeneration was also studied regarding the -alumina after the ODH
reaction under identical conditions. It is expected that -alumina can be easily
regenerated. The isotherms and pore size distribution curves of GA-R5 and GA-MR24
are nearly identical to the fresh -alumina (Fig. 9), meanwhile the isotherm and PSD
for GA-MR5 show less adsorption due to the incomplete regeneration. The TGA-based regeneration efficiency for the MR5 was about 93% but this 7% left residue
impacted on the BET with 41% lower than the fresh counterpart (Table 3). However, the regeneration in air for 5 h or 24 h in air gives rise to 95 and 94% recovery of BET surface area and approximately 100% pore volume’s recovery, with no residual
coke (Table 3).
Figure 7. Nitrogen sorption isotherms at -196 oC for the fresh, spent, and regenerated MWCNT at different conditions. See Table 1 for description of the treatment conditions. Inset: BJH pore size distribution.
0,5 0,6 0,7 0,8 0,9 1,00
200
400
600
800
1000
1200
0 20 40 60 80 100 120 140 1600,0
0,5
1,0
1,5
2,0
Pore width / nm
MW-F
MW-S1
MW-MR2
MW-MR5
MW-MR8
MW-MR24
Qua
ntity
Adsorb
ed (
cm
³/g
ST
P)
Relative Pressure (p/p°)
Oxidative dehydrogenation of ethylbenzene to styrene over MWCNT. Regeneration: myth or reality?
121
Figure 9. Nitrogen sorption isotherms at -196 oC for the fresh, spent, and mild regenerated -alumina. Inset: BJH pore size distribution.
5.4 Conclusions
MWCNT and -alumina show similar performance in the oxidative dehydrogenation of
ethylbenzene to styrene and demonstrate good stability under the applied
conditions; the selectivity is however relatively lower compared to the conventional process. Both types of materials suffer from coke deposition, under the reaction conditions, that depletes the textural features notably. -alumina can be fully
regenerated by calcination in air, or longer times under milder conditions, with the
0,5 0,6 0,7 0,8 0,9 1,00
50
100
150
200
250
300
350
400
450
500
0 10 20 300,0
0,5
1,0
1,5
2,0
Pore width / nm
GA-F
GA-S1
GA-R5
GA-MR5
GA-MR24
Qua
ntity
Adsorb
ed (
cm
³/g
ST
P)
Relative Pressure (p/p°)
Figure 8. Oxidation rate patterns (TPO) for the fresh, spent, and mild-regenerated samples for the MWCNT. = (Wo-W)/Wo; where Wo is the initial weight. Conditions: synthetic air, 100
ml.min-1, heating rate of 3 °C.min-1.
300 400 500 600 700 8000,0
0,2
0,4
0,6
0,8
1,0
1,2
1,4
1,6
MW-F
MW-S1
MW-MR2
MW-MR5
MW-MR24
Temperature / oC
- d
/dT
/ (
%.
oC
-1)
Chapter 5
122
complete recovery of its textural features and no residual coke. MWCNT can also be
regenerated under mild conditions. This was possible because the deposited ODH coke has a lower reactivity, i.e. appears in the TPO profiles at lower temperatures, than the MWCNT backbone. In order to quantify the regeneration efficiency, the
weight loss during the regeneration and reaction conditions have been taken into account. At the optimal conditions, 70% of its initial surface area can be recovered. It was also found that the specific surface area, for prolonged regeneration times,
can be even higher than the starting material. This was explained by the cleaning effect of the MWCNT backbone from amorphous domains, at the expense of a substantial weight loss. Therefore, MWCNT can not be considered a single-use
material, but can be partially recycled.
5.5 References
[1] Meyers RA. Handbook of petrochemicals production processes. New York NY:
McGraw-Hill. 2005: 11.3-11.34. [2] Cavani F, Trifiro F. Alternative processes for the production of styrene. Appl
Catal A-Gen 1995;133(2):219-39.
[3] Bogdanova OK, Belomestnykh IP, Voikina NV, Balandin AA. The oxidative dehydrogenation of ethylbenzene to styrene. Petrol Chem 1967;7(3):186-90.
[4] Raitmeier RE, Meyfield FD. Catalytic dehydrogenation process and
compositions. US patent 3437703, 1969. [5] Lisovskii AE, Aharoni C. Carbonaceous deposits as catalysts for
oxydehydrogenation of alkylbenzenes. Catal. Rev. Sci. Eng. 1994;36(1):25-74,
and references therein. [6] Lisovskii AE, Alkhazov TG, Dadasheva AM, Feizullaeva SA, Oxidative
dehydrogenation of alkylaromatic hydrocarbons on alumina catalysts. 4. Effect
of alkalis on acid properties of alumina and its catalytic activity in oxidative dehydrogenation of ethyl benzene, Kinet Catal 1975;16(2):385-9.
of oxidational dehydrogenation of ethylbenzene on an aluminum oxide catalyst. Rus Chem Bul 1977;26(3):477-80.
[8] Iwasawa Y, Nobe H, Ogasawara S. Reaction mechanism for styrene synthesis
over polynaphthoquinone. J. Catal 1973;31(3):444-9. [9] Alkhazov TG, Lisovskii AE. Role of condensation products in oxidative
dehydrogenation process of ethylbenzene on aluminium-oxide catalyst. Kinet
Catal 1976;17(2):375-9. [10] Fiedorow R, Przystajko W, Sopa M, Dalla Lana IG. The nature and catalytic
influence of coke formed on alumina: Oxidative dehydrogenation of ethylbenzene J Catal 1981;68:33-41.
[11] Schraut A, Emig G, Sockel H-G. Composition and structure of active coke in the
oxydehydrogenation of ethylbenzene. Appl Catal 1987;29(2):311-26. [12] Cadus LE, Arrua LA, Gorriz OF, Rivarola JB. Action of activated coke as a
catalyst: Oxydehydrogenation of ethylbenzene to styrene. Ind Eng Chem Res
1988;27:2241-6. [13] Vrieland GE, Menon PG. Nature of the catalytically active carbonaceous sites for
the oxydehydrogenation of ethylbenzene to styrene: A brief review. Appl Catal
1991;77(1):1-8. [14] Nederlof C, Kapteijn F, Makkee M. Catalysed ethylbenzene dehydrogenation in
CO2 or N2 - Carbon deposits as the active phase. Appl Catal A-Gen
2012;417:163-73. [15] Nederlof C, Zarubina V, Melián-Cabrera IV, Heeres HJ, Kapteijn F, Makkee M.
Oxidative dehydrogenation of ethylbenzene to styrene over alumina: effect of
calcination. Catal Sci Technol 2013; 3(2):519-26.
Oxidative dehydrogenation of ethylbenzene to styrene over MWCNT. Regeneration: myth or reality?
123
[16] Zarubina V, Nederlof C, Van der Linden B, Kapteijn F, Heeres HJ, Makkee M, Melián-Cabrera I. Making coke a more efficient catalyst in the oxidative dehydrogenation of ethylbenzene using wide-pore transitional aluminas. J Mol
ethylbenzene over a charcoal catalyst. React Kinet Catal Lett 1979;12(2):189-
93. [18] Grunewald GC, Drago RS. Oxidative dehydrogenation of ethylbenzene to
styrene over carbon-based catalysts. J Mol Catal 1990;58(2):227-33.
[19] Guerrero-Ruiz A, Rodriguez-Reinoso F. Oxydehydrogenation of ethylbenzene to styrene catalyzed by graphites and activated carbons. Carbon 1994;32(1):23-9.
[20] Pereira MFR, Orfão JJM, Figueiredo JL. Oxidative dehydrogenation of ethylbenzene on activated carbon catalysts I. Influence of surface chemical groups. Appl Catal A-Gen 1999;184(1):153-60.
[22] Pereira MFR, Orfão JJM, Figueiredo JL. Oxidative dehydrogenation of ethylbenzene on activated carbon catalysts 3. Catalyst deactivation. Appl Catal
A-Gen 2001;218(1-2):307-18. [23] Pereira MFR, Orfão JJM, Figueiredo JL. Influence of the textural properties of an
activated carbon catalyst on the oxidative dehydrogenation of ethylbenzene.
Colloid Surface A 2004;241:165-71. [24] Drago RS, Jurczyk K. Oxidative dehydrogenation of ethylbenzene to styrene
over carbonaceous catalysts. Appl Catal A-Gen 1994;112(2):117-24.
[25] Kane MS, Kao LC, Mariwala RK, Hilscher DF, Foley H. Effect of porosity of carbogenic molecular sieve catalysts on ethylbenzene oxidative dehydrogenation. Ind Eng Chem Res 1996;35(10):3319-31.
[26] Badstube T, Papp H, Kustrowski P, Dziembaj R. Oxidative dehydrogenation of ethylbenzene with carbon dioxide on alkali-promoted Fe active carbon catalysts. Catal Lett 1998;55(3-4):169-72.
[27] Ikenaga N, Tsuruda T, Senma K, Yamaguchi T, Sakurai Y, Suzuki T. Dehydrogenation of ethylbenzene with carbon dioxide using activated carbon-supported catalysts Ind Eng Chem Res 2000;39(5):1228-34.
[28] Zhang J, Su D, Zhang A, Wang D, Schlögl R, Hebert C. Nanocarbon as robust catalyst: Mechanistic insight into carbon-mediated catalysis. Angew Chem Int Ed 2007;46:7319-23.
[29] De Oliveira SB, Barbosa DP, De Melo Monteiro AP, Rabelo D, Do Carmo Rangel M. Evaluation of copper supported on polymeric spherical activated carbon in the ethylbenzene dehydrogenation. Catal Today 2008;133-135:92-8.
[30] Zhang J, Su DS, Blume R, Schlögl R, Wang R, Yang X, et al. Surface chemistry and catalytic reactivity of a nanodiamond in the steam-free dehydrogenation of ethylbenzene. Angew Chem Int Ed 2010;49(46):8640-4.
[31] Keller N, Maksimova NI, Roddatis VV, Schur M, Mestl G, Butenko YV, et al. The catalytic use of onion-like carbon materials for styrene synthesis by oxidative dehydrogenation of ethylbenzene. Angew Chem Int Ed 2002;41(11):1885.
[32] Su DS, Maksimova N, Delgado JJ, Keller N, Mestl G, Ledoux MJ, et al. Nanocarbons in selective oxidative dehydrogenation reaction. Catal Today 2005;102-103:110-4.
[33] Su D, Maksimova NI, Mestl G, Kuznetsov VL, Keller V, Schlögl R, et al. Oxidative dehydrogenation of ethylbenzene to styrene over ultra-dispersed diamond and onion-like carbon. Carbon 2007;45(11):2145-51.
Chapter 5
124
[34] Mestl G, Maksimova NI, Keller N, Roddatis VV, Schlögl R. Carbon nanofilaments
in heterogeneous catalysis: An industrial application for new carbon materials? Angew Chem Int Ed 2001;40(11):2066-8.
[35] Pereira MFR, Orfão JJM, Figueiredo JL. Oxidative dehydrogenation of
ethylbenzene on activated carbon fibers. Carbon 2002;40(13):2393-401. [36] Delgado JJ, Vieira R, Rebmann G, Su DS, Keller N, Ledoux MJ, et al. Supported
carbon nanofibers for the fixed-bed synthesis of styrene. Carbon
2006;44(4):809-12. [37] Delgado JJ, Su DS, Rebmann G, Keller N, Gajović A, Schlögl R. Immobilized
carbon nanofibers as industrial catalyst for ODH reactions. J Catal
2006;244(1):126-9. [38] Li P, Li T, Zhou JH, Sui ZJ, Dai YC, Yuan WK, et al. Synthesis of carbon
nanofiber/graphite-felt composite as a catalyst. Micropor Mesopor Mat 2006;95(1-3):1-7.
[39] Zhao TJ, Sun WZ, Gu XY, Rønning M, Chen D, Dai YC, et al. Rational design of
the carbon nanofiber catalysts for oxidative dehydrogenation of ethylbenzene. Appl Catal A-Gen 2007;323:135-46.
[40] Su DS, Chen X, Liu X, Delgado JJ, Schlögl R, Gajović A. Mount-etna-lava-
supported nanocarbons for oxidative dehydrogenation reactions. Adv Mater 2008;20(19):3597.
[41] Delgado JJ, Chen XW, Frank B, Su DS, Schlögl R. Activation processes of highly
ordered carbon nanofibers in the oxidative dehydrogenation of ethylbenzene. Catal Today 2012;186(1):93-8.
[42] Zhang J, Su DS, Blume R, Schlögl R, Wang R, Yang X, et al. Surface chemistry
and catalytic reactivity of a nanodiamond in the steam-free dehydrogenation of ethylbenzene. Angew Chem Int Ed 2010;49(46):8640-4.
of carbon nanotubes in the oxidative dehydrogenation of ethylbenzene. Carbon 2004;42(14):2807.
[44] Nigrovski B, Zavyalova U, Scholz P, Pollok K, Müller M, Ondruschka B.
Microwave-assisted catalytic oxidative dehydrogenation of ethylbenzene on iron oxide loaded carbon nanotubes. Carbon 2008;46(13):1678-86.
[45] Rinaldi A, Zhang J, Mizera J, Girgsdies F, Wang N, Hamid SBA, et al. Facile
synthesis of carbon nanotube/natural bentonite composites as a stable catalyst for styrene synthesis. Chem Commun 2008;48:6528-30.
[46] Nigrovski B, Scholz P, Krech T, Qui NV, Pollok K, Keller T, et al. The influence of
microwave heating on the texture and catalytic properties of oxidized multi-walled carbon nanotubes. Catal Commun 2009;10(11):1473-7.
[47] Frank B, Zhang J, Blume R, Schlögl R, Su DS. Heteroatoms increase the
selectivity in oxidative dehydrogenation reactions on nanocarbons. Angew Chem Int Ed 2009;48(37):6913-7.
[48] Delgado JJ, Chen X, Tessonnier JP, Schuster ME, Del Rio E, Schlögl R, et al.
Influence of the microstructure of carbon nanotubes on the oxidative dehydrogenation of ethylbenzene to styrene. Catal Today 2010;150(1-2):49-54.
[49] Qui NV, Scholz P, Krech T, Keller TF, Pollok K, Ondruschka B. Multiwalled carbon nanotubes oxidized by UV/H2O2 as catalyst for oxidative dehydrogenation of ethylbenzene. Catal Commun 2011;12(6):464-9.
[50] Du Y, Li J, Ya X. Polyaniline as nonmetal catalyst for styrene synthesis by oxidative dehydrogenation of ethylbenzene. Catal Commun 2008;9(14):2331-3.
[51] Su DS, Delgado JJ, Liu X, Wang D, Schlögl R, Wang L, et al. Highly ordered
mesoporous carbon as catalyst for oxidative dehydrogenation of ethylbenzene to styrene. Chem Asian J 2009;4(7):1108-13.
Oxidative dehydrogenation of ethylbenzene to styrene over MWCNT. Regeneration: myth or reality?
125
[52] Wang L, Delgado JJ, Frank B, Zhang Z, Shan Z, Su DS, et al. Resin-derived hierarchical porous carbon spheres with high catalytic performance in the oxidative dehydrogenation of ethylbenzene. ChemSusChem 2012;5(4):687-93.
[53] Zarubina V, Talebi H, Nederlof C, Kapteijn F, Makkee M, Melián-Cabrera I. On the stability of conventional and nano-structured carbon-based catalysts in the oxidative dehydrogenation of ethylbenzene under industrially relevant
conditions. Carbon 2014;77:329-340. [54] Schuth F, Sing KSW, Weitkamp J. Handbook of Porous Solids. Weinheim: Wiley-
VCH. 2002, 4.
[55] Hughes R, Shettigar UR. Regeneration of silica-alumina catalyst particles. J Appl Chem Biotechn 1971;21(2):35-38.
[56] Kruk M, Jaroniec M. Gas adsorption characterization of ordered organic-
inorganic nanocomposite materials. Chem Mater 2001;13(10):3169-83.
Chapter 5
126
Screening of bare inorganic supports and phosphorous modified catalysts in the oxidative dehydrogenation of ethylbenzene to styrene
Many catalysts have been tested in the oxidative dehydrogenation of ethylbenzene to styrene
reaction and have been demonstrated relatively good performance for short operation time using
O2 or CO2 as an oxidant. The deactivation of the catalysts due to excessive coking is still the major
concern as well as enhancing the selectivity; the conventional process achieves extremely high
selectivity to ST.
There are three types of catalyst based on phosphorous: metal pyrophosphates and phosphates or
P-supported silicas have been reported to be active and selective for oxidative dehydrogenation of
ethylbenzene. However, no systematic investigations about the performance (conversion,
selectivity) and stability under industrially relevant conditions have been reported.
In this chapter, a screening of various bare supports (silicas, alumino-silicate, zeolites, and zeolites
with low alumina content) in comparison with the corresponding phosphorous-based catalysts are
studied. Preliminary characterization of the fresh, spent, and regenerated catalysts is presented.
Chapter 6
128
6.1 Introduction
Direct dehydrogenation of ethylbenzene (EB) is a current method for a styrene (ST) production in industry. The process is carried out at temperatures 580-600 oC in the
presence of steam using promoted iron-oxide catalyst. One of the main drawbacks is that conversion per pass is low due to the equilibrium restrictions [1]. Another way to produce ST is oxidative dehydrogenation (ODH) of EB. The ODH process is not
equilibrium limited [2] and very importantly it can be operated at lower temperatures. This would have significant advantages over the conventional dehydrogenation process including lower energy cost by elimination the excessive
steam use and higher conversions per pass (reduced separation). Considerable research effort has already been done into the development of ODH styrene production catalysts. Many catalysts have been tested and have been demonstrated
relatively good performance [3-8] for short operation time using or CO2 as an oxidant. The deactivation of the catalysts due to excessive coking is still the major concern [9] as well as enhancing the selectivity; the conventional process achieves
extremely high selectivity to ST. There are three types of catalyst based on phosphorous: metal
pyrophosphates [7,10,11], phosphates [7,11-20], or P-supported silicas [7,21] have
been reported to be active and selective for EB ODH. However, no systematic investigations about the performance (conversion, selectivity) and stability under industrially relevant conditions have been reported.
In this chapter, a screening of various bare supports (silicas, alumino-silicates, and high-silica zeolites with low alumina content) in comparison with the corresponding phosphorous-based catalysts are studied. Preliminary characterization
of the fresh, spent, and regenerated catalysts is presented.
6.2 Experimental methods
6.2.1 Materials
The support materials used in this study are zeolites, alumina-silicate, and silicon oxides. These materials and their description are listed in the Table 1.
G-10 Silica Fuji Silysia SS61138 HSA silica Saint Gobain NorPro SS61155 Silica-Alumina, 25% Al2O3 Saint Gobain NorPro
T-4536 High silica BEA Süd-Chemie/Clariant T-4573 High silica MFI Süd-Chemie/Clariant T-4842 Silicate MFI Süd-Chemie/Clariant * Information is not available.
Screening of bare inorganic supports and phosphorous modified catalysts in the oxidative dehydrogenation of ethylbenzene to styrene
129
6.2.2 Preparation of 5 wt.% P/support catalysts
The extrudates that were used as support material were crushed and sieved into a 212-425 µm fraction. Typically 3.0 g of material were accurately weighed in a 4-digit
balance (Mettler Toledo) and placed in a glass tubes. The samples were dried at 150 °C under vacuum for 4 hours. After drying, a neoprene stoppers were put in the test tube to prevent reabsorption water from the air in the sample. After samples
were cooled down to room temperature the phosphorous solution was added onto materials. A solution (10 mL) was prepared by mixing 0.651 mL (or 1.345 ml) of H3PO4 (Merck, 85%) with water. This corresponds to a phosphorous loading of ca. 5
wt.% (as P). Afterwards, the samples were shaken during 4 min at 2500 rpm using a multi-tube vortexer (VWR DVX-2500) at room temperature and dried at 70 oC overnight in an atmospheric oven. Afterwards, the samples were calcinated at
500 °C for 8 hours with a heating rate of 4 °C/min in a Nabertherm P330 muffle furnace.
6.2.3 Regeneration of spent catalysts
The regeneration was performed in a tubular oven in two steps. The regeneration was carried out in a quartz-tube housed tubular oven (Nabertherm RT 50/250-11).
The sample was loaded in a flat quartz crucible that was placed horizontally in the centre of the furnace heating zone. Catalysts were regenerated at temperature 500 oC in 150 ml/min of 1% O2 in Ar flow at a heating rate of 3 oC/min and held for 16h.
Afterwards, the gas flow was switched to synthetic air with a flow 150 ml/min and held 8h at 500 oC. Current regeneration method was chosen due to its mild conditions (e.g. initial low oxygen concentration in argon as inert gas) and slow
heating rate. Synthetic air was chosen for the guarantee that the remaining coke was burned.
6.2.4 Catalysts characterization
6.2.4.1 Nitrogen physisorption analysis
The textural properties of the catalysts and the bare supports were analysed by N2-physisorption at -196 °C using a Micromeritics ASAP 2420. Before analysis the bare supports and fresh catalysts were degassed for 4h at 300 °C, spent samples were
degassed at 200 °C for 10h to ensure that the coke layer was unimpaired. The specific surface area (SBET) was calculated with the BET method [24] in the relative pressure range p/p0 = 0-0.25. The pore volume was estimated using the single point
total desorption pore volume at p/p0 = 0.97. The external surface area and the micropore surface area were calculated using the Harkins and Jura t-plot. Pore size distributions were calculated using the BJH-model [25].
6.2.4.2 Thermogravimetric analysis (TGA)
The amount of coke on the spent catalyst was measured by thermogravimetric analysis (TGA) with a Mettler-Toledo analyzer (TGA/SDTA851e) using a flow of
synthetic air N2:O2 = 80:20. Crucibles made of -alumina were filled with 5-10 mg of
sample. The weight loss was monitored for a temperature program from 30 to 900 °C at a heating rate of 10 °C/min in a flow of 100 ml/min (NTP) of synthetic air.
Chapter 6
130
Calibration was performed using an empty crucible, a mass profile was subtracted from the measured weight.
6.2.4.3 CHN elemental analysis
CHN elemental analyses were carried out in a EuroVector 3000 CHNS analyzer.
Approximately 2 mg of sample was accurately weighed in a 6-digit analytic balance (Mettler Toledo). The samples were burned at 1800 oC in the presence of an oxidation catalyst and decomposed into CO2, H2O, and N2. Then, these gases were
separated in a Porapak QS column at 80 oC and quantified with a TCD detector. Acetonitrile (99.9%) was used as an external standard.
6.2.3 Catalysts testing
The catalytic tests were performed in a setup with six parallel quartz fixed bed
reactors (inner diameter 4 mm) in down-flow operation. The reactors were loaded – from top to bottom – with a quartz wool plug, 10 cm glass beads (0.5 mm diameter), ~ 65 mm catalyst bed (0.80 ml), 10 cm glass beads (0.5 mm diameter), and a
quartz wool plug. Each reactor gas feed has a flow of 36 ml/min (STP) and consist of a mixture
of nitrogen, oxygen, and ethylbenzene. A liquid ethylbenzene flow of 1 g/h
evaporates (3.6 ml/min vapour at STP) resulting in a 1:10 volume ratio of ethylbenzene and gas (10 vol.% EB) with an ethylbenzene gas hourly space velocity (GSVH) of (2160 ml/h)/0.8 ml = 2700 h-1. The EB liquid evaporates in a -Al2O3
filled tube in a synchronized flow with the gas feed. Pressure in the reactor system is
typically 1.2-1.3 bars and with an atmospheric outlet pressure, the pressure drop is typically 0.2-0.3 bars.
The reactor outlet flows were analyzed using an online two channel gas chromatograph with a TCD (columns: 0.3m Hayesep Q 80-100 mesh with back-flush, 25m × 0.53mm Porabond Q, 15 m × 0.53mm molsieve 5A) for permanent gasses
analysis (CO2, H2, N2, O2, CO) and a FID column (30 m × 0.53 mm, Df = 3 mm, RTX-1) for hydrocarbon analysis (methane, ethane, ethene, benzene, toluene, ethylbenzene, styrene, and heavy aromatics). The streams from the different
reactors and a reference gas flow were analysed every 1¾ h, given that analysis time is 15 minutes per flow.
The EB conversion, ST selectivity, ST yield and O2 conversion were calculated
according to equations (1-1)-(1-4). Note that the factor 8 in equation (1-5) comes from stoichiometry; eight COx molecules can be formed from one ethylbenzene or styrene molecule.
(1-1)
(1-2)
(1-3)
(1-4)
Screening of bare inorganic supports and phosphorous modified catalysts in the oxidative dehydrogenation of ethylbenzene to styrene
131
(1-5)
where: = Conversion of ethylbenzene
= Selectivity to styrene
= Styrene yield
= Conversion of O2
= Selectivity to COx
= Molar inlet flow component i, mol/min
= Molar outlet flow component i, mol/min
The primary testing protocol used, comprises 7 reaction conditions to test the
catalysts samples as shown in Table 2. The protocol consists of 3 temperatures
(500 °C, 475 °C and 450 °C) going from high to low, to ensure a short catalyst activation time, as is discussed in [26]. Two O2:EB molar feed ratios were used (0.2 and 0.6) to see the influence on the conversion and selectivity. One condition is used
twice with different time on stream (condition 4 and 7) to see the deactivation effect of the catalyst. The complete catalytic test has duration of 68 hours. The bare supports required a lower temperature to get the psueodo-stationary condition, as
reflected by Table 2. Table 2. Primary catalytic testing conditions
Condition Time on stream, h Temperature, °C (a) O2 : EB
6. 50-56 450 (425) 0.2 7. 56-68 475 (450) 0.6 (a) In parenthesis the temperature conditions applied for the bare supports are given.
6.3 Results and discussion
6.3.1 Bare supports screening
6.3.1.1 Catalyst performance
In order to investigate the activity, the various available bare supports were catalytically screened in the six-flow reactor. All the supports have an EB conversion
in the range of 15 to 30% after 60 h time on stream, as shown in Figure 1-a. The performances of the bare supports are only slightly lower than the reference catalyst (-alumina). This is surprising, since the zeolites have a large fraction of micropores
which can be blocked by coke easily. It means that the external surface area plays a
role in the activity of the zeolites for this reaction. Figure 1-b shows that three of the support materials have a good selectivity
The values of the selectivity are slightly lower than for the reference -alumina. Y-
zeolite CBV100 gives substantially low conversion and selectivity, zeolite ZSM-5 T-4573 gives also a low selectivity.
Chapter 6
132
The difference in performance of the Y-zeolites CBV100 and CBV 500 is the
subject of interest, regarding the fact that both zeolites have the same crystal structure and almost the same Si:Al ratios (2.55 vs. 2.60). It is evident that the structures on a molecular scale are very similar. Thus, this small difference in the
Si:Al ratio cannot be the cause of the differing EB conversion and selectivity to ST. The difference in performance of the two catalysts can probably be ascribed to the variations in Na2O content as CBV100 and CBV500 have 13 wt.% Na2O and 0.2 wt.%
Na2O respectively. The high Na2O content of the former support probably explains the various behaviour of CBV100 and it high COx selectivity.
The selectivity towards COx and dealkylation, forming benzene and toluene
(BT), is also recorded by the GC analyzer. Figure 1-c shows that all supports have selectivity towards COx to some extent. The Y-zeolite CBV100 even has selectivity
towards COx over 50%. During the 7th condition all the catalysts except CBV100 perform similar regarding selectivity to COx (~20%), although slightly worse than γ-alumina. Evidently, the formation of COx is undesirable since this is a valueless
product in the current process, and feedstock is actually burned in the process of COx formation.
The Figure 1-d demonstrates that ZSM-5 zeolite T-4573 gives a continuous
selectivity towards benzene and toluene, although declining during time on stream. This decline might be due to a change of structure similar to ZSM-5 zeolite CBV28014. The figure also shows some selectivity towards benzene and toluene in
the first hours of the reaction, which might be caused by the lack of active coke on the surface of the catalyst.
The second run of bare supports in Figure 2 shows additional results for the
zeolite CP7146 and the silicon oxide G-10. The conversion of the G-10 is low due to the very low acidity of silicas [27]. However, the selectivity of the G-10 looks promising (Figure 2-b). At a certain point of the run the selectivity becomes 94%,
which is more than the -alumina reference sample has. It must be noted that due to
the different behaviour at T=425°C and O2:EB=0.6, this high value can be caused by a measurement error. Also, the CP7146 zeolite has selectivity comparable to -
alumina. The ZSM-5 zeolite CBV28014 shows a low selectivity towards the
production of styrene, despite the rather high conversion of EB. The selectivity towards COx in the second run is quite similar for the most
supports. Figure 2-c shows that that COx selectivity is especially low for the silicon
oxide G-10, all the other supports perform similar to the -alumina reference catalyst
(~15%). All the catalysts except ZSM-5 zeolite CBV28014 show a low BT selectivity. Note that the graph of the BT selectivity of CBV28014 in Figure 2-d looks very similar
to the one of T-4573 (Figure 1-d). Both samples are high silica ZSM-5 zeolites. It seems that the lack of Al slows down the formation of active/selective coke, and the dealkylation is observed in the complete run (i.e. > 60 h TOS). Sample CBV2314 is
also a ZSM-5 structure with a high Al content; dealkylation can be seen in the first few hours TOS. The development of the coke makes the catalyst more selective to styrene, and the dealkylation disappears probably because the pores are blocked.
Support CP7146, though the structure and composition are unknown, can be ascribed to be a ZSM-5 zeolite with high Al-content, regarding the selectivity to BT behaviour. In addition, it was found that the high silica ZSM-5 structure changes
during the reaction.
Screening of bare inorganic supports and phosphorous modified catalysts in the oxidative dehydrogenation of ethylbenzene to styrene
133
Figure 1. Time on stream EB conversion (a), selectivity to ST (b), selectivity to COx (c), BT
selectivity (d) of supports at various temperatures (475, 450, 425, and 450 oC) and O2/EB= 0.6 and 0.2 (vol.); GHSV of 3000 l/l/h; 10 vol. % EB. The top headings correspond to the O2:EB (vol.) ratio and reaction temperature in oC.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 20 40 60
CO
×sele
ctivity
Time [h]
γ-alumina
CBV100 bare
CBV500 bare
CBV2314 bare
T-4536 bare
T-4573 bare
Temp.
O2:EB
475
C 450
C
425
C450
C
0.4 0.4 0.60.40.60.6 0.6
Temp.
O2:EB
475
C 450
C425
C450
C
0.2 0.2 0.60.20.60.6 0.6
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 20 40 60
BT
sele
ctivity
Time [h]
γ-alumina
CBV100 bare
CBV500 bare
CBV2314 bare
T-4536 bare
T-4573 bare
Temp.
O2:EB
475
C 450
C
425
C450
C
0.4 0.4 0.60.40.60.6 0.6
Temp.
O2:EB
475
C 450
C425
C450
C
0.2 0.2 0.60.20.60.6 0.6
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 20 40 60
EB
convers
ion
Time [h]
γ-alumina
CBV100 bare
CBV500 bare
CBV2314 bare
T-4536 bare
T-4573 bare
Temp.
O2:EB
475
C 450
C
C
425
C450
C
0.2 0.2 0.60.20.60.6 0.6
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 20 40 60
ST
sele
ctivity
Time [h]
γ-alumina
CBV100 bare
CBV500 bare
CBV2314 bare
T-4536 bare
T-4573 bare
Temp.
O2:EB
475
C 450
C
425
C450
C
0.4 0.4 0.60.40.60.6 0.6
Temp.
O2:EB
475
C 450
C425
C450
C
0.2 0.2 0.60.20.60.6 0.6
a b
c d
Chapter 6
134
Figure 2. Time on stream EB conversion (a), selectivity to ST (b), selectivity to COx (c), BT selectivity (d) of supports at various temperatures (475, 450, 425, and 450 oC) and O2/EB= 0.6 and 0.2 (vol.); GHSV of 3000 l/l/h; 10 vol. % EB.
6.3.1.2 Bare supports characterizations
To investigate the effect of the active coke on the textural properties, N2-physisorption experiments were performed on the fresh and spent catalysts. From
this data, shown in Table 3, it becomes clear that the pore volume decreases substantially, especially for the samples having more micropores. The micropore volume for the spent catalysts is nearly negligible, as the remaining surface area is
external, meaning that the micropores are completely blocked. The SBET is reduced for all samples by the same percentage (~93%), except for the zeolite ZSM-5 T-
4573. These results clearly show that coke has approximately the same influence on the textural properties of the catalysts that is why physisorption experiments were performed only for the first run of the bare supports.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 20 40 60
EB
convers
ion
Time [h]
G-10 bare
T-4842 bare
CP7146 bare
γ-alumina
CBV28014 bare
Temp.
O2:EB
475
C 450
C425
C450
C
0.2 0.2 0.60.20.60.6 0.6
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 20 40 60
ST
sele
ctivity
Time [h]
G-10 bare
T-4842 bare
CP7146 bare
γ-alumina
CBV28014 bare
Temp.
O2:EB
475
C 450
C
425
C450
C
0.4 0.4 0.60.40.60.6 0.6
Temp.
O2:EB
475
C 450
C425
C450
C
0.2 0.2 0.60.20.60.6 0.6
0%
10%
20%
30%
40%
50%
60%
70%
80%
0 20 40 60
BT
sele
ctivity
Time [h]
G-10 bare
T-4842 bare
CP7146 bare
γ-alumina
CBV28014 bare
Temp.
O2:EB
475
C 450
C
425
C450
C
0.4 0.4 0.60.40.60.6 0.6
Temp.
O2:EB
475
C 450
C425
C450
C
0.2 0.2 0.60.20.60.6 0.6
0%
5%
10%
15%
20%
25%
30%
0 20 40 60
CO
x s
ele
ctivity
Time [h]
G-10 bare
T-4842 bare
CP7146 bare
γ-alumina
CBV28014 bare
Temp.
O2:EB
475
C 450
C
425
C450
C
0.4 0.4 0.60.40.60.6 0.6
Temp.
O2:EB
475
C 450
C425
C450
C
0.2 0.2 0.60.20.60.6 0.6
a
c d
b
Screening of bare inorganic supports and phosphorous modified catalysts in the oxidative dehydrogenation of ethylbenzene to styrene
135
Table 3. Textural properties of the various bare supports (fresh and spent catalyst)
T-4573 fresh 0.418 0.080 303 11.9 T-4573 spent 0.110 (-74%) 0.018 84 (-72%) >200 (a) In parentheses the relative difference between the fresh and the spent material is given.
6.3.2 5P-based catalyst performance
Although the Y-zeolite CBV500 showed a promising results as bare support, when
PO43- was added the conversion was nearly half of the best supports. The selectivity
seems to be low, despite the high noise of the measurements, as shown in the Figure 4-a. The best catalysts found were based on beta zeolite T-4536 and
NorPro55. The selectivities were in the range of 90-95%, however, conversions were only approximately 20%.
Looking at the selectivity to COx in the Figure 5-c, it can be seen that
5P/NorPro55 and beta zeolite based 5P/T-4536 have the lowest selectivity to COx approximately 10-15%. Catalysts 5P/CBV28014 and 5P/T-4573 based on ZSM-5 support have the selectivity to COx 10-15% and high selectivity to BT (Figure 5-d).
Other catalysts do not show significant selectivity to BT. The second run with 5 wt.% phosphorous addition based on silicas and three
additional zeolites shown in the Figure 6. Two catalyst of 5P/NorPro38 (labelled as
5P/SiO2) both have the same performance, which confirms the reproducible of catalysts preparation. All the silica samples show a large decline in performance, especially for the conversion while the selectivity remains relatively steady. From the
silica samples the 5P/G-10 performs the best, and it is the more stable catalyst. The performance of 5P/CBV100 is significantly worse than for the most of
other catalysts based on zeolites. For both 5P/CP7146 and 5P/CBV2314 there is
hardly any difference between the performance of the bare support and the 5 wt.% phosphorous added catalysts.
Figure 6-c shows that the 5P samples from the second run have a varying COx
selectivity. Although the Y-zeolite 5P/CBV100 has a lower selectivity towards COx than the bare support CBV100, it is still exceptionally high (30-35%). The silica based catalysts (5P/NorPro38 and 5P/G-10) have the lowest tendency towards COx
formation. Also, these silica based samples show the lowest BT selectivity. 5P/CBV2314 catalyst has higher selectivity to BT first 40 h TOS in comparison with other catalysts tested in this run (Figure 6-d).
Chapter 6
136
Figure 5. Time on stream EB conversion (a), selectivity to ST (b), selectivity to COx (c), BT selectivity (d) of various 5P/supports catalysts at various temperatures (475, 450, 425, and 450 oC) and O2/EB= 0.6 and 0.2 (vol.); GHSV of 3000 l/l/h; 10 vol. % EB. 5P/SiO2 corresponds to the silica SS61138.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
0 20 40 60
EB
convers
ion
Time [h]
5P/CBV500
5P/CBV28014
5P/NorPro55
5P/T-4536
5P/T-4573
5P/T-4842
Temp.
O2:EB
500
C 475
C450
C475
C
0.2 0.2 0.60.20.60.6 0.6
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 20 40 60
ST
sele
ctivity
Time [h]
5P/CBV500
5P/CBV28014
5P/NorPro55
5P/T-4536
5P/T-4573
5P/T-4842
Temp.
O2:EB
500
C 475
C450
C475
C
0.2 0.2 0.60.20.60.6 0.6
0%
10%
20%
30%
40%
50%
60%
0 20 40 60
CO
x s
ele
ctivity
Time [h]
5P/CBV500
5P/CBV28014
5P/NorPro55
5P/T-4536
5P/T-4573
5P/T-4842
Temp.
O2:EB
500
C 475
C450
C475
C
0.4 0.4 0.60.40.60.6 0.6
0%
10%
20%
30%
40%
50%
60%
70%
80%
0 20 40 60
BT
sele
ctivity
Time [h]
5P/CBV500
5P/CBV28014
5P/NorPro55
5P/T-4536
5P/T-4573
5P/T-4842
Temp.
O2:EB
500
C 475
C450
C475
C
0.4 0.4 0.60.40.60.6 0.6
a b
c d
Screening of bare inorganic supports and phosphorous modified catalysts in the oxidative dehydrogenation of ethylbenzene to styrene
137
Figure 6. Time on stream EB conversion (a), selectivity to ST (b), selectivity to COx (c), BT selectivity (d) of 5P/supports catalysts at various temperatures (475, 450, 425, and 450 oC) and O2/EB= 0.6 and 0.2 (vol.); GHSV of 3000 l/l/h; 10 vol. % EB. 5P/SiO2 corresponds to the silica
SS61138.
6.3.3 Comparison between bare supports and 5P-based catalysts
6.3.3.1 Performance comparison
For comparison the performance of the bare supports and the 5P catalyst the same conditions should be compared in order to draw valid conclusions. Because the bare
supports were tested in another temperature program, the 7th condition of bare supports testing run was compared with the 5th condition in the 5P catalysts run. In this way the samples were compared in the same conditions (i.e. 450 °C, O2:EB ratio
= 0.6), and the performance has already achieved the pseudo-stationary conditions. As can be seen from Table 4, the result of 5 wt.% phosphorous addition varies
in terms of improvement of catalyst performance. The most extreme difference is
Fuji G-10 silica support with 1632% improvement from 5P addition. The reason for this extreme improvement is that silicas have almost no acidity [28], hence coke
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 20 40 60
EB
convers
ion
Time [h]
5P/CP7146
5P/SiO2
5P/G-10
5P/CBV100
5P/SiO2
5P/CBV2314
Temp.
O2:EB
500
C 475
C450
C475
C
0.4 0.4 0.60.40.60.6 0.6
30%
40%
50%
60%
70%
80%
90%
100%
0 20 40 60
ST
sele
ctivity
Time [h]
5P/CBV2314
5P/SiO2
5P/G-10
5P/CBV100
5P/SiO2
5P/CP7146
Temp.
O2:EB
500
C 475
C450
C475
C
0.4 0.4 0.60.40.60.6 0.6
0%
10%
20%
30%
40%
50%
60%
0 20 40 60
CO
x s
ele
ctivity
Time [h]
5P/CBV2314
5P/SiO2
5P/G-10
5P/CBV100
5P/SiO2
5P/CP7146
Temp.
O2:EB
500
C 475
C450
C475
C
0.4 0.4 0.60.40.60.6 0.6
0%
10%
20%
30%
40%
50%
60%
70%
80%
0 20 40 60
BT
sele
ctivity
Time [h]
5P/CBV2314
5P/SiO2
5P/G-10
5P/CBV100
5P/SiO2
5P/CP7146
Temp.
O2:EB
500
C 475
C450
C475
C
0.4 0.4 0.60.40.60.6 0.6
c d
a b
Chapter 6
138
does not build-up. The addition of phosphorous greatly increases the acidity and,
thus, the performance. For the other support materials the effect of phosphorous differs greatly. Some supports show virtually no difference in performance by phosphorous addition (CP7146, T-4573), others show a strong positive (CBV100, T-
4536, T-4842) or strong negative (CBV500, CBV2314, CBV28014) effect. Looking at the stability, the performances of catalysts in the 7th and 3rd
reaction condition were compared. Both conditions are equal at T=475°C (T=450°C
for the bare supports) and O2:EB = 0.6. The yield is chosen because this reflects the stability of both, the conversion and selectivity as Y = X∙S.
The yield in both the 3rd and 7th condition and the ratio y7/y3 are shown in
Figure 7. The addition of phosphorous increases the stability of most catalyst systems except for the T-4536 supported catalyst. Although the addition of
phosphorous does not necessarily increase the performance of the zeolite supported catalysts, it seems that the stability of these catalysts is increased by 5P-addition. However, one must note that there is a difference between the temperature program
for the bare supports runs and the 5P runs. It is possible that this has influence on the stability of the supports during the reaction.
Table 4. Yield in condition 7 of the bare material and in condition 5 for the 5P added material and the increase in yield by phosphorous addition
NorPro55 N/A* 24.4 N/A * *Information is not available
Screening of bare inorganic supports and phosphorous modified catalysts in the oxidative dehydrogenation of ethylbenzene to styrene
139
Figure 7. Styrene yield of supports and 5P/supports catalysts in the 3rd and 7th reaction condition and the ratio Y7/Y3.
6.3.3.2 Relation between yield and coke content
The relation between styrene yield and coke content are presented in the Figure 9.
All bare and 5P spent samples were plotted except for CBV100 and 5P/CBV100 as these catalyst gave exceptionally high COx selectivity. Figure 9 evidences that a small amount of coke (~10 wt.%) can already give a significant yield for styrene. For
higher coke contents the slope of the curve becomes less pronounced.
6.3.3.3 Regeneration of spent catalysts
Relevant catalysts were analyzed by N2-phyisisorption to evaluate the effect of the regeneration on the textural properties of the catalyst samples. The results from Table 5 show that for both catalysts the BET surface area decreases only minor
(~3%) when the samples are regenerated. However, the micropore surface area (obtained from the t-plot) for 5P/NorPro55 decreased by 25%. The decrease in micropore surface area is even larger than the total decrease in total surface area,
i.e. SBET. On the other hand, the micropore surface area of 5P/T-4536 increases, although with only 1 m2/g. Hence this can be within the experimental error. Nevertheless, it might be concluded that the micropore structure of the BEA zeolite
remains constant during regeneration. The difference in micropore stability between 5P/NorPro55 and 5P/T-4536 can be ascribed to the higher crystallinity of the T-4536 support, being a zeolite.
The pore volume of both samples before and after regeneration is nearly the same. TGA results show that the regenerated samples have very little coke left on
0 0,2 0,4 0,6 0,8 1
0% 10% 20% 30% 40% 50%
CBV100 bare
CBV2314 bare
CBV28014 bare
CBV500 bare
CP714 bare
G-10 bare
T-4536 bare
T-4573 bare
T-4842 bare
5P/CBV100
5P/CBV2314
5P/CBV28014
5P/CBV500
5P/CP7146
5P/G-10
5P/NorPro55
5P/T-4536
5P/T-4573
5P/T-4842
Ratio y7/y3
Styrene yield
y7 y3 y7/y3
Chapter 6
140
the surface. The peak in pore size distribution shifts a small fraction to larger
diameters. The coke content of the regenerated catalysts measured by TGA does not give
a reliable value because there is no distinctive coke burning step, as the weight loss
between 300 °C and 700 °C takes into account the coke burned out and the OH condensation at the silica surface [20]. For this reason, a selective technique as CHN analysis was performed for both, the spent and regenerated catalysts. The spent
samples were analyzed to compare CHN values to the coke content given by TGA. As Table 6 shows, the weight percentage of carbon is lower than the coke
content given in Table 5. It was expected, since part of the coke consists of oxygen
and hydrogen, and the OH condensation effect is omitted. The coke content of 5P/NorPro55 is 20% higher than the mass fraction of carbon, for 5P/T-4536 the coke
content is 28% higher than the carbon mass fraction. This difference can be ascribed to the difference in coke composition.
The CHN analysis gives a very precise figure for the weight percentage of
carbon, as well at low carbon content; it can be concluded that TGA slightly underestimates the coke content of the regenerated samples.
By comparing the isotherms and pore size distribution graphs of the fresh and
regenerated catalyst, it can be seen that there are almost no changes in the texture. The graphs of the fresh and the regenerated samples are practically overlapping in the isotherms and the PSD graphs. This is certain for both, 5P/T-4536 (Figure 10-a,
b) and 5P/NorPro55 (Figure 10-c, d). An overview of the performance (selectivity, conversion, and yield) of all
tested catalysts is given in Figure 8. The best performing catalysts are 5P added to
the zeolite BEA T-4536, silica-alumina NorPro55, and pure silica Fuji G-10.
Figure 8. Selectivity, conversion, and yield of all tested bare and 5P catalysts in the 7th reaction condition.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%SELECTIVITY CONVERSION YIELD
Screening of bare inorganic supports and phosphorous modified catalysts in the oxidative dehydrogenation of ethylbenzene to styrene
141
Figure 9. Yield of the various catalysts plotted versus the coke content of the spent samples.
Table 5. Results from the TGA and physisorption analyses for fresh, spent,and regenerated samples
5P/T-4536 reg. 0.7 158.9 (-3.2%) 109.6 (+0.9%) 0.388(-0.8%) 29.6 (a) TGA has been performed over the temperature range of 300-700°C to assess the amount of coke (b) In parentheses is given the percentage change compared to fresh 5P-catalyst.
Table 6. CHN elemental analysis of spent and regenerated samples as an average of two measurements
Figure 10. Physisorption isotherms of fresh and regenerated 5P/T-4536 (a) and 5P/NorPro55 (c) and pore size distribution of 5P/T-4536 (b) and 5P/NorPro55 (d)
6.4 Conclusions
The zeolite based catalysts have a varying performance in the ODH of ethylbenzene to styrene. Nevertheless, the results show that all tested zeolites have at least some
activity in the concerning reaction; the EB conversion and ST selectivity are not superior to the reference catalyst -alumina. Some ZSM-5 zeolites had a large
selectivity towards dealkylation of ethylbenzene to benzene and toluene, when the
rate of coke build-up is low. With the addition of 5 wt.% P to the bare supports by means of dry
impregnation, the improvement of the catalysts performance varies per support.
Some catalyst performances radically improved by adding phosphorous, others catalysts perform similar to the bare materials or even have a lower activity than the corresponding bare support. Although the addition of phosphorous does not
necessarily increase the performance of catalysts, it, however, increases the stability of almost all supports.
TGA experiments show that almost all of the coke was removed with the
optimal regeneration method, thus it can be concluded that this method is adequate regarding coke removal of the spent catalysts. CHN analysis confirmed that almost all coke is burned off during regeneration. The remaining coke content is around 1
wt.%.
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
10 100 1000
dV
/dlo
g(w
) P
ore
Vo
lum
e, cm
³/g
·Å
Pore Width, Å
fresh regenerated
0
50
100
150
200
250
300
0 0,5 1
Quantity
Adsorb
ed,
cm
3/g
ST
P
Relative pressure p/p0
fresh regeneated
0
50
100
150
200
250
300
0 0,5 1
Qu
an
tity
Ad
so
rbe
d,
cm
3/g
ST
P
Relative pressure p/p0
fresh regeneated
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
10 100 1000
dV
/dlo
g(w
) P
ore
Vo
lum
e, cm
³/g
·Å
Pore Width, Å
fresh regenerated
a b
c d
Screening of bare inorganic supports and phosphorous modified catalysts in the oxidative dehydrogenation of ethylbenzene to styrene
143
The physisorption data indicates that the mesoporosity of both 5P/T-4536 and 5P/NorPro55 is stable during the catalytic test and regeneration. Looking at the microporosity data, it shows that the microporosity of the 5P/T-4536 is retained.
However, the microporosity of 5P/NorPro55 is partly lost (-5%). The SBET for both samples decreases with 3%.
This combined data from the characterization by physisorption, TGA, and CHN
and show that the chosen method for regeneration works sufficiently well. However, some questions remain regarding the stability of the crystalline structure during the regeneration process and what is the reason of decrease in the textural properties:
in the deactivation under reaction conditions or within the experimental error.
6.5 References
[1] Cavani F, Trifiro F. Alternative processes for the production of styrene. Appl
Catal A-Gen 1995;133(2):219-39. [2] Sun AL, Qin ZF, Wang JG. Reaction coupling of ethylbenzene dehydrogenation
with water-gas shift. Appl Catal A-Gen 2002;234(1-2):179-89.
[3] Kozharov AI, Makhlis LA, Lisovskii AE, Alkhazov TG, Vasserberg BE. Mechanism of oxidational dehydrogenation of ethylbenzene on an aluminum oxide catalyst. Rus Chem Bul 1977;26(3):477-80.
[4] Sakurai Y, Suzaki T, Ikenaga N, Suzuki T. Dehydrogenation of ethylbenzene with an activated carbon-supported vanadium catalyst. Appl Catal A-Gen 2000;192(2):281-8.
[5] Sato S, Ohhara M, Sodesawa T, Nozaki F. Combination of ethylbenzene dehydrogenation and carbon dioxide shift-reaction over a sodium oxide/alumina catalyst. Appl Catal 1988;37(1-2):207-15.
[6] Fedorov GI, Sibgatullin SG, Solodova NL, Izmailov RI. Dehydrogenation of ethylbenzene on oxide catalysts in the presence of carbon dioxide. Petrol Chem+ 1976;16(3):157-62.
[7] Vrieland GE. Oxydehydrogenation of ethylbenzene to styrene over metal pyrophosphates. 1. Catalyst composition and reaction variables. J Catal 1988;111(1):1-13.
[8] Vrieland GE. Oxydehydrogenation of ethylbenzene to styrene over metal pyrophosphates. 2. Microbalance studies of carbon deposition and burnoff. J Catal 1988;111(1):14-22.
[10] Emig G, Hofmann H. Action of zirconium phosphate as a catalyst for the
oxydehydrogenation of ethylbenzene to styrene. J Catal 1983;84(1):15-26. [11] Vrieland GE, Friedli HR. Method of oxydehydrogenation of ethyl benzene. US
patent 3933932, 1976.
[12] Bautista FM, Campelo JM, Luna D, Marinas JM, Quirós RA, Romero AA. Screening of amorphous metal-phosphate catalysts for the oxidative
dehydrogenation of ethylbenzene to styrene. Appl Catal B-Environ 2007;70(1-4):611-20.
[13] Murakami Y, Iwayama K, Uchida H, Hattori T, Tagawa T. Study of the oxidative
dehydrogenation of ethylbenzene.1. Catalytic behavior of SNO2-P2O5. J Catal 1981;71(2):257-69.
[14] Murakami Y, Iwayama K, Uchida H, Hattori T, Tagawa T. Screening of catalysis
for the oxidative dehydrogenation of ethylbenzene. Appl Catal 1982;2(1-2):67-74.
[15] Schraut A, Emig G, Sockel H-G. Composition and structure of active coke in the
oxydehydrogenation of ethylbenzene. Appl Catal 1987;29(2):311-26.
Chapter 6
144
[16] Bagnasco, P. Ciambelli, M. Turco, A. La Ginestra, P. Patrono. Layered
zirconium-tin phosphates: II. Catalytic properties in the oxydehydrogenation of ethylbenzene to styrene. Appl Catal 1991;68(1):69-79.
[17] Arrúa LA, Ardissone DE, Quiroga OD, Rivarola JB. Oxidehydrogenation of
ethylbenzene on P-O-Ni catalyst. React Kinet Catal L 1995;56(2):383-9. [18] Dziewiecki Z, Jagiello M, Makowski A. Investigation of polymer organic deposit
formed on nickel phosphate in oxidative dehydrogenation of ethylbenzene.
React Funct Polym 1997;33:185. [19] Vrieland GE, Friedli HR. Method of oxydehydrogenation of ethyl benzene. US
patent 3923916, 1975.
[20] Hofmann H, Emig G, Ruppert W, From ethylbenzene, zirconium phosphate catalyst. US patent 4400568, 1983.
[21] Tagiyev DB, Gasimov GO, Rustamov MI. Carbon deposits on the surface of CaO/SiO2 as active catalysts for the oxidative dehydrogenation of ethylbenzene. Catal Today 2005;102-103:197-202.
[22] Nederlof C, PhD Thesis dissertation; URL: http://repository.tudelft.nl/ [23] Minachev KM, Kharlamov V, Tagiyev D, Zulfugarov Z. Catalytic properties of
zeolites in oxidative dehydrogenation of hydrocarbons. Zeolites 1984;4(3):270.
[24] Brunauer S, Emmett PH, Teller E. Adsorption of gases in multimolecular layers. J Am Chem Soc 1938;60(2):309-19.
[25] Barrett EP, Joyner LG, Halenda PH. The determination of pore volume and area
distributions in porous substances. 1. Computations from nitrogen isotherms. J Am Chem Soc 1951;73(1):373-80.
[26] Zarubina V, Nederlof C, Van der Linden B, Kapteijn F, Heeres HJ, Makkee M,
Melián-Cabrera I. Making coke a more efficient catalyst in the oxidative dehydrogenation of ethylbenzene using wide-pore transitional aluminas. J Mol Catal A: Chem 2014;381:179-87.
[27] Parry E. An infrared study of pyridine adsorbed on acidic solids. Characterization of surface acidity. J Catal 1963;2(5):371-9.
[28] Tamura M, Shimizu K, Satsuma A. Comprehensive IR study on acid/base
properties of metal oxides. Appl Catal A-Gen 2012;433-434:135-45.
