chemicals: methane production · A thermodynamic analysis of the CO 2 hydrogenation using as feed a...

CO2 capture and valorisation to

chemicals: methane production

Dissertation presented to the

University of Porto

for the degree of

Doctor of Philosophy in Chemical and Biological Engineering

by

CARLOS EDUARDO GERALDES DE VASCONCELOS MIGUEL

Supervisor: Prof. Dr. Luís Miguel Palma Madeira

Co-Supervisor: Prof. Dr. Adélio Miguel Magalhães Mendes

Department of Chemical Engineering Faculty of Engineering

University of Porto

2018

I

Abstract

In this work, a carbon capture and utilization (CCU) unit for CO2 capture

and conversion to methane was studied. A layered packed bed consisting in a

high temperature CO2 sorbent and a methanation catalyst was considered to

perform both capture and conversion in the same unit, in a perspective of

process intensification. The proof-of-concept of the sorptive reactor was

performed using commercial materials in a lab-scale experimental setup which

was built for such purpose allowing also to perform sorption or reaction

experiments.

A thermodynamic analysis of the CO2 hydrogenation using as feed a post-

combustion stream was performed. Hydrogenation to methane is considerably

less demanding in terms of pressure compared to the methanol option,

particularly in the operating temperature window of available industrial catalysts.

Regarding CO2 methanation, the negative effect caused by the presence of

oxygen on CO2 conversion recommends its removal from flue gas. Further

separation of CO2 from the remaining species present in flue gas is also

recommended to enrich the outlet stream with CH4.

Hydrotalcite sorbents were synthesized considering the partial substitution

of aluminum by gallium and the promotion with K, Cs and Sr; their effect on CO2

sorption at high temperature (200-300 ºC) was studied. Cesium and, especially,

potassium enhanced the sorption capacity of the hydrotalcite, while strontium

decreased. A kinetic model assuming two parallel contributions was proposed

for the sorption kinetics, which fitted very well the experiment data. Ga-

substitution improved the sorption capacity of the hydrotalcite that performed

well under cyclic operation. However, gallium incorporation increases the

sorbent cost, which certainly restrains its utilization at industrial scale.

The intrinsic kinetics of CO2 methanation over an industrial nickel catalyst

was determined in a temperature range of 250-350 ºC and at atmospheric

pressure. The selected kinetic model was adapted from the literature (among

others also tested) and has only 4 fitting parameters. The mechanism assumes

II

a formyl intermediate species (CHO*) obtained after H2 and CO2 dissociation,

and hydroxyl (OH*) as the most abundant species at the catalyst surface. Such

model was validated through the simulation of an isothermal plug-flow reactor

and a reasonable agreement between model predictions and the observed

values was obtained.

The integrated sorption-reaction process for CO2 capture and conversion

to CH4 was studied between 300 ºC and 350 ºC and at low pressure (P ≤ 2.5

bar). The sorptive reactor had inside a layered bed of a commercial hydrotalcite

sorbent (Pural MG30-K from Sasol) and a commercial nickel catalyst (METH 134

from Clariant). The materials proved their compatibility and stability under cyclic

operation, at least under the considered operation conditions. The sorption

equilibrium was well described by the dual-site Langmuir model, which considers

that CO2 is sorbed through physical sorption (exothermic contribution) and

chemical reaction (endothermic contribution). Reactive regeneration mode (i.e.

with a H2 stream) improved desorption kinetics, when compared to normal

regeneration (i.e. with a N2 stream), and allowed to completely recover the

sorbent capacity in each cycle. The steam produced in situ by the methanation

reaction causes this positive effect. The synthetic flue gas used was made of

CO2 and N2 only. CO2 conversion was high in batch operation (ca. 92 %) and

almost complete (99 %) when the reactor operated under continuous mode. The

major drawback found in the operation of the sorptive reactor was the need for

improving the methane purity; CO formation could be avoided by decreasing the

temperature and/or increasing the pressure.

The use of highly permeable and selective metal membranes can be

considered for feeding radially hydrogen to the methanation packed-bed reactor.

The permeance towards H2 of a 50-µm thick PdAg membrane was then

measured between 200 ºC and 300 ºC considering a pure H2 stream or binary

H2/CO or H2/CO2 mixtures. The Sieverts law was observed for the former

conditions, while for the binary mixtures the Sieverts-Langmuir model explained

the inhibitory effect on H2 permeance caused by CO and CO2. A modified

III

Sieverts-Langmuir equation was proposed to account for changes in the

permeation driving-force under high H2 recovery conditions.

V

Resumo

Neste trabalho estudou-se uma unidade de captura e utilização de

carbono (CCU) para a captura e conversão de CO2 em metano. Para realizar a

captura e conversão de CO2 na mesma unidade, foi considerado um leito

estratificado contendo um adsorvente de CO2 de alta temperatura e um

catalisador para a reação de metanação, numa perspetiva de intensificação de

processos. A prova de conceito do reator adsortivo foi efetuada utilizando

materiais comerciais e teve lugar numa unidade experimental à escala

laboratorial que foi construída para este fim, a qual permite também realizar

ensaios de adsorção e reação separadamente.

A reação de hidrogenação de CO2 foi estudada do ponto de vista

termodinâmico considerando como alimentação uma corrente de pós-

combustão. A hidrogenação de CO2 para formação de metano é

consideravelmente menos exigente em termos de pressão de operação quando

comparada com a opção do metanol, em particular na gama de temperatura de

operação dos catalisadores industriais disponíveis. Em relação à reação de

metanação, o efeito negativo causado pela presença de oxigénio na conversão

do CO2 recomenda a sua remoção do gás de exaustão. A separação do CO2

das restantes espécies presentes no gás de exaustão também é recomendada,

essencialmente para enriquecer a corrente de saída em CH4.

Foram sintetizadas hidrotalcites considerando a substituição parcial de

alumínio por Ga e a modificação com K, Cs e Sr, tendo-se estudado o seu efeito

na capacidade de adsorção de CO2 entre 200 ºC e 300 ºC. A impregnação com

césio e potássio, e deste último em particular, aumenta a capacidade de

adsorção da hidrotalcite, enquanto o estrôncio diminui. Foi proposto um modelo

para a cinética de adsorção considerando duas contribuições em paralelo, o

qual ajustou muito bem os dados experimentais. A substituição com Ga

melhorou a capacidade de adsorção e o adsorvente mostrou boa capacidade

durante operação cíclica. Contudo, a incorporação de gálio aumenta

VI

consideravelmente o custo do adsorvente, o que certamente pode restringir a

sua utilização a uma escala industrial.

A cinética verdadeira da metanação de CO2 foi determinada usando um

catalisador de níquel industrial entre 250 ºC e 350 ºC à pressão atmosférica. O

modelo cinético selecionado foi adaptado da literatura (entre outros igualmente

testados) e apresenta apenas 4 parâmetros de ajuste. O mecanismo assume a

formação de uma espécie de formilo (CHO*) intermediária obtida após a

adsorção dissociativa de H2 e CO2, bem como o grupo hidroxilo (OH*) como

espécie mais abundante na superfície do catalisador. O modelo foi validado

através da simulação de um reactor isotérmico de leito fixo incluindo a cinética

determinada e os resultados ajustam bem os dados experimentais.

O processo integrado de adsorção-reação para a captura e conversão de

CO2 em CH4 foi estudado entre 300 ºC e 350 ºC a baixa pressão (P ≤ 2.5 bar).

O leito do reator adsortivo foi estruturado em camadas intercaladas de uma

hidrotalcite comercial (Pural MG30-K da Sasol) e de um catalisador de níquel

comercial (METH 134 da Clariant). Os materiais mostraram serem compatíveis

e estáveis em operação cíclica, pelo menos nas condições de operação usadas.

O equilíbrio de adsorção foi bem descrito pelo modelo de Langmuir de dois

centros (dual-site), o qual considera que o CO2 é adsorvido através de adsorção

física (contribuição exotérmica) e reação química (contribuição endotérmica). A

regeneração reativa (i.e. com uma corrente de H2) melhorou a cinética de

dessorção, quando comparada com a regeneração normal (i.e. com N2), tendo

permitido recuperar completamente a capacidade de adsorção em cada ciclo.

O vapor de água produzido in situ pela reação de metanação provocou este

efeito positivo. A corrente de gás de exaustão considerada foi feita apenas com

CO2 e N2. A conversão de CO2 foi elevada em operação descontínua (ca. 92 %)

e quase total (99 %) quando o reator foi operado em modo contínuo. A maior

contrariedade registada na operação do reator adsortivo foi a necessidade de

aumentar-se a pureza do metano na corrente de saída; a formação de CO pode

ser evitada diminuindo a temperatura ou aumentando a pressão.

VII

O uso de membranas metálicas altamente permeavéis e seletivas pode

ser considerado para se alimentar radialmente hidrogénio ao reactor de

metanação de leito fixo. Mediu-se então a permeância ao H2 de uma membrana

de PdAg com 50 µm de espessura entre 200 ºC e 300 ºC considerando uma

corrente pura de H2 ou mistura binárias de H2/CO ou H2/CO2. A lei de Sieverts

foi observada para o primeiro caso, enquanto o modelo de Sieverts-Langmuir

descreve o efeito inibitório na permeação de H2 provocado pelo CO e CO2. Foi

proposta uma versão modificada deste modelo que considera as alterações na

força diretriz da permeação e que deve ser aplicada em condições de elevada

recuperação de H2.

IX

Acknowledgments

In this section I would like to acknowledge all of you who, in many ways,

gave me support throughout this journey.

I am grateful to my thesis supervisors, Prof. Luis Miguel Madeira and Prof.

Adélio Mendes, for their guidance and encouragement. Both showed me how to

think as a chemical engineer researcher. Thanks Professor Madeira for your

confidence on my work, your comprehension during these years and for assuring

me the best conditions to do my work. Thanks Professor Adélio for being always

available and for your valuable comments.

To Eng. Silvano Tosti, from ENEA (Frascaty, Italy), I acknowledge for his

collaboration (namely for providing the membranes) and sharing his expertise

on Pd-based membranes.

I am grateful to Prof. Vicente Rives, Prof. Miguel Angel Vicente and Prof.

Raquel Trujillano for welcoming me twice in the University of Salamanca (Spain)

and for providing me guidance and sharing their knowledge in the synthesis and

characterization of hydrotalcite materials.

I am grateful to Prof. Alírio Rodrigues for giving me the opportunity to finish

the adsorption experiments at LSRE and move forward with my work when I

faced endless problems with the gas sorption analyzer at LEPABE. I am grateful

to Dr. Alexandre Ferreira for his help on the adsorption experiments and for

sharing his knowledge during my stay at LSRE.

I would like to thank to Dr. Miguel Soria for helping me assembling the

experimental unit, for his useful advices on experimental questions and for his

scientific inputs in the works we have performed together. I am also grateful for

our Monday morning talks about Sporting CP and River Plate performances

during the weekend, which were very pleasant, especially when both teams had

won.

To all my lab mates, FEUPsal team colleagues and friends outside the

“academic and scientific world” I acknowledge for their camaraderie and

X

friendship, for the relevant and less relevant discussions which have made these

years very enjoyable.

I acknowledge companies Sasol and Clariant for supplying the materials

that I used in this work. I am grateful to Ana Taborda (Clariant) for her attention

and sympathy.

To my parents António and Clarisse and my brothers Vanessa and

Armando for their unconditional love.

To my wife Cristina for her love, patience, and generosity. This thesis is

dedicated to her… and to our son Gustavo.

I acknowledge the Portuguese Foundation for Science and Technology

(FCT) for my PhD grant (SFRH/BD/110580/2015) financed by national funds of

the Ministry of Science, Technology and Higher Education and the European

Social Fund (ESF) through the Human Capital Operational Programme (POCH).

Furthermore, I would like to acknowledge also partial financial support of:

● Project PTDC/EQU/ERQ/098730/2008 financed by the European Fund for

Regional Development (ERDF) through COMPETE - Operational Programme

for Competitiveness Factors and through national funds by FCT;

● Project NORTE-07-0124-FEDER-000026 - RL1_ Energy funded by the ERDF

through the COMPETE and ON.2 – O Novo Norte - North Portugal Regional

Operational Programme and through national funds by FCT;

● Project PEst-C/EQB/UI0511 funded by national funds through FCT;

● Portugal-Spain Cooperation Integrated Action Nº E-59/10 funded by CRUP.

● Project POCI-01-0145-FEDER-006939 (Laboratory for Process Engineering,

Environment, Biotechnology and Energy – UID/EQU/00511/2013) - funded by

the ERDF through COMPETE2020 - Programa Operacional Competitividade e

Internacionalização (POCI) – and by national funds through FCT;

● Project “LEPABE-2-ECO-INNOVATION”, with the reference NORTE‐01‐

0145‐FEDER‐000005, supported by Norte Portugal Regional Operational

Programme (NORTE 2020), under the Portugal 2020 Partnership Agreement,

through the ERDF.

XI

XIII

Contents

Abstract I

Resumo V

Acknowledgments IX

Chapter 1. General Introduction 1

1.1 Relevance and Motivation 2

1.1.1 Options for CO2 capture 3

1.1.1.1 Absorption 5

1.1.1.2 Membranes 6

1.1.1.3 Cryogenic distillation 9

1.1.1.4 Adsorption 10

1.1.2 Options for CO2 conversion 12

1.1.3 The Power-to-Gas concept 14

1.2 Overview of the Portuguese energy sector 17

1.2.1 Energy dependence 17

1.2.2 Energy from fossil fuels 19

1.2.2.1 Oil 19

1.2.2.2 Coal 20

1.2.2.3 Natural Gas 21

1.2.3 Energy from renewables 24

1.2.3.1 Geothermal 27

1.2.3.2 Photovoltaic 27

1.2.3.3 Biomass 28

1.2.3.4 Hydropower 29

1.2.3.5 Wind 31

1.2.4 Prospects for Power-to-Methane 34

XIV

1.2.4.1 Surplus renewable power 34

1.2.4.2 CO2 sources and availability 35

1.2.4.3 Natural gas grid 39

1.2.4.4 Summary 45

1.3 Objective and outline of the thesis 48

1.4 Acronyms 51

1.5 References 52

Chapter 2. Direct CO2 hydrogenation to methane and methanol from post-

combustion exhaust streams – a thermodynamic study 61

2.1 Introduction 62

2.2 Methodology 64

2.3 Results and Discussion 67

2.3.1 Strategies for CO2 valorisation: CH4 or CH3OH? 67

2.3.2 CO2 methanation: effect of pressure, temperature and H2/CO2

ratio 70

2.3.3 Direct CO2 methanation from coal-fired power plant (CF-PP)

flue gas streams 72

2.3.3.1 Effect of H2O 72

2.3.3.2 Effect of O2 74

2.3.3.3 Simultaneous effect of H2O and O2 75

2.4 Technological implementation 76

2.5 Conclusions 78

2.6 References 79

Chapter 3. High temperature CO2 sorption with gallium-substituted and

promoted hydrotalcites 83

3.1 Introduction 84

3.2 Experimental 86

3.2.1 Chemicals and gases 86

3.2.2 Sorbents preparation 86

XV

3.2.3 Sorbents characterization 88


3.3.1 Sorbents physicochemical characterization 89

3.3.1.1 TG / DTG 89

3.3.1.2 XRD 90

3.3.1.3 SEM / EDS 92

3.3.1.4 Physical adsorption of nitrogen at 77 K 94

3.3.1.5 FTIR 95

3.3.2 CO2 sorption experiments 97

3.3.2.1 Determination of adsorption equilibrium isotherms at 573

K – sorbent screening 97

3.3.2.2 Sorption-desorption experiments 100

3.4 Conclusions 107

3.5 References 108

Chapter 4. Intrinsic kinetics of CO2 methanation over an industrial nickel-

based catalyst 113

4.1 Introduction 114

4.2 Mechanisms for CO2 methanation 116

4.2.1 “Carbon intermediate” mechanism 116

4.2.2 “Formate intermediate” mechanism 118

4.2.3 “Formyl intermediate” mechanism 119

4.3 Experimental 121

4.3.1 Experimental setup 121

4.3.2 Kinetic experiments 122

4.3.3 Computational methods 124


4.4.1 Kinetic experiments 125

4.4.1.1 Isothermal regime and catalyst stability 125

4.4.1.2 Identifying the region of kinetic rate control 126

4.4.1.3 Calculation of observed reaction rates 128

XVI

4.4.2 Modelling work 129

4.4.2.1 Model discrimination and parameter estimation 129

4.4.2.2 Model validation 132

4.5 Conclusions 138

4.6 Nomenclature 139

4.7 References 141

Chapter 5. A sorptive reactor for CO2 capture and conversion to

renewable methane 145



5.2.1 Experimental setup and hybrid unit 148

5.2.2 Sorption-desorption/reaction experiments 151

5.2.3 Process indicators: definitions and equations 152


5.3.1 Sorptive reactor under discontinuous operation mode 155

5.3.1.1 CO2 sorption working capacity 155

5.3.1.2 Effect of steam produced during reactive regeneration

156

5.3.1.3 Effect of temperature 165

5.3.1.4 Effect of pressure 168

5.3.2 Sorptive reactor under continuous operation mode 171

5.3.3 Reactor design considerations 174

5.4 Conclusions 175

5.5 Nomenclature 177

5.6 References 179

Chapter 6. Effect of CO and CO2 on H2 permeation through finger-like Pd-

Ag membranes 183


6.2 Theoretical background 186

XVII


6.3.1 Membrane preparation and experimental set up 189

6.3.2 Gas permeation experiments 190


6.4.1 Pure H2 permeance experiments 191

6.4.2 Effect of CO and CO2 on H2 permeance 192

6.4.2.1 Comparison between Sieverts-Langmuir model and

rearranged equation 192

6.4.2.2 Modelling by the Sieverts-Langmuir equation 195

6.5 Conclusions 201

6.6 References 202

Chapter 7. Conclusions and Future Work 205

7.1 Conclusions 205

7.2 Future work 211

7.3 References 214

Appendix A. Supplementary information for Chapter 3 215

Appendix B. Supplementary information for Chapter 4 223

Appendix C. Supplementary information for Chapter 5 233

Appendix D. Experimental Setup 241

1

Chapter 1. General Introduction

This chapter presents the relevance and motivation behind this PhD

study, listing the available options for CO2 capture and conversion. Among

these options, the Power-to-Gas concept is introduced and its

implementation in Portugal forecasted based on the situation of the

Portuguese energy sector. The chapter ends listing the PhD work objectives

and providing the thesis outline.

Chapter 1

2

1.1 Relevance and Motivation

The increase of CO2 concentration in the atmosphere, resulting from

human activity and excessive burning of fossil fuels, is considered the major

contributor to global warming [1]. In this perspective, it is essential to develop

strategies that lead to improved technologies regarding CO2 capture,

sequestration and recycling [2]. Although there are several examples of systems

for capturing CO2 that have been used for several years [3], there are clear

technical and economic advantages in contemplating its conversion, particularly

if substances that can bring added value to the process/company can be

produced [2]. In these cases, CO2 can be seen as a raw material with null (or

even negative) cost, if used to produce chemicals and fuels, a market that opens

potentially gigaton-level applications for CO2 utilization [4].

There are several products into which CO2 can be converted: fuels, the

intermediates of such substances or other value-added products, a topic that

has been the subject of several reviews [4-8]. When choosing the technology to

use for CO2 recycling several criteria should also be considered, such as the

associated costs, the market needs or the easiness of storage and transportation

of the produced fuels/intermediates. For these reasons, the choice of the

technology for CO2 recycling fell in the synthesis of CH4, which can be easily

integrated with the existing natural gas infrastructures [4]. In this process there

is consumption of H2, and so this route should be viable when the H2 proceeds

from non-fossil sources [4, 7]. This is the case, for instance, when H2 is obtained

via water electrolysis. Additionally, the energy required for H2 production must

be also renewable (such as solar, wind, etc.) to approach a sustainable process

from the carbon footprint point of view.

This thesis addresses therefore the sustainable production of methane

while simultaneously reducing carbon dioxide emissions into the atmosphere by

recycling carbon using renewable energy for “green” hydrogen production.

General Introduction

3

1.1.1 Options for CO2 capture

The International Energy Agency (IEA) estimated that CO2 emissions

reached 32.4 Gt in 2014 and this value is expected to grow up to 37.5 Gt of CO2

by 2035 [9]. Additionally, IEA also shows that the emissions resulting from

electricity and heat generation in power plants represented in 2014 about 42 %

of the global CO2 emissions, whereas the transport and industry sectors

accounted for 23 % and 19 %, respectively [9]. Developing more efficient and

economical capture technologies is required to minimize the impact on the

environment caused by greenhouse gases (GHG) emissions when the world’s

economy still intensively relies on the combustion of fossil fuels for energy

production.

The efficiency of capture technologies depends on several conditions such

as the exhaust stream composition, temperature and pressure. CO2 content in

a post-combustion exhaust stream from a natural gas power plant can be as low

as 4 %, up to 15 % in a coal power plant, and between 20-30 % in cement and

steel production plants [10]. Higher values are obtained in pre-combustion

streams, such as those observed in syngas production processes, like

hydrocarbons reforming or gasification. Syngas is commonly produced via the

steam reforming of methane (SRM) (Eq. (1.1)) and water-gas shift (WGS)

reactions (Eq. (1.2)). In these processes, CO2 can be captured by adsorption

during reaction - the so-called Sorption-Enhanced Reaction Processes (SERP)

(e.g. [11-14]).

-1

4 2 2 298 KCH + H O CO + 3H 205.9 kJ molH (1.1)

-1

2 2 2 298 KCO + H O CO + H 41.1 kJ molH (1.2)

Since both SRM and WGS are equilibrium-limited reactions and because

CO2 (product) is being adsorbed, the reaction is shifted and proceeds in the

Chapter 1

4

direction of product formation so that the overall performance is improved.

Therefore, in theory this concept allows obtaining:

I. An higher reactant conversion when compared to a traditional packed-

bed reactor in the same operating conditions;

II. Two separate H2- and CO2-rich streams during the adsorption and

desorption stages, respectively.

In gasification processes, namely in Integrated Coal Gasification

Combined Cycle (IGCCC) systems, CO2 is typically captured by absorption with

amines prior to H2 combustion for electricity production in a gas turbine

combined-cycle [3, 15]. From a thermodynamic point of view, and for streams at

the same temperature and pressure, less work is required for the separation in

streams with high CO2 contents, which is the case found in pre-combustion

approaches as illustrated in Fig. 1.1.

Fig. 1.1 – Minimum thermodynamic work for various coal or gas-to-electricity conversions. NGCC: natural gas combined cycle; PCC: Post-combustion capture; IGCC: Integrated Coal Gasification combined cycle. Reprinted by permission from Springer Nature: Springer, Carbon Capture by Jennifer Wilcox, Copyright (2012).


5

Fig. 1.2 shows a diagram presenting the available CO2 separation

technologies. Each technology will be briefly addressed in the following sections.

Fig. 1.2 - Carbon dioxide separation technologies. Reprinted from Chemical Engineering Research and Design, 89, M. Wang, A. Lawal, P. Stephenson, J. Sidders, C. Ramshaw, Post-combustion CO2 capture with chemical absorption: A state-of-the-art review, 1609-1624, Copyright (2011), with permission from Elsevier.

1.1.1.1 Absorption

CO2 capture by absorption with amines is the most mature and established

solution among the various options available, taking place for over 70 years [3].

Absorption technology is based on the transfer of a solute from a gas- to a liquid-

phase and requires two connected columns. On one column the solute (i.e. CO2)

is being separated due to its affinity with the solvent (amine) and in the second

one it is released while the solvent is regenerated for further usage (Fig. 1.3).

Chapter 1

6

Fig. 1.3 - Typical CO2 absorption process scheme. Reprinted from Separation Science and Technology, 40, D. Aaron, C. Tsouris, Separation of CO2 from flue gas: a review, 321-348, Copyright (2005), with permission from Taylor & Francis.

The efficiency of the removal is based on the CO2 solubility, diffusivity and

chemical reactivity with the solvent. The most common solvent used is

monoethanolamine (MEA), particularly for low CO2 concentrations (<15 vol. %)

[16]; the industrial standard for CO2 removal using MEA is ca. 60 gCO2·kgMEA-1 (or

1.36 molCO2 kgMEA-1) [17]. The operation and maintenance cost including solvent

replacement is about $40-$70 per ton of CO2. Temperature requirements for

solvent regeneration represent ca. 70-80 % of the operating cost [16].

1.1.1.2 Membranes

Membranes can be porous or non-porous and the dominant related

transport mechanisms for each type are illustrated in Fig. 1.4.


7

a) b) c) d)

Fig. 1.4 – Transport mechanisms for (a-c) porous and (d) non-porous membranes: a) convective flow, b) Knudsen diffusion, c) activated transport/molecular sieving and d) solution-diffusion (adapted from [18]).

As illustrated in Fig. 1.4, separation in porous membranes relies on

molecular size difference between species present in the mixture (Fig. 1.4c)

and/or their diffusivity rates (Fig. 1.4b). In the former, only molecules having a

smaller size than the pores are able to pass through the membrane, while on the

latter the fastest molecules are firstly recovered at the permeate side. Depending

on the pore size range, porous membranes can be classified according to the

classes presented in Table 1.1.

Table 1.1 - Porous membranes classifications [19].

Classification Pore size range (nm)

Microfiltration 100-5000

Ultrafiltration 1-100

Nanofiltration 0.5-10

Reverse osmosis < 0.5

However, the kinetic diameter of species typically present in flue gas

streams are very similar (cf. Table 1.2). Thus, preparing membranes with a

narrow pore size distribution is complex and, generally, high selectivity towards

CO2 is difficult to achieve. Besides, water vapour condensation may block the

membrane porous strucuture [3].

Chapter 1

8

Table 1.2 – Kinetic diameter of species typically found in flue gas streams [18].

Molecule Kinetic diameter (nm)

CH4 0.380

CO 0.376

N2 0.364

O2 0.346

CO2 0.330

H2 0.289

H2O 0.280

Examples of porous membranes for CO2 separation includes zeolite

membranes, carbon molecular sieve and metal-organic frameworks (MOF’s)

based membranes [3]. The most common non-porous membranes are based on

polymers (e.g. poly-acetylenes, polyaniline, polyarylates, polycarbonates,

polyetherimides, polyimides, polysulfones) [20]. In the work by Robeson [21], the

relationship between permeability and separation for several gas pairs and

polymeric membranes is given; a trade-off relationship shows that, in general,

the higher the permeability, the lower is the selectivity.

Another class of non-porous membranes are the facilitated-transport

membranes (also referred as carrier mediated transport membranes). These

membranes can be divided in two classes: mobile- and fixed-carrier [22]. The

working principle is based on the reaction of CO2 with a carrier (e.g. amine) to

form a complex at the feed side, being regenerated in the permeate side (c.f.

Fig. 1.5). The remaining feed gases that do not react with the carrier agent

permeate by the solution-diffusion mechanism or remain in the retentate side

[23].

The applicability of membrane separation for CO2 capture from a power

plant exhaust stream is mostly compromised due to the simultaneous low CO2

content and pressure, not providing the required driving-force for effective

permeation [3]. For this reason, membrane systems are more suitable for

application in pre-combustion capture, where a higher CO2 fraction and pressure

can be found. However, to this technology become competitive, cheap,

selective, and stable membranes at high temperatures are required. If these


9

requirements are fulfilled, membrane systems present important advantages

such as the easiness of operation, low maintenance work, easy scale-up and,

eventually, the exemption of a regeneration stage.

Fig. 1.5 – Illustration showing the facilitated-transport mechanism. Reprinted with permission from Industrial & Engineering Chemistry Research, 47, J. Huang, J. Zou, W. S. Winston Ho, Carbon dioxide capture using a CO2-selective facilitated transport membrane. Copyright (2008) American Chemical Society.

1.1.1.3 Cryogenic distillation

Cryogenic distillation is used to separate CO2 from N2 (cf. Fig. 1.6). Thus,

the remaining flue gas species must be previously removed [16].

Fig. 1.6 – Cryogenic distillation process for CO2 separation. Reprinted from Separation Science and Technology, 40, D. Aaron, C. Tsouris, Separation of CO2 from flue gas: a review, 321-348, Copyright (2005), with permission from Taylor & Francis.

As depicted in Fig. 1.6, while CO2 is being liquefied at the bottom of the

column, gaseous N2 is recovered at the top. The major advantage of cryogenic

distillation is that CO2 is obtained in a liquefied state with a purity up to 99.95 %

(i.e. CO2 ready for transportation). The major disadvantage is related to the high

Chapter 1

10

energy consumption necessary to maintain the cool temperature (CO2 triple

point is 216.55 K at 5.185 atm), making this technology costly and thus less

interesting than other options available [16].

1.1.1.4 Adsorption

Adsorption processes showed already their effectiveness in the separation

of dilute streams [3]. Different categories of materials can be used for CO2

adsorption depending on the operating temperature as shown in Fig. 1.7.

Fig. 1.7 – CO2 adsorbent materials options as a function of the temperature [24].

Zeolites, activated carbons and organic-inorganic hybrids are limited to

temperatures up to 400 K. Recently, metal-organic frameworks (MOF’s) have

attracted the attention for CO2 capture. This type of sorbents are made of

transition metal ions and organic linkers resulting in highly organized and porous

structures with high surface areas, but their performance towards CO2 capture

is also substantially detrimentally affected by the presence of water vapour [25,

26].


11

Hydrotalcites are cheap materials that can be used at temperatures

between 473 and 673 K, a range typically found in both post-combustion and

pre-combustion streams. Although hydrotalcites use at temperatures above 673

K has also been reported [27], lithium zirconates and calcium oxides are options

more suited for CO2 capture at high temperatures (673-873 K).

Exhibiting a reasonably good adsorption capacity and stability,

hydrotalcites are easily regenerated by temperature or pressure swing [25, 28].

Moreover, water vapour benefits adsorption because it promotes the

reconstruction of their layered structure (see Fig. 1.8), a characteristic of these

materials known as memory effect [29, 30]. This is a distinctive feature of these

materials compared to other adsorbents, making them suitable for CO2 capture

in processes were water vapour is present (e.g. flue gas from fossil fuels

combustion or in sorption enhanced processes for H2 production (i.e. pre-

combustion capture)).

Fig. 1.8 – Illustration showing the layered structure of a hydrotalcite. Reprinted with permission from Chemistry of Materials, 16, N.D. Hutson, S.A. Speakman, E.A. Payzant, Structural effects on the high temperature adsorption of CO2 on a synthetic hydrotalcite. Copyright (2004) American Chemical Society.

Studies addressing some aspects that influence the hydrotalcite thermal

stability, CO2 adsorption/desorption kinetics and equilibrium such as the

Mg2+/Al3+ ratio, the substitution of the compensating anion, the calcination

temperature, alkali metal doping and the water vapour presence can be found

elsewhere (e.g. [28-33]).

Chapter 1

12

1.1.2 Options for CO2 conversion

Some reviews provide a comprehensive insight about emerging

technologies for CO2 recycling (e.g. [4, 34]). Since CO2 molecule has low

reactivity, its conversion requires high-energy reaction “partners” (e.g. H2). For

this reason, although some existing technologies for CO2 conversion are well-

established (e.g. urea, methanol or methane production), the development of

innovative routes with industrial potential will be surely closely linked to the

design of novel catalysts, particularly for methanol synthesis [2]. Fig. 1.9 shows

the options for CO2 conversion into chemical commodities along with the

corresponding market volume.

Fig. 1.9 – CO2 conversion options into chemicals and market volume. Reproduced from Ref. [35] with permission from the Royal Society of Chemistry.

Among the options presented in Fig. 1.9, the use of CO2 as precursor in

organic carboxylation reactions or in reduction reactions are the most promising

routes. In organic carboxylation reactions, CO2 is used for the production of

carbonates, acrylates, carboxylic acids and polymeric materials [36]. For

instance, polyurethane contains as much as 40 wt. % of CO2 and if used as an


13

insulating material (e.g. in buildings), it is claimed that can avoid up to 70 kg of

CO2 emissions for every kilogram of fixated CO2 [4]. However, it is claimed that

the plastics market is not large enough in order to restrain CO2 emissions as

needed [4].

Using CO2 to produce algae is another open topic in the literature, although

it has been reported that the production costs of algal biomass are still too high

to cultivate algae for biomass-to-energy applications and that, for now, the

balance appears unfavourable when compared to fossil fuel alternatives [2, 4].

On the other hand, CO2 reduction to produce methanol, methane, ethanol,

formic acid, syngas and dimethyl ether opens gigaton-level applications for CO2

utilization [4, 37]. However, these reactions require H2 which is widely produced

by steam methane reforming process [38]. For this reason, H2 production should

use renewable resources, or electrons as reducing agent in electrochemical

applications, so that these CO2 recycling options can be sustainable from the

global carbon footprint point of view (cf. Fig. 1.10).

Fig. 1.10 – Renewable power and CO2 storage routes. Reprinted from Renewable and Sustainable Energy Reviews, 69, M. Bailera, P. Lisbona, L. M Romeo, S. Espatolero, Power to Gas projects review: lab, pilot and demo plants for storing renewable energy and CO2, 292-312, Copyright (2017), with permission from Elsevier.

Chapter 1

14

1.1.3 The Power-to-Gas concept

The “Power-to-Gas” (PtG) concept is based on the conversion of surplus

electricity obtained from a renewable source (sun, wind, etc.) into a gas easily

transported and stored. The conversion to CH4 benefits from the existence of a

well-established natural gas infrastructure into where methane can be injected,

enabling the integration/optimization of power and gas grids; this allows the

stabilization of the power grid driven by electricity consumption through the

reconversion of the energy carrier produced, in this case methane [39].

The PtG concept can thus be subdivided in 2 domains: 1) Energetic,

aiming to improve renewable energy technologies/processes and 2)

Environmental, since promoting the efficiency of renewables and CO2 capture

and utilization, it tackles climate changes resulting from CO2 emissions and

contributes for reducing the carbon footprint. Since the current energy

infrastructure relying on fossil fuels is unlikely to change in the forthcoming

years, the production of valuable CH4 from waste CO2 can thus be an interesting

carbon management solution.

From the technological viewpoint, the process consists on the production

of CH4 from the catalytic reaction between CO2 and H2 through the methanation

(or Sabatier) reaction (Eq. (1.3)).

-12 2 4 2 298 KCO + 4H CH + 2H O = - 165 kJ molH (1.3)

The surplus renewable electricity is used for H2O electrolysis (Eq. (1.4))

to obtain “green” H2 necessary for the methanation.

-12 2 2 298 K

1H O H + O = 285.8 kJ mol

2H (1.4)


15

During electrolysis, the application of an electric potential to two

electrodes leads to the splitting of H2O molecule into hydrogen and oxygen at

the cathode and anode, respectively. The electrolysers can be classified

regarding to the used electrolyte as: i) alkaline water electrolysis (AEL), ii) proton

exchange membrane electrolysis (PEM) and iii) solid oxide electrolysis (SOEC),

also known as high-temperature water electrolysis [40]. Currently, AEL is the

cheapest and most reliable technology, PEM the most promising (in the near

future) for transient operation and SOEC is still in the development phase [41].

With respect to CO2 sources, several processes may provide the

necessary feedstock: coal or biomass gasification (for syngas production),

biomass anaerobic fermentation (for biogas production), or flue gas streams of

varied industrial processes (e.g. coal and natural gas power plants, refineries,

cement and steel industries, etc.), requiring however its separation from other

coexisting species (e.g. N2, H2O, O2, etc.) prior to methanation.

CO2 purification requirements may differ substantially depending on the

carbon source considered, which has direct influence on PtG process

configuration and flexibility.

In the European context, due to the increasing share of renewable energy

across the electricity mix, the demand for balancing the electricity supply and the

demand over spatial and temporal distances is increasing, constituting the major

drivers towards the implementation of PtG processes [42].

The largest PtG plant worldwide is the Audi e-gas plant in Wertle,

Germany, where 3 AEL (total electric power of 6 MW) are used to produce

hydrogen, whereas carbon dioxide comes from a biogas plant [41]. The

substitute natural gas (SNG) production capacity of this plant is, in average,

around 1.4x106 m3/year [43]. A picture of the Audi e-gas plant is shown in Fig.

1.11.

Chapter 1

16

Fig. 1.11 – Audi e-gas plant in Wertle (Germany). Reproduced from Ref. [44]. © Audi AG.

Another demonstration activity in Germany is the KIC-Project, DemoSNG,

where the methanator is fed with a volumetric feed flow rate of 10 Nm3/h [45].

The pilot plant is enclosed in a standard shipping container and is currently under

operation in a biomass gasification plant located in Köping, Sweden.

In Denmark, the BioCatProject developed by Eletrochaea uses biological

methanation as a part of the Power-to-Gas process chain. The plant uses an

advanced version of Hydrogenics’ S1000 alkaline electrolyser to produce H2 that

is fed to the biological methanation system together with CO2 from: raw biogas

from an adjacent anaerobic digester with a composition of approximately 60%

CH4 and 40% CO2, or a pure stream of CO2 supplied by an on-site biogas

upgrading system [46]. The plant is located at the BIOFOS Avedøre wastewater

treatment plant near Copenhagen.

A full list of Power-to-Gas projects in the lab, pilot and demonstration

stages was recently presented in the review by Bailera et al. [47], showing that

most activities take place in Europe, particularly because of the support of

governments from Germany, Denmark and Switzerland.


17

1.2 Overview of the Portuguese energy sector

1.2.1 Energy dependence

The energy dependence of a country is an indicator that expresses the

extent to which an economy relies upon imports to meet its energy needs. The

indicator is calculated as net imports of primary energy (IMP-EXP) divided by

the sum of gross inland energy consumption (GIC) plus international maritime

bunkers (IMB), which are quantities of fuels delivered to ships of all flags that

are engaged in international navigation (cf. Eq. (1.5)) [48].

100IMP EXP

Energy DependenceGIC IMB

(1.5)

Hence, a country that exports more primary energy than imports has a

negative energy dependence or, in a simpler way, it is said to be energetically

independent. The Portuguese energy dependence and of the Euro-economic

area (EU-19) and European Union countries (EU-28) are shown in Fig. 1.12.

Fig. 1.12 – Portuguese energy dependence: a) along recent years and b) comparison

with EU-19 countries in year 2015. Data taken from Eurostat online database [48].

In 2015, Portugal was the 7th country with higher energy dependence

among the EU-19 and EU-28 countries (cf. Fig. 1.12b). None of the EU-19

Chapter 1

18

countries has a negative energy dependence (cf. Fig. 1.12b), all depending on

primary energy imports to satisfy the energetic needs.

The normalized consumption of primary energy (CPE) per type of source

in Portugal is depicted in Fig. 1.13a, for the period 2000-2015. The country

situation is compared with those from EU-19 and EU-28 group countries for the

year 2015 in Fig. 1.13b.

The National Energy Strategy approved by the Portuguese government in

2005 settled strategic policies such as the energy market liberalization, the

promotion of renewable energies and technologies with improved efficiencies

[49]. Consequently, in the following decade the oil share remarkably declined

(i.e. 14.6 %), being replaced by natural gas and renewable energies, whose

values increased 4 % and 9 %, respectively, while the coal share practically

remained constant in the same period (rise of only 1.6 %) (cf. Fig. 1.13a). Still,

oil, natural gas and coal together represented 78 % of the consumed primary

energy in 2015, a value slightly above EU-19 (72 %) and EU-28 (73 %) group

countries, whose patterns are identical (cf. Fig. 1.13b). Notably, the remaining

primary energy consumed was exclusively based on renewable energies (22 %),

which turned Portugal the country with 5th highest share of renewables amongst

EU-28 countries [50]. If only the consumption of primary energy for electricity

generation is considered, then the share of renewables is even higher; in 2015,

44.6 % of all electricity produced in Portugal was obtained from renewable

resources [51].

Nuclear has almost the same weight as renewable energies (ca. 13-15 %)

in EU-19 and EU-28 groups, although it is absent in some members like

Portugal.


19

Fig. 1.13 – Normalized primary energy consumption per type of source in Portugal: a) from 2000 to 2015 and b) comparison with EU groups in 2015. Author calculations based on data taken from Eurostat online database [48] and excluding the negligible contribution of non-renewable waste sources.

A description of the current situation regarding fossil fuels and renewable

energy contributions to the Portuguese energy sector is given in sections 1.2.2

and 1.2.3, respectively, with particular emphasis on renewable power

production, as it is one of the main building blocks of Power-to-Gas technologies.

1.2.2 Energy from fossil fuels

1.2.2.1 Oil

Portugal does not have indigenous oil reserves although regular onshore

and offshore exploration activities have been carried out since 1940. Therefore,

all the oil consumed by the country is imported. Table 1.3 lists the top-5 supplier

countries from 2014 to 2016.

Table 1.3 – Top-5 oil suppliers and corresponding share (based on data taken from [52]).

Top-5 2014 2015 2016

1st Angola (26.1 %) Angola (22.9 %) Angola (24.9 %) 2nd Saudi Arabia (12.6 %) Saudi Arabia (14.2 %) Russia (19.7 %) 3rd Algeria (9.9 %) Kazakhstan (10.6 %) Azerbaijan (11.1 %) 4th Kazakhstan (9.7 %) Algeria (9.5 %) Saudi Arabia (10.8 %) 5th Azerbaijan (9.2 %) Azerbaijan (9.0 %) Kazakhstan (9.3 %)

Imp. oil (106 ton) 7.5 (out of 11.17) 9.1 (out of 13.73) 10.7 (out of 14.09) Nr. oil suppliers 14 15 13

Year

2000 2005 2010 2015

CP

Ei /

CP

Eto

tal x 1

00

0

20

40

60

80

100

a)

b)

EU-28 EU-19 Portugal

CP

Ei /

CP

Eto

tal x 1

00

0

20

40

60

80

100

Oil

Natural gas

Coal

Renewables

Nuclear

Chapter 1

20

In the listed years, Portugal imported oil from 13 to 15 countries and the

top-5 oil suppliers were responsible for around 66-76 % of the total imported oil.

Angola is the major oil supplier with a contribution of ca. 25 %. Diversification of

oil suppliers along the years has contributed to assure reliable and secure

access to energy resources [53].

Final consumption of oil by activity sector is shown in Fig. 1.14. It shows

that the transportation sector is responsible for the largest share of final oil

consumption (ca. 75-79 %). The use of oil by all sectors declined in the 5-year

period, except the Agriculture/Forestry that remained practicaly constant. Within

the Industry sector, the non-metallic minerals industries (e.g. cement, glass, etc.)

are responsible for ca. 50-60 % of oil consumed by the industry sector.

Fig. 1.14 – Evolution of oil products final energy consumption in Portugal by activity

sectors from 2010 to 2015 (data taken from [54]).

1.2.2.2 Coal

After national coal production has ceased in 1994, Portugal dependence

on imported coal to secure its energy needs increased. Portugal imported 4.5 Mt

of coal from Colombia (88.1 %), United States (6.6 %), South Africa (3.5 %) and


21

Ukraine (1.8%) in 2014 [50]. Coal consumed in Portugal is of the bituminous type

and is used essentially for electricity generation in two coal-fired power plants

located in Sines (1250 MW) and Pego (620 MW). These plants act as a backup

system guaranteeing that power demand is fulfilled in periods of low renewable

power production. Coal consumption is particularly dependent on hydrological

conditions, namely when hydropower output is compromised in drought periods.

The remaining imported coal that is used for final energy consumption is

negligible when compared to the quantity used for electricity production. For

instance, in 2015, the quantity of coal used for electricity and for final energy

consumption was 3246 ktoe vs. 12 ktoe (136 PJ vs. 502 TJ), respectively. Only

the industry sector consumes coal for final energy purposes, namely the iron

and steel, chemical/petrochemical and non-metallic minerals industries (e.g.

cement industry), as shown in Fig. 1.15.

Fig. 1.15 – Coal use for final energy consumption by industry sub-sectors from 2010 to 2015 [54].

1.2.2.3 Natural Gas

Portugal absence of natural gas resources requires that the country needs

are completely satisfied though importation, similarly to oil and coal. Table 1.4

Year

2010 2011 2012 2013 2014 2015

Fin

al consum

ption

/ k

toe

0

10

20

30

40

Non-metallic minerals

Iron & Steel

Chemical/Petrochemical

Chapter 1

22

shows natural gas import origins and corresponding volumes regarding years

2014 and 2015.

Table 1.4 – Natural gas imports (106 m3) in the years 2014 and 2015 [55].

Origin 2014 2015

Natural gas (pipeline) 2736 3002 Algeria 2196 2111 Spain 535 891 Not specified 5

Liquefied natural gas (ships) 1523 1687 Algeria 102 210 Qatar 687 224 Nigeria 352 1166 Norway 80 80 Spain* 6 7 Trinidad and Tobago 223 Not specified 73

TOTAL 4259 4777 * LNG imported using tanker trucks.

Around 62-64 % of supplies were received through pipeline, while the

remaining part, in the liquefied state, was transported to Portugal in ships that

unload at the liquefied natural gas (LNG) terminal of Sines port, on the southern

part of Portugal. Only a negligible quantity (6-7x106 m3) was imported, using

tanker trucks, exclusively from Spain. The most important supplier is Algeria,

with a share ranging 49-54 % in the presented years, while Qatar and Nigeria

were the major suppliers of LNG.

Final energy consumption of natural gas by activity sector is shown in Fig.

1.16. The Industry sector accounts for the largest amount of natural gas

consumption (67-74 %), followed by Residential (16-19 %) and Services (13-14

%) sectors. The use of natural gas in Agriculture/Forest and Transportation

sectors is negligible and both sectors represent only ca. 1 % of the total final

consumption. Energy for transportation is secured essentially by oil products (as

shown in section 1.2.2.1) and natural gas represents a negligible quantity; for

instance, in 2015, the quantity of oil and natural gas consumed for transportation

were 6245 ktoe vs. 13 ktoe, respectively.


23

Fig. 1.16 - Natural gas for final energy consumption in Portugal by activity sectors from

2010 to 2015 (data taken from [54]).

Table 1.5 lists the natural gas consumption by the different industrial

sectors during 2005, 2014 and 2015.

Table 1.5 – Final energy consumption (ktoe) of natural gas by the Portuguese Industry (data taken from [54]).

Industry 2005 2014 2015 Δ

(2015/2005)

Paper, Pulp, and Print 38.1 90.2 111.6 2.93 Construction 5.8 13.1 14.5 2.48 Chemical and Petrochemical industry 64.4 142.1 152.7 2.37 Food and Tobacco 66.5 124.6 147.2 2.21 Non-ferrous metal industry 7.6 12.9 16.0 2.09 Machinery 21.3 32.6 36.0 1.69 Iron & steel industry 41.4 47.4 51.1 1.23 Textile and Leather 128.6 131.4 131.9 1.03 Non-metallic Minerals (e.g. cement) 516.6 426.3 441.6 0.85 Wood and Wood Products 9.7 9.0 7.8 0.81 Mining and Quarrying 6.3 5.1 4.6 0.74 Transport Equipment 28.8 14.9 16.6 0.58 Non-specified (Industry) 20.9 7.2 5.8 0.28

Total 956.1 1056.8 1137.4 1.19

Chapter 1

24

The non-metallic minerals sector is the most relevant consumer of natural

gas amongst the industries listed in Table 1.5. Its consumption is practically the

same as the sum of the 2nd, 3rd and 4th industrial sectors with higher natural gas

consumption. Table 1.5 highlights the growing importance of natural gas over

time and across several industrial activities. Natural gas consumption

significantly increased since 2005 in almost every industrial activity (see relative

variation in last column). Amongst them, due to their weight in current natural

gas consumption, it is noteworthy the remarkable demand increase by the

Paper, Pulp and Print (2.93), Chemical and Petrochemical (2.37) and Food and

Tobacco (2.21) industries.

Globally, the natural gas demand increased 19 % in the 2005-2015

decade, which reflects its growing importance in almost every activities of the

Portuguese industry sector.

1.2.3 Energy from renewables

The Renewable Energy Roadmap 21 sets a share for renewables

dissemination in the overall European energy mix of 20 % by 2020 [56]. Some

countries, where Portugal is included, have already reached or even surpassed

this target [41, 51]. The strategic effort to replace fossil fuels by renewable

energies has made Portugal to become one of Europe’s leaders in terms of use

of renewable energy sources (RES) [50]. Table 1.6 lists the amount (in ktoe) of

renewable energy produced in Portugal in the 2005-2015 decade.

Since 2006 that Portugal produces biodiesel which is incorporated almost

completely in diesel and only a small fraction (ca. 1 %) is directly sold to the

market. Soybean and, particularly, colza oils are the most used raw materials

[57].


25

Table 1.6 – Portuguese annual production of renewable energy (ktoe) from 2005 to 2015 [51].

Renewable energy type

2005 2007 2009 2011 2013 2015 Δ

(2015/2005)

Biofuels 0 162 226 330 274 321 - Electricity1 599 1265 1456 1872 2369 1927 3.2 Biomass2 2773 2891 3019 2571 2812 2781 1.0 Other renewables3 20 23 36 61 74 82 4.1

Total 3392 4342 4737 4835 5530 5110 1.5

More than a half of renewable energy produced in Portugal comes from

biomass, although that share declined from 82 % to 54 %, when comparing the

values of 2005 and 2015, respectively. The amount of biomass-based energy

produced remained practically constant along the 2005-2015 decade, while the

production of electricity increased 3.2 times, reaching a share of 38 % of the total

renewable energy produced in 2015 (cf. Table 1.6). It is note to worth that

electricity values shown in Table 1.6 consider only hydro, wind, photovoltaic and

geothermal contributions, although an important share of biomass is also used

for electricity generation in thermoelectric and co-generation plants.

Since renewable power is one of the main building blocks of the Power-to-

Gas concept, it will be specifically addressed in more detail below. The current

status of energy production from biofuels and biomass can be found elsewhere

(e.g. [50]).

The investment made on the different RES for power production is

highlighted through the analysis of the installed capacity (MW) values listed in

Table 1.7. The most established RES’s for electricity production in Portugal are

hydro and wind, both totaling 90 % of the installed capacity. Biomass is the third

RES with higher installed capacity, followed closely by photovoltaic, which

remarkably increased 150 times in the 2005-2015 decade. During this period,

1 Includes the contribution of hydro, wind, photovoltaic and geothermal power. 2 Includes the contribution of biogas. 3 Includes solar (for thermal purposes) and (low enthalpy) geothermal sources.

Chapter 1

26

wind energy installed capacity increased 4.7 times and was by far the type of

RES with the highest absolute variation (i.e. 3971 MW).

Table 1.7 – RES installed capacity (MW) for electricity production and corresponding variation in the 2005-2015 decade [51, 58].

RES 2005 2007 2009 2011 2013 2015 Δ

(2015/2005)

Geothermal 18 29 29 29 29 29 1.6 Photovoltaic 3 15 110 175 299 451 150.3 Biomass 429 449 518 712 718 726 1.7 Wind 1063 1699 3564 4378 4731 5034 4.7 Hydro 4816 4853 4883 5330 5533 6053 1.3

Total 6329 7045 9104 10624 11310 12293 1.9

Among biomass, it should be mentioned the evolution of biogas

production, whose installed capacity increased from 8 MW (in 2005) to 85 MW

(in 2015), while the capacity for energy generation from urban solid wastes only

increased 3 MW, reaching a total capacity of 89 MW in 2015.

The exploitation of the installed capacity to produce renewable power from

the different RES is provided in Table 1.8.

Table 1.8 – Annual renewable power production (GWh) [51, 58].

RES 2005 2007 2009 2011 2013 2015 Δ

(2015/2005)

Geothermal 71 201 184 210 197 204 2.9 Photovoltaic 3 24 160 282 479 799 266.3 Biomass 1651 1883 2086 2924 3052 3104 1.9 Wind 1773 4036 7577 9162 12015 11608 6.5 Hydro 5118 10449 9009 12114 14868 9800 1.9

Total 8616 16593 19016 24692 30610 25514 3.0

The values listed in Table 1.8 shows that renewable power production has

considerably increased in the 2005-2015 decade for all RES, as discussed in

the following sections.


27

1.2.3.1 Geothermal

Among the RES, high temperature geothermal resources are confined to

the Azores archipelago where this kind of energy plays an important role. Two

geothermal power plants in operation at S. Miguel island, corresponding to a

global installed capacity of 23 MW, are responsible for the production of 42 % of

the consumed electricity (i.e. around 22 % of the archipelago total demand).

Plans to increase the installed capacity up to 28.5 MW until 2019 have been

reported [59]. Fig. 1.17 shows the geothermal plant in Ribeira Grande.

The IEA reported that enhanced geothermal systems technology, which

uses thermal energy from high-temperature rocks (dry rocks) located at great

depths, may be suitable to explore the potential geothermal resources in

mainland and be tested in the future [50]. Still, Portugal is the 5th country among

IEA-29 members with the highest share of geothermal energy used for power

production [50].

Fig. 1.17 – Geothermal plant (13 MW) in Ribeira Grande (S. Miguel island, Azores) [60].

1.2.3.2 Photovoltaic

Power production in photovoltaic plants was negligible in 2005 but, after

an exceptional improvement, it reached 799 GWh in 2015. Portugal has the best

Chapter 1

28

yearly solar radiation in Europe after Cyprus, particularly in Alentejo region, in

the south of the territory, where the country has a current installed capacity of

162 MW (out of a total of 467 MW) [51, 61]. Moura photovoltaic plant (Fig. 1.18)

is in fact one of the biggest in world with an installed capacity of 46 MW.

Fig. 1.18 – Photovoltaic plant (46 MW) in Moura (Alentejo, south of Portugal) [62].

It is expected that solar energy will play an important role in decentralised

power production. A mini-generation programme created in 2011 has a target to

install approximately 250 MW of new capacity by 2020 in the country [50].

1.2.3.3 Biomass

Regarding biomass, the most common resources available in Portugal are

wood residues, animal waste and municipal solid waste [61]. It was estimated

that the country’s total biomass potential is 42.5 TWh/year, being municipal solid

wastes the main resource (17.0 TWh/year) [61]. Based on the estimated

potential, it was also reported that the use of municipal solid wastes, animal

manure and waste waters are still underexploited [61]. In 2015, 586 GWh of

power was generated from biogas and urban solid wastes (ca. 294 GWh each),

together representing 19 % of the total power produced from biomass (i.e. 3.10

TWh).


29

Power production from biomass is more developed in the center region of

the country, representing 62 % of the total power produced from biomass in

2015. Fig. 1.19 shows the largest biomass power plant of the country (30 MW)

located in Figueira da Foz.

Fig. 1.19 – Celbi biomass (forest residues) power plant (30 MW) located in Figueira da Foz (center of Portugal) [63].

1.2.3.4 Hydropower

Hydropower production takes place in 184 hydropower plants, considering

both large (≥ 10 MW) and small plants (<10 MW) [64]. Portugal mainland most

important river basins are: Lima, Cávado, Mondego, Tejo, Guadiana and,

particularly, Douro, which is responsible for more than a half of the hydropower

generated in the country [51]. Portugal has storage, run-of-the-river and

pumped-storage type hydropower plants. Storage plants accumulate large

quantities of water that can be used on the driest months, while run-of-river may

include a small storage capacity, but generally the turbines are in operation

depending on the river flow. Pumped-storage plants are conventional plants that

were modified to include a system for pumping water from a lower elevation

reservoir to a higher elevation, allowing to store energy in the form of

gravitational potential energy. It is used for load balancing, i.e. to bring back

water to the dam which was used previously in a period of high demand. This

Chapter 1

30

type of plants has recently gained a renewed attention as a way to store off-peak

electricity produced by renewable resources such as wind or photovoltaic.

Portugal has a pump-storage installed capacity of 2.44 GW [65].

Table 1.9 lists hydropower generation per type of plant and river basin in

the year of 2015. Up to the author’s knowledge, the values of power production

through pumped-storage plants per river basin are not publicly available, but a

global production of 1.16 TWh in 2015 was reported by Redes Energéticas

Nacionais (REN) [65].

Table 1.9 – Hydropower generation (GWh) by type of plant and river basin in 2015 [51]. River basin Storage Run-of-River Total %

Lima 484 5 489 5.0 Cávado 1180 29 1209 12.3 Douro 366 5422 5788 59.1 Mondego 322 88 410 4.2 Tejo 415 320 735 7.5 Guadiana 812 0 812 8.3 Others 0 355 355 3.6

Total 3581 6219 9800 100

Hydropower production in 2015 was affected by the hydrological

conditions and the amount of generated power was considerably lower than

typical values found, representing only 60 % of the power produced in 2014. Still,

Table 1.9 highlights the importance to the global hydropower production of

storage and run-of-river plants located in Cávado and Douro river basins,

respectively. Fig. 1.20 and Fig. 1.21 shows the biggest plants in Cávado and

Douro River, respectively.


31

Fig. 1.20 – Photograph of Caniçada plant (storage) in Cávado river (63 MW) [63].

Fig. 1.21 – Photograph of Miranda plant (run-of-river) in Douro river (369 MW) [63].

1.2.3.5 Wind

Portugal had 255 wind parks with 2604 turbines in operation in 2015,

corresponding to a total installed capacity of 5034 MW [51]. Fig. 1.22 shows

how the installed capacity, wind power production and annual equivalent hours

at full capacity (HFC) were distributed countrywide in 2015.

Chapter 1

32

Fig. 1.22 – Wind power installed capacity, generation and annual equivalent hours at full capacity (HFC) for year 2015 (adapted from [51]).

The HFC is determined dividing the generated output (MWh) by the

installed capacity (MW), providing a way to compare the performance of wind

turbines at different locations.

Fig. 1.22 highlights that wind power production is massively obtained in

the Center region, followed by the North, together representing 87% of the global

production [51]. Globally, wind power was generated in 2305 equivalent hours


33

at full capacity, being the North, Azores and Madeira the only regions with an

HFC lower than the global.

The performance of wind turbines can also be expressed in terms of a

capacity factor, where the number of equivalent hours at full load is normalized

by the number of hours available in a certain period, allowing to assess the

amount of power that was produced to the maximum possible. Table 1.10 lists

the capacity factor of each region for an ideal availability of 8760 hours (i.e. hours

available in one year and without excluding shutdown periods, for instance, for

maintenance).

Table 1.10 – Annual wind power capacity factor obtained in different regions of Portugal in 2015 (author calculations based on data taken from [51]).

Region Capacity factor

Center 0.44 Algarve 0.30 Alentejo 0.30 Lisbon 0.29 North 0.26 Azores 0.25 Madeira 0.19

The annual capacity factor ranged from 0.19 (Madeira) up to 0.44 (Center),

while the remaining regions had a value between 0.25 and 0.30. However, it is

noteworthy to mention that the capacity factor can be considerably different

depending on the time frame considered (year, month, etc.). Silva et al. [66] used

historic wind power generation time series (up to 5 years) and reported that the

capacity factor in Portugal can be 1.5 times higher in winter than in summer. The

analysis on a hourly basis also showed that wind generation is higher during

base-load and off-peak periods. Unfortunately, the amount of wind power that

was lost during these periods was not reported.

Chapter 1

34

1.2.4 Prospects for Power-to-Methane

1.2.4.1 Surplus renewable power

The IEA recently reported that instantaneous and daily renewable

electricity output in Portugal regularly exceeds national demand and that the

surpluses are either used in pumped-storage plants or exported [50].

Table 1.11 lists the amount of power that was consumed and produced in

the country through pumped-storage in the period between 2010 and 2015.

Table 1.11 – Pumped-storage power consumption and production (GWh) (author calculations based on data taken from [54]).

Pumped-storage power (GWh)

2010 2011 2012 2013 2014 2015

Consumed 512 737 1331 1459 1081 1460 Produced 399 575 1038 1138 843 1139

Table 1.11 shows that the amount of surplus power that was consumed in

pumped-storage plants increased 2.85 times in 2015 compared to the 2010

value; the storage round-trip efficiency is 78 %. However, it should be recalled

that pumped-storage plants represent huge capital investments and have

associated several environmental impacts that constitute important barriers

towards their adoption [67]. Besides, the storage capacity declines in periods of

high rainfall and that may compromise the ability to store all the exceeding power

being generated.

Another option is to export the surpluses to Spain. However, it has been

reported that when feed-in tariff supported technologies (like wind) generate

moments with high exports, Portugal provides cheap electricity to Spain, but

covers its high production costs internally, resulting in net costs to the

Portuguese Electricity System [68]. The feed-in tariff, a mechanism used by

countries to foster the diffusion of renewables, is always paid to power producers

independently of the power generated being used in Portugal or outside.

Moreover, the tariff was guaranteed to power producers for a 20-year period, the

longest reported among several European countries [68]. Therefore, it was


35

recommended that exports should be reduced at high-generation moments,

releasing the condition of feeding all renewables to the grid, and allowing for spill

of wind generation and/or invest in storage technologies [68].

Pumped-storage plants have been to date the main solution adopted in

Portugal. However, to avoid the associated limitations and low extension

potential of these large-scale plants, the use of decentralized units is foreseen

to increase the storage capacity of the country and to avoid exportations in

unfavorable terms. To this end, Power-to-Gas technologies, particularly Power-

to-Methane could be selected, offering the possibility to integrate the power and

gas grids. The methane produced in this way, often called substitute natural gas,

can thus be injected into the grid or stored in dedicated reservoirs, such as salt

caverns or in LNG tanks after being compressed, and used later when the

demand is higher.

1.2.4.2 CO2 sources and availability

The data presented in this section refers exclusively to GHG emissions of

CO2 and not to other GHG species and their corresponding CO2 equivalents.

Fig. 1.23 illustrates the evolution of CO2 emissions in the country from 2005 to

2015, using data taken from the GHG emissions inventory regularly performed

by the Portuguese Environmental Agency.

Fig. 1.23 shows that CO2 emissions are essentially divided in two

categories. The most relevant class is related to CO2 generated from the

combustion of fuels for energy production (ca. 90 % of all CO2 emitted in 2015)

and the other to CO2 produced in industrial processes. It should be recalled that

emissions from biomass combustion are excluded from the national emissions

totals since carbon released had been in fact fixed from atmosphere by the

photosynthetic process and when is burnt and returns to atmosphere does not

increase the atmospheric/biosphere CO2 pool [69]. However, CO2 emitted from

biomass combustion accounted 11.5 Mt in 2015 [69].

Chapter 1

36

Fig. 1.23 – CO2 emissions evolution resulting from energy production and industrial

processes and product use (adapted from [69]).

Regarding to the evolution of the emissions along time, while CO2

production from industrial processes remained almost constant in the reported

period, the emissions resulting from energy production declined from 69.1 Mt to

47.7 Mt between 2005 and 2014, due to a combination of increased renewable

power production and economic slowdown [50]. A slight increase of emissions

was observed in 2015 (total of 52 Mt), which was a reflex of higher primary

energy consumption for power production, namely of coal and natural gas.

Additionally, final energy consumption also increased, particularly in road

transport, natural gas and electricity [69].

Fig. 1.24 shows CO2 emissions related to energy production in 2015 by

type of sector and sub-sector. The largest point emission sources for energy

production considered were: 16 power plants, 2 oil refineries plants, 1 iron and

steel industry, 1 petrochemical unit, 1 carbon black industrial plant, 8 paper pulp

plants and 6 cement plants.


37

Fig. 1.24 – CO2 emissions due to energy production by sector (2015) (adapted from [69]).

The amount of CO2 emissions from Public Electricity and Heat Production

and Road Transportation stand out amongst other sub-sectors. Together, these

sub-sectors are responsible for 31 Mt of emitted CO2, which corresponds to 60

% of the national totals. Afterwards, the non-metallic minerals sub-sector was

responsible for the emission of 2981 kt of CO2, being the principal emitter

amongst the Manufacturing Industries and Construction sector. Besides energy

production, the mineral industry also stands out in what concerns CO2 emissions

resulting from industrial processes (cf. Fig. 1.25).

2015

Pu

blic

ele

ctr

icity

an

d h

ea

t p

rod

uctio

n

Pe

tro

leu

m r

efin

ing

Iro

n a

nd

ste

el

Ch

em

ica

ls

Pu

lp, p

ap

er

an

d p

rin

t

Fo

od

pro

ce

ssin

g, b

eve

rag

es a

nd

to

ba

cco

No

n-m

eta

llic m

ine

rals

Oth

er

Do

me

stic a

via

tio

n

Ro

ad

tra

nsp

ort

atio

n

Ra

ilwa

ys

Do

me

stic n

avig

atio

n

Co

mm

erc

ial/in

stitu

tio

na

l

Re

sid

en

tia

l

Ag

ricu

ltu

re/f

ore

str

y/fish

ing

CO

2 e

mis

sio

ns (

kt)

0

500

1000

1500

2000

2500

3000

14000

16000

Energy industries

Manuf. ind. & construction

Transport

Other sectors

Chapter 1

38

Fig. 1.25 – CO2 emissions from industrial processes and product use (non-energy use) in 2015 (adapted from [69]).

The mineral industry sector was responsible for the majority of emissions

(3794 kt), being cement production the most relevant activity with 2921 kt,

followed by other processes such as in fertilizers production (354 kt), lime

production (351 kt), and glass production (167 kt).

The second largest CO2 emitter sector is the Chemical Industry, namely

by the petrochemical and carbon black production activity with 650 kt. In the

past, ammonia production process was also an important contributor within this

sector, but since 2008 the only plant operating in the country was closed [69].

The non-energy products from fuels and solvent use comprises emissions

resulting from solvents, lubricants and paraffin wax uses by several industries

(e.g. plastics, wood products, rubber industry, metalworking industry, etc.). CO2

emissions associated to the metal industry come from secondary steel making.

2015

Mineral Industry

Chemical Industry

Metal Industry

Non-energy products from fuels and solvent use

CO

2 e

mis

sio

ns (

kt)

0

250

500

750

1000

3000

4000


39

From the analysis of Fig. 1.24 and Fig. 1.25, it can be concluded that

Portugal offers a wide variety of CO2 sources, essentially diluted in flue gas

streams, with a content between 5 vol. % (natural gas combustion) to 40 vol. %

(e.g. cement) that can be selected to couple in Power-to-Methane applications,

requiring however a previous CO2 capture/purification stage to separate it from

other contaminants. Criteria to select the best CO2 source for Power-to-Methane

applications include:

- Continuous access to CO2 in a stream having low concentrations of

severe catalyst and sorbent poisons (e.g. H2S);

- Proximity to the national natural gas grid for methane injection to

avoid/minimize storage and transportation costs;

- Proximity to renewable electricity plants that will power the water

electrolysis unit in periods where production exceeds demand, minimizing

distribution losses;

- Interest on recycling of the methane produced, for instance, if the

selected site has a natural gas co-generation plant;

- Interest on recycling oxygen produced during water electrolysis to the

process, which would further benefit the whole process from, at least, the

economic point-of-view.

1.2.4.3 Natural gas grid

Portugal has a well-established natural gas storage, transportation and

distribution infrastructure (cf. Fig. 1.26).

The main pipeline goes next to the coast from Sines until Valença do

Minho (and onwards to Spain), where the main natural gas consumption points

are located. It has several branches and 2 lines towards the interior of the

country, one of which that ends in Campomaior, making the connection to the

Spanish pipeline in Badajoz (cf. Fig. 1.26). This interconnection allows Portugal

to receive up to 3.5x109 m3/year of natural gas from Spain with origin in Algeria,

while the interconnection in the north (Valença do Minho/Tuy) has a lower

Chapter 1

40

capacity (0.8x109 m3/year). Still, the highest entry capacity to the grid is provided

by the LNG terminal in Sines (i.e. 5.3x109 m3/year) [50].

Fig. 1.26 – Portugal natural gas storage and transportation infrastructure [65].

Both interconnections with Spain are fully reversible and a third one in

Bragança/Zamora is planned. This connection (dashed line in Fig. 1.26) would

allow to comply with the European Regulation 994/2010 that establishes the

dispositions required for security of supply and the correct functioning of the

natural gas market, being identified by the European Commission as a project

of common interest [50, 55]. Currently, the national pipeline network has an


41

extension of 1375 km and contains 202 pipeline stations: 46 block valve stations,

70 junction stations, 84 gas regulation and metering systems and 2 custody

transfer stations [50].

Sines terminal receives LNG from large vessel ships with a capacity from

45x103 up to 216x103 m3. These ships unload into 3 LNG storage tanks having

a combined capacity of 390x103 m3, corresponding to ca. 242x106 m3 of natural

gas [50, 65] – cf. Fig. 1.27. The terminal is equipped with 5 vaporizers that use

sea water as thermal fluid to gasify LNG, which is further compressed to 77 barg

and injected into the gas grid [50]. The terminal facilities also includes a filling

station that may load up to 4500 tanker trucks a year to transport LNG to

locations which are not covered by the pipeline transmission network [50].

Fig. 1.27 – Two aerial views of the LNG reception, storage and regasification facilities at Sines terminal [70].

Another fundamental component of the national natural gas grid is the

combined underground storage capacity of 333x106 m3 provided by 6 salt

caverns located in Carriço, strategically placed in the middle of the main high-

pressure pipeline (cf. Fig. 1.26) [65, 71]. The reasons for its construction were:

1) the storage of strategic reserves and 2) to balance supply and demand,

namely it’s seasonal and daily fluctuations, thus securing natural gas supply [71].

Chapter 1

42

The caverns are located in the Lusitanian basin, in particular in the Monte

Real salt structure, which has a length of 27 km, a width of 6 km and 2 km of

height (cf. Fig. 1.28) [71].

Fig. 1.28 – Illustration of the Monte Real salt structure in the Lusitanian basin. Reproduced with permission from Ref. [71]. © KBB Underground Technologies GmbH.

Carriço underground storage facilities are positioned next to the sea

because during the construction of the caverns large amounts of brine were

produced and could be discharged to the sea. Around 60-80 % of the produced

brine was disposed to the Atlantic Ocean and the remaining purified by Renoeste

to produce salt with 99.5 % purity [71]. Fig. 1.29 shows an aerial view of the

Carriço facilities next to the coast and the situation of 4 existing salt caverns.


43

Fig. 1.29 – Aerial view of Carriço storage facilities and illustration showing the situation of 4 existing salt caverns (out of 6). Reproduced with permission from Ref. [71]. © KBB Underground Technologies GmbH.

Carriço storage facilities allows for gas injection and withdrawal at 110x103

Nm3/h and 300x103 Nm3/h, respectively. Before injection into the grid, the gas is

filtered, to remove solid and liquid particles, compressed and dehydrated in a

vertical absorber (the maximum final gas moisture content is 40 ppmv) [71].

The LNG terminal and Carriço salt caverns provide a total storage capacity

of 575x106 m3. Considering that consumption in 2016 was 4.6x109 m3 [65], the

existing combined capacity allows to stock the equivalent to the amount

consumed by the country in 46 days.

Chapter 1

44

The development plan of REN, the operator of the gas network, considers

the construction of 25 caverns in Carriço, which would increase the storage

capacity up to 1.25x109 m3 [72], but expansion plans were reported to be

currently under review [50].

The country underground storage capacity may be even higher based on

the available area of salt deposits (cf. Fig. 1.30), particularly next to the coast,

but up to the author knowledge few studies were done or are publicly available

where such potential is assessed (e.g. [72]).

Fig. 1.30 – European salt geological map showing salt deposits, gas storage and brine production sites. Reproduced with permission from Ref. [71]. © KBB Underground Technologies GmbH.

In the work by Nunes [72], the author assessed the underground storage

potential in caverns belonging to other salt structures of the Lusitanian basin

(see Fig. 1.28). Several criteria were adopted to choose the best locations.


45

These included rejecting zones that were in close distance to airports, roads and

houses, inside protected areas, far away to the sea and gas grid or not in a plain

field. Afterwards, 3 zones were elected: Nazaré, Caldas da Rainha and Peniche.

The study considered a similar volume and distance between caverns as it

occurs in Carriço and an underground storage potential of 1x109 m3 was

estimated. Still, only in one of the areas belonging to Peniche was possible to

have more than one cavern next to each other (in this case 2 caverns). In another

scenario, the minimum distance to roads was limited to highways and railways,

which changed completely the storage potential to a global value of 1.65x109

m3, through 24 caverns in Nazare (1.2x109 m3), 8 in Peniche (0.4x109 m3) and 1

in Caldas da Rainha (0.05x109 m3) [72]. Hence, a potential storage capacity of

3.14x109 m3 is envisaged, safeguarding 249 days of consumption (based on

2016 data). However, the preliminary assessment of the underground storage

potential made by Nunes [72] should be complemented with the necessary

environmental impact and economic studies.

1.2.4.4 Summary

The present analysis of the Portuguese energy sector highlights that the

country remains with a high dependence on fossil fuels to afford its energetic

needs. However, renewables contribution, particularly for electricity generation,

has made Portugal the 4th EU-18 country with the highest incorporation of

renewables in power production in 2015 (i.e. 44.6 %), a value that reached 57

% in 2016 [65].

So far, the country’s options to manage with the inevitable energy

surpluses that are generated by renewables were to invest in pumped-hydro

storage plants or through exportation to Spain. While these options have their

own pros and cons, as mentioned previously, another option would be to

chemically store those surpluses as a gas, being methane an appropriate choice

following the increasing natural gas consumption of the country (i.e. particularly

by the industry sector with an increase of 19 % from 2005 to 2015).

Chapter 1

46

Methane could be later reconverted in natural gas power plants in periods

of high electricity demand, used in the chemical industry or exported to Spain.

In the latter case, the third interconnection with Spain of the gas grid (see Fig.

1.26) would increase the exporting capacity, which could be further increased as

soon as projected connections are built to link the Iberian Peninsula with France

(see Fig. 1.31) [50].

Fig. 1.31 – Map of the European gas pipeline network and LNG terminals. Reproduced with permission from Ref. [71]. © KBB Underground Technologies GmbH.

Still, natural gas dependence for power production will increase in Portugal

following the decommissioning of Sines and Pego coal power plants by 2021

[73]. Storing surplus renewable electricity as methane would also allow to

diversify natural gas provision, minimizing the dependence and risk of shortage

supply from foreign countries, as it is advised by the Portuguese Directorate-

General for Energy and Geology [73]. Besides, Power-to-Methane applications

in Portugal would benefit from the high quantity of installed wind power capacity


47

within close distance to potential gas storage facilities (cf. Fig. 1.32), making

Portugal a predestined country to implement PtG technologies, as long as

adequate and nearby CO2 sources are also available [74].

Fig. 1.32 – Map showing the wind power parks (WPP) situated within 5 km of the natural gas network; salt deposit formations of the Lusitanian basin are also shown. White colour shows the most attractive power-to-gas plant positions. Reproduced with permission from Ref. [74]. © 2015 IEEE.

Chapter 1

48

1.3 Objective and outline of the thesis

The main objective of this PhD work was the study and development of a

Carbon Capture and Utilization (CCU) technology that can be inserted in a

Power-to-Gas process able to, simultaneously, remove CO2 from flue gas, and

further convert it to methane in a hybrid sorptive reactor unit. The in situ

conversion of captured CO2 avoids the costs associated with its compression

and transport to a methanation reactor.

The reactor bed contains a sorbent material to separate CO2 from other

species present in the flue gas and also a catalyst to accelerate the methanation

reaction during the desorption stage (via reactive regeneration with renewable-

based H2). For the process to run continuously, at least two reactors in parallel

working under cyclic operation are required; while one reactor is performing the

sorption stage, the other is running the reactive regeneration step. This approach

is illustrated in Fig. 1.33.

Fig. 1.33 – Power-to-Gas concept: system boundaries with proposed technological approach for CO2 capture and conversion.


49

Each step (sorption and reaction) was firstly studied individually and,

afterwards, merged in the same unit to proof-the-concept of such integrated

process.

The following paragraphs provide the thesis outline.

Chapter 2 presents a thermodynamic analysis performed using the

software Aspen Plus to check the limitations and optimal conditions when

producing methane (or methanol) using CO2 from flue gas exhaust streams

without previous removal of unnecessary species.

Chapter 3 addresses the synthesis, characterization and testing of

hydrotalcite-based sorbents for CO2 capture at high temperature. After a detailed

physical and chemical characterization, the best sorbent was selected through

the determination of the sorption equilibrium isotherms of all materials at 300 ºC.

Afterwards, its stability towards cyclic operation was checked and the sorption

and desorption kinetics modelled.

Chapter 4 concerns the determination of the reaction kinetics over a

commercial Ni-based catalyst supplied by Clariant. Kinetic tests were performed

using a fixed-bed reactor and an ideal feed stream of CO2 and H2 diluted in N2.

Conversion was measured against the space velocity at different temperatures

and after obtaining reaction rates under different conditions, non-linear

regression was used to fit the kinetic data to different rate expressions and to

obtain the corresponding parameters. Finally, the fixed-bed reactor was

modelled to validate the selected rate equation.

Chapter 5 deals with the proof-of-concept of the sorptive reactor for CO2

capture and conversion to methane using pellets of a commercial potassium

promoted hydrotalcite (from Sasol) and spheres of the commercial Ni-catalyst

tested before (from Clariant). The sorptive reactor was assessed in terms of CO2

sorption capacity, CO2 conversion, CH4 productivity and CH4 purity under

different operation conditions. The major drawbacks encountered were listed

and possible solutions for further reactor improvement suggested.

Chapter 1

50

Chapter 6 focuses on the assessment of the effect of CO and CO2 on H2

permeation through a dense palladium-based membrane. This kind of

membranes can be used for radial feeding of H2 along the sorptive reactor tube

length, to better control the H2/CO2 ratio during reactive regeneration (and

accelerate this stage). This was one of the suggestions made in Chapter 5 to

avoid undesirable CO formation.

Chapter 7 presents the general conclusions of this work and some

suggestions for future work.


51

1.4 Acronyms

Acronym Description

AEL Alkaline water electrolysis CCU Carbon capture and utilization EU-19 Euro economic zone countries EU-28 European Union countries EXP Primary energy exports GHG Greenhouse gas GIC Gross inland energy consumption HFC Annual equivalent hours at full capacity IEA International Energy Agency IGCCC Integrated Coal Gasification combined cycle IMB International Maritime bunkers IMP Primary energy imports LNG Liquefied natural gas MEA Monoethanolamine MOF Metal-organic framework NGCC Natural gas combined cycle PCC Post-combustion capture PEM Proton exchange membrane electrolysis PtG Power-to-Gas REN Redes Energéticas Nacionais RES Renewable energy source SERP Sorption-enhanced reaction processes SNG Substitute (or synthetic) natural gas SOEC Solid oxide electrolysis SRM Steam reforming of methane reaction WGS Water gas shift reaction

Chapter 1

52

1.5 References

[1] G.A. Olah, A. Goeppert, G.K.S. Prakash, Beyond Oil and Gas: The Methanol

Economy, Wiley - VCH Verlag GmbH & Co, Weinhein, 2006.

[2] G. Zakrzewska-Trznadel, European Workshop "CO2: From Waste to Value",

available at http://ec.europa.eu/research/industrial_technologies/pdf/co2-

workshop-30032011_en.pdf, date of last access: 17/01/2018.

[3] J. Wilcox, Carbon capture, Springer, New York, 2012.

[4] E.A. Quadrelli, G. Centi, J.L. Duplan, S. Perathoner, Carbon dioxide

recycling: Emerging large-scale technologies with industrial potential,

ChemSusChem, 4 (2011) 1194-1215.

[5] G. Centi, S. Perathoner, Opportunities and prospects in the chemical

recycling of carbon dioxide to fuels, Catalysis Today, 148 (2009) 191-205.

[6] J. Ma, N. Sun, X. Zhang, N. Zhao, F. Xiao, W. Wei, Y. Sun, A short review of

catalysis for CO2 conversion, Catalysis Today, 148 (2009) 221-231.

[7] C. Song, Global challenges and strategies for control, conversion and

utilization of CO2 for sustainable development involving energy, catalysis,

adsorption and chemical processing, Catalysis Today, 115 (2006) 2-32.

[8] M. Aresta, A. Dibenedetto, Utilisation of CO2 as a chemical feedstock:

opportunities and challenges, Dalton Transactions, (2007) 2975-2992.

[9] International Energy Agency, CO2 emissions from combustion of fossil fuels

: highlights 2016., 2016.

[10] E. Favre, Membrane processes and postcombustion carbon dioxide

capture: Challenges and prospects, Chemical Engineering Journal, 171 (2011)

782-793.

[11] E.R. Van Selow, P.D. Cobden, P.A. Verbraeken, J.R. Hufton, R.W. Van Den

Brink, Carbon capture by sorption-enhanced water-gas shift reaction process

using hydrotalcite-based material, Industrial and Engineering Chemistry

Research, 48 (2009) 4184-4193.

http://ec.europa.eu/research/industrial_technologies/pdf/co2-workshop-30032011_en.pdf

http://ec.europa.eu/research/industrial_technologies/pdf/co2-workshop-30032011_en.pdf


53

[12] E.L.G. Oliveira, C.A. Grande, A.E. Rodrigues, Effect of catalyst activity in

SMR-SERP for hydrogen production: Commercial vs. large-pore catalyst,

Chemical Engineering Science, 66 (2011) 342-354.

[13] K.B. Lee, M.G. Beaver, H.S. Caram, S. Sircar, Novel thermal-swing

sorption-enhanced reaction process concept for hydrogen production by low-

atemperature steam-methane reforming, Industrial and Engineering Chemistry

Research, 46 (2007) 5003-5014.

[14] A.E. Rodrigues, L.M. Madeira, Y.-J. Wu, R. Faria, Sorption-Enhanced

Reaction Processes, World Scientific Publishers, 2017.

[15] A.A. Olajire, CO2 capture and separation technologies for end-of-pipe

applications – A review, Energy, 35 (2010) 2610-2628.

[16] D. Aaron, C. Tsouris, Separation of CO2 from flue gas: A review, Separation

Science and Technology, 40 (2005) 321-348.

[17] H. Gupta, L.S. Fan, Carbonation-calcination cycle using high reactivity

calcium oxide for carbon dioxide separation from flue gas, Industrial and

Engineering Chemistry Research, 41 (2002) 4035-4042.

[18] R.H. Perry, D.W. Green, Perry's Chemical Engineers' Handbook, McGraw

Hill, New York, 1999.

[19] Z.F. Cui, Y. Jiang, R.W. Field, Fundamentals of pressure-driven membrane

separation processes, in: Z.F. Cui, H.S. Muralidhara (Eds.) Membrane

Technology: a practical guide to membrane technology and applications in food

and bioprocessing, Butterworth-Heinemann, Oxford, Uk, 2010.

[20] C.E. Powell, G.G. Qiao, Polymeric CO2/N2 gas separation membranes for

the capture of carbon dioxide from power plant flue gases, Journal of Membrane

Science, 279 (2006) 1-49.

[21] L.M. Robeson, The upper bound revisited, Journal of Membrane Science,

320 (2008) 390-400.

[22] M. Mulder, Basic Principles of Membrane Technology, 2nd ed., Kluwer

Academic Publishers, Dordrecht, 1996.

Chapter 1

54

[23] J. Huang, J. Zou, W.S.W. Ho, Carbon dioxide capture using a CO2-selective

facilitated transport membrane, Industrial and Engineering Chemistry Research,

47 (2008) 1261-1267.

[24]

www.netl.doe.gov/File%20Library/research/coal/carbon%20capture/FE000243

8.pdf, date of last access: 17/01/2018

[25] Q. Wang, J. Luo, Z. Zhong, A. Borgna, CO2 capture by solid adsorbents and

their applications: current status and new trends, Energy and Environmental

Science, 4 (2011) 42-55.

[26] P. Li, J. Chen, J. Zhang, X. Wang, Water stability and competition effects

toward CO2 adsorption on metal organic frameworks, Separation & Purification

Reviews, 44 (2015) 19-27.

[27] M.G. Beaver, H.S. Caram, S. Sircar, Sorption enhanced reaction process

for direct production of fuel-cell grade hydrogen by low temperature catalytic

steam-methane reforming, Journal of Power Sources, 195 (2010) 1998-2002.

[28] E.L.G. Oliveira, C.A. Grande, A.E. Rodrigues, CO2 sorption on hydrotalcite

and alkali-modified (K and Cs) hydrotalcites at high temperatures, Separation

and Purification Technology, 62 (2008) 137-147.

[29] Y. Ding, E. Alpay, Equilibria and kinetics of CO2 adsorption on hydrotalcite

adsorbent, Chemical Engineering Science, 55 (2000) 3461-3474.

[30] M.K. Ram Reddy, Z.P. Xu, G.Q. Lu, J.C. Diniz Da Costa, Influence of water

on high-temperature CO2 capture using layered double hydroxide derivatives,

Industrial and Engineering Chemistry Research, 47 (2008) 2630-2635.

[31] Z. Yong, V. Mata, A.E. Rodrigues, Adsorption of carbon dioxide onto

hydrotalcite-like compounds (HTlcs) at high temperatures, Industrial and


[32] M.K. Ram Reddy, Z.P. Xu, G.Q. Lu, J.C.D. Da Costa, Layered double

hydroxides for CO2 capture: Structure evolution and regeneration, Industrial and


http://www.netl.doe.gov/File%20Library/research/coal/carbon%20capture/FE0002438.pdf

http://www.netl.doe.gov/File%20Library/research/coal/carbon%20capture/FE0002438.pdf


55

[33] C.T. Yavuz, B.D. Shinall, A.V. Iretskii, M.G. White, T. Golden, M. Atilhan,

P.C. Ford, G.D. Stucky, Markedly improved CO2 capture efficiency and stability

of gallium substituted hydrotalcites at elevated temperatures, Chemistry of

Materials, 21 (2009) 3473-3475.

[34] M. Peters, B. Köhler, W. Kuckshinrichs, W. Leitner, P. Markewitz, T.E.

Müller, Chemical technologies for exploiting and recycling carbon dioxide into

the value chain, ChemSusChem, 4 (2011) 1216-1240.

[35] N. MacDowell, N. Florin, A. Buchard, J. Hallett, A. Galindo, G. Jackson, C.S.

Adjiman, C.K. Williams, N. Shah, P. Fennell, An overview of CO2 capture

technologies, Energy & Environmental Science, 3 (2010) 1645-1669.

[36] E.A. Quadrelli, G. Centi, Green Carbon Dioxide, ChemSusChem, 4 (2011)

1179-1181.

[37] M. Aresta, A. Dibenedetto, A. Angelini, Catalysis for the valorization of

exhaust carbon: from CO2 to chemicals, materials, and fuels. Technological use

of CO2, Chemical Reviews, 114 (2014) 1709-1742.

[38] C. Song, Introduction to hydrogen and syngas production and purification

technologies, in: K. Liu, C. Song, V. Subramani (Eds.) Hydrogen and syngas

production and purification technologies, John Wiley & Sons, Inc., New Jersey,

2010, pp. 1-12.

[39] M. Specht, J. Brellochs, V. Frick, B. Stürmer, U. Zuberbühler, The Power to

Gas Process: Storage of Renewable Energy in the Natural Gas Grid via Fixed

Bed Methanation of CO2/H2, in: T.J. Schildhauer, S.M.A. Biollaz (Eds.) Synthetic

Natural Gas from Coal, Dry Biomass, and Power-to-Gas Applications, John

Wiley and Sons, Inc., Hoboken, New Jersey, 2016.

[40] S. Schiebahn, T. Grube, M. Robinius, V. Tietze, B. Kumar, D. Stolten, Power

to gas: technological overview, systems analysis and economic assessment for

a case study in Germany, International Journal of Hydrogen Energy, 40 (2015)

4285-4294.

Chapter 1

56

[41] M. Götz, J. Lefebvre, F. Mörs, A. McDaniel Koch, F. Graf, S. Bajohr, R.

Reimert, T. Kolb, Renewable Power-to-Gas: A technological and economic

review, Renewable Energy, 85 (2016) 1371-1390.

[42] T.J. Schildhauer, Introdutory remarks, in: T.J. Schildhauer, S.M.A. Biollaz

(Eds.) Synthetic Natural Gas from Coal, Dry Biomass, and Power-to-Gas

Applications, John Wiley and Sons, Inc., Hoboken, New Jersey, 2016.

[43] News: Germany/Power-to-gas: Clariant supplies SNG catalyst for first

commercial CO2 methanation plant, Oil Gas European Magazine, 39 (2013) 217.

[44] www.audi-mediacenter.com/en/press-releases/audi-e-gas-plant-stabilizes-

electrical-grid-4499, date of last access: 17/01/2018.

[45] http://www.kit.edu/kit/english/pi_2014_16120.php, date of last access:

17/01/2018.

[46] http://biocat-project.com, date of last access: 17/01/2018.

[47] M. Bailera, P. Lisbona, L.M. Romeo, S. Espatolero, Power to Gas projects

review: Lab, pilot and demo plants for storing renewable energy and CO2,

Renewable and Sustainable Energy Reviews, 69 (2017) 292-312.

[48] Eurostat, Environment and energy database,

http://ec.europa.eu/eurostat/data/database, date of last access 17/01/2018.

[49] International Energy Agency, Energy policies of IEA countries: Portugal

2009 Review.

[50] International Energy Agency, Energy policies of IEA countries: Portugal

2016 Review.

[51] Direção Geral de Energia e Geologia, Renováveis - Estatísticas Rápidas,

abril 2017 (en: Portuguese Directorate-General for Energy and Geology,

Renewables- brief statistics, April 2017).

[52] Direção Geral de Energia e Geologia, Fatura Energética Portuguesa 2016

(en: Portuguese Directorate-General for Energy and Geology, Portuguese

Energetic Invoice 2016).

[53] J. Amador, Energy production and consumption in Portugal: stylized facts,

Economic Bulletin and Financial Stability Report Articles, (2010) 69-83.

http://www.audi-mediacenter.com/en/press-releases/audi-e-gas-plant-stabilizes-electrical-grid-4499

http://www.audi-mediacenter.com/en/press-releases/audi-e-gas-plant-stabilizes-electrical-grid-4499

http://www.kit.edu/kit/english/pi_2014_16120.php

http://biocat-project.com/

http://ec.europa.eu/eurostat/data/database


57

[54] Eurostat, Energy balances,

http://ec.europa.eu/eurostat/web/energy/data/energy-balances, date of last

access: 17/01/2018.

[55] Direção Geral de Energia e Geologia, Relatório sobre Avaliação dos Riscos

que afetam o aprovisionamento de Gás Natural em Portugal- Período 2017-

2025 (en: Portuguese Directorate-General for Energy and Geology, Assessment

Report of Natural Gas provision in Portugal - 2017-2025 period).

[56] European Commission, Renewable Energy Road Map - Renewable

energies in the 21st century: building a more sustainable future, Brussels, 2007.

[57] Entidade Nacional para o Mercado de Combústiveis, Mercado de

Biocombustíveis em Portugal: A atuação da ENMC desde abril de 2015. (en:

National Entity for Fuels Market, Biofuels market in Portugal: ENMC action since

April 2015).

[58] Direção Geral de Energia e Geologia, Renováveis - Estatísticas Rápidas,

outubro 2014 (en: Portuguese Directorate-General for Energy and Geology,

Renewables- brief statistics, October 2014).

[59] L.C. J.M. Carvalho, J.C. Nunes, M.R Carvalho, J. Garcia, R. Cerderia,

Portugal Country Update 2015, in: Proceedings World Geothermal Congress

2015 Melbourne, Australia 2015.

[60] http://siaram.azores.gov.pt/energia/geotermia/_texto.html, date of last

access: 17/01/2018.

[61] S. Ferreira, E. Monteiro, P. Brito, C. Vilarinho, Biomass resources in

Portugal: Current status and prospects, Renewable and Sustainable Energy

Reviews, 78 (2017) 1221-1235.

[62] http://www.portal-energia.com/a-maior-central-solar-do-mundo-no-

alentejo, date of last access: 17/01/2018.

[63] http://www.a-nossa-energia.edp.pt/centros_produtores, date of last access:

17/01/2018.

[64] C. Teotónio, P. Fortes, P. Roebeling, M. Rodriguez, M. Robaina-Alves,

Assessing the impacts of climate change on hydropower generation and the

http://ec.europa.eu/eurostat/web/energy/data/energy-balances

http://siaram.azores.gov.pt/energia/geotermia/_texto.html

http://www.portal-energia.com/a-maior-central-solar-do-mundo-no-alentejo

http://www.portal-energia.com/a-maior-central-solar-do-mundo-no-alentejo

http://www.a-nossa-energia.edp.pt/centros_produtores

Chapter 1

58

power sector in Portugal: A partial equilibrium approach, Renewable and

Sustainable Energy Reviews, 74 (2017) 788-799.

[65] Redes Eletricas Nacionais, Technical Data 2016,

http://www.centrodeinformacao.ren.pt/PT/InformacaoTecnica/DadosTecnicos/

REN%20Dados%20T%C3%A9cnicos%202016.pdf, date of last access:

17/01/2018.

[66] B. Silva, M. Marques, J.C. Matos, A. Rodrigues, R. Pereira, A. Bastos, P.

Cabral, F. Saraiva, Wind power in Portugal: from the potential to the integration

in the electric grid, in: European Wind Energy Conference and Exhibition 2010,

EWEC 2010, 2010, pp. 4508-4515.

[67] J.A.R. Marques, Hidroelectricidade e Barragens Reversíveis: Panorama

actual, in: Departamento de Geociências, Ambiente e Ordenamento do

Território, Universidade do Porto, Porto, 2015.

[68] I. Peña, I. Azevedo, L.A.F. Marcelino Ferreira, Lessons from wind policy in

Portugal, Energy Policy, 103 (2017) 193-202.

[69] T.C. Pereira, T. Seabra, A. Pina, P. Canaveira, A. Amaro, M. Borges, R.

Silva, Portuguese National Inventory Report on Greenhouse Gases, 1990-2015

submmitted under the United Nations Framework Convention on Climate

Change and the Kyoto Protocol, Portuguese Environmental Agency, Amadora,

2017.

[70] http://www.ign.ren.pt/en/terminal-de-gnl3, date of last access: 17/01/2018.

[71] http://www.kbbnet.de/wp-content/uploads/2015/10/150821-Carrico-

Brochure.pdf, date of last access: 17/01/2018.

[72] P.V.C. Nunes, Potencial de armazenamento subterrâneo em cavidades

salinas de gás natural em Portugal, in: Departamento de Engenharia de Minas

e Georrecursos, Instituto Superior Técnico, Lisboa, Portugal, 2010.

[73] Direção Geral de Energia e Geologia, Relatório de Monitorização da

Segurança de Abastecimento do Sistema Elétrico Nacional 2013-2030 (en:

Portuguese Directorate-General for Energy and Geology, Monitoring Report of

the Eletric National System Security of Supply 2013-2030).

http://www.centrodeinformacao.ren.pt/PT/InformacaoTecnica/DadosTecnicos/REN%20Dados%20T%C3%A9cnicos%202016.pdf

http://www.centrodeinformacao.ren.pt/PT/InformacaoTecnica/DadosTecnicos/REN%20Dados%20T%C3%A9cnicos%202016.pdf

http://www.ign.ren.pt/en/terminal-de-gnl3

http://www.kbbnet.de/wp-content/uploads/2015/10/150821-Carrico-Brochure.pdf

http://www.kbbnet.de/wp-content/uploads/2015/10/150821-Carrico-Brochure.pdf


59

[74] F. Heymann, R. Bessa, Power-to-Gas potential assessment of Portugal

under special consideration of LCOE, in: 2015 IEEE Eindhoven PowerTech,

PowerTech 2015, 2015.

61

Chapter 2. Direct CO2 hydrogenation to methane and methanol from post-combustion exhaust streams – a thermodynamic study

The contents of this chapter were adapted from: C.V. Miguel, M.A. Soria, A. Mendes,

L.M. Madeira, Direct CO2 hydrogenation to methane or methanol from post-combustion

exhaust streams – A thermodynamic study. Journal of Natural Gas Science and

Engineering, 2015, 22, 1-8.

The conversion/utilization of waste carbon dioxide is seen as a

complementary option to the well-known capture and storage strategies

(CCS) to substantially reduce atmospheric CO2 (environmental concern).

This approach is attractive regarding CCS strategies because CO2 can be

transformed into a valuable chemical (economic benefit). Among the

options available, methane and methanol are important chemicals that

could be obtained from CO2 hydrogenation and used for energy

production/storage or as intermediaries to other chemicals.

A thermodynamic analysis regarding the hydrogenation of CO2 into

CH4 or CH3OH was carried out. The analysis was performed to check the

limitations and optimal conditions when converting CO2 from flue gas

exhaust streams without previous removal of unnecessary species present

in significant amounts (e.g. N2, H2O and O2). The present analysis supports

that, from the thermodynamic point of view, the conversion of CO2 into CH4

is favoured in comparison to the CH3OH valorisation strategy, for the

considered pressure and temperature ranges.

Chapter 2

62

2.1 Introduction

Carbon dioxide is the end-product of the largest-volume and most globally

applied chemical reaction, the combustion of hydrocarbons and biomass, and it

is well known the growing concern about reducing CO2 emissions due to its

enormous contribution to the greenhouse effect [1]. The Kyoto Protocol has

created the market for carbon credits, a crucial mechanism for valuating CO2

emissions and thus incorporating the pollution effect in the cost structure of the

corporation’s economy [2]. However, top-polluting countries such as Canada

and USA are out of the agreement; moreover, according to the International

Energy Agency, the top 10 polluting countries represent around 2 3 of world CO2

emissions [3]. This means that any solution to solve the carbon dioxide problem

will always depend on the compromise of these countries. So, economic benefits

should be considered together with environmental concerns. In this regard, in

recent years there has been a focus on developing different possibilities for CO2

recycling as complement of the well-known capture, sequestration and storage

approaches, particularly, its conversion into added value products [4-6]. This

new paradigm considers CO2 as chemical feedstock (value) and not only as a

waste that needs to be treated (cost) [7].

Recent works provide together a comprehensive state-of-the-art of the

options available for CO2 valorisation and utilization, including the necessary

timeframe for development, the time of sequestration, the economic

perspectives, etc. [4-7]. Among the options presented in those works are CO2

hydrogenation into methane (Eq. (2.1)) or methanol (Eq. (2.2)), the first being

also known as Sabatier reaction or CO2 methanation.

-12 2 4 2 298 KCO + 4H CH + 2H O Δ = - 165 kJ molH (2.1)

-12 2 3 2 298 KCO + 3H CH OH + H O Δ = - 49.4 kJ molH (2.2)

Direct CO2 hydrogenation to CH4 and CH3OH - a thermodynamic study

63

These two options, however, require expensive H2 which, in turn, is

preferentially produced worldwide using non-renewable feedstock’s, being the

steam methane reforming the most developed and commercialized technology

[8]. In this regard, these routes should be viable in view of CO2 emissions

abatement only when H2 is produced from renewable resources, such as water

electrolysis. Moreover, the energy required for the electrolysis should be also

renewable for the global process to truly allow reducing CO2 emissions [5, 9].

So, in this case, important chemicals such as methane or methanol could be

produced using renewable resources (for H2 production) and waste CO2.

The conversion of CO2 into methanol (reaction 2) has, compared to the

methanation process, the advantage of consuming less hydrogen (see also the

stoichiometry of reaction 1). Moreover, methanol has a higher energy density, is

easier to store and can be used, for example, in the synthesis of important

chemicals such as formaldehyde, methyl tertiary-butyl ether (MTBE), among

others [10]. These advantages are pointed by various personalities as the driving

force for the conversion/recycling of CO2 into methanol, thus alleviating the

dependence on fossil fuels while simultaneously reducing the emission of

greenhouse gases. Among such personalities is the winner of the Nobel Prize in

Chemistry in 1994, Prof. Olah, who clearly supports a strategy of "Methanol

Economy" [11]. On the other hand, the “Power-to-Gas” concept can be a very

interesting way to chemically store the off-peak electricity generated in wind

power stations in the form of methane, which can be further integrated with the

already existing natural gas infrastructures [12-14], as long as the exit process

stream properties complies with the specifications required for natural gas

transport in pipelines.

The necessary CO2 is available from a large variety of emission sources.

However, the International Energy Agency reported that the majority of the world

CO2 emissions arise from post-combustion sources related to electricity and

heat production (41 % in 2010), particularly, from coal-fired power plants and the

combustion of oil or gas, respectively 43 %, 36 % and 20 % of the electricity

Chapter 2

64

related CO2 emissions. Previous works addressed, from the thermodynamic

standpoint, the CO2 valorisation into CH4 [15] or CH3OH [16]; in particular, Gao

et al. [15] studied the effect of species present in syngas produced by coal or

biomass gasification, where CO is the major species present (rather than CO2).

In this work, however, CO2 valorisation was assessed considering its direct

conversion from a real coal-fired power plant exhaust stream. Moreover, the

effect of pressure, temperature, H2/CO2 ratio and the presence of major co-

existing species present in flue gas streams (N2, O2 and H2O) was systematically

assessed regarding CO2 conversion, product yield and selectivity.

2.2 Methodology

Thermodynamic analysis was performed using the Gibbs reactor model

(RGibbs) available in the Aspen Plus software from AspenTech. RGibbs models

simultaneous phase and chemical equilibria minimizing the Gibbs free energy

and does not require the specification of the reactions involved and their

stoichiometry [17, 18].

The total Gibbs free energy of a system is given by the sum of the chemical

potential of all the N species [15, 19, 20]:

1

N

T i i

i

G n

(2.3)

where in is the number of moles of species i, which chemical potential,

i , is

given by:

0

0ln i

i fi

i

fG RT

f (2.4)


65

where 0

fiG is the standard Gibbs function of species i formation, R is the ideal

gas constant, T is the absolute temperature and if and 0

if stand for the fugacity

and standard fugacity, respectively. For the reaction equilibrium in the gas

phase:

i i if y P (2.5)

0 0

if P (2.6)

where iy is the mole fraction of species i, i is the fugacity coefficient, P is the

pressure of the system and P0 is the standard pressure. So, combining Eqs.

(2.3)-(2.6) and applying Lagrange multipliers, used to incorporate the constraints

related to conservation of the total amount of individual chemical elements into

the body of the problem, the constrained function to be minimized, objf , is [18]:

obj

1 1

0

01 1 1

ln

m N

T j j i ji

j i

N m Ni i

i fi j j i ji

i j i

f G b n a

y Pn G RT b n a

P

(2.7)

where j is the Lagrange multiplier, jb the total amount of element j in the

mixture, jia the number of atoms of element j in species i. Whenever the

presence of solid carbon was considered in simulations, Eq. (2.7) should be

changed to the following one [20]:

0 0

obj ( )01 1 1

lnN m N

i ii fi j j i ji C fC s

i j i

y Pf n G RT b n a n G

P

(2.8)

Chapter 2

66

Fugacity was estimated using the Soave-Redlich-Kwong (SRK) equation of state

contained in the Aspen Plus database, as suggested elsewhere for similar

conditions [15].

The starting gas composition (mixture 1 in Table 2.1) was obtained from a

typical coal-fired power station flue gas stream. The inlet gas compositions used

for calculations (mixtures 2-10, Table 2.1) were established to independently

analyse, in realistic scenarios, the effects of the H2/CO2 and H2O/CO2 molar

ratios, as well as the influence of the presence of H2O and/or O2, which is

discussed in section 3. One should consider that H2 is not originally present in

flue gas streams (mixture 1 in Table 2.1) and should be added. So, the resulting

feed stream (mixtures 2-10) has a composition different from the considered flue

gas one.

Table 2.1 – Inlet compositions (mol. %) of the Gibbs reactor used in simulations.

Mixture CO2 H2O N2 O2 H2 2

2

H

CO

2

2

N

CO

2

2

H O

CO

2

2

O

CO

1a) 13.0 20.5 63.0 3.5 0.0 0 4.8 1.6 0.3 2 10.2 0.0 49.2 0.0 40.6 4 4.8 0.0 0.0 3 11.3 0.0 54.8 0.0 33.9 3 4.8 0.0 0.0 4 12.8 0.0 61.8 0.0 25.5 2 4.8 0.0 0.0 5 8.8 13.8 42.4 0.0 35.0 4 4.8 1.6 0.0 6 9.9 0.0 47.9 2.7 39.5 4 4.8 0.0 0.3 7 8.6 13.5 41.4 2.3 34.2 4 4.8 1.6 0.3 8 9.6 15.1 46.4 0.0 28.9 3 4.8 1.6 0.3

9b) 8.8 4.8 51.0 0.0 35.4 4 5.8 0.5 0.0 10b) 9.7 5.2 56.0 0.0 29.1 3 5.8 0.5 0.0

a) Flue gas composition taken from [21]. b) Stream composition based on flue gas composition from [22].

Table 2.2 shows the main reactions considered for the analysis of our

results (reactions 1-7) as well as other possible reactions that may occur in small

extent.


67

Table 2.2 – Reactions considered in the present thermodynamic study.

Reaction formula ΔH298 K (kJ mol-1) Reaction description

Main reactions

1 CO2+4H2 ⇌ CH4+2H2O -165.0 CO2 hydrogenation to CH4

2 CO2+3H2 ⇌ CH3OH+H2O -49.4 CO2 hydrogenation to CH3OH

3 CO2+2H2 ⇌ C+2H2O -90.1 CO2 reduction

4 CO2+H2 ⇌ CO+H2O 41.2 Reverse water-gas shift

5 CH4+2O2 ⇌ CO2+2H2O -803.0 CH4 oxidation

6 CH4+ 1 2⁄ O2 ⇌ CO+2H2 -36.0 CH4 partial oxidation

7 H2+ 1 2⁄ O2 ⇌ H2O -241.8 H2 oxidation

Other possible reactions

8 C + 1 2⁄ O2 ⇌ CO -110.5 Coke partial oxidation

9 C+O2 ⇌ CO2 -393.5 Coke complete oxidation

10 CO + 1 2⁄ O2 ⇌ CO2 -283.0 CO oxidation

11 CH4+CO2 ⇌ 2CO+2H2 247.4 Dry reforming of CH4

12 CH4 ⇌ C+2H2 74.9 CH4 cracking

13 CO + 3H2 ⇌ CH4+H2O -206.2 CO hydrogenation to CH4

14 CO + H2 ⇌ C+H2O -131.3 CO reduction

15 2CO ⇌ CO2+C -172.5 Boudouard reaction

2.3 Results and Discussion

2.3.1 Strategies for CO2 valorisation: CH4 or CH3OH?

As stated before, thermodynamic analysis was performed using the

continuous Gibbs reactor model. Carbon dioxide equilibrium conversion (Eq.

(2.9)) was determined for hydrogenation into methane or methanol as a function

of pressure and temperature (Fig. 2.1),

2 2

2

2

100

in out

CO CO

CO in

CO

F FX

F

(2.9)

Chapter 2

68

In this equation, F stands for the molar flow rate at the inlet (in) or outlet

(out) of the Gibbs reactor. In this section only the main reactions (reactions 1

and 2, Table 2.2) were considered, which means that the occurrence of

secondary reactions was, at this stage, discarded. Broad ranges of pressure and

temperature were set for the calculations, including those found in industrial

catalytic reactors operating these reactions.

As shown in Fig. 2.1, for either route of CO2 valorisation its conversion

decreases with reaction temperature, because both processes are exothermic.

In addition, total pressure has a positive effect, because in either case there is a

decrease in the total number of moles (from reactants to products – cf. equations

1 and 2). However, data presented in Fig. 2.1 clearly evidences that CO2

conversion into CH3OH (Fig. 2.1b) requires high pressures, particularly in the

temperature range where active catalysts operate in industry (see dashed areas

in Fig. 2.1) so that significant conversions can be achieved. Thus, since post-

combustion flue gases are typically at the atmospheric pressure, the CH3OH

route requires compression of the feed stream, which increases operation costs,

when compared to the methane route. For instance, at 250 ºC the conversion of

CO2 in CH4 production is almost complete at the atmospheric pressure while for

the CH3OH route it is practically null. Based on these evidences, in the following

sections it was chosen to analyse into more detail only the valorisation strategy

of CO2 hydrogenation into CH4.


69

Fig. 2.1 – Carbon dioxide conversion obtained for hydrogenation reactions into: a) CH4

and b) CH3OH. Dashed areas show typical operation temperature ranges of industrial

catalysts.

Chapter 2

70

2.3.2 CO2 methanation: effect of pressure, temperature

and H2/CO2 ratio

The effects of pressure, temperature and H2/CO2 molar ratio on CO2

conversion (Eq. (2.9)), product selectivity (Eq. (2.10)) and yield (Eq. (2.11)) were

investigated on CO2 methanation. In this section, the methanation main reaction

1 and secondary reactions 3 and 4 (see Table 2.2) were considered. Reactions

5-7 (Table 2.2) were not accounted because O2 was not considered to be

present in the feed stream (this will be addressed in Section 2.3.3); other

reactions present in Table 2.2 occur in small/very small extent.

Selectivity to carbon-containing species (CH4, CO and C):

2 2CO CO

100out

ii in out

FS

F F

(2.10)

Yield of carbon-containing species (CH4, CO and C):

2

i 100out

i

in

CO

FY

F (2.11)

Since H2 is an expensive compound, the highest H2/CO2 ratio considered

was 4, corresponding to the stoichiometry of the desired reaction (Eq. (2.1)).

Fig. 2.2a shows that, for a given pressure, CO2 conversion decreases with

temperature but increases with the H2/CO2 ratio. In Fig. 2.2b it can be observed

that CH4 selectivity increases with both the temperature and the H2/CO2 ratio,

except at low pressures and for H2/CO2=2. For a H2/CO2 ratio of 4 the selectivity

is almost complete (Fig. 2.2b), except for low pressures and high temperatures

where a small fraction of carbon monoxide is formed (YCO< 1 %) through the

endothermic reverse water gas shift reaction. So, when a ratio of 4 is used, CO2

methanation is the most favoured reaction. This is also supported by the fact


71

that the outlet molar flow rate of H2O is twice the value of CH4, which obeys to

the stoichiometry of the CO2 methanation and means that these species are not

being produced nor consumed through other reactions.

Fig. 2.2 – Effect of the H2/CO2 ratio (mixtures 2, 3 and 4 in Table 2.1), temperature and pressure on: a) CO2 conversion, b) CH4 selectivity, c) CH4 yield and d) Carbon yield.

40

50

60

70

80

90

100

510

1520

2530

150200

250300

350

XC

O2

/ %

P /

bar

T / ºC

a)

40

50

60

70

80

90

100

H2/CO2=4

H2/CO2=3

H2/CO2=2

20

30

40

50

60

70

80

90

100

510

1520

2530

150200

250300

350

SC

H4 / %

P / ba

r

T / ºC

b)

20

30

40

50

60

70

80

90

100

H2/CO2=4

H2/CO2=3

H2/CO2=2

20

30

40

50

60

70

80

90

100

5

10

15

2025

30

150200

250300

350

YC

H4 / %

P / ba

r

T / ºC

c)

20

30

40

50

60

70

80

90

100

H2/CO2=2

H2/CO2=4

H2/CO2=3

0

10

20

30

40

50

60

5

10

15

2025

30

150200

250300

350

YC

/ %

P /

bar

T / ºC

d)

0

10

20

30

40

50

60

H2/CO2=2

H2/CO2=3

H2/CO2=4

Chapter 2

72

As mentioned above, when lower H2/CO2 ratios (i.e. 3 and 2) are used,

CO2 conversion decreases and a similar trend is observed for both ratios along

the temperature and pressure values. This was somehow expected because

more CO2 will remain available (unconverted) since H2 is not fed with the

required stoichiometry for CO2 methanation (Eq. (2.1)). Interestingly, for a ratio

of 3 it is observed that for temperatures below ca. 250 ºC the conversion of CO2

starts to increase (Fig. 2.2a), which is accompanied by a decrease of CH4

selectivity (Fig. 2.2b) and yield (Fig. 2.2c). This trend suggests that below 250

ºC another compound is being produced using CO2 as reactant; from Fig. 2.2d

it becomes clear that such species is solid carbon (coke) through CO2 reduction

(reaction 3 in Table 2.2). Its exothermic nature also supports that carbon

formation is favoured at lower temperatures, as observed. Moreover, this is also

corroborated by the fact that the H2O molar flow rate is higher than twice the

value of CH4, meaning that H2O is produced not only through CO2 methanation

(Eq. (2.1)) but also via the CO2 reduction reaction.

2.3.3 Direct CO2 methanation from coal-fired power plant

(CF-PP) flue gas streams

2.3.3.1 Effect of H2O

Water vapour is an important component present in coal-fired power plant

(CF-PP) flue gas streams. Herein, its effect was studied considering that H2O is

coming exclusively from the post-combustion stream. One should note that H2

is not present in CF-PP flue gas streams (cf. mixture 1 in Table 2.1) and so it

should be added. To observe the effect of H2O in the methanation reaction it was

considered that oxygen was absent. Still, in Section 2.3.3.3 the simultaneous

effect of H2O and O2 will be addressed.

Fig. 2.3 shows contour plots illustrating the effect of H2O content on CO2

conversion as a function of pressure, temperature and H2/CO2 ratio; H2O content

was varied to address realistic limits. For both H2/CO2 ratios of 4 and 3, the


73

addition of H2O decreases the equilibrium conversion, which was expected

because H2O is a product of the CO2 methanation reaction (Eq. (2.1)), thus

limiting the forward reaction. The decrease of CO2 conversion due to the

presence of H2O is more notorious for a H2/CO2 ratio of 3. Still more importantly,

CH4 was the only carbon-containing product formed (nearly 100 % selectivity

was observed) in the presence of H2O for the pressure and temperature ranges

considered when a H2/CO2 ratio of 4 or 3 is used.

Fig. 2.3 – Contour plots showing the effect caused by H2O content on CO2 conversion considering different H2/CO2 ratios of 4 (left column) and 3 (right column). For the compositions of the different mixtures, please refer to Table 2.1.

Chapter 2

74

In Fig. 2.4 it is shown that when a H2/CO2 molar ratio of 3 is used, the

presence of H2O markedly inhibits carbon formation at temperatures below 250

ºC. In this regard, industrial operation can be carried out using a H2/CO2 ratio of

3 without carbon formation if H2O is present. Obviously, this option represents a

CO2 equilibrium conversion decrease of at least 25 % as compared to the

situation of H2/CO2=4 without H2O (cf. Fig. 2.3a and Fig. 2.3f). Moreover, due to

the exothermic nature of the CO2 hydrogenation into CH4 (ΔH298 K = -165.0 kJ

mol-1), the addition/presence of H2O can be also an interesting strategy to control

the heat produced in a catalytic reactor, as suggested elsewhere [23].

Fig. 2.4 – Carbon yield (YC) obtained in the absence (mixture 3) or in the presence of H2O (mixture 8) considering a feed stream with a H2/CO2 ratio of 3. For the compositions of the different mixtures, please refer to Table 2.1.

2.3.3.2 Effect of O2

Oxygen is commonly found in flue gas streams despite in small contents

(< 5 %). In this regard it is important to assess its possible impact on CO2

methanation. A stream with 2.7 mol % of O2 (mixture 6 in Table 2.1) was

0

5

10

15

20

25

30

35

5

10

15

20

25

30

150200

250300

350

YC /

%

P /

bar

T / ºC

0

5

10

15

20

25

absence of H2O (mix. 3)

(surface above)

presence of H2O (mix. 8)

(surface below)


75

considered at the inlet of the Gibbs reactor. The possible oxidations (reactions

5-7) are depicted in Table 2.2. Oxygen can react with H2 and CH4, both species

participating in CO2 methanation, the first as a reactant and the latter as a

product. The presence of only 2.7 mol % of O2 in the feed stream produces a

significant decay of CO2 conversion, as shown in Fig. 2.5, which can be mainly

explained by the formation of CO2 and H2O. In fact, in certain conditions the CO2

conversion shifts from 95 % to ca. 82 %, while in the range of industrial nickel

catalysts operation (i.e. at 400 ºC and atmospheric pressure) the conversion can

be as low as 70 %.

Fig. 2.5 – Contour plots showing the effect of the presence of O2 on CO2 conversion. For the compositions of the different mixtures, please refer to Table 2.1.

2.3.3.3 Simultaneous effect of H2O and O2

Fig. 2.6 shows that 2COX decreases when both H2O and O2 are present

in the feed stream (trends for 4CHY are the same as for

2COX because methane

selectivity was nearly 100 % in both plots – data not shown). The presence of

H2O, as mentioned in Section 2.3.3.1, hinders CO2 methanation in the forward

direction (Eq. (2.1)) because it is a product of the reaction. Additionally, the

presence of O2 promotes the oxidation of species such as CH4 or H2 (reactions

Chapter 2

76

5-6 and 7 in Table 2.2, respectively), leading to the formation of H2O and CO2.

Therefore, the presence of O2, although in a small percentage (2.3 mol % - cf.

mixture 7 in Table 2.1) should be avoided because it leads to the parallel

consumption of a reactant (H2) and of the desired product (CH4). In fact, in the

conditions tested, all the O2 fed is consumed. Above 300 ºC and at the

atmospheric pressure a slight formation of CO (YCO < 1 %) through reverse water

gas shift (reaction 4 in Table 2.2) is observed (data not shown for brevity

reasons). However, CO formation can be suppressed increasing the pressure.

In this case, CO2 methanation (Eq. (2.1)) is favoured and overlaps the reverse

water-gas shift (reaction 4), which is not influenced by the pressure since the

reaction takes place without change in the number of moles.

Fig. 2.6 – Contour plots showing the effect of simultaneous presence of H2O and O2 on CO2 conversion for a H2/CO2 ratio of 4. For the compositions of the different mixtures, please refer to Table 2.1.

2.4 Technological implementation

From the technological point of view, i.e. for process implementation, two

related problems are identified: i) the presence of substances in the flue gas,


77

namely O2 (as above-mentioned) and N2 (that simply acts as diluents), and ii)

the existence of un-reacted CO2 in the reactor outlet, due to thermodynamic

restrictions. Integration of a methanation catalyst with a CO2-selective sorbent in

a single mixed bed is anticipated to allow overcoming all these drawbacks

simultaneously. In a first stage, the unit (operating in sorption mode) is fed with

the flue gas, so that CO2 is selectivity retained while the other species leave the

bed, up to almost sorbent saturation (in fact up to CO2 breakthrough from the

column). In the second stage, (renewable) H2 is fed to the bed, reacting with the

previous concentrated CO2, in a so-called reactive regeneration approach. To

operate on a continuous basis of CO2 capture and conversion at least two beds

are thus necessary operating in complementary stages: when one CO2-

saturated bed is being regenerated (with a hydrogen-containing stream) and

CH4 is being produced, the other one is capturing carbon dioxide; after

regeneration of the 1st column, the bed is able again to capture more carbon

dioxide (cf. Fig. 2.7).

Fig. 2.7 – Illustration of the integrated reactive regeneration process for CO2 capture and conversion to CH4 with two beds operating 180º out of phase with each other.

Chapter 2

78

The reactive regeneration concept was already proved for other

applications, namely in sorption-enhanced reactors for H2 production through

steam methane reforming [24], and the proof-of-concept towards CO2

methanation is presented in Chapter 5. Finally, it should be mentioned that water

must be removed from the exit stream to obtain a product with quality compatible

with existing natural gas infrastructures.

2.5 Conclusions

The present work compared, from the thermodynamic standpoint, the

carbon dioxide valorisation to methanol and methane. The option for CO2

conversion into CH3OH requires harsh operation conditions when compared to

the CH4 route, namely in terms of pressure. Thus, in the near term, CO2

methanation seems to be an easier pathway for CO2 valorisation, while research

on the development of active catalysts at lower temperatures for CO2

hydrogenation to CH3OH is required.

This study also allowed concluding that CO2 methanation can take place

with complete (~100%) methane selectivity and with high methane yields in the

temperature and pressure ranges of industrial catalysts operation, as long as the

H2/CO2 ratio is 4. A preliminary stage for O2 removal from post-combustion

exhaust streams is required due to its detrimental impact on CO2 conversion,

apart from security reasons. On the other hand, the effect caused by H2O is not

so pronounced as for O2. In fact, water presence can substantially inhibit coke

formation whenever a H2/CO2 ratio of 3 is used, thus opening a wider range of

operation conditions available for the catalytic conversion of CO2 into CH4.

Moreover, the addition of water can bring additional advantages regarding

temperature control of the methanation reactor due to the exothermic nature of

the Sabatier reaction.


79

2.6 References

[1] E.A. Quadrelli, G. Centi, Green carbon dioxide, ChemSusChem, 4 (2011)

1179-1181.

[2] Kyoto Protocol to the United Nations Framework Convention on Climate

Change, (1998).

[3] International Energy Agency, CO2 emission from fuel combustion: highlights,

OECD/IEA, 2012.



ChemSusChem, 4 (2011) 1194-1215.

[5] M. Peters, B. Köhler, W. Kuckshinrichs, W. Leitner, P. Markewitz, T.E. Müller,

Chemical technologies for exploiting and recycling carbon dioxide into the value

chain, ChemSusChem, 4 (2011) 1216-1240.

[6] A.A. Olajire, Valorization of greenhouse carbon dioxide emissions into value-

added products by catalytic processes, Journal of CO2 Utilization, 3-4 (2013) 74-

92.

[7] M. Aresta, A. Dibenedetto, A. Angelini, The changing paradigm in CO2

utilization, Journal of CO2 Utilization, 3–4 (2013) 65-73.

[8] V. Subramani, P. Sharma, L. Zhang, K. Liu, Catalytic steam reforming

technology for the production of hydrogen and syngas, in: K. Liu, C. Song, V.

Subramani (Eds.) Hydrogen and Syngas Production and Purification

Technologies, John Wiley & Sons, Inc., 2010, pp. 14-126.

[9] S.K. Hoekman, A. Broch, C. Robbins, R. Purcell, CO2 recycling by reaction

with renewably-generated hydrogen, International Journal of Greenhouse Gas

Control, 4 (2010) 44-50.

[10] C. Higman, M.v.d. Burgt, Gasification, 2nd ed., Elsevier Inc., Burlington,

2008.

[11] G. A. Olah, A. Goeppert, G.K.S. Prakash, Beyond Oil and Gas: The

Methanol Economy, Wiley - VCH Verlag GmbH & Co, Weinhein, 2006.

Chapter 2

80

[12] L. Grond, P. Schulze, J. Holstein, Systems analyses Power to Gas: A

technology review, KEMA Nederland B.V., Groningen, 2013.

[13] News: Germany/Power-to-gas: Clariant supplies SNG catalyst for first

commercial CO2 methanation plant, Oil Gas European Magazine, 39 (2013) 217.

[14] S. Walspurger, G.D. Elzinga, J.W. Dijkstra, M. Sarić, W.G. Haije, Sorption

enhanced methanation for substitute natural gas production: Experimental

results and thermodynamic considerations, Chemical Engineering Journal, 242

(2014) 379-386.

[15] J. Gao, Y. Wang, Y. Ping, D. Hu, G. Xu, F. Gu, F. Su, A thermodynamic

analysis of methanation reactions of carbon oxides for the production of

synthetic natural gas, RSC Advances, 2 (2012) 2358-2368.

[16] W.-J. Shen, K.-W. Jun, H.-S. Choi, K.-W. Lee, Thermodynamic investigation

of methanol and dimethyl ether synthesis from CO2 Hydrogenation, Korean

Journal of Chemical Engineering, 17 (2000) 210-216.

[17] S. Adhikari, S. Fernando, S.R. Gwaltney, S.D. Filip To, R. Mark Bricka, P.H.

Steele, A. Haryanto, A thermodynamic analysis of hydrogen production by steam

reforming of glycerol, International Journal of Hydrogen Energy, 32 (2007) 2875-

2880.

[18] S.M. Walas, Phase equilibria in Chemical Engineering, Butterworth

Publishers, Boston, United States of America, 1985.

[19] M.A. Soria, C. Mateos-Pedrero, A. Guerrero-Ruiz, I. Rodríguez-Ramos,

Thermodynamic and experimental study of combined dry and steam reforming

of methane on Ru/ ZrO2-La2O3 catalyst at low temperature, International Journal

of Hydrogen Energy, 36 (2011) 15212-15220.

[20] Y. Li, Y. Wang, X. Zhang, Z. Mi, Thermodynamic analysis of autothermal

steam and CO2 reforming of methane, International Journal of Hydrogen Energy,

33 (2008) 2507-2514.

[21] K.A. Mumford, K.H. Smith, C.J. Anderson, S. Shen, W. Tao, Y.A.

Suryaputradinata, A. Qader, B. Hooper, R.A. Innocenzi, S.E. Kentish, G.W.

Stevens, Post-combustion capture of CO2: results from the solvent absorption


81

capture plant at Hazelwood power station using potassium carbonate solvent,

Energy & Fuels, 26 (2011) 138-146.

[22] M.T. Ho, G.W. Allinson, D.E. Wiley, Comparison of MEA capture cost for

low CO2 emissions sources in Australia, International Journal of Greenhouse

Gas Control, 5 (2011) 49-60.

[23] T.T.M. Nguyen, L. Wissing, M.S. Skjøth-Rasmussen, High temperature

methanation: Catalyst considerations, Catalysis Today, 215 (2013) 233-238.

[24] G.H. Xiu, P. Li, A.E. Rodrigues, Sorption-enhanced reaction process with

reactive regeneration, Chemical Engineering Science, 57 (2002) 3893-3908.

83

Chapter 3. High temperature CO2 sorption with gallium-substituted and promoted hydrotalcites

The contents of this chapter were adapted from: C.V. Miguel, R. Trujillano, V. Rives,

M.A. Vicente, A.F.P Ferreira, A.E. Rodrigues, A. Mendes, L.M. Madeira, High

temperature CO2 sorption with gallium-substituted and promoted hydrotalcites.

Separation and Purification Technology, 2014, 127, 202-211.

Hydrotalcites (HTC) and gallium-substituted hydrotalcites (HTC-

10Ga) were prepared by co-precipitation and modified by impregnation

with alkali (K and Cs) and alkaline-earth (Sr) metals. The materials were

tested and screened for CO2 sorption at 573 K. The results indicate that

the modification with potassium greatly enhanced the sorption capacity

when compared with the original materials (both HTC and HTC-10Ga). An

outstanding sorption capacity (2.01 mmol·cm-3 at 1.08 bar) was obtained

for the sample partially substituted with gallium and modified with

potassium (HTC-10Ga-20K). Moreover, this material was also submitted

to sorption-desorption cycles towards CO2 under low-pressure conditions

(<0.0001 bar to 0.15 bar) to assess its use in cyclic operation. An average

working capacity of 0.97 mmol·cm-3 and 0.25 mmol·cm-3 was obtained after

repeated cycles at 573 K and 473 K, respectively. The sorption kinetics

was assessed through uptake measurements, showing two parallel

resistances, which were well described by the proposed model.

Chapter 3

84

3.1 Introduction

The excessive anthropogenic greenhouse gases (GHG) emission is

causing negative changes on the planet climate such as global warming,

frequent events of heavy precipitation and heat waves [1]. Policies have been

undertaken by developed countries to control, reduce and tax the emission of

GHG like CO2. Carbon dioxide is a relevant GHG; in fact, around 30 x 1012 kg of

CO2 per year are emitted to the atmosphere arising from the combustion of fossil

fuels (coal, petroleum and natural gas) [2]. In this regard, there is a great interest

in developing technological solutions for large-scale CO2 capture, sequestration

or conversion/utilization, from both the environmental and economic point of

views [3].

Depending on the operating temperature, several materials (e.g. activated

carbons, zeolites, metal-organic frameworks, hydrotalcites, etc.) may be used to

capture CO2 by adsorption [4, 5]. Among them, hydrotalcites can be used at

intermediate temperatures (473-673 K). Hydrotalcite materials are layered

double hydroxides, which general formula is:

2+ 3+ -1- 22

M M OH A H Onx x x n

y (3.1)

where M2+ and M3+ are divalent (e.g. Mg2+, Ni2+, Zn2+, etc.) and trivalent (e.g.

Al3+, Ga3+, Mn3+, etc.) metal cations, respectively. An- is a charge balancing anion

(CO32-, Cl-, SO4

2-, etc.) located in the interlayer space and x is generally between

0.2–0.4 [6], while y is the number of moles of hydration water located also in the

interlayer space. The mineral known as hydrotalcite has Mg2+, Al3+ and CO32- as

divalent and trivalent cations and balancing anion, respectively. Hydrotalcites

can be found in nature or be synthesized. Currently, several thousands of tons

of hydrotalcites are yearly produced by several chemicals companies (e.g.

BASF, SASOL, Clariant, Kisuma Chemicals, Sakai Chemicals, etc.) which found

High temperature CO2 sorption with gallium-substituted and promoted hydrotalcites

85

application as catalysts, precursors, catalyst support, adsorbents, polymer

stabilizers and antacids.

Regarding their application as an adsorbent for CO2 capture, it has been

reported that hydrotalcites exhibit a reasonably good sorption capacity, stability

and easy regeneration by temperature or pressure swing [5, 7-9]. Additionally,

results found in the literature show that the presence of water vapour also

benefits carbon dioxide sorption [10, 11]. This is particularly important for CO2

capture in applications were water vapour is present, as it is the case of coal-

fired power station exhaust streams (post-combustion), or in sorption enhanced

processes for H2 production (pre-combustion) [12-14]. On the other hand, it has

been reported that SOx species, also present in those streams, have a stronger

affinity to hydrotalcites when compared to CO2, despite exhibiting slower kinetics

[15]; in fact, in the work by Beaver [16] it was reported that CO2 sorption capacity

on a K-promoted hydrotalcite decreased ca. 80 % after exposure to CO2/SO2

mixtures at the end of eight successive sorption-desorption cycles.

In the work by Yavuz et al. [8], the effect of gallium substitution and

modification with potassium was firstly addressed and a markedly improved CO2

sorption capacity was observed (1.4 mmol∙g-1) at 473 K for a CO2 partial

pressure of 0.7 atm under dry conditions.

In this work the effect on CO2 sorption capacity caused by gallium partial

substitution and modification with potassium, cesium and strontium was

evaluated and compared to that shown by the original formulation. Along with a

detailed physicochemical characterization of all the prepared materials, sorption

equilibrium isotherms were determined at 573 K, a typical temperature found in

both post- and pre-combustion streams. Additionally, the most promising

material was also submitted to sorption-desorption cycles in the low-pressure

region (< 0.0001 bar to 0.15 bar) at 473 K and at 573 K. Finally, the sorption

kinetics was assessed for this material.

Chapter 3

86

3.2 Experimental

3.2.1 Chemicals and gases

Magnesium nitrate 6-hydrate (Sigma-Aldrich, 99 %), aluminium nitrate 9-

hydrate (Panreac, 99.82 %) and gallium nitrate hydrate (Aldrich, 99.99 %) have

been used as Mg, Al and Ga precursors, respectively. Anhydrous sodium

carbonate (Panreac, 98.88 %) was used in excess to assure that the charge

balancing anion was carbonate instead of nitrate. Sodium hydroxide (Panreac,

98.93 %) was used to keep a basic pH during synthesis.

Potassium acetate (Panreac, 99 %), cesium acetate (Aldrich, 99.9 %) and

strontium acetate (Aldrich, 99.995 %) were used as chemical promoters for

modifying the prepared hydrotalcite and the 10 % (mol percentage) gallium-

substituted hydrotalcite samples.

Carbon dioxide (99.99 %) and helium (99.999 %), both from L’Air Liquide,

were used in the sorption experiments.

3.2.2 Sorbents preparation

Hydrotalcites (HTC) and hydrotalcites partially substituted with gallium

(herein called HTC-10Ga) were modified by impregnation with aqueous

solutions of alkaline or alkaline-earth metal cations such as K+, Cs+ and Sr2+.

The targeted not promoted hydrotalcites were Mg2Al(OH)6(CO3)0.5·yH2O and

Mg2(Al0.9Ga0.1)(OH)6(CO3)0.5·yH2O. The two previously mentioned samples,

respectively HTC and HTC-10Ga, were prepared by the co-precipitation method.

In the work by Yong et al. [9] the CO2 sorption capacity was determined at 573

K for three commercial hydrotalcites from SASOL having different Mg/Al ratios:

0.5, 1.27 and 3. It was observed that an Mg/Al ratio of 1.27 had the highest

sorption capacity followed by 3 and, finally, by 0.5. However, the lowest limit of

the Mg/Al ratio for obtaining crystallographically pure hydrotalcite (2.0) is

attributed to electrostatic repulsion between neighbouring, octahedrally


87

coordinated, trivalent metals in the layers, which is unavoidable if Mg/Al is lower

than 2 [6]. For this reason, and based on the results obtained by Yong et al,

[9],molar ratios of 2 were chosen for Mg2+/Al3+ and Mg2+/(Al3++Ga3+), while for

the Ga-containing sample an Al:Ga ratio of 9:1 was also considered (see method

details in [6]) to check the effect of gallium presence.

The solution containing the divalent and trivalent cations was drop-wise

added to the basic solution containing NaOH and Na2CO3. The resulting solution

was continuously stirred during 17 h at room temperature and then the

suspension was centrifuged and the solids washed with distilled water for

removing nitrate anions and sodium cations. The samples were then dried at

308-323 K and crushed. Both samples (HTC and HTC-10Ga) were then calcined

at 673 K during 2 h in air and the loss of mass assessed in order to further

prepare the alkali modified samples with a pre-established stoichiometry. So,

part of the remaining HTC sample was modified with K while three parts of the

HTC-10Ga sample were modified with K, Cs or Sr (cf. Table 3.1) by impregnation

to accomplish a 20 wt. % loading. This content is typically found in other works

[7, 14, 17] following the important work by Hufton et al. [18], although it has been

reported that the sorption capacity at 673 K is not appreciably changed when

using different loadings [13, 18]. Impregnation was followed by drying (313-323

K), crushing, and calcination at 673 K during 2 h in air.

Table 3.1 – List of prepared samples.

Sample M2+ M3+ An- Promoter (wt. %)

HTC Mg Al - CO3 - -

HTC-20K Mg Al - CO3 K 20

HTC-10Ga Mg Al Ga CO3 - -

HTC-10Ga-20K Mg Al Ga CO3 K 20

HTC-10Ga-20Cs Mg Al Ga CO3 Cs 20

HTC-10Ga-20Sr Mg Al Ga CO3 Sr 20

Chapter 3

88

3.2.3 Sorbents characterization

The HTC and HTC-10Ga materials were analysed by thermogravimetric

analysis (TG/DTG) in a TA Instruments apparatus, Model SDT Q600, to study

their thermal stability. Each sample was heated at 10 K∙min-1 up to 1273 K in a

dynamic air atmosphere.

Powder X-ray diffraction (PXRD) patterns were recorded in a Siemens

D500 diffractometer equipped with DIFFRACT-AT software, using Cu K

radiation (λ= 1.5405 Å), a Ni filter and a graphite monochromator. The analysis

conditions were: 40 kV, 30 mA, a step size of 0.05° and a scanning speed of 2º

(2 theta) per minute.

Fourier Transformed Infrared Spectroscopy (FTIR) (Perkin-Elmer, 1600

Series) was performed in the range 4000-400 cm-1 using the KBr technique.

Scanning Electron Microscopy (SEM) was performed on a JEOL JSM 6301F

instrument coupled with Energy Dispersive Spectroscopy (EDS) (JEOL JSM

6301F/Oxford INCA Energy 350) to investigate the materials structure and

composition. The textural properties of the prepared materials were obtained by

means of nitrogen adsorption at 77 K (Micrometrics, model Gemini II), after

degassing the samples for 2 h at 383 K in a Flowprep 060 apparatus

(Micrometrics).

High temperature CO2 sorption measurements were performed using a

magnetic suspension balance from Rubotherm. First, the sample (ca. 1 g) was

heated overnight up to 585 K under vacuum (< 0.001 bar) followed by buoyancy

determination with He at 313 K, which allowed to determine the solid density

after the regeneration step. Equations for buoyancy correction can be found in

Appendix A. Subsequently, the sample was again heated up to the desired

temperature and CO2 fed to the chamber until reaching the required pressure for

measuring the first sorption equilibrium value. The uptake curve was recorded

by an automated data acquisition system. The equilibrium was considered to be

reached when a variation of less than 0.1 mg was observed during 20 min. For


89

screening purposes, the sorption equilibrium values were obtained at 573 K and

for CO2 pressures of ca. 0.05, 0.15, 0.45 and 1.10 bar. Afterwards, the most

promising material was submitted to sorption-desorption cycles at 573 K and 473

K and the reversible/steady working capacity determined. Typically, the carbon

dioxide partial pressure in post-combustion flue gases is 0.15 bar and so each

cycle consisted in increasing the pressure up to 0.15 bar followed by desorption

down to vacuum, except for cycle number 3 where the pressure was increased

up to 1.1 bar to check the effect of the feed pressure on the cyclic stability.


3.3.1 Sorbents physicochemical characterization

3.3.1.1 TG / DTG

Thermogravimetric analysis was performed to study the thermal

decomposition of HTC and HTC-10Ga fresh samples (Fig. 3.1); a total mass loss

of 55 % and 47 % was found, respectively. Both samples show three DTG peaks

related to the removal of hydration water and hydroxyl groups (at 353-373 K and

463 K) and carbonate anions (at 653 K), as discussed in detail elsewhere [19,

20]. As mentioned in Section 3.2.2, the fresh samples were calcined at 673 K for

further studies, as this treatment ensures that all water existing in the fresh

samples is removed. The used calcination temperature is in line with several

works [7, 8, 21] and was also chosen following the study by Ram Reddy et al.

[22], where it was observed that samples calcined at 673 K provided a higher

CO2 sorption capacity. Moreover, this is the highest operation temperature

expected to be used in most final applications (e.g. for pre-combustion CO2

capture).

Chapter 3

90

Fig. 3.1 – Thermogravimetric (TG) (solid lines) and derivative thermogravimetry (DTG)

(dotted lines) analysis of the fresh samples HTC and HTC-10Ga.

3.3.1.2 XRD

The XRD patterns of the samples calcined at 673 K corresponded to

mostly amorphous materials, where only weak and broad signals due to MgO

were observed; the pattern for the Ga-containing sample is shown in Fig. A.0.1

presented in Appendix A. This result agrees with the TG curve (Fig. 3.1), which

showed that at 673 K most of the removable mass has been eliminated (water

vapour and carbonates), leaving the mostly amorphous solid. However, the XRD

patterns evidenced that one year after their synthesis, calcination at 673 K and

storage, all the materials have a typical hydrotalcite structure, suggesting that

upon exposure to humid air the materials have reconstructed their layered

structure [23] (Fig. 3.2a).

The partial substitution with gallium led to an increased height intensity of

the d(003) peak (see patterns of samples HTC and HTC-10Ga in Fig. 3.2a).

Impregnation with cesium (HTC-10Ga-20Cs) did not result in large enough

crystals that could be identified by XRD, meaning that cesium is well dispersed


91

on the material. For the case of impregnation with strontium, a strontianite

crystalline phase (SrCO3) was identified (JCPDS file 5-0418), while K2CO3∙1.5

H2O (JCPDS file 11-0655) was verified in both samples impregnated with

potassium.

After CO2 sorption the XRD patterns showed important differences.

Representative diagrams are included in Fig. 3.2b. First of all, the maxima due

to the layered hydrotalcite structure (easily observed in the sample stored for

one year) have vanished completely for sample HTC-10Ga-20Sr or decreased

markedly their intensities for other samples (although the bands due to SrCO3

remained unaltered). This behaviour is due to two complementary facts: (i) the

CO2 capture experiments were carried out at 573 K (after outgassing the sample

at 585 K), a temperature high enough to destroy the layered structure, as

concluded from the TG results above mentioned (Fig. 3.1); (ii) despite the

presence of a high pressure of CO2 would hopefully give rise to recovering the

layered structure (the so-called “memory effect”) [24, 25], it should be stressed

that the experiments were carried out in the absence of water vapour, and thus

the hydrotalcite-like structure could not be recovered. When the sample HTC-

10Ga-20K was submitted to five consecutive CO2 sorption-desorption cycles the

XRD patterns did not show any significant difference (cf. Fig. 3.2b), although

some minor sharpening of the maxima was observed when the cycles were

carried out at 573 K, but not at 473 K. Probably, this treatment at the higher

temperature in some sort of way favoured an incipient crystallization of the

existing crystalline phases.

Chapter 3

92

Fig. 3.2 – XRD patterns of samples: a) one year after being calcined at 673 K and stored, and b) after CO2 sorption.

3.3.1.3 SEM / EDS

SEM images evidenced that the samples exhibit a not defined and/or

regular shape (Fig. 3.3), excepting HTC-10Ga-20Sr (pattern e)) and, particularly,

HTC-20K and HTC-10Ga-20K (patterns c) and f)), which also revealed the

presence of zones of the needle-like (Z1) or flat surface (Z3) type. This was

already observed by Oliveira et al. [7] using commercial hydrotalcites modified

with potassium. Interestingly, EDS analysis allowed to verify that the potassium

mass content at the flat surface zones is the same for both HTC-20K and HTC-


93

10Ga-20K samples (26 wt. %). This content is also much higher than on the bulk

material (7 wt. %) and needle-like zones (5 wt. % vs. 14 wt. %, respectively) (see

Fig. A.0.2 in Appendix A). Although incipient wetness impregnation usually leads

to a homogeneous distribution of the active phase, potassium clearly has a

marked preference to deposit in certain zones of the supports.

Fig. 3.3 – SEM images of the prepared materials: a) HTC, b) HTC-10Ga, c) HTC-20K, d) HTC-10Ga-20Cs, e) HTC-10Ga-20Sr and f) HTC-10Ga-20K.

Chapter 3

94

3.3.1.4 Physical adsorption of nitrogen at 77 K

The specific surface areas were determined by physical adsorption of N2

at 77 K (cf. Table 3.2). Type II isotherms were obtained (see Fig. A.0.3 in

Appendix A), which means that the resulting materials are

macroporous/mesoporous solids; the determined volume of microporous (<

0.005 cm3∙g-1) can be neglected.

The specific surface area increased after calcination, particularly in the

case of the HTC sample (ca. 50 % of increase). This is related to dehydration

followed by collapse of the layered structure that occurs upon calcination at 673

K [26]. Nevertheless, the modification with alkali metals (K and Cs) resulted in

lower specific surface areas, being this fact more evident when the hydrotalcites

were modified with potassium. In fact, both samples modified with potassium

(HTC-20K and HTC-10Ga-20K) have nearly the same specific surface areas,

being the lowest observed. Probably this is related to pore blocking caused by

the K2CO3 phase, which is in agreement with SEM images (cf. Fig. 3.3c and Fig.

3.3f).The modification with strontium resulted in the material with the highest

specific surface area (SBET = 237 m2g-1). Although a straightforward explanation

is lacking for this behaviour, it might be tentatively related to subtle changes in

the basicity of the media upon addition of the doping cations.

Table 3.2 – BET specific surface area of the prepared materials determined by physical adsorption of nitrogen at 77 K.

Sample SBET [ m2∙g-1]

HTC (fresh) 122

HTC (calcined) 183

HTC-20K (calcined) 61

HTC-10Ga (fresh) 156

HTC-10Ga (calcined) 175

HTC-10Ga-20K (calcined) 58

HTC-10Ga-20Cs (calcined) 165

HTC-10Ga-20Sr (calcined) 237


95

3.3.1.5 FTIR

FTIR analyses obtained for the as prepared (uncalcined) HTC and HTC-

10Ga samples indicate the presence of carbonate as interlayer anion (cf. Fig.

3.4).

Fig. 3.4 - FTIR spectra of the starting and modified samples.

The spectra show an intense, broad band centered at ca. 3460 cm-1, due

to the stretching mode of the hydroxyl groups, both from the brucite-like layers

and the interlayer water molecules, as well as the water molecules physisorbed

on the external surface of the crystallites. The shoulder at 2900-3000 cm-1 has

been ascribed [27] to the stretching mode of interlayer hydroxyl groups

hydrogen-bonded to interlayer carbonate species. The bending mode of the

water molecules is responsible for the rather weak band at ca. 1630 cm-1, also

Chapter 3

96

recorded for both samples. The ν3 mode of interlayer carbonate is responsible

for the intense band at 1384 cm-1. This is a degenerated mode for the D3h

symmetry of the original carbonate anion; under this symmetry, the ν1 mode

expected at ca. 1080 cm-1 is inactive and it is not recorded. However, it can be

noticed that the band at 1384 cm-1 shows a clearly identified shoulder at 1500

cm-1, which can be considered because of the splitting of the ν3 mode. However,

as the band due to the ν1 mode is not recorded, the decrease in the symmetry

should be discarded and the two bands at 1384 and 1500 cm-1 should be

ascribed to carbonate species in two sites with different interaction with their

local environment. Weaker bands below 1000 cm-1 correspond mainly to

vibration of lattice bonds (Mg-OH, Al-OH, etc.).

When the samples were calcined at 673 K the spectra showed some

differences, but unfortunately it seems that some interlayer carbonate species

are still present (it should be noticed that at this temperature not all carbonate

has been removed, cf. the TG curve – Fig. 3.1); the hydroxyl groups have not

been completely removed either. Consequently, the spectra show the band at

ca. 3450 cm-1 due to the stretching mode of the hydroxyl groups (although the

band is much narrower than for the uncalcined samples), and the hydrogen

bonding between the hydroxyl groups and the carbonate species seem to be

absent. The remaining carbonate species show a lower symmetry and the ν3

mode has lost its original degeneration, probably on changing the carbonate

anion from a D3h to a C2v symmetry, the corresponding band is now split into two

bands at ca. 1480 and 1410 cm-1 and the band due to mode ν1 is now recorded

at 1110 cm-1.

Impregnation with Sr, K, or Cs and calcination at 673 K leads to some

changes in the spectra, concerning almost exclusively to the region below 1600

cm-1. The band due to mode ν3 of carbonate species splits in all cases and the

band due to mode ν1 of carbonate is recorded in all cases as well, indicating that

in this case a decrease in the local symmetry of the carbonate anions actually

has taken place. However, both components of the split ν3 band show a change


97

in their relative intensities from the monovalent (K and Cs) to the divalent (Sr)

doping cation; this might be tentatively related to the different nature (mono- or

divalent) of the doping cations. Probably, the increased basicity of the solids after

impregnation with the alkaline or alkaline-earth cations delays decomposition of

the solid and evolution of CO2, and thus remaining of carbonate species above

the temperature where decomposition, in the absence of the doping cations,

would be expected. Moreover, if carbonate species still exist after calcination at

673 K, this would account for the shoulder around 2900 cm-1 due to the νOH

stretching mode of hydroxyl groups hydrogen-bonded to the carbonate species.

The absence of bands around 1700-1750 cm-1 indicates that although

acetate salts were used for modifying the hydrotalcites with K, Cs and Sr, no ion

exchange between acetate and interlayer carbonate ions, within the hydrotalcite

structure, took place.

3.3.2 CO2 sorption experiments

3.3.2.1 Determination of adsorption equilibrium isotherms at 573 K

– sorbent screening

Fig. 3.5 shows CO2 sorption isotherms obtained at 573 K under dry

conditions for the prepared materials. The results are very promising, particularly

for the sample substituted with gallium and modified with potassium (HTC-10Ga-

20K); a sorption equilibrium value of 1.82 mmol·g-1 (i.e., 2.01 mmol·cm-3) at 1.08

bar was obtained which is, up to the authors knowledge, clearly above typical

results found in the literature under similar operation conditions (see Table 3.3)

[9, 22, 28, 29].

Chapter 3

98

Table 3.3 – CO2 sorption equilibrium value on HTC-10Ga-20K and comparison with other sorbents reported in the literature.

Material T [K] pCO2 [bar] q [mol·kg-1] Ref.

HTC-10Ga-20K 573 1.08 1.82 This work

Hydrotalcite 573 1 0.52 [29]

Hydrotalcite 573 1 0.50 [9]

Hydrotalcite 573 1.1 0.25 [22]

K-promoted hydrotalcite 673 1 0.79 [28]

Generally, the modification with alkali metals increases the sorption

capacity of the original hydrotalcite samples (HTC and HTC-10Ga). However,

adding potassium provided higher sorption values when compared to cesium

and this should be related to its higher basicity. On the contrary, the sorption

capacity decreased for the sample modified with an alkaline-earth metal (Sr).

Moreover, the sorption capacity of the materials decreased in general with the

increase of their specific BET surface area (cf. Table 3.2 and Fig. 3.5).

Fig. 3.5 – CO2 sorption equilibrium isotherms at 573 K for the prepared materials. Lines shows fitting using the Freundlich equation.


99

This means that the available BET surface area does not play a crucial

role on the sorption capacity, which was also reported in other works [7, 30]; the

sorption capacity is more likely to be related to the chemical nature of the

exposed surface. In the case of HTC-10Ga-20Sr it seems that the phase

obtained (strontianite), which has a completely different morphology (see Fig.

3.3e), almost does not sorb CO2, at least under the tested operation conditions.

The results herein presented are in line with the ones obtained by Yavuz

et al. [8], where they showed that the simultaneous modification with potassium

and gallium markedly improved the sorption capacity of hydrotalcites, the first

acting as a chemical promoter and the latter providing a superstructure robust

and stable at elevated temperatures (a sort of synergic effect). However, in our

work the sorption isotherm was obtained at 573 K and up to 1.1 bar. Moreover,

this promising sample (HTC-10Ga-20K) was also submitted to sorption-

desorption cycles at 573 K and 473 K to determine its reversible working

capacity, checking its suitability to be used in cyclic operation for post-

combustion CO2 capture/valorisation applications (Section 3.3.2.2).

The sorption data of each sample was fitted to the Freundlich equation

(Eq. (3.2)), which has also been proposed elsewhere to describe CO2 sorption

on hydrotalcites [21, 31]:

1

nq K p (3.2)

where q (mmol·cm-3) is the concentration of the sorbed species, and K

(mmol·cm-3·bar-(1/n)) and n (dimensionless) are Freundlich equation parameters.

Parameter n is normally greater than unity and as it increases the isotherm

approaches to the so-called irreversible isotherm. In this case, the pressure

needs to be very low in order to the sorbate start desorbing from the surface

[32].

Chapter 3

100

The parameters of the Freundlich equation (along with a 90 % interval of

confidence) are included in Table 3.4. The experimental data of the starting and

modified samples showed good adhesion to the Freundlich model within the

tested pressure range (see also Fig. 3.5).

Table 3.4 – Freundlich type isotherm parameters and their confidence limits (90 %) obtained after fitting the experimental data of Fig. 3.5 (T = 573 K) and solid density (determined with He at 313 K after regeneration) for each material.

Sample log(K) 1/n R2 ρ [g·cm-3]

HTC - - - 0.831

HTC-20K 0.26 ± 0.034 0.16 ± 0.038 0.9875 0.841

HTC-10Ga -0.30 ± 0.055 0.20 ± 0.074 0.9685 0.860

HTC-10Ga-20K 0.29 ± 0.035 0.10 ± 0.039 0.9685 1.107

HTC-10Ga-20Cs 0.14 ± 0.046 0.17 ± 0.054 0.9774 1.187

HTC-10Ga-20Sr -0.48 ± 0.11 0.25 ± 0.15 0.9251 0.762

3.3.2.2 Sorption-desorption experiments

To determine the working capacity of the best sorbent (HTC-10Ga-20K),

a series of sorption-desorption cycles at 573 K and 473 K was carried out. As it

can be seen in Fig. 3.6, in what we called cycle 0, CO2 was fed to the

microbalance chamber and the pressure allowed to increase up to 0.05 bar and

then to 0.15 bar. Each cycle ended with the sample being submitted to vacuum.

In this cycle, the sorption capacity determined at 573 K and 0.15 bar (Fig. 3.6a)

was 1.57 mmol·cm-3. However, when the microbalance chamber was evacuated

it was not possible to remove all the sorbed CO2; around 0.38 mmol·cm-3

remained sorbed – see open symbol. In the following cycles the pressure was

again switched between 0.15 bar and vacuum, except in cycle 3 where the

pressure was increased up to 1.1 bar before the regeneration stage. It is

important to note that the sorption values presented in each cycle (black

symbols) correspond to the working capacity, i.e., it was considered a starting

mass of the sample which included the previous irreversibly sorbed mass of CO2.


101

For instance, in cycle 1 at 573 K, the sample was able to effectively sorb only

1.13 mmol·cm-3, which is smaller than the value in cycle 0 because some carbon

dioxide remained irreversibly sorbed.

Fig. 3.6 – Working capacity of HTC-10Ga-20K towards CO2 (see guiding solid line) obtained after consecutive sorption-desorption cycles at: a) 573 and b) 473 K.

By adopting this approach, it is thus possible to determine the working

capacity in each cycle. Fig. 3.6 shows that just after cycle 2 a steady working

Chapter 3

102

capacity is mostly attained; at 573 K this working capacity is ca. 0.97 mmol∙cm-3

whereas at 473 K it is ca. 0.25 mmol∙cm-3.

However, at the end of the first cycle, around 24 % and 77 % of the total

sorption capacity is irreversibly lost at 573 K and 473 K, respectively (cf. Fig. 3.7

and Fig. 3.8). In Fig. 3.7 the irreversible fraction at the end of each cycle is

determined as the difference between the black and white bars (for instance in

cycle 0 at 573 K such value is the one mentioned above, 0.38 mmol·cm-3).

Fig. 3.7 – History of sorbed (black bars) and desorbed (white bars) amount of CO2 for each cycle at a) 573 K and b) 473 K.


103

The irreversible part was found to be also pressure dependent as shown

in Fig. 3.8, where the accumulated irreversible fraction is shown; it is noteworthy

that when the pressure was increased up to ~1.1 bar (i.e. during cycle 3), the

number of moles of CO2 that were irreversible sorbed increased 51 % and 72 %

at 573 K and 473 K, respectively. However, in the subsequent cycles the working

capacity remained nearly constant at each temperature.

Fig. 3.8 – Irreversible CO2 moles sorbed per cycle considering the mass of solid obtained after the first regeneration step at 473 K and 573 K. Lines are for eye guidance.

The obtained results suggest that the CO2 sorption takes place through an

activated process; otherwise, it would be expected to obtain a higher sorption

capacity at 473 K. In this regard, several authors proposed different models to

describe the sorption equilibrium of carbon dioxide over hydrotalcites [7, 28, 33],

although in the present work a good fit was obtained using a simple Freundlich

equation (Section 3.3.2.1).

The changes observed in the CO2 sorption capacity during the

implemented cyclic process in the absence of water vapour can be, in principle,

related to the structural changes observed by XRD, in which the hydrotalcite-like

structure could not be recovered. Further work should thus aim assessing the

behaviour of these materials in the presence of water vapour, which is found in

Chapter 3

104

both typical post- and pre-combustion streams. In fact, other authors found that

in presence of water the working capacity loss practically does not occur (< 10

%) as compared to that exhibited by the same material under dry conditions [12].

Regarding the sorption mechanism, Ritter and co-workers [34] observed,

from in situ FTIR spectroscopy, the carbonate transformations occurring in K-

promoted hydrotalcites during CO2 adsorption and desorption at 723 K. In their

study, it was observed that one irreversible and three reversible processes took

place simultaneously, that could somehow be represented by two distinct kinetic

contributions. The slow and irreversible process was related to the formation of

polydentate carbonate, whereas the fast and reversible uptakes were related to

the formation of bridged (surface), uni- and bidentate carbonate species. Ritter

and co-workers concluded that the formation of bidentate carbonate species is

favoured at very short times. Unidentate and bridged carbonate species start to

disappear after one hour, eventually due to their transformation into polydentate

carbonate. In order to preliminary assess the sorption kinetics of sample HTC-

10Ga-20K, based on the above mentioned findings, the following expression is

herein proposed to tentatively describe the total amount of substance entering

or leaving the sorbent particle [35]:

2 2 2 21 2

2 2

1 2

1 2

2 2 2 21 1

0

0

6 1 6 11 1

D Di t i t

r r

i i

q t q tF F F

q q t

q qe e

q qi i

(3.3)

Eq. (3.3) assumes constant surface concentration in the sorbent particle

and two kinetic contributions occurring in parallel, one responsible for the fast

uptake at the beginning and, the other one, by the slow increase observed for

longer times. The model parameters 1

2

D

r(s-1), 2

2

D

r(s-1) and 1q (mmol·cm-3) were

obtained through non-linear fitting of Eq. (3.3) to experimental data. One should


105

note that q (mmol·cm-3) is known, since it is equal to the difference between

the sorbed (or desorbed) quantity at equilibrium and the value for t = 0.

Additionally, the quantity related to the slow uptake, 2q (mmol·cm-3), is equal to

the difference between q (known a priori) and 1q (fitted parameter).

Fig. 3.9 shows the experimental uptake (F), downtake (1-F) and model

curves based on Eq. (3.3), where F is the sum of curves F1 and F2 (please see

Fig. A.0.4 in Appendix A for the total and partial loading uptake curves).

Fig. 3.9 – Uptake (and downtake) curves for CO2 sorption at 573 K (first row) and 473 K (second row) for sample HTC-10Ga-20K.

t [s]

0 1000 2000 3000 4000 5000 6000

1-F

0.0

0.2

0.4

0.6

0.8

1.0

1-Fexp

1-F1

1-F2

1-Fmod

Cycle 4T = 573 Kp < 0.001 bar

t [s]

0 1000 2000 3000 4000

F

0.0

0.2

0.4

0.6

0.8

1.0

Fexp

F1

F2

Fmod

Cycle 5T = 573 Kp = 0.05 bar

t [s]

0 2000 4000 6000 8000 10000

1-F

0.0

0.2

0.4

0.6

0.8

1.0

1-Fexp

1-F1

1-F2

1-Fmod

Cycle 4T = 473 Kp < 0.001 bar

t [s]

0 500 1000 1500 2000 2500 3000

F

0.0

0.2

0.4

0.6

0.8

1.0

Fexp

F1

F2

Fmod

Cycle 5

T = 473 K

p = 0.05 bar

t [s]

0 1000 2000 3000 4000

F

0.0

0.2

0.4

0.6

0.8

1.0

Fexp

F1

F2

Fmod

Cycle 0T = 573 Kp = 0.05 bar

t [s]

0 1000 2000 3000 4000 5000

F

0.0

0.2

0.4

0.6

0.8

1.0

Fexp

F1

F2

Fmod

Cycle 0T = 473 Kp = 0.05 bar

Chapter 3

106

The model predicts quite well the experimental data both during sorption

and desorption at 573 K and 473 K; the obtained parameters are shown in Table

3.5. Generally, the inverse of the time constant 1

2

D

r during CO2 sorption at 473 K

is two-fold higher than at 573 K.

Table 3.5 – Kinetic parameters of Eq. (3.3) obtained after fitting CO2 uptakes and downtakes at 573 K and 473 K on sample HTC-10Ga-20K.

Parameter

Cycle 0 Cycle 4 Cycle 5

p = 0.05 bar (uptake)

< 0.001 bar (downtake)

p = 0.05 bar (uptake)

473 K 573 K 473 K 573 K 473 K 573 K

4 11

210

Ds

r

15.55 7.57 117 1.00 9.08 4.88

6 12

210

Ds

r

7.41 12.26 0.55 2.54 6.01 9.91

100q

q

1 48 44 63 11 61 40

-3

1 mmol cmq 0.422 0.603 0.141 0.100 0.109 0.355

-3

2 mmol cmq

0.467 0.754 0.084 0.916 0.069 0.524

-3 mmol cmq

0.889 1.358 0.225 1.016 0.178 0.879

When compared the uptakes performed at the beginning of the experiment

on the reactivated sample (cycle 0) and at the end (cycle 5), it is observed that

the inverse of time constants 1

2

D

r and 2

2

D

r decreased for both temperatures.

The sorbed fraction related to the first contribution 1 q q increased

between cycles (13 percentage points) at 473 K, while at 573 K practically

remained constant. Desorption is much faster at 473 K where ca. 63 % of the

total desorbed quantity is related to the fast contribution whereas at 573 K

represents only 11 %.


107

3.4 Conclusions

Hydrotalcite-based materials were prepared and their sorption capacity

towards CO2 at high temperature was determined. The working capacities

obtained under low CO2 pressures and high temperatures are lower than those

typically observed in absorption units using amines operating at low

temperatures (i.e. 1.36 mmol·g-1), which however require that the flue gas

stream should be cooled (and later heated up to for the solvent regeneration

stage). Nevertheless, at 573 K the hydrotalcites capacities are clearly above the

minimum value (i.e. 0.3 mmol·g-1) considered for their application in pre-

combustion capture applications, particularly in sorption-enhanced reaction

processes. The sorption capacity was enhanced by partially substituting

aluminium with gallium and, particularly, after modifying with alkali metals (K and

Cs). The sorption equilibrium isotherms for all the materials were adequately

fitted using a simple Freundlich-type equation.

The best material, i.e. hydrotalcite substituted with gallium and promoted

with potassium (HTC-10Ga-20K), showed to be suitable to be used in cyclic

operation with regeneration under low CO2 pressure, particularly at 573 K, where

a very good working capacity (0.97 mmol·cm-3) was obtained. At 473 K, more

than 75 % of the sorption capacity was irreversibly lost and a low working

capacity is thus obtained (0.25 mmol·cm-3). However, as mentioned before, it is

known that in presence of water vapour such working capacity should be almost

not affected.

Sorption kinetics of the HTC-10Ga-20K material was preliminary assessed

and a model proposed considering two resistances in parallel. The model

showed good adherence to experimental data both during sorption and

desorption experiments; the inverse of the time constant related to the fast

uptake at 473 K is generally two-fold higher than at 573 K.

Chapter 3

108

3.5 References

[1] IPCC: Summary for Policymakers, in: T.F. Stocker, D. Qin, G.-K. Plattner, M.

Tignor, S.K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex, P.M. Midgley (Eds.)

Climate Change 2013: The Physical Science Basis. Contribution of Working

Group I to the Fifth Assessment Report of the Intergovernmental Panel on

Climate Change, Cambridge University Press, Cambridge, United Kingdom and

New York, NY, USA, 2013.

[2] J. Wilcox, Carbon capture, Springer, New York, 2012.



ChemSusChem, 4 (2011) 1194-1215.

[4] S. Choi, J.H. Drese, C.W. Jones, Adsorbent materials for carbon dioxide

capture from large anthropogenic point sources, ChemSusChem, 2 (2009) 796-

854.



Science, 4 (2011) 42-55.

[6] A. de Roy, C. Forano, J.P. Besse, Layered double hydroxides: synthesis and

post-synthesis modification, in: V. Rives (Ed.) Layered double hydroxides:

present and future, Nova Science Publishers, Inc., New York, 2001, pp. 1-39.







Materials, 21 (2009) 3473-3475.


109

[9] Z. Yong, V. Mata, A.E. Rodrigues, Adsorption of carbon dioxide onto

hydrotalcite-like compounds (HTlcs) at high temperatures, Industrial and


[10] M.K. Ram Reddy, Z.P. Xu, G.Q. Lu, J.C. Diniz Da Costa, Influence of water

on high-temperature CO2 capture using layered double hydroxide derivatives,

Industrial and Engineering Chemistry Research, 47 (2008) 2630-2635.



[12] J.R. Hufton, S. Mayorga, S. Sircar, Sorption-enhanced reaction process for

hydrogen production, AIChE Journal, 45 (1999) 248-256.

[13] H.T.J. Reijers, S.E.A. Valster-Schiermeier, P.D. Cobden, R.W. Van Den

Brink, Hydrotalcite as CO2 sorbent for sorption-enhanced steam reforming of

methane, Industrial and Engineering Chemistry Research, 45 (2006) 2522-2530.

[14] Y.J. Wu, P. Li, J.G. Yu, A.F. Cunha, A.E. Rodrigues, K-Promoted

hydrotalcites for CO2 capture in sorption enhanced reactions, Chemical

Engineering and Technology, 36 (2013) 567-574.

[15] M.K.R. Reddy, Z.P. Xu, G.Q. Lu, J.C. Diniz da Costa, Effect of SOx

adsorption on layered double hydroxides for CO2 capture, Industrial &


[16] M.G. Beaver, High temperature CO2 chemisorbents: applications,

characterization, and study of the chemical nature of chemisorbent surfaces, in,

Lehigh University, 2010.

[17] E. Van Dijk, S. Walspurger, P. Cobden, R. Van Den Brink, Testing of

hydrotalcite based sorbents for CO2 and H2S capture for use in sorption

enhanced water gas shift, International Journal of Greenhouse Gas Control, 5

(2011) 505-511.

[18] J.R. Hufton, S. Mayorga, T. Gaffney, S. Nataraj, M. Rao, S. Sircar, Sorption

enhanced reaction process for production of hydrogen, Proceedings of the 1998

U.S. DOE Hydrogen Program Review., 1998.

Chapter 3

110

[19] V. Rives, Comment on "direct observation of a metastable solid phase of

Mg/Al/CO3-layered double hydroxide by means of high-temperature in situ

powder XRD and DTA/TG", Inorganic Chemistry, 38 (1999) 406-407.

[20] V. Rives, Characterisation of layered double hydroxides and their

decomposition products, Materials Chemistry and Physics, 75 (2002) 19-25.

[21] M.H. Halabi, M.H.J.M. De Croon, J. Van Der Schaaf, P.D. Cobden, J.C.

Schouten, High capacity potassium-promoted hydrotalcite for CO2 capture in H2

production, International Journal of Hydrogen Energy, 37 (2012) 4516-4525.

[22] M.K. Ram Reddy, Z.P. Xu, G.Q. Lu, J.C.D. Da Costa, Layered double

hydroxides for CO2 capture: Structure evolution and regeneration, Industrial and


[23] J. Rocha, M. Del Arco, V. Rives, M.A. Ulibarri, Reconstruction of layered

double hydroxides from calcined precursors: A powder XRD and 27Al MAS NMR

study, Journal of Materials Chemistry, 9 (1999) 2499-2503.

[24] K. Chibwe, W. Jones, Intercalation of organic and inorganic anions into

layered double hydroxides, Journal of the Chemical Society, Chemical

Communications, (1989) 926-927.

[25] T. Kwon, T.J. Pinnavaia, Pillaring of a layered double hydroxide by

polyoxometalates with Keggin-ion structures, Chemistry of Materials, 1 (1989)

381-383.

[26] M. Bellotto, B. Rebours, O. Clause, J. Lynch, D. Bazin, E. Elkaïm,

Hydrotalcite decomposition mechanism: a clue to the structure and reactivity of

spinel-like mixed oxides, The Journal of Physical Chemistry, 100 (1996) 8535-

8542.

[27] E.C. Kruissink, L.L. van Reijen, J.R.H. Ross, Coprecipitated nickel-alumina

catalysts for methanation at high temperature. Part 1.-Chemical composition and

structure of the precipitates, Journal of the Chemical Society, Faraday

Transactions 1: Physical Chemistry in Condensed Phases, 77 (1981) 649-663.


111

[28] K.B. Lee, A. Verdooren, H.S. Caram, S. Sircar, Chemisorption of carbon

dioxide on potassium-carbonate-promoted hydrotalcite, Journal of Colloid and

Interface Science, 308 (2007) 30-39.

[29] Z. Yong, V. Mata, A.E. Rodrigues, Adsorption of carbon dioxide at high

temperature - A review, Separation and Purification Technology, 26 (2002) 195-

205.

[30] N.D. Hutson, B.C. Attwood, High temperature adsorption of CO2 on various

hydrotalcite-like compounds, Adsorption, 14 (2008) 781-789.

[31] J.L. Soares, G.L. Casarin, H.J. José, R.D.F.P.M. Moreira, A.E. Rodrigues,

Experimental and theoretical analysis for the CO2 adsorption on hydrotalcite,

Adsorption, 11 (2005) 237-241.

[32] D.D. Do, Adsorption analysis: equilibria and kinetics, Imperial College

Press, London, 1998.

[33] H. Du, A.D. Ebner, J.A. Ritter, Pressure dependence of the nonequilibrium

kinetic model that describes the adsorption and desorption behavior of CO2 in

K-promoted hydrotalcite like compound, Industrial and Engineering Chemistry

Research, 50 (2011) 412-418.

[34] H. Du, C.T. Williams, A.D. Ebner, J.A. Ritter, In situ FTIR spectroscopic

analysis of carbonate transformations during adsorption and desorption of CO2

in K-promoted HTlc, Chemistry of Materials, 22 (2010) 3519-3526.

[35] J. Crank, The mathematics of diffusion, Second ed., Oxford University

Press, London, 1975.

113

Chapter 4. Intrinsic kinetics of CO2 methanation over an industrial nickel-based catalyst

The contents of this chapter were adapted from: C.V. Miguel, A. Mendes, L.M. Madeira,

Intrinsic kinetics of CO2 methanation over an industrial nickel-based catalyst, Journal of

CO2 Utilization, 2018, in press, http://doi.org/10.1016/j.jcou.2018.03.011.

The intrinsic kinetics of CO2 methanation over an industrial nickel-

based catalyst was determined for a temperature range between 250 ºC

to 350 ºC. The kinetic experiments were performed operating a fixed-bed

reactor far from equilibrium conditions, in the absence of heat and mass

resistances and at the atmospheric pressure. Five mechanistic-based

models for describing the reaction kinetics were taken from literature and

used for fitting the experimental reaction rates. Model discrimination was

based on the assessment of the thermodynamic consistency and

statistical significance of inherent parameters (for 95 % confidence

level). Comparison of the adequacy of fit between accepted models was

done through the determination of the corresponding F-values to select

the best model. The selected model assumes a formyl intermediate

mechanism with a hydroxyl group being the most abundant species. The

proposed reaction kinetics was further validated by the simulation of an

isothermal plug-flow reactor operating at the same experimental

conditions employed in this work, where a reasonable agreement

between model predictions and the observed values was obtained.

http://doi.org/10.1016/j.jcou.2018.03.011

Chapter 4

114

4.1 Introduction

The main application for the methanation reaction (also known as the

Sabatier reaction) has been since long ago the removal of carbon oxides traces

from hydrogen-rich feed streams in ammonia plants [1, 2] (Eq. (4.1)) and

Eq.(4.2)). It has also been proposed to produce synthetic natural gas (SNG),

particularly in the 60’s decade, because of increased natural gas demand. At

that time, SNG production from coal was envisioned as an alternative pathway

to assure security of supply in any event of natural gas shortage and research

on the topic was strongly financed, particularly by the USA [3].

298 K -12 2 4 2 rCO + 4H CH + 2H O Δ = - 165 kJ molH (4.1)

298 K -12 4 2CO + 3H CH + H O Δ = - 206 kJ molrH (4.2)

Nowadays, CO2 methanation reaction has gained a renewed interest in

the scope of Power-to-Gas applications (PtG), a concept where surplus

renewable electricity is transformed into hydrogen (via H2O electrolysis) and

afterwards into methane. This last step makes the process more flexible since

methane can be more easily stored and transported than hydrogen, enabling the

integration and balance of the power grid with the gas grid. Methane wide range

of end-use possibilities (e.g. vehicle fuel, for heat and power production,

intermediate to obtain other chemicals) contributes to process flexibility and

versatility [4].

Carbon dioxide methanation reaction (Eq. (4.1)) is thermodynamically

favoured at low temperatures and high pressures [5]. The equilibrium constant

(Keq) dependence with the absolute temperature (T) can be retrieved from the

equation provided by Lunde and Kester [6]:

Intrinsic kinetics of CO2 methanation over an industrial nickel-based catalyst

115

21.056000 +34633 -16.4ln +0.00557 +33.165

1.987eq=

T T T T

K e

(4.3)

The heat released by the reaction is the major difficulty to handle at

industrial scale [3]; the adiabatic rise in temperature per each percent of CO2

converted is 60 ºC [7]. Hence, methanation usually takes place in a series of

adiabatic fixed-bed reactors with inter-bed gas recycling cooling or in fluidized-

bed reactors [3, 8]. Structured catalysts such as metal coated foams have also

been envisaged for this reaction due to their improved heat transfer capacity

(e.g. [9, 10]).

Ni-based catalysts supported on various solids (e.g. Al2O3, SiO2, CeO2,

etc.) are the most studied and commercialized catalysts at high temperatures

(i.e. > 250 ºC), a range where the formation of dangerous nickel carbonyl

(Ni(CO)4) is avoided, while more expensive Ru-based catalysts are best options

at low temperatures (< 200 °C) [7, 11]. Generally, catalysts which are effective

for CO methanation (Eq. (4.2)) are also effective for CO2 methanation, at least

for streams having low COx concentrations, as found in hydrogen purification

processes [11]. Table 4.1 lists main manufacturers and characteristics of some

industrial methanation catalysts.

The catalysts are available in several shapes and some are supplied in a

pre-reduced form (e.g. Katalco 11-4R or PK-7R), which simplifies process start-

up. Industrial methanation catalysts lifetime ranges from 5 to 10 years, although

some manufacturers report a period up to 24 years [1, 7]. Common poisons are

sulphur and arsenic compounds, particularly for nickel catalysts [2]. Industrial

catalysts for hydrogenation of carbon oxides has been the subject of a detailed

review by Golosman and Efremov [7], covering the available commercial

catalysts and their properties, useful information about their preparation for use,

handling instructions and safety precautions, besides operation issues of some

industrial processes where methanation intervenes.

Chapter 4

116

Table 4.1 - List of some industrial methanation catalysts and related characteristics [7, 12-15].

Commercial reference

Metal Metal cont.

(wt. %) P

(bar) T

(ºC) Shape

Size (mm)

METH 134 a) Ni 20-25 n.a. >200 Ball 3.0 - 6.0

METH 150 a) Ru n.a. n.a. <170 Ball 4.5 - 4.5

PK-7R b) Ni >23 n.a. 190-450 Ring 5.0 x 2.5

Katalco 11-4 c) Ni n.a n.a. n.a Pellet 5.4 x 3.6

Katalco 11-4R c) Ni n.a n.a. >220 Pellet 5.4 x 3.6

Katalco 11-4M c) Ni n.a n.a. n.a Pellet 3.1 x 3.6

Katalco 11-4MR c) Ni n.a n.a. >220 Pellet 3.1 x 3.6

NIAP-07-01 d) Ni 33-39 20-300 180-450 Pellet 5.5 x 4.5

NIAP-07-02 d) Ni 32-38 20-300 180-450 Pellet 5.5 x 5.0

NIAP-07-04 d) Ni 25 n.a n.a Pellet 5.5 x 4.5

TO-2M d) Ni 35-41 1-300 180-450 Pellet 5.5 x 5.0

RKM-3 d) Ru 0.3 300 150 Ball 2.6 - 3.0

Manufacturer

a) Clariant b) Haldor-Topsoe c) Johnson Matthey d) NIAP Katalizator

This work determines the intrinsic reaction kinetics over an industrial

nickel-based catalyst. Knowing the reaction kinetics is fundamental for

modelling, simulation and optimization of conventional or new reactor concepts

(e.g. sorptive reactors (e.g. [16]). To this end, mechanistic-based rate equations

available in the literature were considered and are presented in the following

section.

4.2 Mechanisms for CO2 methanation

4.2.1 “Carbon intermediate” mechanism

Dalmon and Martin [17] proposed a mechanism for CO and CO2

methanation over a Ni/SiO2 catalyst by studying the hydrogenation of

intermediate species. The authors postulated that in both reactions adsorbed

carbon monoxide (CO*) is a common intermediate (apart from O* in the latter

case). CO* then dissociates into C* and O* while CH4 is produced following C*


117

hydrogenation. Since CO* can be formed at lower temperatures in the case of

CO2 adsorption, a lower activation energy could explain the higher

hydrogenation rate of CO2. The higher selectivity of CO2 methanation was

tentatively explained by the lower C* concentration (because of lower CO2

adsorption capacity) and to the higher abundance of O*, which may act as a

geometric diluent decreasing the C-C formation.

Weatherbee and Bartholomew [18] also used a Ni/SiO2 catalyst and the

mechanism proposed by Dalmon and Martin adequately explained their results.

The elementary steps of the mechanism are presented in Table 4.2.

Table 4.2 – Carbon intermediate mechanism (from [18]).

Elementary step Step

2H + 2* 2H* 1

2CO + 2* CO* + O* 2

CO* CO + * 3

CO* + * C* + O* (r.d.s.) 4

C* + 4H* 4CH * + 4* 5

4CH * 4CH + * 6

O* + H* OH* + * 7

OH* + H* 2H O* + * 8

2H O* 2H O + * 9

The kinetic rate equation derived, assuming CO* dissociation to C* and O*

(step 4) as the rate determining step (r.d.s.), is given by Eq. (4.4).

2 2

4

2

2 2

2

0.5 0.5

CH 20.5

0.5 0.51 2 30.5

1

CO H

CO

CO H CO

H

kp pr

pK K p p K p

p

(4.4)

Chapter 4

118

where k is the rate constant, pi is the partial pressure of species i (CO2, H2, CO)

and K1, K2 and K3 are the constants that result from the combination of several

adsorption equilibrium and/or reaction rate constants.

4.2.2 “Formate intermediate” mechanism

Fujita et al. [19] studied the mechanism of CO and CO2 methanation over

an Ni/Al2O3 catalyst at 180 ºC. For CO methanation it was observed that

adsorbed carbon, linear and bridged CO were the predominant species,

whereas for CO2 methanation the presence of two types of adsorbed CO in

bridged structures and a formate species (HCOO) were detected. The presence

of linear CO species markedly retarded the hydrogenation of adsorbed carbon.

Since the latter species has not been observed during CO2 methanation, this

could explain the difference between the rate and selectivity of CO and CO2

methanation reactions [19, 20].

Pan et al. [21] recorded the in-situ FTIR spectra of CO2 methanation using

a Ni/Al2O3 catalyst at different temperatures (75-375 ºC). The presence of

bidentate formate intermediate was observed for temperatures above 225 ºC

and its hydrogenation was considered the main reaction step. Both Fujita et al.

[19] and Pan et al. [21] concluded that formate species was adsorbed at the

alumina support. A mechanism for CO2 hydrogenation over a commercial nickel-

alumina-calcium catalyst (ref.: NKM-4A from NIAP-Katalizator) involving a

formate intermediate was given by Ibraeva et al. [22] (cf. Table 4.3).


119

Table 4.3 – Formate intermediate mechanism (from [22]).


2H + 2* 2H* 1

2CO + * 2CO * 2

2CO * + H* HCOO* + * (r.d.s.) 3

HCOO* + H* HCO* + OH* 4

HCO* + H* CH* + OH* 5

CH* + 3H* 4CH * + 4* 6

OH* + H* 2H O + 2* 7

4CH * 4CH + * 8

In this mechanism, the formate species is obtained through the reaction of

molecularly adsorbed CO2 and atomic hydrogen (step 3), being this the rate

determining step of the reaction. The derived kinetic rate equation is expressed

by Eq. (4.5).

2 2

2 2

0.5

0.5

H CO

H CO

p pr k

p K p

4CH

(4.5)

where K is the adsorption equilibrium constant.

4.2.3 “Formyl intermediate” mechanism

A mechanism comprising a formyl species intermediate has been

proposed recently by Koschany et al. [23] for a coprecipitated NiAl(O)x catalyst

(cf. Table 4.4).

Chapter 4

120

Table 4.4 – Formyl intermediate mechanism (from [23]).


2H + 2* 2H* 1

2CO + 2* CO* + O* 2

CO* + H* CHO* + * (r.d.s.) 3

CHO* + * CH* + O* 4

CH* + 3H* 4CH * + 3* 5

4CH * 4CH + * 6

O* + H* OH* + * 7

OH* + H* 2H O* + * 8

2H O* 2H O + * 9

In this mechanism, the formyl species results from the reaction of adsorbed

CO (obtained after CO2 dissociation) and atomic hydrogen. Eq. (4.6) was

derived by Koschany et al. [23] assuming adsorbed CO hydrogenation (step 3)

as the rate limiting step and CO*, H* and OH* as the most abundant surface

species.

2 2 4 2

4

2 22

2 2 2

2

0 5 0 5 2

CO H CH H

CH 2 4

CO H eqH O 0 5 0 5 0 5

OH H H mix CO0 5

H

1

1

. .

O

. . .

.

k p p p pr

p p KpK K p K p

p

(4.6)

where 2

2

H O

OH 0 5

H

.

pK

p

, 2 2

0 5 0 5

H H

. .K p and 2

0 5

mix CO

.K p stand for the surface coverage of

hydroxyl OH , atomic hydrogen H and carbon monoxide CO ,

respectively. eqK is the reaction equilibrium constant determined through Eq.

(4.3).


121

If water is considered as one of the most abundant surface intermediates

instead of hydroxyl species, than Eq. (4.6) is modified as follows [23].

2 2 4 2

4

2 22 2 2 2 2

0 5 0 5 2

CO H CH H

CH 2 40 5 0 5 0 5

CO HH O H O H H mix CO

1

1

. .

O

. . .eq

k p p p pr

p p KK p K p K p

(4.7)

and the term 2 2H O H OK p stands for the water surface coverage

2H O .

4.3 Experimental

4.3.1 Experimental setup

An illustration of the experimental setup used in the kinetic tests is shown

in Fig. 4.1.

Fig. 4.1 – Scheme of the experimental setup used in this work.

Experiments were performed in a stainless steel packed-bed unit with 12

cm length and 7.75 mm inside diameter, which was placed inside a tubular oven

(model Split from Termolab, Fornos Eléctricos, Lda.) equipped with a 3-zone PID

temperature controller (model MR13 from Shimaden). The 3 type-K

Chapter 4

122

thermocouples used to measure and control the oven temperature were placed

in contact with the reactor wall.

CO2 (99.998 %, Air Liquide), H2 (99.9995 %, Air Liquide) and N2

(99.9995% %, Air Liquide) were fed upwardly to the reactor using mass flow

controllers (model F201 from Bronkhorst-High Tech). The flow rate of the stream

leaving the reactor was measured with a mass flow meter (model F101 from

Bronkhorst-High Tech).

Type-K thermocouples allowed recording the temperature histories in two

axial positions of the reactor. The reactor was placed inside the oven so that the

temperature in the bed was practically the same along its length (maximum

difference of 1 ºC). The pressure drop was measured by means of two pressure

transducers (model PMP 4010 from Druck) placed before and after the reactor.

The water vapour produced during the course of the reaction was condensed in

a home-assembled Peltier cold-trap located after the reactor.

The steady-state composition of the (dry) stream leaving the reactor was

measured using a gas chromatograph (model 1000 from DANI) equipped with a

micro-TCD detector (VICI) and a capillary column (Supelco Carboxen 1010 Plot,

30 m x 0.32 mm i.d. from Sigma Aldrich). Hydrogen fraction was obtained

through the mass balance. The carbon balance error was lower than 10 %.

4.3.2 Kinetic experiments

Kinetic tests were performed using a commercial nickel catalyst (METH

134). The catalyst has a NiO content of ca. 20-25 wt. % supported on calcium

aluminate [7, 24]. Axial dispersion and gas-solid wall effects can be neglected

when L/dp and Dr/dp ratios are above 50 and 10, respectively [25, 26], which was

guaranteed in the present work. Pressure drop along the bed may be assumed

negligible (< 7.4 % of inlet pressure).

Temperature was varied between 250-350 ºC, a range where commercial

catalysts are designed to operate without activity loss, even with occasional


123

temperature excursions up to 650 ºC [2]. The activity of commercial nickel

catalysts does not increase significantly above 350 ºC [27]. The experiments

were performed at the atmospheric pressure. Blank tests using the reactor

packed with only glass beads showed the absence of CO2 conversion in the

considered temperature range.

The first set of kinetic tests was carried at 350 ºC to check the absence of

external mass and heat transport resistances by varying the feed flow-rate while

the value of the time coordinate CO2

in

catW F was kept constant (see sections

4.4.1.1 and 4.4.1.2). In the absence of external mass/heat transfer effects and

in the presence of isothermal operation at 350 ºC, the same premises are

guaranteed at lower temperatures and lower catalyst activities [25].

After assuring the absence of external resistances, the presence of

internal diffusion was assessed using catalyst particles with different sizes (see

section 4.4.1.2). Intrinsic kinetics can thus be determined if the experiments are

conducted in the absence of both external and internal resistances. In this

regard, catalyst pellets were crushed and sieved to different sizes, namely,

between 150-250 µm and 250-350 µm; Kiendl et. al [24] used a similar catalyst

and reported that crushing has no influence on the Ni loading. The catalyst

particles were then mixed with inert glass beads with the same particle size.

Dilution of the catalyst with inert glass beads enhances the heat transfer and

minimizes temperature gradients, which was confirmed through direct

measurement of the bed temperature. Due to the highly exothermic nature of the

Sabatier reaction (Eq. (4.1)), the CO2 content in the reactant mixture was low

and balanced in N2 to minimize the temperature increase. An H2/CO2 feed ratio

of 4 was chosen for all experiments; at low temperature and atmospheric

pressure, H2/CO2 ratios lower than 4 could lead to low CH4 selectivity and

undesired carbon formation, as found in Chapter 2.

The mixture of catalyst and inert particles was loaded into the fixed-bed

tubular unit and framed in both ends by means of two stainless steel discs (mesh:

10-15 μm). At the bottom of the reactor (i.e. at the inlet) a layer of inert particles

Chapter 4

124

allows to even out the flow of the reaction mixture. Catalyst activation consisted

in the following protocol:

1. Purging the entire system with N2 at room temperature;

2. Heating the packed-bed (1 °C∙min-1) from room temperature up to 320

°C at the ambient pressure under a 12.8 % H2/N2 stream and hold 1h at

that temperature;

3. Increasing the temperature under N2 flow to the desired value for the first

kinetic experiment (i.e. 350 ºC).

A 4-way valve allowed to: 1) keep the reactor under a static N2 atmosphere

while the reactant mixture stream could be prepared and the flow stabilized and

2) feed the reactant stream to the reactor. Shutdown consisted on flushing the

reactor with N2 to remove unreacted carbon oxides and afterwards cooling down

the reactor up to room temperature.

4.3.3 Computational methods

The parameters of the kinetic equations were estimated through nonlinear

regression and statistically tested using the fitnlm function available in the Matlab

software as detailed below (cf. section 4.4.2.1). Nonlinear regression algorithm

consisted in the Levenberg-Marquardt method (details available in [28]) for the

minimization of the sum of residuals squares. Statistical testing was performed

for 95 % level of significance. The selected kinetic model was validated based

on the simulation of the fixed-bed reactor. The differential equations and

respective boundary conditions were solved numerically with a fourth-order

Runge-Kutta algorithm also using the Matlab software.


125


4.4.1 Kinetic experiments

4.4.1.1 Isothermal regime and catalyst stability

Reactor layouts were varied to assess the system behaviour regarding the

planned operation conditions (cf. Fig. 4.2).

Fig. 4.2 – Reactor layouts employed to determine the best configuration for kinetic tests.

Isothermal operation must be guaranteed for the determination of the

intrinsic kinetics. Due to the exothermic nature of CO2 methanation, the bed

temperature rise depends on the number of moles of CO2 present in the feed

composition which are converted per unit of time. In preliminary experiments,

the initial bed temperature increased 8.7 ºC vs. 2.8 ºC by changing the CO2 feed

fraction from 12 vol. % to 4 vol. %, respectively, using a reactor with 200 mg of

catalyst (reactor R1 in Fig. 4.2). This effect was not possible to assess for lower

CO2 percentages due to mass flow controllers limitations. In turn, the available

heat transfer area per mass of catalyst was increased changing the dilution

factor (Ψ) and using a feed composition (vol. %) of 4:16:80 (CO2:H2:N2) in all the

subsequent experiments. Isothermal regime was considered when the variation

Chapter 4

126

of the temperature along the bed length was lower than 1.5 ºC, which was

attained for the reactor with configuration R3 in Fig. 4.2, under the considered

range of total feed flow rates (30-100 mln∙min-1). The catalyst mass used in the

subsequent experiments varied between 12-40 mg.

Catalyst stability was checked under reactive conditions for a total period

of approximately 83 hours. During this time the catalyst was submitted to

reactive conditions in the temperature range of 250-350 ºC. The catalyst was

always kept in a nitrogen stream at 350 ºC overnight. The observed difference

of CH4 yield measured at the beginning and the end of this procedure (operation

conditions: T = 350 ºC, P = Patm, CO2

in

catW F = 6.22 gcat·h·mol-1) was negligible (ca.

0.61 %). Therefore, in the kinetic tests it was considered that the catalyst activity

remained unchanged since they were conducted in less time than the reference

period.

4.4.1.2 Identifying the region of kinetic rate control

The external and intraparticle mass and heat transport transfer limitations

were assessed based on the criterion listed in Table 4.5.

Table 4.5 – Criteria for negligible transport limitations in steady state kinetic studies [29].

Transport process Criterion Eq.

External mass transport obs p

b

0 05g

rCa .

k a' C

(4.8)

Intraparticle mass transport (Wheeler-Weisz criterion)

2

obs p2

eff s

10 10

2

p rr L n

.D C

(4.9)

External heat transport g b a

b b

0 05r

e e

k H C ECa Ca .

hT RT

(4.10)

Intraparticle heat transport eff s2 2a

s s

0 05r

i i

eff

D H C E.

T RT

(4.11)


127

The expressions employed for determining the thermodynamic and

transport properties and parameters used throughout Eqs. (4.8)-(4.11) are given

in the Appendix B (see parameters definitions and units in the Nomenclature

section). The criterion listed in Table 4.5 were always verified considering the

extreme situations, i.e.:

1) Conditions of minimum and maximum reaction rate at 350 ºC and;

2) Using either reactor feed or outlet compositions for calculating the

parameters.

The inequalities of Eqs. ((4.8)-((4.11) were all verified and thus both external and

intraparticle heat and mass transport limitations could be neglected. In addition,

the effect of external resistances was also assessed experimentally at 350 ºC

by determining CO2 conversion and varying the feed flow rate, while keeping the

time coordinate CO2

in

catW F constant (cf. Fig. 4.3).

Fig. 4.3 – Effect of total volumetric feed flow rate on CO2 conversion for constant

2

incat COW F = 3.73 g·h·mol-1 at 350 ºC. Dashed line stands for the thermodynamic

equilibrium conversion at 350 ºC, 1 bar and feed composition.

Fig. 4.3 shows the absence of external resistances for a feed flow rate

range between 30-100 mln·min-1. Internal diffusion effects can also be neglected

Qin / ml

n min

-1

0 20 40 60 80 100 120

XC

O2

/ %

0

20

40

60

80

100

150-250 m

250-350 m

Chapter 4

128

since the conversion obtained with catalyst particles of a different size was

nearly the same (Fig. 4.3).

4.4.1.3 Calculation of observed reaction rates

Fig. 4.4 shows the observed methane yield 4 4 2

out inCH CH COY F F as a

function of the ratio between the weight of catalyst catW to the CO2 molar feed

flow rate 2

in

COF at temperatures ranging from 250 ºC to 350 ºC.

Fig. 4.4 – Methane yield as a function of the time coordinate 2

incat COW F at various

temperatures. The continuous line stands for the equilibrium value of 4CHY at 350 ºC, 1

bar and feed composition (i.e. the minimum 4CHY for the temperature range considered).

The observed reaction rates were obtained using the differential method

[26].

(Wcat

/ Fin

CO2

) / g h mol-1

0 2 4 6 8 10 12 14

YC

H4 ,

ob

s

0.0

0.2

0.4

0.6

0.8

1.0

350 ºC

325 ºC

300 ºC

275 ºC

250 ºC


129

4

4

2

CH

CH incat CO

d

d

Yr

W F (4.12)

To this end, the relationship between 4CHY and

CO2

in

catW F was described in terms

of polynomial functions for each temperature (see dashed lines in Fig. 4.4). The

derivative of each polynomial function allowed to extract the observed reaction

rates at a given temperature and CO2

in

catW F (cf. Eq. (4.12)).

4.4.2 Modelling work

4.4.2.1 Model discrimination and parameter estimation

A scaling method consisting in temperature centering was adopted for the

estimation of rate and adsorption parameters of the proposed kinetic equations,

according to Eq. (4.13) and Eq. (4.14), respectively.

m

1 1

0

aE

R T T'k k e

(4.13)

m

1 1

0

H

R T T'K K e

(4.14)

where mT stands for the mean temperature of the considered range (i.e. 573 K)

and 0

'k and 0

'K are the constants evaluated at T = Tm. Temperature centering

reduces the correlation between the pre-exponential factor and Ea or ΔH, making

the numerical least-squares procedure more robust and less sensitive to poor

starting estimates [28]. Afterwards, the “true” pre-exponential factors are

calculated from Eq. (4.15) and Eq. (4.16), by the Arrhenius and Van’t Hoff

equations, respectively.

Chapter 4

130

a

m

0 0

E

RT'k k e

(4.15)

m

0 0

H

RT'K K e

(4.16)

Model discrimination was based on the thermodynamic consistency and

statistical validity of the estimated parameters and the overall adequacy of fitting

given by the F-value with a confidence level , as proposed by Froment and

Bischoff [26] (cf. Eq. (4.17)),

2

21

mod

k

mod obs

k

r

pF value F p,k p;

r r

k p

(4.17)

where k and p stand for the number of experiments and estimated parameters,

respectively. The statistical significance of parameters was assessed through

the t-value, which is equal to the parameter estimate minus zero (null

hypothesis) divided by the standard deviation of the parameter. If the obtained

t-value is lower than the t-value for = 0.95 and (k-p) degrees of freedom, or

has a large confidence interval including zero, it has no significant contribution

to the rate equation and can be deleted [26].

The kinetic models proposed in section 4.2 were then rejected based on

the criteria mentioned above. However, the experimental data could be quite

well described after deleting the hydrogen and carbon monoxide coverage terms

from the denominator of Eq. (4.6) and Eq. (4.7). The simplified reaction rate

equations are as follows:


131

2 2 4 2

4

2 22

2

0 5 0 5 2

CO H CH H

CH 2 4

CO HH O

OH 0 5

H

1

1

. .

O

eq

.

k p p p pr

p p KpK

p

(4.18)

2 2 4 2

4

2 22 2

0 5 0 5 2

CO H CH H

CH 2 4

CO HH O H O

1

1

. .

O

eq

k p p p pr

p p KK p

(4.19)

The kinetic model assuming water as most abundant species (Eq. (4.19))

was however rejected because the obtained F-value was lower than the F-value

for the model expressed by Eq. (4.18) (209 vs. 259).

Fig. 4.5 shows the parity plot of the observed reaction rates against those

obtained by the model of Eq. (4.18).

Fig. 4.5 – Parity of plot of rmod versus robs showing adequacy of fit of proposed model (Eq. (4.18)).

Table 4.6 lists the parameters estimates with a 95 % confidence level (at

reference conditions).

robs

x 100 / mol gcat

-1 h

-1

0 2 4 6 8 10 12 14

r mo

d x

10

0 /

mo

l g

ca

t-1 h

-1

0

2

4

6

8

10

12

14

350 ºC

325 ºC

300 ºC

275 ºC

250 ºC

Chapter 4

132

Table 4.6 – Parameter estimates at reference temperature (573 K) and corresponding lower (LL) and upper (UL) limits of the 95 % confidence interval.

Parameter Units Estimate t-value LL UL

0

'k -1 -1mol g h 100 0.8936 8.959 0.7938 0.9933

ER

K

14275 13.34 13205 15345

0,OH

'K kPa-0.5 0.4326 3.001 0.2885 0.5768

HR

K

7414 3.626 5369 9458

Besides the significance of the estimated parameters, the F value was

also satisfied (i.e. 259 > 2.76). The estimated activation energy was 118.7

kJ·mol-1, a value close to others reported in literature for similar nickel catalysts

(e.g. 105.9 kJ·mol-1 in [30] and 113.5 kJ·mol-1 in [31]). The adsorption enthalpy

of the hydroxyl coverage term is positive, which is in agreement with the results

by Koschany et al. [23]. The obtained hydroxyl enthalpy of adsorption was 61.6

kJ·mol-1, a value which is however 2.75 times higher than the value reported by

Koschany et al. [23].

4.4.2.2 Model validation

A one-dimensional pseudo-homogeneous plug-flow model was used to

validate the selected kinetic expression against the experimental data. The

following main assumptions have been made:

1) Isothermal operation;

2) Negligible axial and radial dispersion;

3) Negligible mass and heat transport limitations within the catalyst

(internal) and between the catalyst and the bulk gas phase (external);

4) Ideal gas behaviour.

5) Reaction takes place on the catalyst surface.

The model included the estimation of pressure (P) along the dimensionless

reactor length (z) by the Ergun equation [32]:


133

2

g g 2o o3 2 3

p p

1 1d150 1 75

d

PL u . u

z d d

(4.20)

where L is the reactor length, is the column porosity, pd is the particle diameter,

g is the gas mixture viscosity, ou is the fluid superficial velocity and g is the

gas mixture density.

The density of the gas mixture g was calculated locally through

g i i1

n

i

Py M

RT

(4.21)

where R is the ideal gas constant, while iy and iM stand for the mole fraction

and molar mass of species i, respectively, and n for the number of species.

The gas mixture viscosity g was estimated locally applying the Wilke

method [33], as follows (Eq. (4.22)):

i ig

j ijij

y

y

(4.22)

with parameter ij being estimated through Eq. (4.23) [33]

21 4

ji

j i

ij 1 2

i

j

1

8 1

M

M

M

M

(4.23)

Chapter 4

134

and the viscosity of the pure components i by Eq. (4.24)

B

i

2

A

C D1

T

T T

(4.24)

where A, B, C and D are coefficients that depend on the species and were

retrieved from Aspen Properties software. Table 4.7 lists the coefficients A-D

used to determine pure viscosity of each species.

Table 4.7 – List of coefficients A-D to determine pure viscosity (in Pa·s) of each species.

Species A B C D

CO2 2.1480 x 10-6 0.46000 290.000 0 H2 1.7970 x 10-7 0.68500 -0.590 140 CH4 5.2546 x 10-7 0.59006 105.670 0 H2O 1.7096 x 10-8 1.11460 0.000 0 N2 6.5592 x 10-7 0.60810 54.714 0

The fluid superficial velocity ou was estimated along z by the following

differential equation obtained from the total mass balance to the reactor in

steady-state (Eq. (4.25)):

o o cati

r

d d

d d i

u u WP RTLr

z P z V P (4.25)

where catW is the weight of catalyst, rV is the reactor volume and i is the

stoichiometric coefficient of species i (excluding N2).

The mole fraction of species i along z was estimated by the following

general equation obtained from the partial mass balance to the reactor in steady-

state (Eq. (4.26)):


135

o cati i ii

o r o

dd d

d d d

u Wy y yRT L Pr

z u z V P u P z (4.26)

The initial conditions for solving the differential equations (4.20), (4.25) and

(4.26) were in0z

P P

, 0

0o ozu u

and 0

i i0zy y

, where the superscript “0”

refers to feed conditions.

Fig. 4.6 shows the model predictions of methane yield compared to the

observed values. A very good agreement was found between model predicted

and observed values along the large methane yield and temperature ranges.

Fig. 4.6 – Model predicted vs. observed values of the methane yield at different temperatures.

Fig. 4.7 shows the simulation profiles of the species mole fractions and

methane formation reaction rate along the dimensionless reactor length

calculated in conditions of maximum rate (i.e. higher temperature and lower

2

incat COW F ). The predicted values at the reactor outlet (i.e. z=1) are compared

with the experimental values (see the red squares in Fig. 4.7).

YCH

4,obs

/ %

0 20 40 60 80

YC

H4,m

od /

%

0

20

40

60

80

350 ºC

325 ºC

300 ºC

275 ºC

250 ºC

Chapter 4

136

Fig. 4.7 – Simulated values along the dimensionless reactor length of: a) the species

mole fractions and b) reaction rate. T=350 ºC and 2

in -1

cat CO 1 52 g h molW F . . Red

squares stand for the experimental values.

Fig. 4.7a shows that H2 and CO2 are consumed along all the reactor length

while H2O and CH4 are continuously produced because one is still far from

equilibrium conditions. N2 fraction slightly increases because of the

stoichiometry of the reaction. The experimental values measured at the reactor

outlet agree well with the predicted values, being the most noticeable deviation

observed for hydrogen. It should be however recalled that hydrogen fraction was


137

determined through the molar balance and it is therefore subject to the errors of

the determination of each species composition.

Regarding to the reaction rate, it decreases steeply during the first half of

the reactor length and then starts declining slowly until a value of 0.10

-1catmol g h is reached at the exit, a value that is very close to the observed

reaction rate (0.13 -1catmol g h ) (Fig. 4.7b).

The simulated methane yield profiles along the reactor length obtained in

the temperature and reaction rate limits are shown in Fig. 4.8.

Fig. 4.8 – Simulated methane yield profile along the dimensionless reactor length for the highest and lowest robs conditions measured at 250 ºC and 350 ºC. Red squares symbols stand for the experimental values.

The model was able to capture the behaviour of the reactor performance

in the considered temperature window limits (250 ºC to 350 ºC) and in both of

minimum and maximum reaction rates conditions.

z

0.0 0.2 0.4 0.6 0.8 1.0

YC

H4

/ %

0

20

40

60

80

350 ºC

250 ºC

2

in

cat CO minW F

2

in

cat CO maxW F

2

in

cat COmax

W F

2

in

cat CO minW F

Chapter 4

138

4.5 Conclusions

The kinetics of the methanation reaction over an industrial nickel-based

catalyst was determined for the relevant 250 ºC to 350 ºC temperature window.

Three reaction mechanisms assuming different intermediate species were

selected from the literature. The mechanism which assumes hydrogen and

carbon dioxide dissociation followed by hydrogenation of adsorbed carbon

monoxide to yield a formyl species, assuming hydroxyl as the most abundant

species, showed a good fit to the experimental data.

An isothermal plug-flow model including the proposed kinetic rate equation

with estimated parameters was used to simulate the fixed-bed reactor. The

reactor model satisfactorily captured the experimental observations in limiting

conditions of temperature and CO2

in

catW F , allowing to validate the proposed

(intrinsic) kinetic rate model.

The selected kinetic rate and estimated model parameters can thus be

transferred and used for modeling, simulation and design of industrial reactors

used for CO2 utilization applications, where methanation is standing out among

the existing options.


139

4.6 Nomenclature

Parameter Description Units

a' Specific external area of the catalyst

particle ( p6a' d for spheres) m-1

Ca Carberry number -

bC CO2 concentration in the bulk phase mol·m-3

sC CO2 concentration at the surface of the catalyst particle

mol·m-3

pd Particle diameter m

effD Effective mass diffusivity in the catalyst m2·s-1

Dr Reactor diameter m

aE Activation energy J·mol-1

4

outCHF CH4 outlet molar feed flow rate mol·s-1

2

inCOF CO2 inlet molar feed flow rate mol·s-1

h Heat transport coefficient between gas and particle

W·m-2·K-1

k Reaction rate constant mol·g-1·s-1·Pa-n kg Mass transport coefficient m·s-1

0k Pre-exponential factor of rate constant mol·g-1·s-1·Pa-n

0K Pre-exponential factor of adsorption equilibrium constant

Pa-1 or Pa-0.5

K Adsorption equilibrium constant Pa-1 or Pa-0.5 Keq Equilibrium constant for methanation Pa-2 L Reactor length m

pL Characteristic catalyst dimension

6 for spheresp pL d m

Mi Molar weight of species i kg·mol-1 nr Reaction order - pi Partial pressure of species i Pa P Total pressure Pa Q Volumetric feed flow rate mLn·min-1

4CHr Reaction rate of methane formation mol·g-1·s-1

obsr Observed reaction rate mol·g-1·s-1

modr Model predicted reaction rate mol·g-1·s-1

R Ideal gas constant 8.314 J·mol-1·K-1 T Absolute temperature K

bT Temperature in the bulk phase K

Chapter 4

140

sT Temperature at the catalyst particle surface

K

Tm Mean temperature of range (Tm=300 ºC) K

ou Fluid superficial velocity m·s-1

rV Reactor volume m3

catW Catalyst weight kg

2COX CO2 conversion -

iy Mole fraction of species i -

4CHY Methane yield -

z Dimensionless reactor length - Greek letters

H Enthalpy of adsorption J·mol-1

rH Heat of reaction J·mol-1

e Dimensionless number for extraparticle heat transport

-

i Dimensionless number for intraparticle heat transport

-

e External Arrhenius number -

i Internal Arrhenius number -

Column porosity (0.36-0.38) -

g Density of the gas mixture kg·m-3

p Density of the catalyst particle g·m-3

Internal effectiveness factor - Generalized Thiele modulus -

2 Wheeler-Weisz modulus -

g Thermal conductivity of the gas mixture W·m-1·K-1

eff Effective thermal conductivity of the catalyst particle

W·m-1·K-1

i Viscosity of pure species i Pa·s

g Viscosity of the gas mixture Pa·s

i Stoichiometric coefficient of species i -

Ψ Inert dilution factor -


141

4.7 References

[1] H.F. Rase, Handbook of commercial catalysts: heterogenous catalysts, CRC

Press, Boca Raton, 2000.

[2] M.V. Twigg, Catalyst Handbook, 2nd ed., Manson Publishing Ltd, London,

1996.

[3] J. Kopyscinski, T.J. Schildhauer, S.M.A. Biollaz, Production of synthetic

natural gas (SNG) from coal and dry biomass - A technology review from 1950

to 2009, Fuel, 89 (2010) 1763-1783.

[4] M. Götz, J. Lefebvre, F. Mörs, A. McDaniel Koch, F. Graf, S. Bajohr, R.

Reimert, T. Kolb, Renewable Power-to-Gas: A technological and economic

review, Renewable Energy, 85 (2016) 1371-1390.

[5] C.V. Miguel, M.A. Soria, A. Mendes, L.M. Madeira, Direct CO2 hydrogenation

to methane or methanol from post-combustion exhaust streams - a

thermodynamic study, Journal of Natural Gas Science and Engineering, 22

(2015) 1-8.

[6] P. Lunde, F. Kester, Carbon dioxide methanation on a ruthenium catalyst

Industrial & Engineering Chemistry Process Design and Development 13 (1974)

27–33.

[7] E.Z. Golosman, V.N. Efremov, Industrial catalysts for the hydrogenation of

carbon oxides, Catal. Ind., 4 (2012) 267-283.

[8] T.T.M. Nguyen, L. Wissing, M.S. Skjøth-Rasmussen, High temperature

methanation: Catalyst considerations, Catalysis Today, 215 (2013) 233-238.

[9] Y. Li, Q. Zhang, R. Chai, G. Zhao, Y. Liu, Y. Lu, Structured Ni-CeO2-Al2O3/Ni-

foam catalyst with enhanced heat transfer for substitute natural gas production

by syngas methanation, ChemCatChem, 7 (2015) 1427-1431.

[10] Z. Wan, J. Jiang, H. Ma, Y. Li, Y. Lu, F. Cao, A study of the heat transfer

characteristics of novel Ni-foam structured catalysts, Canadian Journal of

Chemical Engineering, 94 (2016) 2225-2234.

Chapter 4

142

[11] G.H. Watson, Methanation catalysts, International Energy Agency Coal

Research, London, 1980.

[12] Chemical catalysts: catalysts for olefin processes, in: J. Matthey (Ed.), 2012.

[13] Catalysts for Syngas, in: C.I. Ltd. (Ed.), 2010.

[14] PK-7R low temperature methanation catalyst, in: H. Topsoe (Ed.), 2001.

[15] Alvigo-Matros catalysts, in: Alvigo (Ed.), 2004.

[16] C.V. Miguel, M.A. Soria, A. Mendes, L.M. Madeira, A sorptive reactor for

CO2 capture and conversion to renewable methane, Chemical Engineering

Journal, 322 (2017) 590-602.

[17] J.-A. Dalmon, G.A. Martin, Intermediates in CO and CO2 hydrogenation over

Ni catalysts, Journal of the Chemical Society, Faraday Transactions 1: Physical

Chemistry in Condensed Phases, 75 (1979) 1011-1015.

[18] G.D. Weatherbee, C.H. Bartholomew, Hydrogenation of CO2 on group VIII

metals: II. Kinetics and mechanism of CO2 hydrogenation on nickel, Journal of

Catalysis, 77 (1982) 460-472.

[19] S.i. Fujita, M. Nakamura, T. Doi, N. Takezawa, Mechanisms of methanation

of carbon dioxide and carbon monoxide over nickel/alumina catalysts, Applied

Catalysis A, General, 104 (1993) 87-100.

[20] S.I. Fujita, N. Takezawa, Difference in the selectivity of CO and CO2

methanation reactions, Chemical Engineering Journal, 68 (1997) 63-68.

[21] Q. Pan, J. Peng, T. Sun, S. Wang, S. Wang, Insight into the reaction route

of CO2 methanation: Promotion effect of medium basic sites, Catalysis

Communications, 45 (2014) 74-78.

[22] Z.A. Ibraeva, N.V. Nekrasov, B.S. Gudkov, V.I. Yakerson, Z.T.

Beisembaeva, E.Z. Golosman, S.L. Kiperman, Kinetics of methanation of carbon

dioxide on a nickel catalyst, Theoretical and Experimental Chemistry, 26 (1991)

584-588.

[23] F. Koschany, D. Schlereth, O. Hinrichsen, On the kinetics of the

methanation of carbon dioxide on coprecipitated NiAl(O)x, Applied Catalysis B:

Environmental, 181 (2016) 504-516.


143

[24] I. Kiendl, M. Klemm, A. Clemens, A. Herrman, Dilute gas methanation of

synthesis gas from biomass gasification, Fuel, 123 (2014) 211-217.

[25] J. Horák, J. Pasek, Design of industrial chemical reactors from laboratory

data, Heyden & Son Ltd., London, 1978.

[26] G.F. Froment, K.B. Bischoff, Chemical reactor analysis and design, 2nd ed

ed., John Wiley, New York, 1990.

[27] F. Ocampo, B. Louis, L. Kiwi-Minsker, A.-C. Roger, Effect of Ce/Zr

composition and noble metal promotion on nickel based CexZr1−xO2 catalysts for

carbon dioxide methanation, Applied Catalysis A: General, 392 (2011) 36-44.

[28] B.W. Wojciechowski, N.M. Rice, Experimental methods in kinetic studies,

Revised edition ed., Elsevier Science B.V., Amsterdam, 2003.

[29] F.H.M. Dekker, A. Bliek, F. Kapteijn, J.A. Moulijn, Analysis of mass and heat

transfer in transient experiments over heterogeneous catalysts, Chemical

Engineering Science, 50 (1995) 3573-3580.

[30] T. Van Herwijnen, H. Van Doesburg, W.A. De Jong, Kinetics of the

methanation of CO and CO2 on a nickel catalyst, Journal of Catalysis, 28 (1973)

391-402.

[31] M. Swickrath, M. Anderson, The Development of Models for Carbon Dioxide

Reduction Technologies for Spacecraft Air Revitalization, in: 42nd International

Conference on Environmental Systems, American Institute of Aeronautics and

Astronautics, 2012.

[32] S. Ergun, Fluid flow through packed columns, Chemical Engineering

Progress, 48 (1952) 89-94.

[33] B.E. Poling, J.M. Prausnitz, J.P. O'Connel, The properties of gases and

liquids, 5th ed ed., The McGraw-Hill Companies, New York, 2001.

145

Chapter 5. A sorptive reactor for CO2 capture and conversion to renewable methane

The contents of this chapter were adapted from: C.V. Miguel, M.A. Soria, A. Mendes,

L.M. Madeira, A sorptive reactor for CO2 capture and conversion to renewable methane.

Chemical Engineering Journal, 2017, 352, 590-602.

The aim of this work is to contribute for the scientific

advancement of carbon capture and utilization technologies (CCU),

while exploring the integration of intermittent renewable electricity

production and Substitute Natural Gas (SNG) production, the so-called

“Power-to-Gas” concept. In this regard, a sorptive reactor for carrying

in the same unit both CO2 capture and conversion to SNG using

renewable H2 was studied, in a perspective of process intensification.

The sorptive reactor containing a layered bed of a K-promoted

hydrotalcite to capture CO2 by sorption (step 1) and a nickel-based

catalyst for CO2 hydrogenation (step 2) is operated at 300-350 ºC and

low pressure (≤2.5 bar). Integration of CO2 capture and conversion in

the same unit leads to enhanced sorption capacities and desorption

kinetics promoted by the steam produced in situ during the reactive

regeneration stage (methanation reaction).

The sorptive reactor working under continuous operation mode

allows to: i) capture ca. 0.30 moles of CO2 per kg of sorbent and per

sorption cycle, at 350 ºC and = 0.2 bar; ii) completely convert the

captured CO2 into CH4; iii) reach a productivity of ca. 2.36

; iv) avoid CO formation at 300 ºC and 1.34 bar and

v) reach a CH4 purity of 35 % at 350 ºC after N2 purge.

Chapter 5

146

5.1 Introduction

Curbing CO2 concentration in the atmosphere requires considering

simultaneously a diversity of approaches, such as the development and adoption

of carbon capture and sequestration (CCS) or utilization (CCU) technologies,

and growing policy support towards the development and penetration of energy

from renewable sources across the power market [1-3]. Regarding the adoption

of end-pipe solutions for CO2 emissions mitigation, CCU has relevant

advantages compared to CCS because it avoids the (critical) cost of CO2

transportation to a storage site and CO2 is seen as a feedstock available at zero

cost from which valuable products can be obtained [4, 5].

Among the options for CO2 utilization, its conversion to methane can offer

a way to simultaneously avoid CO2 emissions and chemically store, in the form

of methane, the surplus energy from renewable sources which is produced

during off-peak periods [6]. This concept, also known as “Power-to-Gas”, relies

on the production of H2 making use of the excess electric power for the

electrolysis of water. Further conversion of hydrogen to methane through the

Sabatier reaction (Eq. (5.1)) is envisaged because it benefits from a well-

established transmission, distribution and storage infrastructure that was

developed for natural gas, a fossil fuel which consists primarily of methane in a

concentration up to 70-90 mol % [7, 8].


Consequently, the Power-to-Gas concept brings the possibility of

connecting the power grid to the natural gas grid, thereby facilitating the

integration between the power and energy sectors. Moreover, since methane is

a low-carbon hydrogen carrier and produces the lowest amount of CO2 upon

combustion compared to other hydrocarbons, it provides a secure and efficient

way of store and supply energy from renewable sources, while simultaneously

A sorptive reactor for CO2 capture and conversion to renewable methane

147

reducing the dependence on fossil fuels and supporting the transition towards a

low-carbon economy [3, 6, 7, 9].

Direct methanation of CO2 from flue gas has been addressed in the

literature in catalytic studies using Ni-based catalysts [10] and in Chapter 2,

which showed the necessity to previously separate CO2 from other co-existent

species in the flue gas, particularly O2 (which consumes H2) and SO2 (that

causes catalyst deactivation). Sorption-enhanced methanation processes

relying on the Le Chatelier principle provide also an interesting way of producing

high grade methane through the in situ capture of H2O by an adsorbent (e.g.

zeolite 4A) during the reaction, this way shifting the reversible Sabatier reaction

(Eq. (5.1)) towards the products side [11, 12]. However, it requires a unit for

previous capture of CO2 and its transport to the sorption-enhanced methanation

reactor.

In this perspective, there is a clear advantage on integrating CO2 capture

from flue gas and its conversion in the same unit. This concept was recently

addressed using a dual function material (DFM) that contains 10 % of

nanodispersed CaO (adsorbent) and 5 % of Ru metal (catalyst) dispersed on

different commercial Al2O3 carriers [13]. However, the reported sorption capacity

per sorption cycle (11.18 2CO DFMg kg , i.e. 0.025

2CO CaOmol kg ), and CH4

productivity 4

4

CH-3CH DFM

Ru

mol3.27 g kg , i.e. 7.66 10

kg h

obtained during 10

cycles (T = 320 ºC; 2COp 0.075 atm; tcycle = 80 min) were very low [13]. In this

work, a mixed bed made of a commercial K-promoted hydrotalcite as CO2

sorbent and a commercial Ni-based methanation catalyst were employed. This

sorbent was chosen since hydrotalcites are the most suitable CO2 sorbents at

intermediate temperatures (200 ºC - 400 ºC), showing good sorption stability

under cyclic operation, reasonable kinetics and easy regeneration by pressure

swing [14-20]. Besides, hydrotalcites are also cheap materials as compared to

other available options (e.g. CaO sorbents). A nickel catalyst was selected

Chapter 5

148

because it is suitable for operation at this temperature range; moreover, Ni-

based catalysts are the most widely used materials for CO2 methanation at the

industrial scale due to their good activity and lower price when compared to

noble-based catalysts (e.g. Ru or Rh) [21, 22]. The use of a sorptive reactor for

CO2 capture and methanation is studied assessing the performance of

continuous cyclic operation, controlling the CO2 sorption capacity, CO2

conversion, CH4 productivity and CH4 purity. Given the complexity related to

processes integration, the flue gas composition is simplified to CO2 and N2,

although a pure H2 stream is used for regenerating the sorbent, as expected to

occur in real applications. Overall, this work shows the advantages of using such

a reactor for CO2 capture and utilization.

5.2 Experimental

5.2.1 Experimental setup and hybrid unit

A stainless-steel column with 15 cm length and 2.08 cm internal diameter

was placed inside a tubular oven (model Split from Termolab, Fornos Eléctricos,

Lda.) equipped with a 3-zone PID temperature controller (model MR13 from

Shimaden). The type-K thermocouples used to measure and control the oven

temperature were placed in contact with the reactor wall; the temperature

difference between the reactor wall and the bed was lower than 2 ºC. An

illustration of the setup used in this work is given in Fig. 5.1.

CO2 (99.998 %, Air Liquide), H2 (99.9995 %, Air Liquide) and N2 (99.9995

%, Air Liquide) were fed to the system using mass flow controllers (model F201

from Bronkhorst-High Tech). The flow rate of the stream leaving the system was

measured with a mass flow meter (model F101 from Bronkhorst-High Tech) and

corrected based on stream composition.


149

Fig. 5.1 – Illustration of the experimental setup.

The pressure was measured by means of pressure transducers (model

PMP 4010 from Druck) located before and after the packed-bed unit. The steam

produced during the reaction that left the system was condensed in a home-

assembled Peltier cold-trap.

The CO2 outlet concentration was recorded along time using an infrared

analyser (Servomex, model 4210). The concentration of other species (such as

N2, CH4 and CO) produced during the reactive regeneration stage was

determined by collecting and analysing samples along time using a 16-loop

valve (VICI, model ST) coupled to a gas chromatograph (Dani, model 1000). The

chromatograph was equipped with a micro-TCD detector (VICI, model TCD-C-

220) and a capillary column (Supelco, carboxen 1010 plot).

The reactor column was packed in alternating layers of sorbent pellets

(23.89 g) and methanation catalyst particles (7.58 g), as depicted in Fig. 5.2.

Chapter 5

150

Fig. 5.2 – Photo of the unit inside the oven (left) and a sketch of the layered bed (right).

The column porosity was 0.30. A commercial hydrotalcite and a

commercial Ni catalyst were employed in this study for the following reasons: i)

they can be purchased in quantities that allow using a larger column compatible

with the available test bench; ii) available in millimetre range size, which

minimizes the bed pressure drop and; iii) better performance uniformity of these

materials. The sorbent is synthesized through the metal alcoholate route and its

properties are listed in Table 5.1, whilst the catalyst was supplied under a non-

disclosure agreement.

Table 5.1 – Properties of the hydrotalcite sorbent.

Property Value / Info Units

Commercial reference Pural MG30K - Mg/Al ratio 0.5 wt. % / wt. %

K2CO3 content 17 wt. % Shape Cylindrical -

Dimensions (d x L) 5 x 5 mm

Bulk density b,sorb 846 kg m-3

Solid density s,sorb 2840 kg m-3

Particle porosity p,sorb 0.70 -


151

5.2.2 Sorption-desorption/reaction experiments

The layered bed was activated in situ increasing the temperature up to 320

ºC and holding at this temperature for 1 h under a 10 % H2/N2 stream (total flow

rate: 100 mLn∙min-1) for catalyst reduction. Bed temperature was continuously

recorded using 4 type-K thermocouples inserted laterally and both radially

centred and axially equidistant (cf. Fig. 5.2). The temperature profile along the

column length was negligible (< 2 ºC) under inert atmosphere. Sorption

experiments were performed using a typical post-combustion (dry) stream

composition obtained from coal power plants: 15 % CO2 balanced in N2. The bed

was filled with N2 prior to each sorption breakthrough, the total flow rate was kept

at 100 mLn min-1 and the pressure inside the unit was 1.34 bar. Regeneration

started when the feed was switched using a 4-way valve placed near the column

inlet to a stream containing N2 or H2 (the later for reactive regeneration), keeping

constant the total volumetric flow rate. Both streams used during sorption and

regeneration steps were fed downwardly towards the bed. Sorption-desorption

cycles were performed at 350 ºC until a steady CO2 sorption capacity was

reached and sorption-reaction cycles were carried afterwards. Two series of

sorption-reaction cycles were carried out at 350 ºC and 300 ºC. The total

pressure inside the column during the sorption step was 1.34 bar, except in

cycles 16 and 17 where it was increased to 2.00 and 2.50 bar, respectively.

Table 5.2 lists the operation conditions used in the cyclic experiments.

Table 5.2 – Operation conditions used in the experimental campaign.

Cycle(s) T (ºC) P (bar) Sorption end Regeneration method

0-5 350 1.34 Bed saturation Normal (w/ N2) 6-10 350 1.34 Bed saturation Reactive (w/ H2) 11-15 300 1.34 Bed saturation Reactive (w/ H2) 16 300 2.00 Bed saturation Reactive (w/ H2) 17 300 2.50 Bed saturation Reactive (w/ H2) 18 300 1.34 Breakthrough time Reactive (w/ H2) 19-21 350 1.34 Breakthrough time Reactive (w/ H2) 22-24 300 1.34 Breakthrough time Reactive (w/ H2)

Chapter 5

152

The system was flushed with N2 at the end of the experimental campaign

to remove carbon oxides before the temperature was decreased; this avoids the

formation of extremely toxic nickel carbonyl at temperatures below 150 ºC [21,

23]. Afterwards, since reduced Ni is pyrophoric, the bed was passivated at 50

ºC feeding a stream with increasing amounts of O2 in N2; the process ended

when the temperature remained constant, i.e., when the catalyst bed was

completely oxidised, as suggested elsewhere [21, 24]. Finally, the unit was

brought to room temperature under N2.

Scanning Electron Microscopy (SEM) was performed on a FEI Quanta

400FEG ESEM instrument coupled with Energy Dispersive Spectroscopy (EDS)

(EDAX Genesis X4M) to investigate the structure and composition of the fresh

(i.e. as supplied) and used sorbent.

5.2.3 Process indicators: definitions and equations

Carbon dioxide sorption capacity,2COq , of the reactor was determined as

follows:

,f

2

2

2

2

outCOin

CO inCO0

COads

1 dSt

S

yF t

yq

m

(5.2)

where 2

inCOF is the CO2 molar flow rate at the column inlet, adsm is the sorbent

mass,2

inCOy and

2

outCOy are the CO2 fractions at the system inlet and outlet

streams, respectively; St stands for time during the sorption stage and

,S ft for

the duration of this step.

Carbon dioxide conversion was obtained from:


153

2 2

2

2

sorbed desorbed/unconvertedCO , CO ,

CO , sorbedCO ,

100k k

k

k

n nX

n

(5.3)

where 2

sorbedCOn (Eq. (5.4)) and

2

desorbed/unconvertedCOn (Eq. (5.5)) are the number of

CO2 moles that were sorbed and those which left the unit during the capture and

reactive regeneration stages of a given cycle k, respectively.

,

2

2 2

2

outCOsorbed

CO CO inCO0

1 dS ft

inS

yn F t

y

(5.4)

,

2 2

desorbed/unconverted out outCO CO

0

dR ft

Rn F y t (5.5)

outF is the total molar flow rate at the column outlet, which is calculated from the

volumetric flow rate; Rt stands for time during the regeneration stage and ,R ft for

the duration of this step. Time values excluded the delay related to the tubing

volume; intra-particular amount of CO2 at the end of the sorption runs was found

to be negligible as compared to that retained in the sorbent.

Carbon balance in a sorption-reaction cycle k (Eq. (5.6)) was calculated

as follows:

2 4

2

2 2

desorbed/unconverted produced producedCO , CO,CH ,

CO , 1sorbed sorbedCO , CO , 1

100

(1 ) 100

k kk

kk

k k

n n nC

Xn n

(5.6)

and

Chapter 5

154

,produced out out

0

dR ft

i Rin F y t (5.7)

where i stands for produced species CH4 and CO. The carbon balance equation

of a given cycle k (Eq. (5.6)) includes the unconverted CO2 quantity of the

previous cycle (k-1), since it remains available to be converted in cycle k. The

error of the carbon balance was in average less than 6 % under pseudo-steady

state conditions (i.e. after 5 cycles operating in the same conditions).

The process performance was evaluated considering methane

productivity (Eq. (5.8)) and purity (dry basis, Eq. (5.9)):

4

out outCH

0

cat

d

R,ft

R

S,f R,f

F y t

Prodm t t

(5.8)

4

out outCH

0

out outj

0

d

100

d

R,f

R,f

t

R

t

Rj

F y t

Pur

F y t

(5.9)

where catm and S,f R,ft t stand for the catalyst mass and the duration of a

sorption-reaction cycle, respectively, and j for all the species present in the

system (except H2O).


In Section 5.3.1 the sorptive reactor is studied considering full saturation

of the bed before starting the regeneration step – discontinuous operation mode.


155

However, for the process being implemented on a continuous basis without

emitting (environmental perspective) and wasting (economical perspective) CO2,

the sorption step should end before CO2 breaks through the column; these

experiments are addressed in Section 5.3.2.

5.3.1 Sorptive reactor under discontinuous operation

mode

5.3.1.1 CO2 sorption working capacity

After the activation procedure, the layered bed was subjected to six

sorption-desorption cycles; Fig. 5.3 shows the CO2 breakthroughs and sorption

capacities calculated using (Eq. (5.2)). The sorption capacity decreased 38.5 %

with cycling (from 0.52 mmol·g-1 to 0.32 mmol·g-1), and the steady working

capacity was mostly reached at the end of the first cycle (cf. Fig. 5.3).

Fig. 5.3 – CO2 sorption breakthroughs curves (cycles 0-5) at 350 ºC and 1.34 bar. Insertion shows the sorption capacity in each cycle.

The sorption capacity decrease with sorption-desorption cycling has

already been reported for hydrotalcite-type sorbents (e.g. [16, 25]). The work by

tS

/ s

0 2000 4000 6000 8000 10000

y o

ut

C

O2

0.00

0.05

0.10

0.15

Cycle 0

Cycle 1

Cycle 2

Cycle 3

Cycle 4

Cycle 5

cycle

-1 0 1 2 3 4 5 6

qC

O2

/ m

ol kg-1

0.0

0.1

0.2

0.3

0.4

0.5

0.6

Chapter 5

156

Du et al. [26] provides an interesting insight about the transformations taking

place on the hydrotalcite during CO2 sorption and desorption. Du et al. [26] found

that under dry conditions CO2 was reversibly sorbed in a K-promoted

hydrotalcite in at least two different active sites: those that favour the formation

of bidentate carbonate (more active sites with favourable kinetics and

thermodynamics, but less abundant) and others promoting the formation of

unidentate and bridged carbonate species, also with fast sorption kinetics.

However, it was also observed that unidentate and bridged carbonates start

being slowly converted into irreversibly sorbed bulk polydentate carbonate,

which explains the sorption capacity decrease with cycling. Based on this

mechanism we have proposed on a previous work a kinetic model assuming

two active sites for CO2 sorption with parallel contributions [25]. The kinetic

model provided an excellent fit to the uptake and downtake curves of CO2

sorption on a K-promoted hydrotalcite at 200-300 ºC, clearly supporting the

validity of the sorption mechanism.

5.3.1.2 Effect of steam produced during reactive regeneration

Reactive regeneration experiments were performed after the sorbent bed

achieved its working capacity. The first experiment under these conditions was

performed in cycle 6 (Table 5.2) in which the bed was fed with a pure H2 stream

and CO2 started to desorb by purge gas stripping. Subsequently, desorbed CO2

in the gas phase became available to react with H2 when passing through the

catalyst layers.

Besides the ability to convert CO2 to CH4, a noticeable feature of the

reactive regeneration was observed: the sorbent is regenerated faster in these

conditions than through desorption with N2. Carbon dioxide outlet concentration

decreases sharply under reactive regeneration and reaches 1 % after 554 s

(cycle 6), while it takes 1309 s under non-reactive regeneration conditions (cycle

5) (cf. Fig. 5.4a-b).


157

Fig. 5.4 – Total pressure and CO2 and CH4 outlet fraction histories during regeneration at 350 ºC for the initial 2000 s (cycles 5-7).

Chapter 5

158

It is also noticed that the total pressure drops abruptly ca. 0.2 bar under

reactive regeneration due to H2 consumption through the Sabatier reaction, until

a plateau is reached at ca. 1.14 bar (Fig. 5.4b-c). Afterwards, the pressure

increases when unreacted H2 starts leaving the column (not shown) and a slight

overshoot occurs when CH4 fraction reaches its maximum, ending in a final value

of 1.17 bar. The pressure could not reach its initial value of 1.34 bar (as in cycle

5) because the pressure drop caused by the same flow rate of H2 or N2 is

different (also observed in blank experiments).

Besides fast desorption, reactive regeneration in cycle 6 was more

effective, so that it boosted the CO2 sorption capacity in cycle 7 (to 0.58 mmol·g-

1) and the initial bed sorption capacity in cycle 0 (0.52 mmol·g-1) was recovered

throughout cycles 8 to 10 (Fig. 5.5). However, CO2 sorption in cycle 0 was faster

than in cycle 8 (Fig. 5.6b), which might be related to changes that took place on

the hydrotalcite structure due to the presence of steam formed in the Sabatier

reaction, as observed by scanning electron microscopy analysis (SEM)

performed on the fresh and used sorbent (Fig. 5.7).

Fig. 5.5 – Evolution of CO2 capture and conversion with cycling at 350 ºC and 300 ºC.

cycle

-1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

nso

rbe

d,

de

so

rbe

d /

mm

ol

0

5

10

15

20

25

XC

O2

/ %

50

60

70

80

90

100

nCO2

desorbed/unconverted (Eq. 5)

nCO2

sorbed (Eq. 4)

XCO2

(Eq. 3)

300 ºC 350 ºC


159

Fig. 5.6 – CO2 breakthrough curves for cycles 0 and 8 with “zoom in” at a) beginning of sorption (xx axis) and b) when reaches the saturation (yy axis).

SEM images of the aged hydrotalcite showed that the characteristic

potassium-rich needle-like zones of fresh sorbent, also reported in other studies

(e.g. [15, 25]), had disappeared (Fig. 5.7). Oliveira et al. [15] considered that the

needles might be mainly formed by K2CO3 and attributed this disappearance to

their dissolution by steam, despite that potassium was not lost and thus the same

sorption capacity was obtained here in cycles 0 and 8. Therefore, the lower

sorption kinetics of cycle 8 might be explained by the absence of these type of

sites which, based on in situ FTIR results reported by Du et al. [26] for a similar

hydrotalcite, could be responsible for CO2 sorption in the form of bidentate

carbonate.

The earlier CO2 breakthrough in cycle 8 (Fig. 5.6a) clearly supports the

evidences given by Du et al. [26] that CO2 then started to be sorbed as

unidentate and bridged carbonates, species that later are being slowly converted

into bulk polydentate carbonate, thus explaining the longer tail of the

breakthrough curve (Fig. 5.6b). Nevertheless, the earlier CO2 breakthrough in

cycle 8 compared to cycle 0 (cf. Fig. 5.6a) only represents a decrease of ca. 6.3

% of the dynamic capacity of the bed (i.e. the capacity obtained when sorption

is stopped before CO2 breakthrough - continuous operation mode, as discussed

in Section 5.3.2).

Col 8 vs Col 9

tS / s

0 500 1000 1500 2000

yC

O2

/ y

CO

2,0

0.0

0.2

0.4

0.6

0.8

1.0

tS / s

0 5000 10000 15000 20000 25000 30000

yC

O2

/ y

CO

2,0

0.85

0.90

0.95

1.00

cycle 0

cycle 8

a) b)

Chapter 5

160

Fig. 5.7 – SEM images of the fresh (first row) and used MG30K sorbent (second row) at different magnifications (10000x and 50000x, from left to right).

The enhanced desorption capacity under reactive regeneration is

highlighted through the comparison of 2COy curves of cycles 5 and 6 (Fig. 5.4a-

b); 2COy starts to decrease after 195 s in cycle 6, while under non-reactive

regeneration conditions (cycle 5) it begun after 87 s. The delay observed in cycle

6 is caused by water produced which has the ability to regenerate sorbent sites

containing irreversibly sorbed CO2 (i.e.not able to be regenerated under dry

conditions), as reported in several works dealing with sorption-enhanced

processes for H2 production with pre-combustion CO2 capture [16, 27]. Herein,

2 moles of H2O are obtained per mole of converted CO2 according to the


161

stoichiometry of the Sabatier reaction (Eq. (5.1)). Therefore, in the proposed

hybrid process for CO2 capture and conversion to CH4, steam is produced in

situ, thus not requiring additional costs for its industrial production, which was

reported to dominate the economy of sorption-enhanced processes [27].

Moreover, since some of the produced water molecules may be sorbed by the

hydrotalcite [28], the reaction equilibrium is shifted to the right side allowing to

produce more methane and leading to a desired methane-enrichment of the

outlet stream. The maximum methane concentration obtained at 350 ºC ranged

from 60 % (cycle 6) to 77-78% (cycles 9-10) and was reached after ca. 300 s; in

cycles 7 and 8 a slightly lower value was reached 74 % (cf. Fig. 5.4 and Fig.

C.0.1 in Appendix C).

During cycles 7-10 a roll-up effect was observed immediately before CO2

concentration starts to decrease steeply (Fig. 5.8a). In equilibrium-based

separations, the roll-up effect is observed in multicomponent mixtures when the

effluent composition of a species exceeds its inlet value, being this phenomenon

related to the displacement of the light components (weakly sorbed) by the

heavier ones (strongly sorbed) [29, 30]. Preliminary experiments allowed to

check that the sorbent does not retain CH4 (data not shown). Furthermore, the

work by Boon et al. [28] reported that this hydrotalcite has higher affinity towards

CO2 as compared to H2O. Therefore, CO2 roll-up should be based on a different

event.

Chapter 5

162

Fig. 5.8 – Effect of the cycling process at different temperatures (a-b) and pressures (c) on the roll-up effect.

tR / s

0 100 200 300 400 500 600

FC

O2

/ m

ol s

-1

0.0

2.0e-6

4.0e-6

6.0e-6

8.0e-6

1.0e-5

1.2e-5

cycle 6

cycle 7

cycle 8

cycle 9

cycle 10

tR / s

0 100 200 300 400 500 600

FC

O2

/ m

ol s

-1

0.0

2.0e-6

4.0e-6

6.0e-6

8.0e-6

1.0e-5

1.2e-5

cycle 11

cycle 12

cycle 13

cycle 14

cycle 15

350 ºC

300 ºC

tR / s

0 100 200 300 400 500 600

FC

O2

/ m

ol s

-1

0.0

2.0e-6

4.0e-6

6.0e-6

8.0e-6

1.0e-5

1.2e-5

cycle 15 (P = 1.34 bar)

cycle 16 (P = 2.00 bar)

cycle 17 (P = 2.50 bar)

300 ºC

a)

b)

c)


163

The analysis of the exit composition during reactive regeneration showed

that CO is being formed apart from CH4, which is expected to occur through the

parallel reverse-water gas shift reaction (RWGS) (Eq. (5.10)), as shown in

thermodynamic (e.g. [31, 32]) and kinetic studies for the methanation reaction

using a nickel catalyst (e.g. [33]).

-12 2 2 298 KCO + H CO + H O 41 kJ molH (5.10)

In Chapter 2 it was reported that CO production was slightly favoured at lower

pressures, higher temperatures and lower H2/CO2 ratios; CO production onsets,

however, for ratios as high as H2/CO2 = 4, the stoichiometric ratio of the Sabatier

reaction. In this regard, the observed roll-up effect is tentatively explained with

the production of CO2 through the water-gas shift reaction (WGS) (Eq. (5.11)),

-12 2 2 298 KCO + H O CO + H = - 41 kJ molH (5.11)

which can be catalysed by a nickel-based catalyst [34] and also by the sorbent

[16, 35]. The WGS becomes favoured along the reactor bed due to the

increasing amount of CO and H2O available. The CO2 roll-up was almost

unnoticed in cycle 6 (Fig. 5.4b and Fig. 5.8a) because less quantity of the H2O

produced was available for the WGS reaction (since it was used mainly to

recover the sorbent sorption capacity). This hypothesis is further supported

through the inspection of the temperature profiles of the regeneration step in

cycles 6 and 7-10 (Fig. 5.9a). In cycles 7-10 a small peak corresponding to a

temperature increase of ca. 1 ºC was observed in the last thermocouple placed

in the bed, whereas it was unnoticed in cycle 6, suggesting that another

(moderately exothermic) reaction, such as the WGS, also occurred.

Chapter 5

164

Fig. 5.9 – Normalized temperature histories measured in the last thermocouple (L=12 cm, z=4) during the first 200 s of CO2 regeneration at 350 ºC (a), 300 ºC (b) and different pressures (c).


165

Since CO is an undesired product, even at trace levels, the temperature

and pressure were varied to minimize its formation and their effects are

discussed in sections 5.3.1.3 and 5.3.1.4, respectively.

5.3.1.3 Effect of temperature

To assess the effect of the temperature on CO2 capture and conversion,

the process was tested at 350 ºC (cycles 6-10) and 300 ºC (cycles 11-15) – cf.

Table 5.2. In previous runs of CO2 sorption and regeneration with N2 (cycles 0-

5) it was observed that the shape of temperature histories profile had changed

after cycle 0 and remained similar until cycle 6 as shown in Fig. 5.10 (related

with the decrease of sorbent working capacity – Fig. 5.3).

However, the sorbent capacity was recovered in cycle 7 (Fig. 5.5) because

the bed undergone reactive regeneration with H2 for the first time in cycle 6 and

steam allowed to recover the initial bed capacity, as mentioned in the previous

section. The same profile shape was obtained in cycles performed at 300 ºC (i.e.

cycles 11-15 – cf. Fig. C.0.2 in Appendix C), although with different intensity

because both sorption and reaction kinetics are less favoured at 300 ºC. In these

cycles the temperature peak related to the WGS reaction (i.e. causing CO2 roll-

up) was also observed but in a smaller extent (Fig. 5.9b and Fig. 5.9c).

Fig. 5.5 shows the amount of sorbed and desorbed CO2 from cycles 0 to

15, besides the CO2 conversion obtained in cycles 6-15. The bed reached a

steady capacity in cycle 5 (regeneration with N2) and in the subsequent cycles

the quantity of desorbed CO2 (black bars) measured at the outlet of the bed

considerably decreases, which is related to CO2 consumption to produce CH4.

Therefore, the black bars in cycles 6-15 correspond to CO2 that was desorbed

and unconverted, besides, as mentioned previously, to the small quantity of CO2

that was produced from the WGS reaction, particularly in the later stage of the

reactor length.

Chapter 5

166

Fig. 5.10 – Normalized temperature histories in four bed positions during CO2 sorption and regeneration with N2 (cycles 0 to 5) or H2 (cycle 6); Tz,0 = 350 º C.

The sorption capacity increases after reactive regeneration experiments

(cycles 0-6 vs. cycles 7-15) and decreases with temperature decreasing (cycles

7-10 vs. cycles 11-15), as shown in Fig. 5.5. CO2 conversion slightly increases

in the first cycles at 300 ºC but then reaches a value close to the same obtained

in the end of the experimental campaign at 350 ºC (i.e. in cycle 10), ~ 91 %.

Tz -

Tz,0

/

º C

0

2

4

6

8

10

12

14

16

Tz -

Tz,0

/

º C

-1.4

-1.2

-1.0

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

tS / s

0 500 1000 1500 2000

Tz -

Tz,0

/

º C

0

2

4

6

8

10

12

14

16

tR / s

0 500 1000 1500 2000

Tz -

Tz,0

/

º C

0

5

10

15

20

25

30

Cycle 5 - Sorption Cycle 5 - Regeneration (N2)

Cycle 6 - Sorption Cycle 6 - Regeneration (H2)

Tz -

Tz,0

/

º C

0

2

4

6

8

10

12

14

16

Tz -

Tz,0

/

º C

-1.4

-1.2

-1.0

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

Cycle 0 - Sorption Cycle 0 - Regeneration (N2)


167

However, when analysing the effect of temperature on conversion it is

noteworthy to mention that a sorptive reactor is different from a conventional

reactor, where the temperature mainly influences the reaction equilibrium and

reaction kinetics. For a sorptive reactor, the temperature will also influence

sorption equilibrium and sorption kinetics, namely: i) the amount of the sorbate

previously retained in the bed, ii) the strength of the bonding between the sorbate

and the sorbent and iii) the rate in which the sorbate is transferred from the

sorbent to the gas phase and becomes available for reaction at the catalyst

active sites. In this regard, although the conversion is practically the same after

5 cycles at 350 ºC and 300 ºC, the quantity of methane produced is lower at 300

ºC (12.8 vs. 10.5 mmol per cycle), corresponding to a decrease of the maximum

methane concentration at the reactor’s outlet from 78 % to 56 % (cf. Fig. C.0.1

in Appendix C – cycles 10 and 15, respectively). Nevertheless, decreasing the

temperature also caused a remarkable drop of produced CO, as shown in Fig.

5.11.

Fig. 5.11 – Maximum CO fraction obtained during cycles at 350 ºC (6-10) and 300 ºC at 1.34 bar (11-15) and 2.0-2.5 bar (16-17).

cycle

5 6 7 8 9 10 11 12 13 14 15 16 17 18

yC

O,

má

x. x 1

00

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

nC

Opro

duced

/ m

mol

0.00

0.02

0.04

0.06

0.08

0.10 yCO,máx @ 350 ºC

yCO,máx @ 300 ºC

nCO produced

Chapter 5

168

The number of moles of CO produced decreased from 0.048 to 0.009

mmol in cycles 10 and 15, respectively. The quantity of captured CO2 in cycles

10 (350 ºC) and 11 (300 ºC) was almost the same (11.9 vs. 11.4 mmol,

respectively – Fig. 5.5), but CO formation decreased considerably. Again, this

suggests that CO is produced through the RWGS reaction, which due to its

endothermic nature is not favoured at lower temperatures; moreover, the RWGS

reaction kinetics also decreases in such conditions. The noticeable CO decline

observed from cycle 6 to 7 can be attributed to the WGS reaction, which

consumed CO to produce CO2, as discussed in the previous section.

5.3.1.4 Effect of pressure

During the sorption step, both the CO2 fraction in the feed and total

pressure were kept constant along the experimental campaign (2CO

iny =0.15 and

P=1.34 bar), except in cycles 16 and 17 where the total pressure was increased

up to ca. 2.00 and 2.50 bar, respectively (cf. Table 5.2). The obtained values for

the bed sorption capacity (2COq ) in the sorption stage were compared to those

predicted by the dual-site Langmuir sorption equilibrium equation,2

*COq ,

proposed by Oliveira et al. [15] (Eq. (5.12)).

2 2

2 2 2

2 2

0 0

* max, P max, CCO CO CO

0 01 1

sorb sorb

sorb sorb

H E

P CRT RTCO CO

H E

P CRT RTCO CO

k e p k e p

q q q

k e p k e p

(5.12)

The used model parameters can be found in Table 5.3.


169

Table 5.3 – List of the dual-site Langmuir model parameters taken from Oliveira et al. [16].

Parameter Units Value

2

max, PCO

q mol·kg-1 0.423

bar-1 9.07 x 10-3

kJ·mol-1 -40.0

2

max, CCO

q mol·kg-1 0.351

bar-1 1.01 x 1012

kJ·mol-1 130.8

This model assumes that sorption takes place through a combination of

(exothermic) physical sorption (superscript P) and (endothermic) chemical

reaction (superscript C), where 2

maxCOq stands for the maximum capacity (mol·kg-

1), 0k is the pre-exponential factor (bar-1),

sorbH is the sorption enthalpy

(kJ·mol-1) and sorbE the heat of reaction (kJ·mol-1). R and T stand, as usual, for

the ideal gas constant and absolute temperature, respectively.

Fig. 5.12 shows the adherence of the experimental data to Eq. (5.12)

without fitting any parameter. For cycle 0 a relative deviation between

experimental and predicted (continuous line) sorption capacity was 10 %. The

figure shows that in the first 5 cycles the experimental sorption capacity drops to

the dotted line corresponding to the physical adsorption contribution. This can

be attributed to the absence of steam in the feed stream, since in the work of

Oliveira et al. [15] the isotherms were determined using wet streams, which is

known to promote CO2 sorption [14-16]. In fact, this effect is evidenced in cycle

7 and succeeding cycles, which were performed under reactive regeneration

conditions, where steam is produced during methanation. In cycle 7 the sorption

capacity reached the predicted value of the dual-site Langmuir model

(continuous line), but dropped slightly in cycles 8-10. This difference might be

related to the absence of the WGS reaction in cycle 6 where all the steam

produced was available for sorbent regeneration, whereas in the regeneration

0Pk

sorbH

0Ck

sorbE

Chapter 5

170

stage of cycle 7 a part of the H2O produced is consumed (thus the CO2 roll-up)

and negatively influences the capacity of cycle 8; the same occurs in the

subsequent cycles.

Fig. 5.12 – Adherence of the experimental data to the dual-site Langmuir sorption

equilibrium isotherm. Model parameters taken from Oliveira et al. [15].

Cycle 11 was the first being performed at 300 ºC and here the sorption

capacity slightly exceeded the value predicted by the model (deviation of 7 %

only). In the subsequent cycles the values are lower and the decrease between

cycles is less noticeable. Since regeneration in cycle 10 was performed at 350

ºC, the sorbent had more active sites available during sorption in cycle 11 than

in the subsequent cycles. Fig. 5.12 shows also that the relative relevance of CO2

sorption via chemical reaction (dashed line) to the total capacity (continuous line)

is lower at 300 ºC than at 350 ºC. On the other side, the weight of physical

sorption (dotted line) on the total sorption capacity predicted by the model

increases from 58 % at 350 ºC to 84 % at 300 ºC (cycles 6-10 vs. cycles 11-15).

Therefore, it is expected that the enhancement of CO2 desorption using heat

released during reactive regeneration is more relevant at 300 ºC, where physical

sorption is the dominant process, whereas at 350 ºC both physical (exothermic)


171

and chemical (endothermic) contributions have a similar weight on the total

sorption capacity (58 % vs. 42 %, respectively).

The sorption capacity increased with total pressure from 0.38 mmol·g-1

(average value at 1.34 bar total pressure – cycles 12-15) to 0.43 mmol·g-1 (at

2.5 bar total pressure – cycle 17). CO2 conversion decreased slightly, from 92.5

% to 90.8 %, but the quantity of CH4 produced remained practically constant.

Therefore, the most important consequence of increasing the pressure was

decreasing CO production, as shown in Fig. 5.11. The quantity of CO formed

declined 34 % when the pressure increased from 1.34 bar to 2.50 bar.

5.3.2 Sorptive reactor under continuous operation mode

As mentioned before, in Section 5.3.1 the sorptive reactor was studied

considering full saturation of the bed before starting the regeneration step.

However, to implement the process on a continuous basis without emitting and

wasting CO2, the sorption step should end before the CO2 breakthrough time.

For this end, at least two sorptive reactors should be operated in parallel, i.e.,

while one is undergoing the capture stage, the other should be regenerated and

producing CH4. This means that under continuous operation mode the bed

sorption capacity (also-called dynamic capacity) is lower than its maximum value

(full bed saturation), which obviously has implications on the global performance

of the process. Therefore, the process performance was assessed through CH4

productivity and purity (please refer to Eqs. (5.8) and (5.9)).

Fig. 5.13 shows the outlet concentration profile of all the species (dry

basis) during the reactive regeneration stage at 350 ºC (cycle 21) and 300 ºC

(cycle 24). CO was not detected at 300 ºC (see insertion of Fig. 5.13b) and CO2

conversion was almost complete (ca. 99 %) at both temperatures since only a

small quantity of unconverted CO2 was noticed at the outlet.

Chapter 5

172

Fig. 5.13 – Outlet concentration profiles of all species during the regeneration step (dry basis) at a) 350 ºC and b) 300 ºC. Insertion shows CO2 and CO profile during the initial 500 s.

Fig. 5.14 shows the evolution of both process indicators, CH4 purity and

productivity, along the reactive regeneration time at 350 ºC and 300 ºC. It can

be observed that temperature improved CH4 purity, while the productivity was

nearly the same at both temperatures (maximum value of ca. 2.32-2.36 mol·kg-

1·h-1) because the decrease of the produced methane at 300 ºC was

counterbalanced by the decrease of the cycle time. Fig. 5.14 also shows that

productivity is still increasing when purity reaches its maximum, which in turn

takes place after H2 breaks out of the unit (please see Fig. 5.13 and Fig. 5.14).

According to the German Technical and Scientific Association for Gas and

Water, the maximum allowable concentration of H2 in the gas mixture is 5 vol. %

[6]. In this regard, if the regeneration is stopped when H2 purity reaches this

value (ca. 280 s), at 350 ºC the purity is close to the maximum value (26 % vs.

29 %), while productivity is far below (1.31 mol·kg-1·h-1 vs. 2.32 mol·kg-1·h-1).

tR / s

0 200 400 600 800 1000 1200 1400 1600

yj

0.0

0.2

0.4

0.6

0.8

1.0

1.2

tR

/ s

0 100 200 300 400 500

yj

0.000

0.001

0.002

0.003

0.004

0.005

tR / s

0 200 400 600 800 1000 1200 1400 1600

N2

CH4

H2

CO2

CO

tR

/ s

0 100 200 300 400 500

yj

0.000

0.001

0.002

0.003

0.004

0.005

Cycle 21 - 350 ºC Cycle 24 - 300 ºCa) b)


173

Fig. 5.14 – CH4 purity (circles) and productivity (squares) evolution during regeneration time at 350 ºC (full symbols) - cycle 21 - and 300 ºC (open symbols) – cycle 24.

Nitrogen is the species that breaks first during the regeneration step (as

shown in Fig. 5.13) and it is possible to purge ca. 30-35 % of the total amount

before methane breaks. Nitrogen purge increases methane purity, as shown in

Fig. 5.15. For instance, at 350 ºC, when the H2 purity reaches 5 %, the CH4 purity

increases from 26 % to 35 % (Fig. 5.15a).

Fig. 5.15 – Effect of N2 purge on CH4 purity evolution during regeneration time at a) 350 ºC and b) 300 ºC.

tR / s

0 200 400 600 800 1000 1200 1400 1600

Pu

r /

%

0

10

20

30

40

50

0.0

0.5

1.0

1.5

2.0

2.5

Pro

d /

mo

l kg

-1 h

-1

tR / s

0 200 400 600 800 1000 1200 1400 1600

Pu

r /

%

0

10

20

30

40

N2 purge (YES)

N2 purge (NO)

tR / s

0 200 400 600 800 1000 1200 1400 1600

Pu

r / %

0

10

20

30

40

N2 purge (YES)

N2 purge (NO)

Cycle 21 - 350 ºC Cycle 24 - 300 ºCa) b)

Chapter 5

174

5.3.3 Reactor design considerations

The results obtained when the sorptive reactor was operated under

discontinuous (Section 5.3.1) or continuous mode (Section 5.3.2) provided

insight about the operations conditions, particularly temperature and pressure,

which could be elected for further optimization of the reactor.

Generally, CO formation through parallel RWGS and low CH4 purity level

(affected by all species leaving the reactor and not only by CO) are the main

drawbacks that need to be tackled from the reactor design point of view.

Controlling the temperature and pressure proved their suitability to surpass

these limitations, but not entirely.

CH4 integral selectivity (i.e. nCH4/nCO ratio) obtained in discontinuous

operation mode improves by a factor of 4.1 when the operation temperature was

decreased from 350 ºC to 300 ºC. For this reason, the temperature was kept at

300 ºC in subsequent cycles while the total pressure was varied (cycles 16 and

17). Increasing the total pressure from 1.34 bar to 2.50 bar enhanced the CO2

sorption capacity (by 13 %) and CH4 integral selectivity (by 50 %). However, in

practical application, flue gas is delivered at atmospheric pressure and, if

possible, its compression should be avoided for economic reasons. In this

regard, the pressure was kept at 1.34 bar when the reactor was operated in

continuous mode (cycles 19-24) and CO was not detected in cycles performed

at 300 ºC. During reactive regeneration of these cycles, the amount of H2

available in the gas phase per mole of desorbed CO2 presumably increased

when compared to discontinuous operation, favouring the Sabatier reaction in

detriment of the RWGS (H2/CO2=4 vs. H2/CO2=1 stoichiometric ratios,

respectively). Nevertheless, the maximum CH4 purity level reached at the

reactor outlet is still unsatisfactory (30 %). It is anticipated that further

optimization of CH4 purity and productivity can be achieved by tuning the bed

properties (e.g. catalyst/sorbent ratio, bed layout, H2 feed mode, etc.). However,


175

special care regarding the influence of other species present in the flue gas on

reactor performance should be also systematically evaluated in future work.

5.4 Conclusions

A sorptive reactor was successfully piloted at lab-scale to capture and

convert CO2 into CH4. The process was studied in a temperature window

between 300-350 ºC based on the typical range reported for both CO2 sorption

and methanation using K-promoted hydrotalcites and Ni-based catalysts,

respectively. The process was studied in a low-pressure range (≤ 2.5 bar)

because post-combustion streams are generally at the atmospheric pressure.

This type of reactor allows that both processes (i.e. sorption and reaction)

take place in the same unit and has the economic advantage that it does not

require CO2 transportation from a capture unit to a (methanation) reactor nor to

a storage site. Process intensification allowed enhancing CO2 capture due to

steam formed in situ during the reactive regeneration step of the previous cycle.

The dual-site Langmuir model could well describe the sorption equilibrium.

Reactive regeneration resulted in a synergetic effect taking place along the

reactor length: CH4 formation (desired product) is favoured by the presence of

H2O, which in turn participates on CO2 desorption, making it faster and removing

more CO2 than under dry conditions. Consequently, the sorption capacity in the

subsequent cycle is enhanced until a pseudo steady-state is reached.

Furthermore, since steam may be sorbed by the hydrotalcite, the reaction

equilibrium can be shifted to the right side to produce more CH4 (Eq. (5.1)). CO2

conversion was high in batch operation (ca. 92 %) and almost complete (99 %)

when the reactor was operated under continuous mode.

Although methane concentration in the outlet stream can be as high as 78

% (dry basis), the maximum CH4 purity obtained at 350 ºC was 37 % if N2 is

purged at the beginning of the regeneration step.

Chapter 5

176

The temperature effect on productivity was found to be negligible in the

range studied (maximum value of 2.32 vs. 2.36 mol·kg-1·h-1 at 350 ºC and

300 ºC, respectively). The undesired formation of CO through the RWGS

reaction was observed and could be minimized by decreasing the temperature

and/or increasing the operating total pressure.

The materials employed in this work proved their compatibility and activity

for this application, besides showing good stability under cyclic operation (which

also allowed to assess repeatability). In this regard, the envisaged process

scale-up would benefit by the fact that both materials are already at a

commercial level. Still, the process conditions and reactor design (e.g. bed

layout) should be further studied to improve methane purity up to the level

required for injection in the natural gas grid.


177

5.5 Nomenclature

Parameter Description Units

C Carbon molar balance %

sorbE Heat of chemical reaction of CO2 with the sorbent

kJ·mol-1

2COF CO2 molar flow-rate mol·s-1

F Total molar flow-rate mol·s-1

sorbH Heat of physical sorption kJ·mol-1

0Ck

Arrhenius pre-exponential factor of CO2 sorption equilibrium constant through chemical reaction

bar-1

0Pk

Arrhenius pre-exponential factor of CO2 sorption equilibrium constant through physical adsorption

bar-1

L Sorptive reactor length cm

adsm Sorbent mass kg

catm Catalyst mass kg

2

sorbedCOn Moles of sorbed CO2 mol

2

desorbed/unconvertedCOn Moles of desorbed and unconverted CO2 mol

producedin

Moles of carbon-containing produced species “i”

mol

2COp CO2 partial pressure bar

Pout Total pressure at the outlet of the sorptive reactor

bar

Prod CH4 productivity mol·kg-1·h-1

Pur CH4 purity %

2COq CO2 sorption capacity (experimental) mol·kg-1

2

*COq

CO2 sorption capacity (dual-site Langmuir model)

mol·kg-1

2

max, CCO

q Maximum CO2 sorption capacity for endothermic chemical reaction

mol·kg-1

2

max, PCO

q Maximum CO2 sorption capacity for exothermic physical adsorption

mol·kg-1

R Ideal gas constant (8.314 x 10-3) kJ·mol-1·K-1

T Temperature in the sorptive reactor K

Rt Time during regeneration step s

,R ft Duration of regeneration step s

Chapter 5

178

St Time during sorption step s

,S ft Duration of sorption step s

2COX CO2 conversion %

2COy CO2 molar fraction

Subscripts & Superscripts

i Carbon-containing species formed: CH4 and CO

in Inlet of the sorptive reactor

j Species N2, H2, CO2, CH4 and CO

k Sorption-reaction cycle

out Outlet of the sorptive reactor

R Regeneration step

S Sorption step

z Axial position of the sorptive reactor (1-4)


179

5.6 References



ChemSusChem, 4 (2011) 1194-1215.

[2] M. Peters, B. Köhler, W. Kuckshinrichs, W. Leitner, P. Markewitz, T.E. Müller,

Chemical technologies for exploiting and recycling carbon dioxide into the value

chain, ChemSusChem, 4 (2011) 1216-1240.

[3] S. Kitasei, Powering the low-carbon economy: the once and future roles of

renewable energy and natural gas, Worldwatch Institute, Washington D.C.,

2010.

[4] E.A. Quadrelli, G. Centi, Green Carbon Dioxide, ChemSusChem, 4 (2011)

1179-1181.

[5] M. Aresta, A. Dibenedetto, A. Angelini, Catalysis for the valorization of

exhaust carbon: from CO2 to chemicals, materials, and fuels. Technological use

of CO2, Chemical Reviews, 114 (2014) 1709-1742.

[6] S. Schiebahn, T. Grube, M. Robinius, V. Tietze, B. Kumar, D. Stolten, Power

to gas: technological overview, systems analysis and economic assessment for

a case study in Germany, International Journal of Hydrogen Energy, 40 (2015)

4285-4294.

[7] S. Walspurger, W.G. Haije, B. Louis, CO2 reduction to substitute natural gas:

toward a global low carbon energy system, Israel Journal of Chemistry, 54

(2014) 1432-1442.

[8] J.G. Speight, Natural gas: A Basic Handbook, Gulf Publishing Company,

Houston, 2007.

[9] N. Mellquist, M. Fulton, Natural Gas and Renewables: A Secure Low Carbon

Future Energy Plan for the United States Deutsche Bank Group, 2010.

[10] K. Müller, M. Fleige, F. Rachow, D. Schmeißer, Sabatier based CO2-

methanation of flue gas emitted by conventional power plants, Energy Procedia,

40 (2013) 240-248.

Chapter 5

180

[11] A. Borgschulte, N. Gallandat, B. Probst, R. Suter, E. Callini, D. Ferri, Y.

Arroyo, R. Erni, H. Geerlings, A. Zuttel, Sorption enhanced CO2 methanation,

Physical Chemistry Chemical Physics, 15 (2013) 9620-9625.

[12] S. Walspurger, G.D. Elzinga, J.W. Dijkstra, M. Sarić, W.G. Haije, Sorption

enhanced methanation for substitute natural gas production: experimental

results and thermodynamic considerations, Chemical Engineering Journal, 242

(2014) 379-386.

[13] Q. Zheng, R. Farrauto, A. Chau Nguyen, Adsorption and methanation of

flue gas CO2 with dual functional catalytic materials: a parametric study,

Industrial & Engineering Chemistry Research, 55 (2016) 6768-6776.

[14] M.K. Ram Reddy, Z.P. Xu, J.C. Diniz da Costa, Influence of water on high-

temperature CO2 capture using layered double hydroxide derivatives, Industrial

& Engineering Chemistry Research, 47 (2008) 2630-2635.




[16] M.A. Soria, S. Tosti, A. Mendes, L.M. Madeira, Enhancing the low

temperature water-gas shift reaction through a hybrid sorption-enhanced

membrane reactor for high-purity hydrogen production, Fuel, 159 (2015) 854-

863.



Science, 4 (2011) 42-55.

[18] S. Choi, J.H. Drese, C.W. Jones, Adsorbent materials for carbon dioxide

capture from large anthropogenic point sources, ChemSusChem, 2 (2009) 796-

854.

[19] M.K.R. Reddy, Z.P. Xu, G.Q. Lu, J.C. Diniz da Costa, Effect of SOx

adsorption on layered double hydroxides for CO2 capture, Industrial &



181

[20] M.G. Beaver, High temperature CO2 chemisorbents: applications,

characterization, and study of the chemical nature of chemisorbent surfaces, in,

Lehigh University, 2010.

[21] E.Z. Golosman, V.N. Efremov, Industrial catalysts for the hydrogenation of

carbon oxides, Catalysis in Industry, 4 (2012) 267-283.

[22] W. Wang, S. Wang, X. Ma, J. Gong, Recent advances in catalytic

hydrogenation of carbon dioxide, Chemical Society Reviews, 40 (2011) 3703-

3727.

[23] H.F. Rase, Handbook of Commercial Catalysts: Heterogeneous Catalysts,

CRC Press, Boca Raton, , 2000.

[24] A. Gil, A. Diaz, M. Montes, Passivation and reactivation of nickel catalysts,

Journal of the Chemical Society, Faraday Transactions, 87 (1991) 791-795.

[25] C.V. Miguel, R. Trujillano, V. Rives, M.A. Vicente, A.F.P. Ferreira, A.E.

Rodrigues, A. Mendes, L.M. Madeira, High temperature CO2 sorption with

gallium-substituted and promoted hydrotalcites, Separation and Purification

Technology, 127 (2014) 202-211.




[27] J. Boon, P.D. Cobden, H.A.J. van Dijk, M. van Sint Annaland, High-

temperature pressure swing adsorption cycle design for sorption-enhanced

water-gas shift, Chemical Engineering Science, 122 (2015) 219-231.

[28] J. Boon, P.D. Cobden, H.A.J. van Dijk, C. Hoogland, E.R. van Selow, M.

van Sint Annaland, Isotherm model for high-temperature, high-pressure

adsorption of CO2 and H2O on K-promoted hydrotalcite, Chemical Engineering

Journal, 248 (2014) 406-414.

[29] G. Li, P. Xiao, D. Xu, P.A. Webley, Dual mode roll-up effect in

multicomponent non-isothermal adsorption processes with multilayered bed

packing, Chemical Engineering Science, 66 (2011) 1825-1834.

Chapter 5

182

[30] A.F.P. Ferreira, A.M. Ribeiro, S. Kulaç, A.E. Rodrigues, Methane

purification by adsorptive processes on MIL-53(Al), Chemical Engineering

Science, 124 (2015) 79-95.

[31] C.V. Miguel, M.A. Soria, A. Mendes, L.M. Madeira, Direct CO2

hydrogenation to methane or methanol from post-combustion exhaust streams

- A thermodynamic study, Journal of Natural Gas Science and Engineering, 22

(2015) 1-8.

[32] J. Gao, Y. Wang, Y. Ping, D. Hu, G. Xu, F. Gu, F. Su, A thermodynamic

analysis of methanation reactions of carbon oxides for the production of

synthetic natural gas, RSC Advances, 2 (2012) 2358-2368.

[33] J. Xu, G.F. Froment, Methane steam reforming, methanation and water-gas

shift: I. Intrinsic kinetics, AIChE Journal, 35 (1989) 88-96.

[34] K. Liu, C. Song, V. Subramani, Hydrogen and syngas production and

purification technologies, Wiley; AIChE, New York, 2010.

[35] M. Maroño, Y. Torreiro, D. Cillero, J.M. Sánchez, Experimental studies of

CO2 capture by a hybrid catalyst/adsorbent system applicable to IGCC

processes, Applied Thermal Engineering, 74 (2015) 28-35.

183

Chapter 6. Effect of CO and CO2 on H2 permeation through finger-like Pd-Ag membranes

The contents of this chapter were adapted from: C.V. Miguel, A. Mendes, S. Tosti, L.M.

Madeira, Effect of CO and CO2 on H2 permeation through finger-like Pd-Ag membranes.

International Journal of Hydrogen Energy, 2012, 37(17), 12680-12687.

The permeance of a 50 µm thick PdAg (25 wt.% of silver) tubular

membrane towards hydrogen was evaluated for hydrogen feed streams or

binary mixtures of industrial relevance, namely of H2/CO or H2/CO2.

It was proposed a rearrangement of the Sieverts-Langmuir (SL)

equation to account for the inhibition effect on the membrane permeance

towards H2 due to the presence of CO or CO2. Such modification is based

on the average partial pressure of CO or CO2 and the logarithm-mean

driving-force to account for concentration gradients along the axial length

of the tubular membrane. Comparison with the original SL equation

evidenced relative deviations < 10.7 %. The experimental data showed

good agreement with the SL model.

It was found that CO has a much stronger inhibition effect on H2

permeation than CO2. Considering the values obtained for the adsorption

enthalpy of each gas, it was concluded that both CO and CO2 physically

adsorb on the membrane surface.

Chapter 6

184

6.1 Introduction

Most of the hydrogen production worldwide is based on the conversion of

fossil fuels, being steam reforming of methane (SRM) (Eq. (6.1)) in combination

with water-gas shift (WGS) (Eq. (6.2)) reactions the most commercialized

process [1].

-14 2 2 298 KCH + H O CO + 3H Δ = 205.9 kJ molH (6.1)

-12 2 2 298 KCO + H O CO + H Δ = - 41.1 kJ molH (6.2)

Recently, hydrogen production through water electrolysis (Eq. (6.3)) has

gained a renewed interest as an important building block in Power-to-Gas (PtG)

applications (as addressed in Chapter 2 and Chapter 5).

-12 2 2 298 KH O H +1 2O Δ =285.8 kJ molH (6.3)

In PtG, exceeding renewable power is chemically stored as hydrogen, or

another energy carrier such as methane or methanol, which are more easily

stored and transported. The hydrogen obtained from water electrolysis is further

used in hydrogenation reactions of a carbon source, such as CO2, to produce

methane (Eq. (6.4)) or methanol (Eq. (6.5)).


-12 2 3 2 298 KCO + 3H CH OH + H O Δ = - 49.4 kJ molH (6.5)

The importance of keeping an H2/CO2 stoichiometric ratio of 4 to reach

high CO2 conversion and CH4 selectivity was observed in the methane pathway

Effect of CO and CO2 on H2 permeation through finger-like Pd-Ag membranes

185

(Eq. (6.4)), from the thermodynamic point of view, in Chapter 2. In Chapter 5,

when the sorptive reactor was presented and tested, it was observed

experimentally the formation of carbon monoxide through the reverse water-gas

shift (inverse reaction of Eq. (6.2)). One possible explanation may be the

inexistence of an ideal H2/CO2 ratio of 4 along the layered bed length. It was

suggested that the hydrogen feed mode could be an option to minimize this

effect, for instance, using an H2 permselective material for the reactor wall,

allowing to feed the sorptive reactor along its length with the desired H2 flux so

that the ideal ratio is verified in every axial position of the reactor. The option for

a membrane feeding system would benefit of high pressure hydrogen delivery

from the electrolyser, which commonly can reach up to 30 bar.

The hydrogen-palladium system has been extensively studied along the

years (see for instance [2, 3]), and it is commonly accepted that Pd-based

materials are the most suitable for H2 separation/delivery with membranes,

despite its price. Palladium membranes are known to exhibit high H2 selectivity

and permeability and when alloyed with other metals (e.g. Cu, Ag, etc.) can lead

to materials less susceptible to hydrogen embrittlement and poisoning [4].

Additionally, long term tests, namely thermal and hydrogenation cycles, also

evidenced that self-supported dense and thin wall PdAg23 wt.% tubular

membranes with a “finger-like” configuration have high durability and reliability

(complete selectivity remained constant for at least one year) [5]. The use of this

“finger-like” configuration allows the free elongation and contraction of the

membrane, thus avoiding any mechanical stress that may damage the thin wall

tube. Besides, this kind of membranes allows obtaining ultrapure hydrogen due

to their nearly infinite permselectivity towards H2 [5, 6]. This is particularly

important when using membranes for selective removal of hydrogen, a product,

in thermodynamic equilibrium-limited reactions such as SRM (Eq. (6.1)) and

WGS (Eq. (6.2)), enabling to overcome the equilibrium conversion, which is the

limit of a correspondent conventional fixed bed reactor operated at the same

Chapter 6

186

conditions of the membrane reactor. A pure/ultrapure hydrogen stream can be

obtained, exempting further separation processes [7, 8].

However, for the implementation of these membranes for the above-

mentioned applications (H2 feed or removal), it is crucial to measure permeability

under different conditions, partial pressures, temperatures, besides assessing

the effect on the permeance caused by species such as CO or CO2 which are

present in reactions given by Eqs. (6.1)-(6.2) and Eqs. (6.4)-(6.5). This topic has

been reported by Gallucci et al. [9], whom measured the permeance of a 60 µm

thick finger like membrane towards H2 in binary mixtures of H2 and CO or CO2.

The authors observed that CO2 presence has no significant inhibition effect (at

523 K), whereas CO presence greatly reduced the H2 permeating flux. Other

authors found that inhibition by methane and steam was negligible as compared

to CO and CO2 [10, 11].

In the present chapter, the experimental tests and the development of an

accurate model that describe the effect of CO and CO2 on the hydrogen

permeability is described. The model has been validated through the

experimental results. In this work a rearranged Sieverts-Langmuir equation is

proposed for the first time to account for the change of the permeability driving

force along the membrane length caused by H2 removal in binary mixtures

containing CO or CO2.

6.2 Theoretical background

Generally, it is considered that Pd-based (defect-free) membranes

permeability to hydrogen follows a solution-diffusion mechanism [3, 12] and the

permeating hydrogen flux, 2HJ , can be described by [12, 13]:

2

2 2 2

H n nH H , feed H , permeate

LJ p p

(6.6)


187

where 2HL is the palladium alloy permeability to hydrogen, is the membrane

thickness, 2

nH , feedp and

2

nH , permeatep are the hydrogen partial pressures in the feed

and permeate sides, respectively, and n is the pressure exponent that varies

between 0.5 and 1.0. If diffusion of atomic hydrogen through the metal lattice of

the membrane is the limiting step, the flux can be expressed by the Sieverts Law

[2]:

2

2 2 2

H 1 2 1 2SievH H , feed H , permeate

LJ p p

(6.7)

However, the presence of surface adsorbed chemical species, such as CO

or CO2 that are present in the water gas shift or reforming reactors, may lead to

deviations to the Sieverts law. In this regard, Barbieri et al. [14] proposed a new

equation where the Sieverts law and the Langmuir isotherm were coupled to

describe the decrease on H2 permeation in the presence of CO as follows:

2 2

SL SievCO COH H

CO CO

11

K pJ J

K p

(6.8)

Where 2

SLHJ is the permeated hydrogen molar flux in the presence of CO,

2

SievHJ

is the hydrogen flux assessed accordingly to the Sieverts law (Eq. (6.7)),COp is

the carbon monoxide partial pressure, COK is the adsorption equilibrium

constant, and CO CO

CO CO1

K pCF

K p

is a dimensionless correction factor due to the

adsorption of CO on the membrane surface that depends on CO COCO

CO CO1

K p

K p

,

which stands for the fraction of the membrane surface covered by CO. Finally,

is a parameter that depends only on the temperature and accounts for

Chapter 6

188

additional effects of the adsorbed gas, for instance in the dissociation of H2

molecules. Before, Eq. (6.8) has been reported with reference to CO [14]: similar

considerations and a similar expression is herein proposed for the first time and

written for CO2.

However, since hydrogen is being removed, the driving-force for

permeability changes along the membrane length and consequently so the CO

(or CO2) partial pressure. In this regard, the SL equation (Eq. (6.8)) should be

rearranged as follows

2

2

HSL i iH ln

i i

11

LK pJ P

K p

(6.9)

where ip is the average partial pressure of species i (CO, in this study extended

also to CO2) between the feed and retentate sides, and lnP is the logarithmic

mean driving-force (calculated based on heat-exchanger theory for parallel flow

[15]), expressed as:

2 2 2 2

2 2

2 2

1 2 1 2 1 2 1 2H , feed H , permeate H , retentate H , permeate

ln 1 2 1 2H , feed H , permeate

1 2 1 2H , retentate H , permeate

ln

p p p pP

p p

p p

(6.10)

In this work, it was studied the inhibition effect caused by CO and CO2 on

a 50 µm thick PdAg25 wt.% membrane permeability to H2 in the temperature

range of 473-573 K. The Sieverts-Langmuir model was used to describe not only

H2/CO but also H2/CO2 permeation experiments. Furthermore, Eq. (6.10) that

accounts for the driving-force change along the membrane length was

introduced by improving significantly the accuracy of the analysis. Finally, the

adsorption enthalpies for both CO and CO2 adsorption were determined.


189

6.3 Experimental

6.3.1 Membrane preparation and experimental set up

The membrane used in this work was prepared by diffusion bonding of

cold-rolled Pd-Ag foils, providing a thin wall tubular membrane with a finger-like

configuration (cf. Fig. 6.1).

Fig. 6.1 – Sketch of the module assembly and the cold rolled PdAg25 wt.% membrane with a finger-like configuration inside.

In the preparation, it was used a commercial 50 µm thick PdAg25 wt.% foil

(from Johnson Matthey); detailed information regarding this technique can be

found elsewhere [16]. The permeation module (Fig. 6.1) was placed inside an

electrical oven (Memmert, model UNE200).

The gases (H2, N2, CO and CO2) were fed into the membrane lumen using

four mass controllers (Bronkhorst High-Tech, model F201) and preheated inside

the oven before entering the membrane module; permeate and retentate stream

flow rates were measured with mass flow meters (Bronkhorst High-Tech, model

F201).

Pressure transducers (Druck, ref. 4010) were used to measure the

pressure in the feed and permeate sides; the pressure in the feed side was set

using a back-pressure regulator (Swagelok).

Chapter 6

190

6.3.2 Gas permeation experiments

Before the experiments with pure hydrogen, the palladium/silver

membrane was firstly activated by reduction with hydrogen following a two-step

protocol: i) while the temperature increased until 573 K, nitrogen was fed to the

feed and permeate sides; ii) after reaching the end temperature, H2 was

introduced in the feed side (H2:N2 molar ratio of 1:1) with a total flow rate of 90

mLNmin-1. The activation step ended when maximum and stable H2 permeating

flux was measured, i.e., when steady state conditions were reached.

Pure hydrogen permeation experiments were performed at different

temperatures (473-573 K) and feed pressures (150-223 kPa) for a constant feed

flow rate of 90 mLNmin-1 of H2. Higher pressures were not used due to the

mechanical resistance limitation of the thin wall membrane (maximum operating

total pressure difference across the membrane should not exceed ca. 200 kPa).

The permeate side was kept at the atmospheric pressure in all experiments.

Additionally, permeation experiments were performed with membranes

being submitted to feed streams comprising binary mixtures of H2/CO or H2/CO2;

the total volumetric flow rate was kept constant while the molar ratio of CO/H2 or

of CO2/H2 was varied. The Sieverts driving force (Eq. (6.7)) was kept

approximately constant (at 130 Pa0.5) by adjusting the total feed pressure (up to

a maximum of 253 kPa) when changing the feed composition. These

experiments were performed at low H2 recoveries: the maximum recovery

obtained was ca. 34% in CO2/H2 experiments. The pressure drop along the

membrane tube was negligible and therefore it was assumed that the total

pressure remained constant, being equal in the feed and retentate streams.

After a set of experiments with binary mixtures, the membrane was

poisoned and consequently needing a recovery of the permeating behaviour

before a new test. A regeneration strategy was then implemented to remove the

adsorbed contaminant gases. The membrane regeneration was performed by

increasing the temperature up to 573 K and feeding synthetic air or nitrogen, to


191

both sides of the membrane, after experiments with CO or CO2, respectively. In

the case of treatment with synthetic air aimed at oxidizing the CO to CO2, for

safety reasons the membrane was first flushed with N2 for H2 removal.


6.4.1 Pure H2 permeance experiments

The experimental data for pure H2 permeation were fitted to the Sieverts

law, Eq. (6.7) (Fig. 6.2); in all cases R2 was found to be higher than 0.993.

Fig. 6.2 – Hydrogen molar flux through the membrane as a function of the difference between square root of H2 partial pressure in the feed and permeate sides - Eq. (6.7).

Since permeability follows an Arrhenius-type equation, the activation

energy was determined from the linear fit between 2H

L and the inverse of

temperature (c.f. Fig. 6.3).

P (Pa0.5)

0 20 40 60 80 100 120 140 160

JH

2

(m

ol m

-2 s

-1)

0.000

0.005

0.010

0.015

0.020

0.025

0.030573 K (R2=0.9954)

548 K (R2=0.9932)

523 K (R2=0.9938)

498 K (R2=0.9939)

473 K (R2=0.9936)

Chapter 6

192

Fig. 6.3 – Hydrogen permeability as a function of the inverse of temperature.

According to Yun and Oyama [12], low activation energy values (< 30

kJmol-1) indicate that diffusion is the limiting step, which is observed in the

present case (17.41 kJ∙mol-1). This value and also the pre-exponential factor

(1.16 x 10-5 mol∙m∙m-2∙s-1∙kPa-0.5) were found to be within the ranges reported in

the literature for this kind of membranes, 12.48-48.50 kJ∙mol-1 and 0.38-9.33 x

10-5 mol∙m∙m-2∙s-1∙kPa-0.5, respectively [13].

6.4.2 Effect of CO and CO2 on H2 permeance

6.4.2.1 Comparison between Sieverts-Langmuir model and

rearranged equation

The Sieverts-Langmuir equation (Eq. (6.8)) and the new herein proposed

rearranged one (Eq. (6.9)) were used to predict the membrane permeability to

hydrogen in the presence of carbon monoxide or carbon dioxide. Eq. (6.9) is in

principle more accurate, because it considers the influence of the concentration

gradients along the membrane length, making use of a logarithm-mean pressure

1/T (K-1)

0.0017 0.0018 0.0019 0.0020 0.0021 0.0022

-ln

(LH

2

)

18.4

18.6

18.8

19.0

19.2

19.4


193

difference. However, in practical applications and for design its use is not

straightforward, as it requires information about the exit composition.

Comparison of H2 fluxes predictions by both equations is shown in Fig. 6.4a and

Fig. 6.4b for CO and CO2, respectively.

The results of Fig. 6.4 clearly indicate an almost linear relationship

between the hydrogen fluxes calculated with Eq. (6.8) and Eq. (6.9) for

experiments in the presence of CO or CO2, in the temperature range studied.

However, Fig. 6.4b clearly indicates a systematic deviation. This trend is related

to the CO2 partial pressure that is underestimated in the case of the original

Sieverts-Langmuir equation since it is based on feed conditions. Once hydrogen

permeates through the membrane, the retentate side becomes richer in the non-

permeating species (CO2 in this case) along the tube. Therefore, the carbon

dioxide partial pressure increases, which is only considered in Eq. (6.9) – an

average partial pressure is considered for CO2, as well as a corrected driving

force for H2, based on the logarithmic mean P (Eq. (6.10)). The deviation seen

for CO2 (average deviation of 4.8 %) is almost not observed in the case of CO

(deviation < 0.9 %) because:

- the experiments were performed with low CO concentrations (max. 5 %

for CO vs. 20 % for CO2, as described below),

- the hydrogen recovery was lower for the runs with carbon monoxide (up

to 24% vs. up to 34% with carbon dioxide), thus reducing the need to correct the

gas concentrations along the membrane axis.

Chapter 6

194

Fig. 6.4 – Parity plot for comparing the H2 flux determined with the Sieverts-Langmuir model (Eq. (6.8)) with the H2 flux determined with the rearranged Sieverts-Langmuir model (Eq. (6.9)) in the presence of CO (a) or CO2 (b).

Nevertheless, Eq. (6.9) does not allow predicting H2 flux reduction due to

the presence of CO or CO2 because the hydrogen partial pressure in the

retentate side is not previously known. Therefore, using the conditions based on

the feed side (Eq. (6.8)) is more convenient for design or modelling purposes,

JH

2

SL (mol m-2 s-1), Eq. (6.9)

0.000 0.004 0.008 0.012 0.016 0.020

JH

2

SL (

mol m

-2 s

-1),

Eq.

(6.8

)

0.000

0.004

0.008

0.012

0.016

0.020573 K

548 K

523 K

498 K

a)

JH

2

SL (mol m-2 s-1), Eq. (6.9)

0.000 0.004 0.008 0.012 0.016 0.020

JH

2

SL (

mol m

-2 s

-1),

Eq.

(6.8

)

0.000

0.004

0.008

0.012

0.016

0.020573 K

548 K

523 K

498 K

473 K

b)


195

even if some relaxation/inaccuracy has to be admitted, which might be

compromised at higher recoveries (e.g. with thinner membranes or higher

permeation driving forces). Even so, in the conditions of Fig. 6.4, the deviation

between both equations is below 10.7% (average deviation of 4.8%).

6.4.2.2 Modelling by the Sieverts-Langmuir equation

With the operation conditions used in this work, it was observed that CO

significantly reduces H2 permeability (cf. Fig. 6.5), even at low concentrations.

Fig. 6.5 – Normalized H2 flux as a function of the CO concentration in the feed for different temperatures. The fitted function (dashed lines) based on SL model (Eq. (6.8)) is also shown.

For a CO content of only 5 %, hydrogen permeation flux decreases 57 %

at 573 K, being this effect even more remarkable at lower temperatures – at 498

K, hydrogen permeation flux is less than 18 % of that obtained in pure H2

streams, for the same driving force. Gallucci et. al [9] also reported a significant

inhibition effect of CO on the same type of membranes. They refer that such

effect is due to a combination of surface (at low CO content) and dilution effects

% (v/v) CO

0 1 2 3 4 5 6

JH

2*

0.0

0.2

0.4

0.6

0.8

1.0 573 K

548 K

523 K

498 K

Chapter 6

196

(as the CO content increases). Also, Fig. 6.5 highlights that the experimental

data show good agreement with the Sieverts-Langmuir model (SL) – Eq. (6.8),

for all conditions tested (see the fitting lines).

In the study performed by Barbieri et al. [14], it was stated that CO2 acts

as an in inert and thus H2 permeance is not affected: however, these authors

used a 60 µm thick PdAg23 %wt. membrane, 10 % of CO2 in the feed stream

and operated at higher temperature (647 K). Differently, for a thicker membrane

(250 µm) Chabot et al. [17] reported that a mixture containing 9 % of CO2 and

9.5 % of H2 balanced with He slightly affects H2 permeation (although much less

significantly than CO), particularly at temperatures below 423 K. In this work, it

was found that H2 permeance could be considerably reduced due to CO2

presence; as shown in Fig. 6.6, the hydrogen flux can be reduced by as much

as 40 %, for a CO2 content in the feed of 20 % at 473 K. In this case, however,

the effect of temperature is not as pronounced as noticed for CO (c.f. Fig. 6.5).

Furthermore, the SL model also predicted quite well the experimental data

obtained for CO2, suggesting that this compound also adsorbs on the membrane

surface (Fig. 6.6).

The results reported in Fig. 6.6 are particularly important because CO2 is

a reaction product in most applications concerning hydrogen

production/purification. Therefore, its content will increase along the membrane

axial direction due to both CO2 formation (in membrane reactors) and H2

permeation, thus enriching the retentate stream in non-permeating species. In

this regard, it is noteworthy that at 473 K the H2 flux decreased by ~41 % for a

feed comprising 20 % of CO2 (see Fig. 6.6).


197

Fig. 6.6 – Normalized H2 flux as a function of the CO2 concentration in the feed for different temperatures. The fitted function (dashed lines) based on SL model (Eq. (6.8)) is also shown.

Despite the results found in the literature concerning the effects of these

species on membrane permeance towards H2 (see for instance the review of

Paglieri and Way [4]), it is difficult to compare them due to the variety of operating

conditions used, types of membrane (supported or unsupported, Pd or Pd-alloy,

range of thickness, etc.) and preparation techniques. Therefore, for assessing

the performance of a membrane it is better to use normalized conditions, for

instance using the normalized hydrogen flux, i.e. in the presence of

contaminants vs. pure hydrogen.

Some results taken from literature concerning the effect of CO and CO2 on

the H2 flux reduction are shown on Fig. 6.7 for comparison with those obtained

in the present work. In this figure is shown that the temperature effect on H2

permeation (when in the presence of CO or CO2) depends on the membrane

type (Pd/Pd-alloy), thickness and preparation technique. For instance, in the

works of Mejdell et al. [18] and Gielens et al. [19] were employed membranes

prepared by magnetron sputtering (MS). In the former case, 5 % CO in the feed

% (v/v) CO2

0 5 10 15 20 25

JH

2

*

0.4

0.5

0.6

0.7

0.8

0.9

1.0 573 K

548 K

523 K

498 K

473 K

Chapter 6

198

(with a constant total feed pressure of 3 bar) caused the most drastic H2 flux

reduction, whereas in the latter, it is noteworthy that a feed with 20 % of CO2 at

673 K is responsible for the same decrease on H2 flux as 5 % of CO at 573 K for

the membrane used in this work (~55 %).

Fig. 6.7 – H2 flux reduction as a function of temperature for different CO or CO2 feed concentrations and at constant Sieverts driving force. Values taken from literature for several Pd-based membranes are shown for comparison (the lines are for eye guidance).

As shown in Fig. 6.5 and Fig. 6.6, the temperature effect on the membrane

permeation to hydrogen is more notorious in the case of CO than in the case of

CO2, for which parameter α in the Sieverts Langmuir model, Eq. (6.8), remains

nearly constant and equal to 0.45 (see Table 6.1).

Table 6.1 – Sieverts-Langmuir parameters obtained by fitting the experimental data of Fig. 6.5 and Fig. 6.6 to Eq. (6.8).

T / K CO CO2

α K / kPa-1 α K / kPa-1

573 0.63 0.698 0.38 0.109 548 0.68 0.720 0.41 0.116 523 0.77 1.146 0.45 0.121 498 0.86 1.443 0.45 0.128 473 - - 0.45 0.140

T (K)

448 473 498 523 548 573 598 623 648 673 698 723 748

(1-

JH

2*)

x 1

00 (

%)

10

20

30

40

50

60

70

80

90

1005% (v/v) CO2 - 50 m PdAg25 - CR - This work

20% (v/v) CO2 - 0.9 m Pd - MS - [19]

20% (v/v) CO2 - 50 m PdAg25 - CR - This work

5% (v/v) CO - 50 m PdAg25 - CR - This work

5 % (v/v) CO - 3 m PdAg23 - MS - [18]


199

In the case of CO, however, parameter α decreases considerably when

the temperature increases. For both gases, the adsorption constant decreases

with T, due to the exothermal nature of the process. Besides, these constants

are, as expected, much higher for carbon monoxide than for carbon dioxide.

The enthalpy of adsorption ( iH ) for each gas was determined using the

Van’t Hoff equation, which is expressed as follows:

i idln

d 1

K H

T R

(6.11)

where R stands for the gas constant. Fig. 6.8 shows the fittings obtained for both

species along with data taken from [18] for comparison.

Fig. 6.8 – Linear regression between the determined adsorption constants of CO and CO2 versus the inverse of temperature.

The adsorption enthalpy obtained for CO2 (-5.41 kJmol-1) clearly indicates

a weak (or physical) adsorption of this compound on the membrane surface.

Physical adsorption is normally attributed to van der Waals [20, 21] and

1000/T (K-1

)

1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2

ln(K

i)

-10

-9

-8

-7

-6

-5

-4

CO - This work

CO2 - This work

CO - [18]

Chapter 6

200

electrostatic forces [21]. In the case of CO, a higher value (-25.23 kJmol-1) was

obtained. Still, this value is inferior than the enthalpy value calculated with the

results obtained by Mejdell et al. [18] (74.07 kJmol-1) for a 3 µm thick Pd/Ag23

wt.% membrane with a microchannel configuration. Nevertheless, in all cases

the enthalpy values are far away the region considered for chemical adsorption

(200 kJmol-1) [20, 22].

At the end of the experiments with these gases, i.e. after shutting off the

flow of CO or CO2, the membrane was heated up to 573 K (the last experiment

was always carried at the lowest temperature studied where adsorption is

higher) and it was checked if the initial hydrogen flux was recovered. In the case

of CO2 it was possible to recover completely the initial flux by passing N2 on both

sides of the membrane, while after using CO a treatment with synthetic air was

required. In the latter case, 50 mLNmin-1 of air was fed to each side for 2 hours,

after which the membrane permeance to H2 was restored. It appears that the

reaction of oxygen with the adsorbed CO produced CO2 that is purged, leading

to complete membrane regeneration. Further insight about heat treatment with

air is given in a work by Mejdell et al. [23] whom observed that performing a heat

treatment with air at 573 K allowed to considerably reduce CO inhibition effect

on H2 permeability at 573 and 623 K. The authors considered that this could be

related to changes on the electronic properties of the membrane surface, where

grain growth and surface roughening occurred.


201

6.5 Conclusions

Permeation experiments were performed to assess in which extent PdAg

membrane performance for H2 permeation could be influenced by CO and CO2

presence. A modified Sieverts-Langmuir (SL) equation was proposed based on

the concentration gradients along the membrane length. Although being more

accurate, deviations to SL model with the rearranged equation are negligible in

the case of CO and are below 10.7 % for CO2, in the used conditions.

It was found that CO has a strong inhibition effect on H2 permeation due

to its co-adsorption on the metallic surface. Still, for the used operation

conditions the inhibition effect caused by CO2 exposure must not be neglected

and clearly CO2 cannot be considered as an inert.

The Sieverts-Langmuir equation adequately fitted the experimental data

for CO and, interestingly, also described quite well the results obtained for CO2.

Based on the values of the adsorption enthalpy and for the operating

temperature range employed (473 K – 573 K) it was concluded that CO and CO2

physically adsorb on the membrane. The results here found are important for

modeling purposes since allow predicting the influence caused by these species

in H2 permeance (or recovery).

Finally, it was possible to completely restore H2 permeance by increasing

the temperature from 473 K to 573 K and performing an oxidation procedure with

synthetic air for CO, or simply flowing nitrogen in the case of CO2 exposure.

Chapter 6

202

6.6 References

[1] V. Subramani, P. Sharma, L. Zhang, K. Liu, Catalytic Steam Reforming

Technology for the Production of Hydrogen and Syngas, in: K. Liu, C. Song, V.

Subramani (Eds.) Hydrogen and Syngas Production and Purification

Technologies, John Wiley & Sons, Inc., 2010, pp. 14-126.

[2] G.L. Holleck, Diffusion and solubility of hydrogen in palladium and palladium-

silver alloys, Journal of Physical Chemistry, 74 (1970) 503-511.

[3] E. Wicke, H. Brodowsky, H. Züchner, Hydrogen in palladium and palladium

alloys, in: G. Alefeld, J. Völkl (Eds.) Hydrogen in Metals II, Springer

Berlin/Heidelberg, 1978, pp. 73-155.

[4] S.N. Paglieri, J.D. Way, Innovations in palladium membrane research,

Separation and Purification Methods, 31 (2002) 1-169.

[5] S. Tosti, A. Basile, L. Bettinali, F. Borgognoni, F. Chiaravalloti, F. Gallucci,

Long-term tests of Pd–Ag thin wall permeator tube, Journal of Membrane

Science, 284 (2006) 393-397.

[6] S. Tosti, A. Basile, L. Bettinali, F. Borgognoni, F. Gallucci, C. Rizzello, Design

and process study of Pd membrane reactors, International Journal of Hydrogen

Energy, 33 (2008) 5098-5105.

[7] D. Mendes, V. Chibante, J.M. Zheng, S. Tosti, F. Borgognoni, A. Mendes,

L.M. Madeira, Enhancing the production of hydrogen via water-gas shift reaction

using Pd-based membrane reactors, International Journal of Hydrogen Energy,

35 (2010) 12596-12608.

[8] G. Barbieri, A. Brunetti, G. Tricoli, E. Drioli, An innovative configuration of a

Pd-based membrane reactor for the production of pure hydrogen: Experimental

analysis of water gas shift, Journal of Power Sources, 182 (2008) 160-167.

[9] F. Gallucci, F. Chiaravalloti, S. Tosti, E. Drioli, A. Basile, The effect of mixture

gas on hydrogen permeation through a palladium membrane: Experimental

study and theoretical approach, International Journal of Hydrogen Energy, 32

(2007) 1837-1845.


203

[10] T.A. Peters, M. Stange, H. Klette, R. Bredesen, High pressure performance

of thin Pd-23%Ag/stainless steel composite membranes in water gas shift gas

mixtures; influence of dilution, mass transfer and surface effects on the hydrogen

flux, Journal of Membrane Science, 316 (2008) 119-127.

[11] European Commission, Renewable Energy Road Map - Renewable

energies in the 21st century: building a more sustainable future, Brussels, 2007.

[12] S. Yun, S. Ted Oyama, Correlations in palladium membranes for hydrogen

separation: A review, Journal of Membrane Science, 375 (2011) 28-45.

[13] A. Basile, F. Gallucci, S. Tosti, Synthesis, Characterization, and

Applications of Palladium Membranes, in: M. Reyes, M. Miguel (Eds.) Membrane

Science and Technology, Elsevier, Amsterdam, The Netherlands, 2008, pp.

255-323.

[14] G. Barbieri, F. Scura, F. Lentini, G. De Luca, E. Drioli, A novel model

equation for the permeation of hydrogen in mixture with carbon monoxide

through Pd-Ag membranes, Separation and Purification Technology, 61 (2008)

217-224.

[15] F.P. Incropera, D.P. DeWitt, Fundamentals of Heat and Mass Transfer, 4th

ed., John Wiley & Sons, Inc., 1996.

[16] S. Tosti, L. Bettinali, Diffusion bonding of Pd-Ag rolled membranes, Journal

of Materials Science, 39 (2004) 3041-3046.

[17] J. Chabot, J. Lecomte, C. Grumet, J. Sannier, Fuel clean-up system:

poisoning of palladium-silver membranes by gaseous impurities, Fusion

Technology, 14 (1988) 614-618.

[18] A.L. Mejdell, M. Jøndahl, T.A. Peters, R. Bredesen, H.J. Venvik, Effects of

CO and CO2 on hydrogen permeation through a ~3 μm Pd/Ag 23 wt.%

membrane employed in a microchannel membrane configuration, Sep. Purif.

Technol., 68 (2009) 178-184.

[19] F.C. Gielens, R.J.J. Knibbeler, P.F.J. Duysinx, H.D. Tong, M.A.G.

Vorstman, J.T.F. Keurentjes, Influence of steam and carbon dioxide on the

Chapter 6

204

hydrogen flux through thin Pd/Ag and Pd membranes, Journal of Membrane

Science, 279 (2006) 176-185.

[20] P. Atkins, J. De Paula, Physical Chemistry, 8th ed., Oxford University Press,

2006.

[21] D.M. Ruthven, Principles of Adsorption and Adsorption Processes, 1st ed.,

John Wiley and Sons Inc., 1984.

[22] J.U. Keller, R. Staudt, Gas adsorption equilibria: experimental methods and

adsorptive isotherms, Springer, 2005.

[23] A.L. Mejdell, D. Chen, T.A. Peters, R. Bredesen, H.J. Venvik, The effect of

heat treatment in air on CO inhibition of a ~3 μm Pd-Ag (23 wt.%) membrane, J.

Membr. Sci., 350 (2010) 371-377.

Conclusions and Future Work

205

Chapter 7. Conclusions and Future Work

7.1 Conclusions

This thesis focused on the study and development of a carbon capture and

utilization (CCU) process able to, simultaneously, remove CO2 from flue gas and

afterwards convert it to methane in a sorptive reactor. The proof-of-concept of

such intensified process was performed after gathering important insight

regarding thermodynamics, adsorption and reaction fundamentals, which were

individually studied.

This work reports for the first time the characterization of CO2 sorption on

hydrotalcites, CO2 methanation over nickel-based catalysts, and testing a

sorptive reactor with reactive regeneration at LEPABE. Work was also done

regarding H2 permeation through Pd-based membranes, a topic however

already addressed previously (e.g. [1]).

The first task consisted on the design and equipment selection followed by

the assembly of an experimental setup suitable to perform the sorption, reaction

and sorption-reaction experiments. Some modifications to the original design

had to be made later to comply with certain operation issues during the sorption-

reaction experiments (see photo of the lab-scale bench in Appendix D). Overall,

this task was time consuming, but the reasoning made for designing the setup

allowed being previously aware of the operational limits/difficulties when

planning the experiments. The H2 permeation tests with PdAg membranes were

performed in an existing experimental setup in the group.

In the thermodynamic study (Chapter 2), two routes for flue gas-derived

CO2 valorisation, based on its direct hydrogenation to methane or methanol,

were compared. The study showed that methanol pathway requires harsher

operation conditions than the option of CO2 reduction to methane, particularly in

terms of pressure. In the case of CO2 valorisation to methane, it was concluded

that CH4 selectivity increases with the H2/CO2 molar feed ratio, being near 100

Chapter 7

206

% (for 150 ºC ≤ T ≤ 400 ºC and 1 bar ≤ T ≤ 30 bar) if a stoichiometric H2/CO2

molar feed ratio of 4 is used; a small fraction of CO can however be formed

through the reverse water gas shift reaction, being this pathway favoured at low

pressure and high temperature. Carbon formation is avoided for a H2/CO2 ratio

of 4, but it can become significant for lower ratios. In this case, carbon formation

can be inhibited if the water present in the flue gas is not removed from the

methanation reactor feed, although it decreases CO2 conversion because it is a

product of the methanation reaction.

Oxygen content in flue gas is generally low (< 5 vol. %), but it can cause a

considerable decrease of CO2 conversion due to the consumption of H2 and CH4.

The removal of the heat released by these extremely exothermic oxidation

reactions is another problem, particularly for safety reasons and to prevent

temperature excursions up to a level that can cause catalyst sintering. Moreover,

the presence of O2 leads to the oxidation of the methanation catalyst, thus

requiring additional H2 consumption for catalyst reactivation [2]. For these

reasons, a previous step of flue gas processing for O2 removal is recommended

before being admitted to a methanation reactor.

The aim of the high temperature study of CO2 sorption with hydrotalcites

(Chapter 3) was to synthesize materials with good CO2 sorption capacity and

fast kinetics at high temperature that could be used in the cyclic sorption-reaction

process for CO2 capture and conversion. Hydrotalcite-based materials were

selected based on their ability to operate in the temperature range compatible

with methanation catalysts and their enhanced performance in wet streams (e.g.

[3-5]). Following the works of Yavuz et al. [6] and Oliveira et al. [5], gallium-

substituted hydrotalcites were prepared by co-precipitation and promoted by

impregnation with ca. 20 wt. % of K, Cs or Sr. The sorbents were screened based

on their sorption capacity at 300 ºC under dry conditions. The sorption

equilibrium isotherms of all sorbents were determined from 0.05 bar to 1.1 bar

of CO2 and were well described by the Freundlich equation. Promotion with Cs


207

and particularly with K improved the sorption capacity, while Sr decreased

(compared to the unpromoted hydrotalcite).

Based on its higher sorption capacity, the gallium-substituted and K-

promoted hydrotalcite was selected and submitted to sorption-desorption cycles

to determine its working capacity under low pressure conditions (<0.0001 to 0.15

bar), as found in post-combustion CO2 capture applications. The sorbent’s

working capacity obtained at 300 ºC was 0.97 mmol·cm-3, representing 60 % of

the initial capacity, while the remaining 40 % could not be removed even under

vacuum. The working capacity at 300 ºC was almost fourfold higher than at 200

ºC, which suggests that CO2 sorption takes place through an activated process.

A kinetic model based on the CO2 sorption mechanism on hydrotalcites

suggested by Du et al. [7] was proposed to describe the sorption and desorption

kinetics. The model assumes two parallel contributions, one responsible for the

initial fast uptake and the other by the slow uptake observed for longer times.

The former is related to the reversible fraction of CO2 that is sorbed as bridged,

unidentate and bidentate carbonate, while the latter is related to the fraction of

irreversible sorbed CO2 in the form of polydentate carbonate. The kinetic model

considers 3 fitting parameters 2 2

1 2 1, and D r D r q q , and adequately fitted

the experimental uptake and downtake curves at 200 ºC and 300 ºC. The

temperature did not affect the initial relative contribution of the reversible amount

of sorbed CO2, being nearly the same at 200 ºC and 300 ºC (i.e. 48 and 44 %,

respectively). However, the temperature influences the values of the diffusion

coefficients. The amount of reversibly sorbed CO2 reaches the equilibrium faster

at 200 ºC than at 300 ºC, with the contrary being observed for the irreversible

fraction of sorbed CO2.

Sorption-desorption cycling leads to sorption kinetics slowdown of both the

reversible and irreversible contributions, independently of the temperature.

Globally, desorption was much faster at 200 ºC than at 300 ºC, especially for

removing the CO2 which was reversible sorbed.

Chapter 7

208

The intrinsic methanation reaction kinetics was determined in Chapter 4

over an industrial nickel catalyst (METH 134 from Clariant) that was later

employed for testing the integrated sorption-reaction process. The catalyst

activity remained stable for (at least) 83 hours under reactive conditions.

Extraparticle and intraparticle mass and heat transport resistances were found

negligible based on theoretical criteria and experimental observations. The

kinetic tests were conducted in a relevant temperature window (250-350 ºC) and

the pressure was the atmospheric. An ideal H2/CO2 molar ratio of 4 was used to

minimize CO and carbon formation, based on previous findings of the

thermodynamic study.

The observed reaction rates were extracted applying the differential

method and adjusted through non-linear fitting towards six kinetic models taken

(and adapted) from literature assuming three types of reaction mechanisms:

carbon intermediate, formate intermediate and formyl intermediate. Model

discrimination consisted on assessing the thermodynamic consistency and

statistical testing of model parameters. The statistical F-test allowed to choose

between rival models.

The chosen model has only four fitting parameters and is based on the

formyl intermediate mechanism, which considers that CO2 methanation begins

with H2 and CO2 dissociation and assumes hydroxyl as the most abundant

species at the catalyst surface. The kinetic equation was also successfully

validated through the modelling of an isothermal plug-flow fixed-bed reactor to

describe the experimental data.

In this thesis, the major contribution and novelty was related to the proof-

of-concept of a sorption-reaction process featuring reactive regeneration for

carbon dioxide capture by adsorption, followed of its conversion to methane; this

is addressed in Chapter 5. The sorptive reactor used for the proof-of-concept

was packed with a commercial hydrotalcite sorbent (Pural MG30-K from Sasol)

and a commercial nickel catalyst (METH 134 from Clariant), which are supplied

in millimetre size and in quantities compatible with the test bench. For these two


209

reasons, the gallium-substituted and K-promoted hydrotalcite synthesized and

characterized in Chapter 3 could not be used in this stage, although showing a

higher sorption capacity than the commercial Pural MG30-K employed.

The sorptive reactor was operated between 300 ºC and 350 ºC, the

pressure was low (≤ 2.5 bar), since flue gas is generally at the atmospheric

pressure, and the feed for the capture step consisted in 15 % vol. of CO2

balanced in N2. Given the complexity of the sorption and reaction processes

integration, other species present in flue gas like H2O and O2 were not included

in the feed stream at this stage.

The CO2 sorbent working capacity measured at 350 ºC after 6 sorption-

desorption cycles (purge with N2) was 0.32 mmol·g-1. When a H2 stream was

used for the desorption step instead of N2 (i.e. reactive regeneration mode), the

initial sorption capacity of the sorbent was recovered (which was not the case

with N2), and the desorption kinetics were enhanced because of the steam

produced in situ by the methanation reaction taking place at the catalyst surface.

The sorption equilibrium of the Pural MG30-K hydrotalcite was well

described by the dual-site Langmuir model using the parameters proposed by

Oliveira et al. [5] for the same sorbent. The endothermic term of the dual-site

Langmuir model explains the increased sorption capacity with temperature, as

also observed previously with the in-house prepared gallium-substituted and K-

promoted hydrotalcite. Knowing the dependence with temperature of the

physical (exothermic) and chemical (endothermic) contributions relative to the

total amount of sorbed CO2 is crucial to select the optimum conditions for

operating the cyclic process, namely to find the best compromise in terms of

sorption capacity and desorption kinetics. In addition, it was shown that

controlling the temperature and pressure could help minimizing the CO formed

though the parallel reverse water shift reaction.

The temperature did not affect methane productivity in the hybrid

multifunctional reactor, at least under the considered operation conditions. On

the contrary, methane purity was slightly higher at 350 ºC than at 300 ºC. The

Chapter 7

210

maximum CH4 purity obtained was 37 % and further reactor design should be

considered to improve this value up to the level required for injection in the

natural gas grid. Among the reactor design options, which will be presented and

discussed in the next section, one consists in changing the H2 feed system to

assure that a constant H2/CO2 ratio of 4 (the stoichiometric one) is achieved

along the reactor length during the regeneration stage, for instance through

radial feeding employing a H2-selective membrane. In this respect, the final part

of the thesis (Chapter 6) focused on the evaluation of the permeance of a 50 µm

thick PdAg membrane for a pure H2 stream and for binary mixtures of industrial

relevance, namely H2/CO and H2/CO2, in the temperature range of 200-300 ºC.

The H2 molar flux permeating through the membrane was well described

by the Sieverts law for a pure H2 feed. For a binary H2/CO or H2CO2 feed, the

Sieverts-Langmuir equation proved better. A modified Sieverts-Langmuir

equation assuming a logarithmic mean driving force to take into consideration

the concentration gradients along the membrane length due to H2 permeation

was proposed. This new equation founds relevance in conditions of high H2

recovery, but for design and modelling purposes, it was concluded that the

simpler Sieverts-Langmuir equation is preferred, even if some inaccuracy must

be admitted, because it does not require knowing beforehand the H2 partial

pressure on the retentate side.

It was found that CO has a strong inhibitory effect on H2 permeance

through the Pd-based membrane, even at low feed contents (< 5 vol. %), being

significantly worse than CO2. The inhibition was caused by physical adsorption

on the membrane, by blocking the available surface area for hydrogen

permeation that occurs through the solution-diffusion mechanism. The inhibitory

effect on H2 permeance decreases with temperature, but while CO2 completely

desorbs by simply increasing the temperature and flushing the system with N2,

membrane regeneration to remove adsorbed CO required membrane oxidation

with air.


211

Overall, the work performed in this thesis showed the technical feasibility

of a sorptive reactor for CO2 capture and conversion to methane. The good

performance of the commercial sorbent and catalyst, besides their compatibility

and stability when operating under cyclic operation, allow to be optimistic

regarding the possibility to scale-up this Carbon Capture and Utilization

technology and its potential relevance within the Power-to-Gas applications field.

7.2 Future work

The proof-of-concept of a sorptive reactor for CO2 capture and conversion

to methane was performed in this thesis. Several issues that can influence the

sorptive reactor performance in terms of CO2 capture capacity, CH4 productivity

and purity were identified and therefore future work is suggested.

First, the effect of other species present on flue gas should be

systematically assessed (e.g. steam and O2) and, afterwards, the sorptive

reactor must be tested employing a real flue gas stream. For instance, it is

expected that steam will enhance the CO2 capture step if a hydrotalcite sorbent

is used but, on the other side, steam will probably decrease methane productivity

during the CO2 conversion step. When considering a real flue gas composition,

the study will also provide evidence of the sorbent and catalyst tolerance to

certain impurities (e.g. SOx). This is crucial for process design, namely for

establishing previous flue gas processing requirements and to conclude about

the suitability of the considered materials. Depending on the effect caused by

other flue gas species on performance indicators, it could be considered to test

the sorptive reactor employing other sorbents and/or catalysts, or to find

optimum operation conditions to minimize potential negative effects. In this

respect, it would be interesting to test the concept using Ru-based catalysts,

which exhibit good activity and selectivity towards methane at lower

temperatures. In this way, both undesired CO and carbon formation reactions

would be disfavoured. Besides, some Ru-based catalysts for methanation are

Chapter 7

212

already commercially available, as presented in Chapter 4. Another option could

be including a water-gas shift catalyst layer (e.g. Cu-based) at the end of the

reactor to act as guard bed, converting CO traces and preventing its presence

in the outlet stream.

Other suggestions for future work include testing different H2 feeding

systems (e.g. countercurrent feed flow, radial feed flow with multiple entrances

along the bed length or through a Pd-based membrane), different bed layouts

(layered, mixed, etc.) and different sorbent-to-catalyst ratios. The use of

multifunctional materials (i.e., CO2 sorbents impregnated or somehow mixed

with methanation catalysts) is another strategy to pursue, apart from those

herein developed (namely the modified hydrotalcite) after appropriate

pelletization.

Testing different H2 feeding systems is important since CO2 desorbs during

reactive regeneration by purge gas stripping. Generally, purge gas stripping

method is performed with a non-adsorbing inert gas only when the adsorbate is

weakly held, otherwise it requires a high quantity of purge gas [8]. In this case,

the purge gas being used is H2, which is not retained by the sorbent, but is

consumed during the methanation reaction. Thus, the H2 feeding system should

deliver an H2 flow along the reactor length that allows to, simultaneously, achieve

complete CO2 desorption from the sorbent and the existence of an H2/CO2 molar

ratio of 4 in the gas phase near the catalyst.

Finding the optimum sorbent-to-catalyst ratio and bed layout is also

important to avoid packing the bed with an unnecessary weight of catalyst for

processing the amount of captured CO2.

Finally, it would be important to implement the cyclic batch process,

perhaps starting with a basic two-bed configuration and a modified Skarstrom

cycle to include reactive regeneration. Afterwards, modelling and simulation of

the cyclic process (with experimental validation) would allow for process design

and optimization. For the materials used in this work, some model parameters

were estimated (e.g. intrinsic methanation kinetics over METH 134), while for


213

the sorbent both sorption equilibrium and kinetic parameters are available in

literature (e.g. [5]).

Chapter 7

214

7.3 References

[1] D. Mendes, V. Chibante, J.M. Zheng, S. Tosti, F. Borgognoni, A. Mendes,

L.M. Madeira, Enhancing the production of hydrogen via water-gas shift reaction

using Pd-based membrane reactors, International Journal of Hydrogen Energy,

35 (2010) 12596-12608.

[2] Q. Zheng, R. Farrauto, A. Chau Nguyen, Adsorption and methanation of flue

gas CO2 with dual functional catalytic materials: a parametric study, Industrial &




[4] M.K. Ram Reddy, Z.P. Xu, J.C. Diniz da Costa, Influence of water on high-

temperature CO2 capture using layered double hydroxide derivatives, Industrial

& Engineering Chemistry Research, 47 (2008) 2630-2635.







Materials, 21 (2009) 3473-3475.






Supplementary information for Chapter 3

215

Appendix A. Supplementary information for Chapter 3

Fig. A.0.1 – XRD patterns of samples HTC-10Ga as prepared and after calcination at 673 K.

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70

Inte

nsi

ty[au]

diffraction angle (2θ Cu Kα)

HTC-10Ga (calcined)

HTC-10Ga

Appendix A

216

Fig. A.0.2 – EDS analyses for samples HTC-20K (first row) and HTC-10Ga-20K (second row) showing the K, Mg, Al and Ga distribution across

different zones.


217

Fig. A.0.3 – Adsorption/desorption isotherms of nitrogen over calcined samples: a) HTC, b) HTC-10Ga, c) HTC-20K, d) HTC-10Ga-

20Cs, e) HTC-10Ga-20Sr and f) HTC-10Ga-20K.

Appendix A

218

Fig. A.0.4 – Uptake (and downtake) curves (expressed in mmol·cm-3) for CO2 sorption at 573 K (first row) and 473 K (second row) for sample HTC-

10Ga-20K.


219

Buoyancy correction

The specific amount of gas adsorbed (q) can be calculated from:

ads

s

adsorbedmoles mq

mass of adsorbent m M (A.1)

where mads is the mass of adsorbed gas, M is the molecular weight of the

adsorbate, and ms is the mass of the activated sample.

Buoyancy is the upward force exerted over an object immersed in a fluid

and it is equal to the weight of the fluid displaced by the object. Buoyancy does

not actually affect the mass of an object, however it can affect its measure and

correction might be necessary.

So, for buoyancy correction, the weight of fluid displaced by the activated

sample ( s gV ), the adsorbed gas ( ads gV ) and the sample container i gV must be

determined and is equal to:

( )s ads i gMass of displaced gas V V V (A.2)

where Vs is the volume of the solid adsorbent, Vads is the volume of adsorbed

gas, Vi is the volume of the measuring cell (axis, basket and glass wool) and

g is the density of the gas at measuring conditions: g

PMz

T

, where z is

the acentric factor obtained from the virial equation of state [1].

Then, the real mass of the sample (mreal), i.e., corrected by the buoyancy

effect is equal to sum of the mass given by the magnetic balance (m) and the

mass of displaced gas ( ( )s ads i gV V V ):

giadss VVVmm )(real (A.3)

Appendix A

220

The adsorbed mass is calculated from the mass difference between the

real mass (Eq. (A.3)) and the sum of the initial mass of the activated sample (

sm ) and the measuring cell ( im )

ads s is ads i gm m (V V V ) (m m ) (A.4)

Assuming that the adsorbed phase presents a density similar to that of the

liquid phase:

ads adsads

ads l

m mV

(A. 5)

and

l

M

v (A.6)

where v is the molar volume calculated by the Gunn-Yamada method [1], then:

adsads s i

l

( ) ( )s i g

mm m m m V V

(A.7)

Solving Eq. (A.7) as function of mads the following expression is obtained:

lads

l

s i g

g

m m (V V )

(A.8)

where Δm represents the mass difference between the mass given by the

microbalance (m) and the sum of the mass of the activated sample (sm ) and

the mass of the measuring cell (axis, basket and glass wool) ( im ).

Finally, the adsorption capacity (q) can be calculated from:


221

s i

s

g l

l g

m (V V )q

m M

(A.9)

The volume of the basket with (Vi+Vs) and without (Vi) the sample, and

the respective mass, mi+ms and mi, were calculated by helium picnometry.

References

[1] B.E. Poling, J.M. Prausnitz, J.P. O'Connell, The properties of gases and

liquids, 5th ed., McGraw-Hill, New York ; London, 2000.


223

Appendix B. Supplementary information for Chapter 4

Calculation of transport and thermodynamic

parameters

Table B.1 lists some thermodynamic properties of the species considered

in this study.

Table B.1 – Properties of species used in this study [1].

Property CO2 H2 H2O CH4 N2

Molar weight, M (g·mol-1) 44.01 2.02 18.02 16.04 28.01 Critical temperature, Tc (K) 304.2 33.19 647.1 190.7 126.2 Critical pressure, Pc (bar) 73.83 13.13 220.55 46.00 34.00

Heat capacity of the gas mixture

The specific heat capacity of the gas mixture assuming ideal gas state,

p,mixC (J·kg-1·K-1), was calculated as follows [1]:

i p,i

i 1 i

n

p,mix

yCC

M

(B.1)

where yi and Mi (in kg·mol-1) stand for the mole fraction and molar weight of

species i, respectively, while the ideal gas state molar heat capacity of species

i, p,iC (J·mol-1·K-1), was retrieved from [1]:

p,i 2 2A B C DC

T T TR

(B.2)

Appendix B

224

where A-D are constants that vary for each species and are listed in Table B.2

and R is the ideal gas constant (8.314 J·mol-1·K-1 or m3·Pa·mol-1·K-1).

Table B.2 – Coefficients to estimate the heat capacity of the pure species.

Species A 103 B 106 C 10-5 D

CO2 5.457 1.045 0.000 -1.157 H2 3.249 0.422 0.000 0.083 H2O 3.470 1.450 0.000 0.121 CH4 1.702 9.081 -2.164 0.000 N2 3.280 0.593 0.000 0.040

Heat of reaction

The heat of reaction, rH (J·mol-1), at a given temperature T (K) was

estimated according to [1, 2]:

0

298

298 K

TpK

r r

CH T H R dT

R

(B.3)

where 298K

rH (J·mol-1) stands for the standard heat of the methanation reaction.

The second term of Eq. (B.3) was estimated as follows [1]:

0

2 2 3 3

0 0 0298

0

B C D 1A 1 1 1

2 3

T p

K

CdT T T T

R T

(B.4)

with 0 298 T K , 0T T and 0

pC is the standard heat capacity change of the

gas mixture given in J·mol-1·K-1. Coefficients A, B, C and D are listed in Table

B.2.


225

Density of the gas mixture

Assuming ideal gas behavior, the density of the gas mixture, g (kg·m-3),

was calculated according to:

g i i

1

n

i

Py M

RT

(B.5)

where P stands for the total pressure (Pa).

Viscosity of the gas mixture

The viscosity of the gas mixture, g (Pa·s), can be estimated using the

Wilke method [2] through Eq. (B.6)),

i ig

1j ij

=1

n

ni

j

y

y

(B.6)

where n is the number of species, and the binary term, ij , is calculated through

Eq. (B.7).

20 5 0 25

i j ij

ij 0 5

i j

1

8 1

. .

.

M M

M M

(B.7)

The viscosity of the pure species, i (Pa·s), is estimated through Eq. (B.8) [1]:

Appendix B

226

B

i

2

A

C D1

T

T T

(B.8)

where A, B, C and D are coefficients that depend on the species and were

retrieved from Aspen Properties software. Table B.3 lists the coefficients A-D

used to determine pure viscosity of each species.

Table B.3– List of coefficients A-D of each species to determine pure viscosity in Pa·s.

Species A B C D

CO2 2.1480 x 10-6 0.46000 290.000 0 H2 1.7970 x 10-7 0.68500 -0.590 140 CH4 5.2546 x 10-7 0.59006 105.670 0 H2O 1.7096 x 10-8 1.11460 0.000 0 N2 6.5592 x 10-7 0.60810 54.714 0

Molecular diffusivity of the gas mixture

The molecular diffusivity of species i in the gas mixture, i,mixD (m2·s-1), was

estimated through Eq. (B.9), as proposed by Fairbanks and Wilke [3]:

,mix

1

1 ii n

j

j ijj

yD

y

D

(B.9)

where ijD (cm2·s-1) is the binary diffusivity of species i when diffusing into a

component j. To estimate ijD , the Fuller correlation was considered (cf. Eq.

(B.10)) [4].


227

0 5

3 1 75

21 3 1 3

1 110

.

.

i j

ij

i j

TM M

D

P

(B.10)

with T in K, P in atmospheres and Mi in g·mol-1. The diffusion volume of a given

molecule is equal to the sum of the atomic diffusion volumes (i.e. ). The

values for the molecules considered in this study are given in Table B.4.

Table B.4 – Diffusion volumes for simple molecules, (cm3·mol-1), of species

considered in this work [1].

Molecule Diffusion volume,

CO2 26.9 H2 7.07 H2O 12.7 CH4 24.4 N2 17.9

The binary diffusivities calculated by Eq. (B.10) must be converted to SI

units before substitution in Eq. (B.9).

Effective mass diffusivity in the catalyst

particle

The effective mass diffusivity in the catalyst particle, effD (m2·s-1), was

calculated through Eq. (B.11) [1]:

2CO ,mix p

eff

p

DD

(B.11)

Appendix B

228

where 2CO ,mixD (m2·s-1) is the CO2 diffusivity in the gas mixture calculated from

Eq. (B.9); p and p are the porosity and tortuosity of the catalyst particle,

respectively. The particle porosity was assumed equal to 0.4. The tortuosity

generally is ranged between 2 and 6 [5]. For a catalyst particle, it is

recommended to consider a tortuosity value of 4 when no other information is

available [1, 6].

Thermal conductivity of the gas mixture

The thermal conductivity of the gas mixture, g (W·m-1·K-1), was estimated

following the Wassiljewa method (cf. Eq. (B.12)) [2]:

g,

g

i 1

1

ni i

n

j ij

j

y

y A

(B.12)

where ijA is the binary interaction parameter calculated by Eq. (B.13), as

suggested by Mason and Saxena [2],

20 5 0 25

0 5

1

8 1

. .

tr ,i tr , j i j

ij .

i j

M MA

M M

(B.13)

with tr ,i tr , j being the ratio of the translational thermal conductivities

calculated from Eq. (B.14) [2].


229

exp 0 0464 exp 0 2412

exp 0 0464 exp 0 2412

j r ,i r ,itr ,i

tr , j i r , j r , j

. T . T

. T . T

(B.14)

r ,i c ,iT T T stands for the reduced temperature and i (m·K·W-1) is the reduced,

inversed thermal conductivity of species i (or j) according to Eq. (B.15) [2],

1 63

4

210 c,i i

i

c ,i

T M

P

(B.15)

where c ,iT (K), c ,iP (bar) and Mi (kg·mol-1) are the critical temperature, critical

pressure and molar weight of species i (see Table B.1).

The thermal conductivity of species i, g,i (W·m-1·K-1), was calculated

through the Eucken model (Eq. (B.16)) [2]:

g,

5

4

p,i

i i

i i

CR

M M

(B.16)

with i in Pa·s (cf. Eq. (B.8)), Mi in kg·mol-1 (cf. Table B.1) and p,iC in J·mol-1·K-

1 (cf. Eq. (B.2)).

Mass and heat transport coefficients

The mass and heat transport coefficients between the gas and the catalyst

particle, kg (m·s-1) and h (W·m-2·K-1), respectively, were estimated from a

correlation expressed in terms of the Colburn J factor analysis [7]:

Appendix B

230

2 3g

D

o

kJ Sc

u (B.17)

2 3

, g

PrH

p mix o

hJ

C u (B.18)

2 3

p1.09ReD HJ J

(B.19)

where ou stands for the fluid superficial velocity (m·s-1); the Schmidt (Sc),

Prandtl (Pr) and Reynolds pRe numbers are given by Eqs. (B.20)-(B.22),

respectively.

2

g

g ,CO mix

ScD

(B.20)

, g

g

Prp mixC

(B.21)

g

g

Reo p

p

u d (B.22)

Effective thermal conductivity of the catalyst

The effective thermal conductivity of the catalyst, eff (W·m-1·K-1), was retrieved

using the correlation expressed by Eq. (B.23) [8], where s (W·m-1·K-1) is the

solid thermal conductivity, which was estimated from information found

elsewhere [8-10].

1

sg

g

p

eff

(B.23)


231

Parameters used to assess mass and heat

transport limitations (Section 4.1.2)

Table B.5 – Parameters used to assess mass and heat transport limitations*.

Parameter Value Units

n 1

pd 256 x 10-6 m

,p mixC 3263.3 J·kg-1·K-1

rH -179.7 kJ·mol-1

g 0.47262 kg·m-3

g 2.9943 x 10-5 Pa·s

aE 118.0 kJ·mol-1

p 4

p 0.4

2 ,CO mixD 6.8062 x 10-5 m2·s-1

effD 6.8062 x 10-6 m2·s-1

gk 0.138 m·s-1

h 161.6 W·m-2·K-1

g 6.946 x 10-2 W·m-1·K-1

s 13.9 W·m-1·K-1

eff 1.667 W·m-1·K-1

* Parameters estimated based on feed conditions at the higher temperature (350

ºC) and higher 2CO

in

catW F .

Appendix B

232

References

[1] R.H. Perry, D.W. Green, Perry's Chemical Engineers' Handbook, McGraw

Hill, New York, 1999.

[2] B.E. Poling, J.M. Prausnitz, J.P. O'Connel, The properties of gases and

liquids, 5th ed ed., The McGraw-Hill Companies, New York, 2001.

[3] D.F. Fairbanks, C.R. Wilke, Diffusion Coefficients in Multicomponent Gas

Mixtures, Industrial & Engineering Chemistry, 42 (1950) 471-475.

[4] J.R. Welty, C.E. Wicks, R.E. Wilson, G.L. Rorrer, Fundamentals of

Momentum, Heat and Mass Transfer, John Wiley & Sons, Inc., 2008.



[6] C.N. Satterfield, Heterogeneous Catalysis in Practice

McGraw-Hill, 1980.

[7] A.E. Rodrigues, Scientific basis for the design of two phase catalytic reactors,

in: A.E. Rodrigues, J.E. Calo, N.H. Sweed (Eds.) Multiphase Chemical Reactors,

Volume II – Design Methods, Sijthoff & Noordhoff International Publishers B.V.,

Alphen aan den Rijn, The Netherlands, 1981.

[8] M. Soomro, R. Hughes, The thermal conductivity of porous catalyst pellets,

The Canadian Journal of Chemical Engineering, 57 (1979) 24-28.

[9] P. Auerkari, Mechanical and physical properties of engineering alumina

ceramics, Technical Research Centre of Finland, VTT Manufacturing

Technology, Research Notes 1792, 3-36, Espoo, 1996.

[10] J.E. Keem, J.M. Honig, Selected electrical and thermal properties of

undoped nickel oxide, Center for Information and Numerical Data Analysis and

Synthesis, Purdue University, Cindas Report 52, West Lafayette, Indiana, 1978.


233

Appendix C. Supplementary information for Chapter 5

(cont.)

Appendix C

234

(cont.)


235

Fig. C.0.1 – History of pressure, CO2 and CH4 outlet fraction during regeneration at

350ºC (cycles 8-10) and 300 ºC (cycles 11-15) for the initial 2000 s.

Appendix C

236

(cont.)


237

Fig. C.0.2 – Normalized temperature histories in four bed positions during CO2 sorption and regeneration with H2 (cycles 11 to 15); Tz,0 = 300 º C.

Appendix C

238

(cont.)


239

Fig. C.0.3 – Normalized temperature histories in four bed positions during CO2 sorption and regeneration with H2 (cycles 7 to 10); Tz,0 = 350 º C.

Experimental Setup

241

Appendix D. Experimental Setup

Fig. D.0.1 – Assembled experimental setup for the PhD work.

Date post:	03-Aug-2020
Category:	Documents
Upload:	others
View:	0 times
Download:	0 times

chemicals: methane production · A thermodynamic analysis of the CO 2 hydrogenation using as feed a...

Documents