Download as pdf or txt
Download as pdf or txt
You are on page 1of 216

Catalytic CO2 conversion: a techno-economic analysis and

theoretical study

Thomas Savaete

Supervisor: Prof. dr. ir. Mark Saeys


Counsellor: ir. G.T. Kasun Kalhara Gunasooriya

Master’s dissertation submitted in order to obtain the academic degree of


Master of Science in Chemical Engineering

Department of Chemical Engineering and Technical Chemistry


Chair: Prof. dr. ir. Guy Marin
Faculty of Engineering and Architecture
Academic year 2015-2016
Catalytic CO2 conversion: a techno-economic analysis and
theoretical study

Thomas Savaete

Supervisor: Prof. dr. ir. Mark Saeys


Counsellor: ir. G.T. Kasun Kalhara Gunasooriya

Master’s dissertation submitted in order to obtain the academic degree of


Master of Science in Chemical Engineering

Department of Chemical Engineering and Technical Chemistry


Chair: Prof. dr. ir. Guy Marin
Faculty of Engineering and Architecture
Academic year 2015-2016
Acknowledgements
First of all, I would like to thank my coach, Kasun, for his constant guidance, support and encouragement
throughout my master thesis. I also sincerely thank my promotor, Professor Mark Saeys, for giving me the
opportunity to work on this topic. The discussions have been a source of inspiration and guidance, although
I’ve got the chance to set my own goals and challenges. A special thank you goes out to Riujin and Esteban,
for helping me with my report; and to Luis, for initiating me in the field of kinetic modeling.

To my colleagues and friends at the lct I would like to say that they have been a great help and support
throughout this master thesis. The amusing moments have certainly made up for the busy and stressful
moments. My fellow ‘Chemical Brothers’ I would like to thank for the awesome years in Ghent. I wish them
good luck in all their future pursuits.

My family has been very important for me. My parents I would like to thank for encouraging me to pursue
my dreams and have high ambitions; my brother Simon for showing anything is possible if you have the
proper motivation; and my sisters Céline and Leonie for making me realize that fun and laughter will always
be an important aspect in life. Finally I would like to thank my girlfriend - I love you- for the movies and
board games, and for her patience and motivation throughout this work.
FACULTY OF ENGINEERING AND ARCHITECTURE

Department of Chemical Engineering and Technical Chemistry


Laboratory for Chemical Technology
Director : Prof. Dr. Ir. Guy B. Marin
Laboratory for Chemical Technology

Declaration concerning the accessibility of the master thesis

Undersigned,

Thomas Savaete

Graduated from Ghent University, academic year 2015-2016 and is author of the master thesis
with title:

Catalytic CO2 conversion: a techno-economic analysis and theoretical study.

The author gives permission to make this master dissertation available for consultation and to copy
parts of this master dissertation for personal use.
In the case of any other use, the copyright terms have to be respected, in particular with regard to
the obligation to state expressly the source when quoting results from this master dissertation.

June 01, 2016

Laboratory for Chemical Technology • Technologiepark 914, B-9052 Gent • www.lct.ugent.be


Secretariat : T +32 9 331 17 57 • F +32 9 331 17 59 • Petra.Vereecken@UGent.be
Catalytic CO2 conversion: a techno-economic analysis
and theoretical study

Thomas Savaete

Supervisor: Prof. dr. ir. Mark Saeys

Counsellor: ir. G.T. Kasun Kalhara Gunasooriya

Master’s dissertation submitted in order to obtain the academic degree of Master of Science in Chemical
Engineering

Department of Chemical Engineering and Technical Chemisty

Chairman: Prof. dr. ir. Guy Marin

Faculty of Engineering and Architecture

Academic year 2015-2016

Summary
Reduction of CO2 emissions is one of the most important challenges mankind is facing today, and human
actions are required to reduce the CO2 concentration in the atmosphere. Catalytic conversion processes with
renewable energy represent a possible pathway and are capable of creating a carbon-neutral fuel and
chemicals cycle. To evaluate the feasibility of these processes, a techno-economic analysis is performed for
(i) hydrogen production as a high energy reductant for conversion of CO 2; (ii) production of methanol via
conversion of CO2 with hydrogen, and further conversion of methanol via the methanol-to-olefins process;
(iii) ethylene synthesis via electrochemical reduction of CO 2 and water, and a modified Fischer-Tropsch
synthesis (FTS) for CO2 and H2; and (iv) electrochemical reduction of CO2 to formic acid. Rapid activation
of the stable CO2 molecule is a key challenge and via computational catalysis effective catalyst systems can
be sought. The Co(111) surface is evaluated for CO2 activation, with inclusion of coverage effects and the
effect of step sites, i.e. the Co(211) surface. CO2 is then compared with the Cu(111), Ni(111), and Au(111)
surfaces. An initial kinetic model is composed to describe the CO activation and chain propagation reactions
in FTS up to C2 species. Preliminary analysis of CO2 activation is done by incorporation of CO2 activation
steps in the reaction mechanism for FTS.

Keywords
Techno-economic analysis, CO2 activation, hydrogen, methanol, density functional theory, cobalt, μ-kinetic
modeling, Fischer-Tropsch synthesis
Catalytic CO2 conversion: a techno-economic
analysis and theoretical study.
Thomas Savaete

Supervisor: prof. dr. ir. Mark Saeys, counsellor: ir. G.T. Kasun Kalhara Gunasooriya

Abstract Reduction of CO2 emissions is one of the most 2015. However, government commitment will be necessary
important challenges mankind is facing today, and human for the integration and stimulation of renewable energy via
actions are required to reduce the CO 2 concentration in the investments and budgetary support [4].
atmosphere. Catalytic conversion processes with renewable Essentially four approaches exist for CO2 emission
energy represent a possible pathway and are capable of creating reduction: (i) improved actual energy efficiency; (ii) use of
a carbon-neutral fuel and chemicals cycle. To evaluate the non-carbon or renewable energy sources (e.g. nuclear, solar or
feasibility of these processes, a techno-economic analysis is
performed for (i) hydrogen production as a high energy
wind energy); (iii) carbon capture and sequestration (CCS);
reductant for conversion of CO2; (ii) production of methanol via and (iv) carbon capture and utilization (CCU). CCU has the
conversion of CO2 with hydrogen and further conversion of potential of securing future energy supply and decrease net
methanol via the methanol-to-olefins process; (iii) ethylene CO2 emissions to the atmosphere, and is therefore further
synthesis via electrochemical reduction of CO 2 and water, and a examined. CO2 utilization consists of three general pathways:
modified Fischer-Tropsch synthesis (FTS) for CO2 and H2; and (i) conversion of CO2 into fuels (storage of renewable
(iv) electrochemical reduction of CO 2 to formic acid. Rapid energy); (ii) production of chemical feedstock; and (iii) non-
activation of the stable CO2 molecule is a key challenge and via conversion use of CO2 (e.g. supercritical solvents, EOR, etc.)
computational catalysis effective catalyst systems can be sought.
[5].
The Co(111) surface is evaluated for CO2 activation, with
inclusion of coverage effects and the effect of step sites, i.e. the
Conversion processes require carbon capture from high
Co(211) surface. CO2 is then compared with the Cu(111), concentration point emission sources of CO2, such as power
Ni(111), and Au(111) surfaces. An initial kinetic model is plants or cement factories. Energy addition is required, and
composed to describe the CO activation and chain propagation should be from carbon-neutral sources, as no net reduction of
reactions in FTS up to C2 species. Preliminary analysis of CO2 CO2 emission would be obtained for energy supply from high
activation is done by incorporation of CO2 activation steps in the emission sources such as fossil fuel driven power plants [1].
reaction mechanism for FTS. Important driving factors for the processes of CO2 recycle
Keywords Techno-economic analysis, CO2 activation, are the product/process value and the economics related to
hydrogen, methanol, density functional theory, cobalt, μ-kinetic carbon taxes. CO2 is freely available in the atmosphere (its
modeling, Fischer-Tropsch synthesis price solely determined by the cost for capture and
separation), and high value-added compounds (e.g. fine
I. INTRODUCTION chemicals) are reported to be economically feasible without
As the concentration of CO2 becomes an environmental carbon taxes, but production scales do not allow for CO2
problem, the catalytic conversion of CO2 to petrochemicals emission control. Large capacity compounds such as methanol
and fuels represents a possible pathway to reduce the CO2 or light olefins have the potential of CO2 emission control, but
emissions and create a carbon-neutral fuel and chemicals production is highly dependent on carbon taxes [6].
cycle. Rapid activation of the very stable CO2 molecule is
however a key challenge for conversion pathways to be A. Hydrogen production
feasible. Using computational catalysis, more effective Renewable energy generating plants (e.g. solar or wind
catalyst systems for CO2 activation reactions can be sought. energy) has an increasing impact on the current energy
Catalytic conversion processes strive to accelerate the natural production, but produce rather fluctuating supply of electricity
conversion processes of CO2 and directly produce fuels and and excess electricity from off-peak periods is difficult to
chemicals from CO2. Utilization of kinetic modeling can give store. Hydrogen production is suggested as a viable option for
additional insight in the reaction mechanism, product energy storage and can be an aid for the integration of
selectivity and indications on specific catalytic properties. renewable energy in our energy economy. For hydrogen to be
competitive with commercial fuels, increased production
II. TECHNO-ECONOMIC ANALYSIS capacities and reduced production costs are required and the
Increasing CO2 emissions has detrimental consequences Department of Energy of the USA reports that a production
such as global warming and the acidification of the oceans; cost for hydrogen of 4 €/kgH2 would allow hydrogen to be
and to reduce the CO2 concentration in our atmosphere, competitive [7]. Furthermore, carbon-neutral H2 production is
human actions are required [1]. Awareness has already grown necessary, which will strongly depend on the cleanliness of
worldwide with international agreements such as the Kyoto the electricity source. Renewable energy fulfills this
protocol [2] in 1997 and recently the Paris agreement [3] in requirement and moreover, allows for greater domestic energy
production and less dependence on foreign oil imports [8].
Electrolysis of water is the main process for production of H2
T. Savaete is with the Chemical Engineering Department, Ghent with renewable energy;
University (UGent), Gent, Belgium. E-mail: Thomas.Savaete@UGent.be .
1 chemical in our society and used in numerous applications,
𝐻2 𝑂 ⇌ 𝐻2 + 𝑂2 (1)
2 making it an interesting compound for research on large scale
and is evaluated for the most developed electrolysis production processes involving carbon recycling [11]. It has
techniques: alkaline electrolysis and polymer electrolyte the advantage of higher added-value, and has the potential to
membrane (PEM) electrolysis. store CO2 over a longer period of time in polymer materials
Actual and future production costs for hydrogen are [6, 12]. Ethylene production is evaluated for the electro-
evaluated for cases of centralized production, and as chemical reduction of CO2 and H2O, a modified Fischer-
renewable energy sources are widely spread over the globe, Tropsch synthesis (FTS) process for CO2 and H2, and the
also for small scale, distributed production facilities. The methanol-to-olefins (MTO) process.
analysis takes in account capital, feedstock, and operating and The overall reaction for electrochemical reduction is given
maintenance (O&M) costs [8]. Table 1 summarizes the H2 by;
production costs for PEM and alkaline electrolysis. Electricity 2 𝐶𝑂2 + 2𝐻2 𝑂 → 𝐶2 𝐻4 + 3𝑂2 (3)
feedstock (0.13 €/kWh) was found to be the main cost
and catalysts are mainly copper-based, although research is
contributor (>65%), followed by the capital cost. The future
ongoing for improved catalyst selectivity [16]. For the
central production via the PEM technology is found to be
reduction process to be feasible, actual energy efficiencies
most advantageous for H2 production.
(65-70%) need to be increased, via reduction of the currently
Table 1 Comparison of alkaline and PEM electrolysis hydrogen high overpotential; and high reaction rates are required.
production for current (2011-2013) and future (2020-2025), The modified FT process consists of the combination of the
distributed (1500 kgH2/day) and central (50 000 kgH2/day) production RWGS reaction and consecutive FT synthesis and mainly iron
facilities. Costs are shown on a €/kgH2 basis [7, 9]. and cobalt catalysts are used.
Case
Current Future Current Future 𝐶𝑂2 + 𝐻2 ⇌ 𝐶𝑂 + 𝐻2 𝑂 (4)
Distributed Distributed Central Central
𝐶𝑂 + 𝐻2 ⇌ 𝐶𝑛 𝐻2𝑛 + 𝐶𝑛 𝐻2𝑛+2 + 𝐻2𝑂 + 𝐶𝑂2 (5)
Alkaline 5.60 4.00 4.90 3.70
PEM 4.52 3.72 4.50 3.69 Actual applications show rather limited olefin selectivity and
high methane selectivity, as well as severe carbon deposition.
Hydrogen is suggested as a possible transportation fuel, but
Hence, catalyst optimization is required before industrial
has some important setbacks, such as its high volatility, low
applications are possible [12]. Electrochemical reduction of
volumetric energy density, and storage and safety issues.
CO2 and H2O in SOE cells to yield syngas and consecutive FT
Therefore, utilization of H2 to inject renewable energy in our
synthesis, is reported to be most promising for ethylene
energy economy (via conversion of CO2) could be a possible
production. The main cost contributor is the electricity
pathway to solve the problems concerning hydrogen [10-12].
feedstock for reduction, but is limited due to the high
efficiency of SOE cells (>95%) [17]. Capital costs are
B. Methanol synthesis relatively high, and only a small cost contribution is attributed
Production of methanol from CO2 and H2 is believed to be a to the feedstock. A production cost of approximately 2100
viable option for renewable fuel production. €/ton is calculated, still significantly higher than the actual
CO2 + 3H2 ⇌ CH3OH + H2 O (2) selling price of ethylene (1200 €/ton). Approximately three
Methanol production is favorable for its compatibility with ton of CO2 is recycled per ton of ethylene, allowing for a
existing distribution infrastructure and the ability to use it as a significant effect of carbon taxes.
replacement for current liquid fuels, as well as an intermediate The MTO process has the advantage of being already partly
in chemical production processes. It can be produced at commercially available and has a tuned and narrow product
moderate operating conditions, and requires only a moderate distribution from the use of zeolite catalysts [12].
±𝐻2 𝑂 −𝐻2 𝑂 𝑙𝑜𝑤𝑒𝑟
investment cost; providing a high acceptance for methanol 2𝐶𝐻3𝑂𝐻 ↔ 𝐶𝐻3 𝑂𝐶𝐻3 → (6)
production [13]. 𝑜𝑙𝑒𝑓𝑖𝑛𝑠
Modeling of the production cost of methanol is done via In a first step DME is produced, which further reacts to form
case studies simulating the conversion process with light olefins [18], but debate is still ongoing on the exact
specialized software such as Aspen [14]. CO2 is assumed to be nature of the process. Modeling of the MTO process indicates
available from high emission sources and H2 is produced from a high dependency on the methanol feedstock price. A
water electrolysis with renewable energy. The electricity to production cost of 3700 €/ton is obtained for ethylene via
methanol efficiency was found to be in the order of 70%, MTO, significantly higher than the actual selling price and
leading to a production cost for renewable methanol of price from the modified FT process.
approximately 1400 €/ton, which is significantly higher than
actual selling prices of 275 €/ton [15]. The main cost D. Electrochemical reduction to formic acid
contributor is the electricity feedstock for water electrolysis, Formic acid is mainly attractive for its use in hydrogen
and for methanol to become competitive the cost for storage and fuel cell applications. It has the advantage of low
renewable electricity should be reduced to approximately 0.03 chemical consumption, lower energy requirement and mild
€/kWh. In case carbon taxes of €100 per ton CO2 would be operating conditions. However, current production is twice the
set, reduction to 0.06 €/kWh would suffice. demand, although the global market is expected to grow [5].
The general reaction for electrochemical reduction of CO2 and
C. Ethylene production H2 is given by;
Due to their high energy of formation, short-chain olefins 2𝐶𝑂2 + 2𝐻2 𝑂 → 2𝐻𝐶𝑂𝑂𝐻 + 𝑂2 (7)
are an excellent opportunity to store renewable energy and performed in electrolytic cells, mainly with metal catalysts
incorporate it in the value chain of chemical production (e.g. Pb, Sn, etc.) [19]. Assuming an average renewable
processes. Ethylene, along with propylene, is the major base electricity price of 0.13 €/kWh and a cost for CO2 capture of
Table 2 Comparison of CO2 dissociation on different catalyst surfaces. Activation energies and Gibbs free activation energies are given in
kJ/mol.
CO2*+* ↔ CO*+O* Co(111) Cu(111) Ni(111) Au(111)
Transition state

Activation energy
60 159 79 310
Ef (kJ/mol)
Gibbs free energy
53 153 73 304
∆Gf (kJ/mol)
30 €/ton, the modeled production cost is approximately 1400 the iterative method can require a suitable preconditioner [28,
€/ton, which is significantly higher than commercial formic 29]. Implementation of a thermodynamic data file, gas-phase
acid (700 €/ton) [20]. Again, the electricity feedstock and kinetics file and surface kinetics file, including the properties
capital cost are the main cost contributors. and desired reactions, is required. Chemkin is then capable of
mathematically describing the reaction mechanism for the
III. METHODS AND MODELS surface coverages and gaseous composition [30-32]. Rate
coefficients are implemented in Chemkin as an Arrhenius
A. Ab-initio calculations equation.
Two catalyst models are constructed for a cobalt catalyst 𝐸𝑎
𝑘 = 𝐴 ∙ 𝑇𝛽 ∙ exp(− ) (10)
surface, consisting of 5 layers: an fcc Co(111) slab, using a 𝑅𝑇
p(3x3); and a Co(211) step surface, using a p(4x4) unit cell. Rate coefficients for adsorption and surface reactions are
For the copper, nickel and gold catalysts; (111) slabs are calculated from the collision and transition state theories
constructed, consisting of 5 layers, using a p(3x3) unit cell. respectively [33]. A homogeneous Co catalyst surface is
Optimized lattice constants of 3.56 Å, 3.68 Å, 3.56 Å, and modeled and adsorption is modeled on single active sites.
4.21 Å were obtained for the Co, Cu, Ni and Au surface Simulation is done for a pure feed stream of CO and H2
respectively; close to the experimental values [21, 22]. (H2/CO = 2) in a plug-flow reactor (PFR).
Density Functional Theory (DFT) calculations were
conducted using the VASP software tool and the vdW-DF2 IV. RESULTS
functional [23, 24]. A cut-off kinetic energy of 450 eV is
implemented for the plane-wave basis set and an inter-slab A. Ab-initio calculations
spacing of 15 Å is defined to minimize the interactions Dissociation of CO2 on the Co(111) surface is initially
between repeated slabs. Also a k-points mesh is generated, evaluated and a relatively low activation energy barrier of 60
with the Brillouin zone sampled with a (3x3x1) Monkhorst- kJ/mol is calculated, significantly lower than previously
Pack grid for both the p(3x3) and p(4x4) unit cells. The reported in literature [34]. Table 2 summarizes the CO2
adsorption energy for an adsorbate on the catalyst surface is dissociation reaction and comparison is made with other
calculated as catalyst surfaces. The Ni(111) surface shows a similar
𝐸𝑎𝑑𝑠 = 𝐸𝑎𝑑𝑠𝑜𝑟𝑏𝑎𝑡𝑒+𝑠𝑙𝑎𝑏 − (𝐸𝑠𝑙𝑎𝑏 + 𝐸𝑎𝑑𝑠𝑜𝑟𝑏𝑎𝑡𝑒(𝑔) ) (8) capability of CO2 activation, but one can conclude that the
Co(111) surface is the preferred catalyst for activation via
With 𝐸𝑎𝑑𝑠 the adsorption energy of the adsorbate in kJ/mol, direct CO2 dissociation.
𝐸𝑎𝑑𝑠𝑜𝑟𝑏𝑎𝑡𝑒+𝑠𝑙𝑎𝑏 the total electronic energy of the adsorbate on
the slab, 𝐸𝑠𝑙𝑎𝑏 the electronic energy of the slab and 250 HCOO* pathway
𝐸𝑎𝑑𝑠𝑜𝑟𝑏𝑎𝑡𝑒(𝑔) the electronic energy of the adsorbate in vacuum COOH* pathway
186

[25]. Transition state calculations are performed using the 200 dissociation
reactions
climbing image Nudged Elastic Band (cNEB) and refined 154
with the dimer method [26, 27]. The thermodynamic stability 150
Energy (kJ/mol)

of the adsorbates is evaluated from the Gibbs free energy 61


∆𝐺𝑎𝑑𝑠 (𝑇, 𝑝), which can be calculated from the vibrational, 100 91 63
110 107
translational and rotational partition functions of the 60
HCO*
adsorbates. 50 + OH*
∆𝐺𝑎𝑑𝑠 (𝑇, 𝑝) = ∆𝐺 ° (𝑇) − 𝑅 ∙ 𝑇 ∙ ln(𝑝𝑎𝑑𝑠(𝑔) ) (9) COOH*
+ H* HCOOH*
HCO* +
0 CO * O* + H*
2
B. Microkinetic modeling + 2H*
-50 HCOO* + H*
Microkinetic modeling is performed with the Chemkin
software, which is capable of solving a large set of ordinary CO* + O* + 2H* CO* + OH* + H*
differential equations, representing the reaction mechanism -100
and reactor model. Chemkin makes use of the DASPK solver, Figure 1 Electronic energy profile calculated with the DFT-vdW-
which employs variable-order variable-stepsize backward DF2 funtional for CO2 activation via direct dissociation, and H-
differentiation formulas for integration of the system of assisted pathways via HCOO* and COOH*. The activation barriers
ordinary differential equations. In case of large-scale systems, are indicated and selected transition state structures are shown.
Moreover, hydrogen-assisted pathways for CO2 activation 2.0E-12 0.0006
on Co(111) were studied and illustrated in Figure 1. The

Rate of production
formate (HCOO*) species pathway has a significant
1.0E-12

TOFCO (s-1)
(mol/cm²s)
advantage over the hydrocarboxyl (COOH*) pathway, as it 0.0004
has the lowest barriers and endothermicity. However, the
direct dissociation path is observed to be most beneficial due
0
to its relatively low energy barrier (60 kJ/mol) and favorable 450 550 650 0.0002
Gibbs free energy of reaction (-72 kJ/mol). -1.0E-12
To obtain more realistic catalyst models, effects of 1/3 ML
CO coverage are evaluated for the different catalyst surfaces. -2.0E-12 0
Temperature (K)
The dissociation barriers are observed to increase in the order
of 10-20 kJ/mol, presumably due to lateral interactions with Figure 3 CO consumption TOFCO (s-1) and the rate of production
CO* species on the surface. Analysis of CO2 dissociation on (mol/cm²s) of HCO* formation (purple), COH* formation (blue),
step sites in the Co(211) surface shows high potential for CO2 HCOH* from HCO* (yellow) and HCOH* from COH* (red) in
activation, with an activation energy slightly lower than function of temperature. Maximal TOFCO is observed at 540 K, and
terraces (54 kJ/mol) and a favorable Gibbs free energy of the preferred pathway goes through HCO* formation.
reaction (-121 kJ/mol). However, due to their low to be dominant for propagation, and the relative importance of
thermodynamic stability and high affinity for reaction CO insertion and C-C coupling is summarized in Table 3, for
intermediates at increased temperature, step sites are expected a base case of 500 K, 20 bar and H2/CO = 2.
not to be available for CO2 activation [35].
Table 3 Rates of CO insertion and C-C coupling (s-1) for the base
case (500 K, 20 bar, H2/CO = 2).
B. Microkinetic modeling
Debate is ongoing on the dominant pathway for CO Rate of CO Rate of C-C
RC-C/RCO
activation and chain propagation in FT synthesis and two insertion (s-1) coupling (s-1)
important mechanisms are proposed: the carbide mechanism, 1.31E-4 -1.85E-6 1.41E-2
suggesting coupling of CHx species [36], and the CO insertion
CO insertion is observed to mainly go through CO* and
mechanism, suggesting CHx and CO coupling [37]. In order to
CH2* coupling as this step has the lowest barrier (64 kJ/mol).
ensure fast chain growth for the carbide mechanism, high
By modification of the energy barriers in the activation
surface coverages of CHx should be present, which requires a
mechanism, the CH* coverage is varied and for sufficiently
sufficiently fast CO dissociation. Calculated C-O dissociation
barriers on terraces however are found to be approximately high CH* coverages, the carbide mechanism is found to be the
dominant propagation mechanism. Meaning the CH*
200 kJ/mol, suggesting a low CHx coverage [38].
coverage for the base case is too low for the carbide
A first model was drafted for the evaluation of CO
mechanism to be competitive with CO insertion.
activation under FT conditions (500 K, 20 bar), taking in
Temperature has a similar effect on the propagation
account the main reactions from both mechanisms. The
mechanism as for activation. CO* is observed to be removed
dominant pathway for CO activation is observed to go through
from the surface and H* coverage is maximal for high
hydrogenation of CO* with OH* to obtain COH* species and
temperature. At intermediate temperatures a maximal OH*
consecutive hydrogenation with H* to HCOH*, which
coverage is observed, causing TOFCO to be maximal (see
dissociates to form CH* on the surface (Figure 2). The DFT
Figure 4), linked to the increased initiation at high OH*
calculated activation energies are lowest, explaining the
preference for this pathway. coverage. A maximal TOFCO of 0.16 s-1 is obtained, close to
experimentally determined values [35]. Model predictions
HCO* show an increased selectivity for CH4 and C2H6 species, due
to the higher H* coverage favoring hydrogenation at higher
CO(g) CO* HCOH* CH*
temperatures. Different H2/CO partial pressure ratios are
evaluated for its influence on the propagation mechanism.
COH*
Increased ratios yield a higher H* coverage, which enhances
Figure 2 Dominant pathways for the activation of CO under FT the CH4 formation and selectivity. Rather limited effects are
conditions on the Co(111) catalyst surface. further encountered from variation of the H2/CO ratio.
The effect of OH* coverage is examined via addition of 0.20 2.5E-05
H2O to the feed. At low H2O partial pressure the CO
CH4 formation

2.0E-05
TOFCO (s-1)

consumption turnover frequency TOCCO (s-1) is observed to be 0.15


(mol/s)

enhanced, but is reduced for pH2O above 0.02. The dominant 0.10 1.5E-05
pathway is shifted to consecutive hydrogenation with H* and 1.0E-05
OH* through HCO*. Increasing temperature leads to a 0.05
0.5E-05
removal of CO* from the surface and a high H* coverage. 0
This favors hydrogenation steps with H*, and the dominant 0
pathway is observed to go through HCO*, but due to the low 500 600 700 800 900 1000
barrier HCOH* is formed from hydrogenation with OH*. The Temperature (K)
maximal TOFCO is observed at 540 K, illustrated in Figure 3, Figure 4 CO consumption TOFCO (s-1) and CH4 formation (mol/s) in
explained by a maximal OH* coverage. function of temperature for the chain propagation mechanism (20
Extension of the reaction mechanism to chain propagation bar; H2/CO = 2).
up to C2 yields a high selectivity (99%) for ethylene, and A preliminary analysis is performed for CO2 activation, by
chain growth probability α of 0.5. CO insertion is calculated incorporation of the DFT calculated activation steps for CO2
in the initiation mechanism of CO. Similar process conditions Carbon for the Production of Light Olefins. ChemSusChem, 2011.
are used as for CO activation, with CO2/H2=2. As the CO2 4(9): p. 1265-1273.
7. Multi-Year Research, Development, and Demonstration Plan,
dissociation barrier is minimal, a high selectivity for CO is U.S.D.o. Energy, Editor. 2015. p. 1-44.
calculated. Further, hydrogen-assisted activation is preferred 8. Saur, G., Wind-to-hydrogen project: electrolyzer capital cost
to go through formate species. The formation of CH4 was study. 2008: Citeseer.
9. Ainscough, C.P., D.; Miller, E., Hydrogen Production Cost From
observed to be higher than for CO activation, denoting the
PEM Electrolysis, D.o. Energy, Editor. 2014: United States of
potential for CO2 activation. America. p. 11.
10. Olah, G.A., G.K.S. Prakash, and A. Goeppert, Anthropogenic
V. CONCLUSION Chemical Carbon Cycle for a Sustainable Future. Journal of the
American Chemical Society, 2011. 133(33): p. 12881-12898.
A techno-economic analysis has been performed for H2 11. Aresta, M., A. Dibenedetto, and A. Angelini, The changing
production via electrolysis of water, methanol and ethylene paradigm in CO2 utilization. Journal of Co2 Utilization, 2013. 3-
synthesis from CO2 and electrochemical reduction to formic 4: p. 65-73.
12. Centi, G., E.A. Quadrelli, and S. Perathoner, Catalysis for CO2
acid. Hydrogen production is currently restrained by too high conversion: a key technology for rapid introduction of renewable
renewable electricity cost, but future production costs for energy in the value chain of chemical industries. Energy &
PEM electrolysis are estimated <4 €/kg, reportedly sufficient Environmental Science, 2013. 6(6): p. 1711-1731.
13. Tremel, A., et al., Techno-economic analysis for the synthesis of
to be competitive with fossil fuels. Methanol synthesis is
liquid and gaseous fuels based on hydrogen production via
favorable for its compatibility with the existing infrastructure, electrolysis. International Journal of Hydrogen Energy, 2015.
but high costs for hydrogen production require increasing 40(35): p. 11457-11464.
energy efficiency. Ethylene production from a modified 14. Mignard, D., et al., Methanol synthesis from flue-gas CO2 and
Fischer-Tropsch process is found to be most economically, renewable electricity: a feasibility study. International Journal of
Hydrogen Energy, 2003. 28(4): p. 455-464.
although significantly more expensive than commercial 15. Methanex posts regional contract methanol prices for North
ethylene. Capital costs and electricity feedstock are the main America, Europe and Asia. 2015 7/2/2016 [cited 2016 8/2/2016];
cost contributors. Formic acid production is currently much Available from: https://www.methanex.com/our-business/pricing.
larger than consumption, but the future market is expected to 16. Ren, D., et al., Selective Electrochemical Reduction of Carbon
Dioxide to Ethylene and Ethanol on Copper (I) Oxide Catalysts.
grow. The production cost for renewable formic acid is ACS Catalysis, 2015. 5(5): p. 2814-2821.
however twice the commercial price and is mainly determined 17. Graves, C., et al., Sustainable hydrocarbon fuels by recycling CO2
by the electricity and capital cost. and H2O with renewable or nuclear energy. Renewable and
Sustainable Energy Reviews, 2011. 15(1): p. 1-23.
DFT calculations show CO2 dissociation is favorable on
18. Sun, X., Catalytic Conversion of Methanol to Olefins over HZSM-
Co(111), but adsorption of CO2 is thermodynamically un- 5 Catalysts. 2013, Universität München.
stable. Hydrogen-assisted pathways mainly go through 19. Lu, X., et al., Electrochemical reduction of carbon dioxide to
formate species, but are higher activated than the direct formic acid. ChemElectroChem, 2014. 1(5): p. 836-849.
dissociation. Coverage effects were observed to increase the 20. Afshar, A.A.N., Chemical profile: Formic acid. 2014, TranTech
Consultants Inc. p. 3.
dissociation barriers, presumably due to lateral interactions, 21. Morrow, P.-S., Contact Magnetoresistance of Multilayered
and step sites on the Co(211) surface are shown to have a high Cobalt/copper Nanostructures Measured by Scanning Tunneling
potential for CO2 activation. Comparison with Cu(111), Microscope. 2008: ProQuest.
Ni(111) and Au(111) indicates cobalt is the preferred catalyst 22. Yoder, B.L., Steric Effects in the Chemisorption of Vibrationally
Excited Methane on Nickel. 2012: Springer Science & Business
for CO2 activation. Media.
Modeling of the activation of CO under FT conditions 23. Kresse, G.M., Martijn; Furthmüller, Jürgen. VASP the Guide. 2015
shows H- and OH-assisted activation is preferred over direct [cited 2015; Available from:
http://cms.mpi.univie.ac.at/vasp/vasp/vasp.html.
dissociation, and the CO insertion mechanism is dominant
24. Lee, K., et al., Higher-accuracy van der Waals density functional.
over the carbide mechanism. Too low coverage of CH* and Physical Review B, 2010. 82(8): p. 081101.
CH2* species compared to high CO* coverages limit the 25. Zhao, Y.-F., et al., Insight into methanol synthesis from CO 2
competitive behavior of the carbide mechanism. Simulations hydrogenation on Cu (111): complex reaction network and the
show a high selectivity for ethylene, with increased CH4 and effects of H 2 O. Journal of Catalysis, 2011. 281(2): p. 199-211.
26. Henkelman, G., B.P. Uberuaga, and H. Jónsson, A climbing image
C2H6 selectivity for high H* coverages. Maximal TOFCO are nudged elastic band method for finding saddle points and
observed at high OH* coverage, denoting its importance in minimum energy paths. The Journal of chemical physics, 2000.
the mechanism. A preliminary analysis for CO2 activation 113(22): p. 9901-9904.
shows a high CH4 formation, denoting the potential for CO2 27. The dimer method. [cited 2016; Available from:
http://theory.cm.utexas.edu/vtsttools/dimer.html#dimer.
activation on Co(111) catalysts. 28. Li, S. and L. Petzold, Design of new DASPK for sensitivity
analysis. University of California at Santa Barbara, Santa Barbara,
REFERENCES CA, 1999.
29. Design, R., CHEMKIN Theory Manual. San Diego, CA, 2007.
1. Aresta, M. and A. Dibenedetto, Utilisation of CO2 as a chemical 30. Kee, R.J., J.A. Miller, and T.H. Jefferson, CHEMKIN: A general-
feedstock: opportunities and challenges. Dalton Transactions, purpose, problem-independent, transportable, FORTRAN
2007(28): p. 2975-2992. chemical kinetics code package. 1980, Sandia Labs.
2. Oberthür, S. and H.E. Ott, The Kyoto Protocol: international 31. Getting stared with CHEMKIN, R. Design, Editor. 2015.
climate policy for the 21st century. 1999: Springer Science & 32. Storsæter, S., D. Chen, and A. Holmen, Microkinetic modelling of
Business Media. the formation of C 1 and C 2 products in the Fischer–Tropsch
3. Commission, E., The Paris Protocol – A blueprint for tackling synthesis over cobalt catalysts. Surface Science, 2006. 600(10): p.
global climate change beyond 2020, E. Union, Editor. 2015: Paris. 2051-2063.
4. Annual Energy Outlook 2015 with projections to 2040, U.S.E.I. 33. Chemistry, R.S.o., Process and Chemical Engineering. 1996:
Administration, Editor. 2015, U.S. Department of Energy: Royal Society of Chemistry.
Washington. 34. Iglesia, E., Design, synthesis, and use of cobalt-based Fischer-
5. Veritas, D.N., Electrochemical Conversion of CO2–Opportunities Tropsch synthesis catalysts. Applied Catalysis A: General, 1997.
and Challenge. 2011, Norway. 161(1–2): p. 59-78.
6. Centi, G., G. Iaquaniello, and S. Perathoner, Can We Afford to
Waste Carbon Dioxide? Carbon Dioxide as a Valuable Source of
35. Gunasooriya, G.K.K., et al., Key role of surface hydroxyl groups
in CO activation during Fischer-Tropsch synthesis. ACS
Catalysis, 2016.
36. Fischer, F. and H. Tropsch, The preparation of synthetic oil
mixtures (synthol) from carbon monoxide and hydrogen.
Brennstoff-Chem, 1923. 4: p. 276-285.
37. Pichler, H. and H. Schulz, Recent results in synthesis of
hydrocarbons from CO and H2. Chemie Ingenieur Technik, 1970.
42(18): p. 1162-&.
38. Zhuo, M., et al., Density functional theory study of the CO
insertion mechanism for Fischer− Tropsch synthesis over Co
catalysts. The Journal of Physical Chemistry C, 2009. 113(19): p.
8357-8365.
I

Table of Contents
List of figures ............................................................................................................................................ IV

List of tables ............................................................................................................................................. XII

Chapter 1 Introduction ................................................................................................................................. 1

1.1. Techno-economic analysis of renewable pathways for CO2 conversion. ............................................. 2


1.2. Ab-initio calculations ........................................................................................................................ 5
1.3. Microkinetic modeling ....................................................................................................................... 6
1.4. Thesis structure ................................................................................................................................. 6
1.5. References ......................................................................................................................................... 7

Chapter 2 Techno-economic analysis of renewable pathways for CO2 conversion. ....................................... 9

2.1. CO2 conversion: a challenging opportunity .......................................................................................10


2.2. Shift to renewable energy/Electricity generation ...............................................................................13
2.3. Hydrogen production ........................................................................................................................16
2.3.1. Biomass gasification and pyrolysis .............................................................................................16
2.3.2. Thermolysis via solar driven thermochemistry ............................................................................17
2.3.3. Photolysis via photelectrochemistry and photobiology ................................................................18
2.3.4. Electrolysis ................................................................................................................................20
2.3.5. Hydrogen Purification and Enrichment .......................................................................................31
2.3.6. Relevance of hydrogen production footprint ...............................................................................31
2.3.7. Hydrogen as an intermediate step towards energy storage ...........................................................32
2.4. CO2 capture and sequestration ..........................................................................................................33
2.5. Liquid fuel and raw chemical production ..........................................................................................37
2.5.1. Methanol synthesis .....................................................................................................................37
2.5.2. Ethylene production ...................................................................................................................45
2.5.3. Electrochemical reduction to formic acid ....................................................................................55
2.6. Conclusion .......................................................................................................................................60
2.7. References ........................................................................................................................................62

Chapter 3 Methods and models ...................................................................................................................68

3.1. Density Functional Theory................................................................................................................69


3.2. Catalyst models ................................................................................................................................69
3.2.1. Cobalt catalyst models ................................................................................................................69
3.2.2. Copper catalyst model ................................................................................................................70
3.2.3. Nickel catalyst model .................................................................................................................70
3.2.4. Gold catalyst model....................................................................................................................71
II

3.3. VASP calculations ............................................................................................................................71


3.3.1. Adsorption energy ......................................................................................................................71
3.3.2. Transition state calculation .........................................................................................................72
3.3.3. Gibbs free energy .......................................................................................................................74
3.4. Chemkin ...........................................................................................................................................76
3.5. Implementation of a Chemkin project ...............................................................................................77
3.5.1. Thermodynamic data file ............................................................................................................77
3.5.2. Gas phase kinetics file ................................................................................................................78
3.5.3. Surface kinetics file ....................................................................................................................79
3.6. Calculation of kinetic parameters ......................................................................................................80
3.7. Chemical reaction rate expressions ...................................................................................................81
3.8. Chemkin simulation..........................................................................................................................81
3.9. Turnover frequency and selectivity ...................................................................................................82
3.10. Sensitivity analysis .........................................................................................................................82
3.11. Challenges involving the set-up of a Chemkin project .....................................................................83
3.12. References ......................................................................................................................................85

Chapter 4 Ab-initio calculations ..................................................................................................................87

4.1. Analysis of the cobalt catalytic surface .............................................................................................88


4.1.1. Adsorption of reaction intermediates on Co(111) ........................................................................88
4.1.2. Reaction mechanism for activation of CO2 on Co(111) ...............................................................90
4.1.3. Coverage effects on Co(111) ......................................................................................................93
4.1.4. Effect of step sites on Co(211) ....................................................................................................95
4.2. Analysis of the Cu(111) surface ........................................................................................................97
4.2.1. Adsorption of reaction intermediates on Cu(111) ........................................................................97
4.2.2. Coverage effects on Cu(111) .................................................................................................... 100
4.3. Analysis of the Ni(111) surface....................................................................................................... 102
4.3.1. Adsorption of reaction intermediates on Ni(111) ...................................................................... 102
4.3.2. Coverage effects on Ni(111) ..................................................................................................... 105
4.4. Analysis of the Au(111) surface ...................................................................................................... 107
4.4.1. Adsorption of reaction intermediates on Au(111) ...................................................................... 107
4.4.2. Coverage effects on Au(111) .................................................................................................... 110
4.5. Conclusion ..................................................................................................................................... 112
4.6. References ...................................................................................................................................... 113

Chapter 5 Microkinetic modeling .............................................................................................................. 115

5.1. Fischer-Tropsch synthesis ............................................................................................................... 116


5.1.1. CO activation ........................................................................................................................... 118
5.1.2. Chain propagation reactions ..................................................................................................... 124
III

5.2. CO2 activation ................................................................................................................................ 133


5.3. Conclusion ..................................................................................................................................... 134
5.4. References ...................................................................................................................................... 136

Chapter 6 Conclusions and future work ..................................................................................................... 138

6.1. Techno-economic analysis .............................................................................................................. 139


6.2. Ab-initio calculations ..................................................................................................................... 140
6.3. Micro-kinetic modeling .................................................................................................................. 141
6.4. Future work .................................................................................................................................... 142

Appendix A Techno-economic analysis of renewable pathways for CO2 conversion. ................................. 143

A.1 References ...................................................................................................................................... 151

Appendix B Ab-initio calculations ............................................................................................................ 152

Appendix C Kinetic modeling ................................................................................................................... 172

C.1 References ...................................................................................................................................... 174

Appendix D List of simulations................................................................................................................. 175


IV

List of figures
Figure 1-1 Different pathways for CO2 utilization [8]. .................................................................................. 2

Figure 1-2 Closed reaction cycle for the conversion of CO2 to fuels and chemicals. Alternative conversion
processes attempt to provide a similar, but accelerated pathway................................................... 4

Figure 1-3 Comparison of the reaction path for dissociation of CO 2* to O* and CO* adsorbed on a free
Co(111) and Cu(111) surface. The different catalyst morphology and properties affect the
adsorption energy and activation barrier for dissociation; and hence, its capability for activation of
CO2. ............................................................................................................................................ 5

Figure 2-1 Left: Mean carbon dioxide concentration measured at the Mauna Loa Observatory, Hawaii [7].
The red line represents the monthly atmospheric CO2 concentration, the black line represents the
yearly average. Right: schematic representation of the perturbation of the global carbon cycle
caused by human activities, averaged globally for the decade 2004- 2013 [7]. ............................10

Figure 2-2 Two mirror-based approaches for focusing sunlight on a thermochemical reactor to produce
temperatures up to 2000°C. (a) A field of heliostat mirrors concentrates sunlight onto a central
reactor tower; (b) dish mirrors focus sunlight onto an attached reactor module. The solar-
generated high-temperature heat can be used to drive thermochemical reactions that produce
hydrogen [47]. ...........................................................................................................................18

Figure 2-3 Two different approaches to PEC solar hydrogen production reactors. (a) Electrode systems
similar to flat-plate photovoltaic panels; (b) particle systems comprised of slurries of PEC
semiconductor particles [50]. .....................................................................................................19

Figure 2-4 Block diagram of the hydrogen production system design. Process water and power are fed to the
electrolyzer stacks in which water is split into oxygen and hydrogen gas [54]. ............................20

Figure 2-5 Simplified representation of an alkaline electrolysis cell. The cell consists of two electrodes,
separated by a diaphragm, immersed in an alkaline electrolyte liquid composed of a solution of
20-30% KOH [58]......................................................................................................................22

Figure 2-6 Schematic representation of the polymer electrolyte membrane electrolysis cell. The cell consists
of two electrodes separated by a membrane that conducts protons from the anode to the cathode
while insulating the electrodes electrically [63]. .........................................................................24

Figure 2-7 Sensitivity analysis on the hydrogen production cost (€/kg H2) for a low, baseline and high value of
the average electricity price over the plant lifetime (c/kWh). Results are shown for the central
V

distributed production (top left); future distributed production (top right), current central
production (bottom left) and future central production (bottom right). Sensitivity limits are taken
as ±50% of the baseline value [54]. ............................................................................................28

Figure 2-8 Sensitivity analysis on the hydrogen production cost (€/kg H2) for a low, baseline and high value of
the uninstalled capital cost (€/kW). Results are shown for the central distributed production (top
left); future distributed production (top right), current central production (bottom left) and future
central production (bottom right) [54]. .......................................................................................29

Figure 2-9 Most important separation techniques for carbon capture [31]. ...................................................33

Figure 2-10 Proposed pathways for the Copper-catalyzed hydrogenation of CO2 to methanol [6, 79]. ..........39

Figure 2-11 Block diagram of the CO2 to methanol conversion process. [22] ...............................................40

Figure 2-12 World market for ethylene and propylene with a breakdown of the main production methods.
Values are given for 2010 and prospect for 2020. Distinction is made between steam cracking,
fluid catalytic cracking (FCC), dehydrogenation from ethane or propane, oxidative
dehydrogenation (ODH), production from syngas and others [12]...............................................46

Figure 2-13 Schematic representation of the changes in the reaction rates of key reaction steps as a function
of metal-carbon (M-C) interaction energies [17]. ........................................................................48

Figure 2-14 Left: illustration of a consecutive pathway involving methylation, cracking and aromatization
reactions [93]. Right: general illustration of the parallel pathway (with hydrocarbon pool) [91]. .49

Figure 2-15 Framework topologies of HZSM-5 (left) and SAPO-34 (right) zeolites [93]. ............................50

Figure 2-16 Left: UOP/INEOS/Total MTO process with integration of an OCP process [95]. Right:
DICP/SINOPEC/SYN Energy Technology MTO process with recycle of C4+ stream [93]. ........50

Figure 2-17 Comparison of the return on investment (ROI) of a naphtha cracker and MTO plant for a range
of methanol feedstock prices [94]. ..............................................................................................51

Figure 2-18 Flowsheet for the MTO process. [1] represents the reactor, [2] to [9] represents the separation
section, [10] is the cracking section of the UOP/Hydro MTO process [97]. .................................52

Figure 2-19 Schematic representation of the modified Fischer-Tropsch process. CO2 is captured and
subsequently dissociated with an appropriate amount of H2O to produce syngas (CO + H2) with an
appropriate composition for FT synthesis [34]. ...........................................................................53
VI

Figure 2-20 Mechanism of CO2 electrochemical reduction on an electrolytic cell cathode [102]. (a)
Adsorption and electronation of CO2 on the cathode surface. (b) Reaction of adsorbed CO2- and
H2O. (c) Electronation of HCO2- on the surface. (d) Desorption of the HCO2- ion. ......................56

Figure 2-21 Basic structure of a CO2 electroreduction electrolytic cell [102]. At the cathode CO2 is reduced
to the desired products. The accompanying oxidation reaction occurs at the anode. ....................56

Figure 2-22 Conceptual flowsheet for the electrochemical reduction of CO2 process [104]. CO2 is reduced
with water in the electrolytic cell. Produced formic acid is separated and the remaining stream is
recycled. ....................................................................................................................................58

Figure 3-1: Top view of the different unit cells for the Co catalyst: left: p(3x3) Co(111), right: p(4x4)
Co(211). ....................................................................................................................................69

Figure 3-2: Top view of the unit cell for the Cu catalyst: p(3x3) Cu(111). ...................................................70

Figure 3-3: Top view of the unit cell for the Ni catalyst: p(3x3) Ni(111)......................................................70

Figure 3-4: Top view of the unit cell for the Au catalyst: p(3x3) Au(111). ...................................................71

Figure 3-5 Left: possible adsorption sites on a closed pack fcc (111) catalyst surface (T-Top, B-Bridge, F-
Hollow-fcc, H-Hollow-hcp); right: 5 layers p(3x3) model slab in the z-direction with interslab
spacing of 15 Å, the top three layers are relaxed while the bottom two layers are constrained at the
DFT optimized bulk positions [7]. ..............................................................................................72

Figure 3-6 Illustration of the climbing image NEB method (cNEB). A constrained energy optimization is
obtained by the addition of spring forces to maintain equal spacing to neighboring images. ........73

Figure 3-7 Possible vibrational modes for an adsorbed CO molecule on the catalyst surface: 1) C-O
stretching, 2) Co-C stretching, 3) frustrated rotation, 4) frustrated translation [7]. .......................74

Figure 4-1 Optimized adsorption configuration for (a) CO* and (b) CO2* on a clean Co(111) surface. ........88

Figure 4-2 Optimized adsorption configuration for (a) H* (fcc) and (b) OH* (fcc) on a clean Co(111)
surface. ......................................................................................................................................90

Figure 4-3 Energy profile for the dissociation of carbon dioxide on a clean Co(111) surface (CO 2* + * 
CO* + O*). An energy barrier of 60 kJ/mol and Gibbs free energy of reaction of -72 kJ/mol is
obtained. Detailed visualization of the transition state is given in Table B-3 in Appendix B. .......90
VII

Figure 4-4 Energy profile for the dissociation of carbon monoxide on a clean Co(111) surface (CO* + * 
C* + O*). An energy barrier of 243 kJ/mol and Gibbs free energy of reaction of 94 kJ/mol is
obtained. Detailed visualization of the transition state is given in Table B-3 in Appendix B. .......91

Figure 4-5 Electronic energy profile calculated with the DFT-vdW-DF2 functional for the activation of CO 2
via (a) direct dissociation; (b) hydrogen-assisted pathway via HCOO*; and (c) hydrogen-assisted
pathway via COOH*. The activation barriers are indicated and selected transition state structures
are shown, a complete overview of the transition state structures is given in Appendix B Table
B-3. ...........................................................................................................................................93

Figure 4-6 Optimized adsorption configuration for (a) CO* and (b) CO 2* on a Co(111) surface with CO
coverage of 1/3 ML. ...................................................................................................................94

Figure 4-7 Energy profile for the dissociation of carbon dioxide on a Co(111) surface with CO coverage of
1/3 ML (CO2* + *  CO* + O*). An energy barrier of 81 kJ/mol and Gibbs free energy of
reaction of -55 kJ/mol is obtained. Detailed visualization of the transition state is given in Table
B-6 in Appendix B. ....................................................................................................................94

Figure 4-8 Optimized adsorption configuration for (a) CO* and (b) CO2* on a clean Co(211) surface. ........96

Figure 4-9 Energy profile for the dissociation of carbon dioxide on a clean Co(211) surface (CO2* + * 
CO* + O*). An energy barrier of 54 kJ/mol and Gibbs free energy of reaction of -121 kJ/mol is
obtained. Detailed visualization of the transition state is given in Table B-9 in Appendix B. .......96

Figure 4-10 Optimized adsorption configuration for (a) CO* and (b) CO2* on a clean Cu(111) surface. ......97

Figure 4-11 Energy profile for the dissociation of carbon dioxide on a clean Cu(111) surface (CO 2* + * 
CO* + O*). An energy barrier of 159 kJ/mol and Gibbs free energy of reaction of 100 kJ/mol is
obtained. Detailed visualization of the transition state is given in Table B-12 in Appendix B. .....99

Figure 4-12 Energy profile for the dissociation of carbon monoxide on a clean Cu(111) surface (CO* + * 
C* + O*). An energy barrier of 546 kJ/mol and Gibbs free energy of reaction of 283 kJ/mol is
obtained. Detailed visualization of the transition state is given in Table B-12 in Appendix B. .....99

Figure 4-13 Optimized adsorption configuration for (a) CO* and (b) CO 2* on a Cu(111) surface with CO
coverage of 1/3 ML. ................................................................................................................. 100

Figure 4-14 Energy profile for the dissociation of carbon dioxide on a Cu(111) surface with CO coverage of
1/3 ML (CO2* + *  CO* + O*). An energy barrier of 156 kJ/mol and Gibbs free energy of
VIII

reaction of 89 kJ/mol is obtained. Detailed visualization of the transition state is given in Table
B-15 in Appendix B. ................................................................................................................ 101

Figure 4-15 Optimized adsorption configuration for (a) CO* and (b) CO2* on a clean Ni(111) surface. ..... 102

Figure 4-16 Energy profile for the dissociation of carbon dioxide on a clean Ni(111) surface (CO 2* + * 
CO* + O*). An energy barrier of 79 kJ/mol and Gibbs free energy of reaction of -43 kJ/mol is
obtained. Detailed visualization of the transition state is given in Table B-18 in Appendix B. ... 104

Figure 4-17 Energy profile for the dissociation of carbon monoxide on a clean Ni(111) surface (CO* + * 
C* + O*). An energy barrier of 289 kJ/mol and Gibbs free energy of reaction of 140 kJ/mol is
obtained. Detailed visualization of the transition state is given in Table B-18 in Appendix B. ... 104

Figure 4-18 Optimized adsorption configuration for (a) CO* and (b) CO2* on a Ni(111) surface with CO
coverage of 1/3 ML. ................................................................................................................. 105

Figure 4-19 Energy profile for the dissociation of carbon dioxide on a Ni(111) surface with CO coverage of
1/3 ML (CO2* + *  CO* + O*). An energy barrier of 95 kJ/mol and Gibbs free energy of
reaction of -18 kJ/mol is obtained. Detailed visualization of the transition state is given in Table
B-21 in Appendix B. ................................................................................................................ 106

Figure 4-20 Optimized adsorption configuration for (a) CO and (b) CO2 on a clean Au(111) surface. ........ 107

Figure 4-21 Energy profile for the dissociation of carbon dioxide on a clean Au(111) surface (CO 2* + * 
CO* + O*). An energy barrier of 310 kJ/mol and Gibbs free energy of reaction of 240 kJ/mol is
obtained. Detailed visualization of the transition state is given in Table B-24 in Appendix B. ... 109

Figure 4-22 Energy profile for the dissociation of carbon monoxide on a clean Au(111) surface (CO* + * 
C* + O*). An energy barrier of 447 kJ/mol and Gibbs free energy of reaction of 414 kJ/mol is
obtained. Detailed visualization of the transition state is given in Table B-24 in Appendix B. ... 109

Figure 4-23 Optimized adsorption configuration for (a) CO* and (b) CO 2* on a Au(111) surface with CO
coverage of 1/3 ML. ................................................................................................................. 110

Figure 4-24 Energy profile for the dissociation of carbon dioxide on a Au(111) surface with CO coverage of
1/3 ML (CO2* + *  CO* + O*). An energy barrier of 314 kJ/mol and Gibbs free energy of
reaction of 234 kJ/mol is obtained. Detailed visualization of the transition state is given in
Appendix B. ............................................................................................................................. 111
IX

Figure 5-1 Reaction scheme of Fischer-Tropsch synthesis according to the carbide mechanism [9]. CO is
adsorbed and dissociates to form carbide species on the surface. Hydrogenation yields methane
and CHx*, proposedly the monomer species for chain growth. .................................................. 116

Figure 5-2 Proposed propagation reaction paths for the CO insertion mechanism by Saeys et al. [12]. The full
arrows indicate the dominant reaction path and the dotted arrows the minor reaction paths. R
represents hydrogen or an alkyl group. ..................................................................................... 117

Figure 5-3 Reaction paths for the activation of CO* on the cobalt catalytic surface; OH-assisted activation is
the dominant pathway and CH* formation preferably goes through COH* and HCOH* formation.
The dominant pathway is indicated by the bold lines. ............................................................... 120

Figure 5-4 Surface coverages of the reaction intermediates in function of the partial pressure of water at 500
K and 20 bar, on a logarithmic scale and detailed linear scale for C*, H*, OH*, and O*. CO*, H*
and OH* are the main surface species. A sharp increase for the OH* and O* coverage is observed
already at low water partial pressures. ...................................................................................... 121

Figure 5-5 (a) Simulated CO consumption rate (s-1) as a function of the partial pressure of H2O at 500 K and
20 bar. Increased partial pressure initially raises the TOF, but stagnates around 0.01-0.02;
maximum TOF is observed at a partial pressure of approximately 0.01. (b) CO consumption
turnover rate (h-1) in function of H2O pressure for FTS on a Ru catalyst at 463 K and 2.9 MPa
[17]. ......................................................................................................................................... 121

Figure 5-6 H2O partial pressure dependency for the rate of production (mol/cm²s) of the COH* species (blue
line), HCO* species (purple line), COH* hydrogenation to HCOH* (red dashed line), and HCO*
hydrogenation to HCOH* (yellow dashed line). The pathway through HCO* is observed to be
dominant, as the COH* formation is blocked due to the high O* coverage. .............................. 122

Figure 5-7 Temperature dependency for the coverage of the surface species over a range of 200 K,
represented on a logarithmic scale with inclusion of detailed linear scale for the coverages of the
most important intermediates. At low temperatures the surface is dominated by CO* species, but
at approximately 510 K, H* becomes the main surface species. OH* coverage increases up to 520
K, but is reduced at higher temperatures. .................................................................................. 123

Figure 5-8 (a) CO consumption turnover frequency (s-1) in function of temperature. (b) Temperature
dependency of the rate of production (mol/cm²s) of the COH* species (blue line), HCO* species
(purple line), COH* hydrogenation to HCOH* (red dashed line), and HCO* hydrogenation to
HCOH* (yellow dashed line). Both the COH* formation and its further hydrogenation, and the
HCO* formation and its further hydrogenation are highly dependent. A shift is observed around
520 K from COH* formation to HCO* formation..................................................................... 124
X

Figure 5-9 Reaction path for chain propagation in FTS on Ru(11𝟐1) (T = 500 K, p = 20 bar, H2/CO = 2),
proposed by Van Santen et al. [19]. The dominant pathway goes through coupling with CH*
species. .................................................................................................................................... 125

Figure 5-10 Representation of the pathways for propagation in FTS up to C 2 species, simulated with the
reactions presented in Table C-1 (500 K, 20 bar and H2/CO = 2). The dominant pathway forms
C2H4 and is indicated by the bold lines. .................................................................................... 126

Figure 5-11 Rate of production for the propagation reactions in the CO insertion mechanism (red line) and
carbide mechanism (blue line) in function of the CH* coverage on a logarithmic scale. ............ 128

Figure 5-12 Comparison of the selectivity ratios CH4/C2H4 (orange line), C2H6/C2H4 (dark blue line),
CH3HCO/C2H4 (purple line), and C2H5OH/C2H4 (light blue line) in function of CH* coverage on
log scale. .................................................................................................................................. 128

Figure 5-13 Temperature dependency of the coverage of the surface species, represented on a logarithmic
scale with inclusion of detailed linear scale for the coverages of the most important intermediates
(20 bar; H2/CO = 2). Initially CO* is the dominant surface species but a fast decrease of coverage
is observed for increased temperature. At higher temperatures hydrogen is found to be dominant.
For intermediate temperatures, OH* is observed to go through a maximum, as also observed for
the activation mechanism. ........................................................................................................ 129

Figure 5-14 (a) CO consumption TOFCO (s-1) and CH4 formation (mol/s) in function of temperature for the
chain propagation mechanism (20 bar; H2/CO = 2). (b) Rate of production for CH*-CH* coupling
(carbide mechanism, blue line) and CH 2*-CO* coupling (CO insertion, red line) in the chain
propagation mechanism. ........................................................................................................... 130

Figure 5-15 Simulated selectivities for the product species in the chain propagation mechanism in function of
temperature (20 bar; H2/CO = 2). Ethylene is found to be the main product, but for increased
temperature an increased selectivity for methane and ethane is observed, also formaldehyde and
acetaldehyde formation is more pronounced at higher temperatures. ......................................... 130

Figure 5-16 Surface species coverages in function of the H 2/CO partial pressure ratio, represented on a
logarithmic scale with inclusion of detailed linear scale for the coverages of the most important
intermediates (500 K; 20 bar). CO* is the dominant surface species at lower ratios, with a shift
towards H* for ratios above 3.5. ............................................................................................... 131

Figure 5-17 (a) TOFCO (s-1) and CH4 formation (mol/s) in function of H2/CO ratio for the chain propagation
mechanism (500 K; 20 bar). Maximum TOF CO is observed at H2/CO = 3; methane production
increases with increasing H2/CO ratio. (b) Rate of production for CH*-CH* coupling (carbide
XI

mechanism, blue line) and CH2*-CO* coupling (CO insertion, red line) in the chain propagation
mechanism. .............................................................................................................................. 132

Figure 5-18 Simulated selectivities of the product species in the chain propagation mechanism in function of
H2/CO ratio (500 K; 20 bar. Increased ratios has a limited effect on the selectivity. Ethylene is
found to be the main product, and methane selectivity increases for increased H 2/CO ratio. ...... 132

Figure 5-19 (a) CO2 consumption TOF (s-1) in function of temperature and (b) CH4 formation (mol/s) for the
CO activation and CO2 activation mechanism, with CO/H2 and CO2/H2 ratio = 2...................... 134

Figure A-1 Aspen simulation flowsheet of the MTO process. The mass balance is given in Table A-20 [8].
................................................................................................................................................ 149
XII

List of tables
Table 2-1 Construction and electricity generation cost of different energy sources [31]. ..............................15

Table 2-2 Hydrogen production cost breakdown for current (2013) and future (2025), distributed (1500
kgH2/day) and central (50 000 kgH2/day) production facilities. Costs are shown on a €/kg H2 basis
[54]. ...........................................................................................................................................27

Table 2-3 Hydrogen production cost breakdown for current (2015) and future (2020), distributed (1500
kgH2/day) and central (50 000 kgH2/day) production facilities. Costs are shown on a €/kgH 2 basis
[44]. ...........................................................................................................................................29

Table 2-4 Comparison of the cost analyses of alkaline and PEM electrolysis hydrogen production for current
and future, distributed and central production facilities [44, 54]. .................................................30

Table 2-5 CO2 emissions from various industrial sectors in Mt CO2 per year [17]. .......................................33

Table 2-6 Estimated energy required to remove and recover CO2 from coal-fired power plants using various
technologies [31]........................................................................................................................34

Table 2-7 Simplified schedule for the availability of renewable electricity. Three periods are assumed during
a 24h day and renewable electricity is assumed to be available only at off-peak periods. Fossil
electricity is used in periods with no excess supply of renewable energy [40]. ............................41

Table 2-8 Cost breakdown for the capital expenditures studies and summary of the variable operating costs
for each option. Calculations of the variable operating costs assume 8000 working hours per year
and unit costs of €0.74/tsteam, €0.06/tcw and €0.05/kWh (fossil power) [40]. .................................42

Table 2-9 Minimum selling price of methanol produced resulting in a nullified NPV over 15 years for each
design option. Option B shows the most advantageous selling price [40]. ...................................42

Table 2-10 Comparison of option B and C for the implementation of an oxygen plant for the valorization of
O2. Oxygen is assumed to be distributed via 20 km pipelines at 35 bar and is sold at €12.54 per
100 Nm³. Option C is highly favorable, due to its lower operating costs [40]. .............................43

Table 2-11 Estimation of utility cost. Distinction is made between values used for the economic analysis by
Mignard and actual values found in literature [40, 81, 82]. .........................................................44
XIII

Table 2-12 Variable operating costs in €10 6/year for the Aspen simulations performed by Van-Dal with
economic parameters by Mignard and literature for current and prospective cases. Increased
energy efficiency of 15% is assumed for the prospective case [40, 81-83]...................................44

Table 2-13 Methanol price per ton (€/ton) for the different cases starting from the simulations from Van-Dal.
Dependence on the electricity feedstock price is visible by comparing results from Mignard and
current literature values. .............................................................................................................44

Table 2-14 Mass balance and product prices for the main products in the MTO process simulations [97]. ...52

Table 2-15 Key economic results12 of the MTO process for the conventional and renewable methanol
feedstock [97]. ...........................................................................................................................52

Table 2-16 Comparison of the CO2 emissions of the MTO process simulation for the conventional and
renewable methanol feedstock [97]. ...........................................................................................53

Table 2-17 Assumptions for cost estimation12 of the modified Fischer-Tropsch process simulation [34]. .....54

Table 2-18 Mass balance and product prices for the main products in the modified Fischer-Tropsch process
simulations.................................................................................................................................54

Table 2-19 Key economic results12 for the modified Fischer-Tropsch process simulations. ..........................54

Table 2-20 Assumptions for cost estimation of the electrochemical reduction process [100]. .......................58

Table 2-21 Mass balance and product prices for the main products in the electrochemical reduction process
simulations [100]. ......................................................................................................................58

Table 2-22 Key economic results12 for the electrochemical reduction process simulation [100]....................59

Table 3-1 Default representation of a chemical species in the thermodynamic data file in Chemkin. ............77

Table 3-2 Structure of the gas phase kinetics file in Chemkin on the basis of an example for a reaction
mechanism including solely surface reactions: conversion of carbon monoxide towards
hydrocarbons and oxygenates. ....................................................................................................78

Table 3-3 Structure of the surface kinetics file in Chemkin on the basis of an example: CO and H 2 adsorption
on the catalyst surface. For every reaction respectively the pre-exponential factor A, the
temperature dependency β and the activation energy Ea need to be inserted for determination of
the rate coefficients from the Arrhenius equation. .......................................................................79
XIV

Table 3-4 Reaction and inlet conditions for the base case simulations performed in Chemkin. Reaction
conditions are taken similar as for Fischer-Tropsch synthesis [31]. .............................................81

Table 4-1 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch
conditions (500 K, 20 bar, 60% conversion) for the most stable configuration of reaction
intermediates on a clean Co(111) surface. An extensive summary is given in Table B-1 in
Appendix B. ...............................................................................................................................89

Table 4-2 Energy barriers (Ef), entropy of activation (∆Sf), Gibbs free energy barriers (∆G f), entropy of
reaction (∆Sr) and Gibbs free energy of reaction (∆G r) for CO* and CO2* scission reactions and
CO2* activation reactions on a clean Co(111) surface. The accompanying transition state
structures are given in Table B-3 in Appendix B.........................................................................92

Table 4-3 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch
conditions (500 K, 20 bar, 60% conversion) for CO* and CO2* adsorbates on a Co(111) surface
with CO coverage of 1/3 ML. A more elaborated summary is given in Table B-4 in Appendix B.
..................................................................................................................................................93

Table 4-4 Energy barrier (Ef), entropy of activation (∆Sf), Gibbs free energy barrier (∆Gf), entropy of
reaction (∆Sr) and Gibbs free energy of reaction (∆G r) for the CO2* scission reaction on a Co(111)
surface with CO coverage of 1/3 ML. .........................................................................................95

Table 4-5 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch
conditions (500 K, 20 bar, 60% conversion) for CO* and CO2* adsorbates on a clean Co(211)
surface. Accompanying structures are illustrated in Figure 4-8. A more elaborated summary is
given in Table B-4 in Appendix B. .............................................................................................95

Table 4-6 Energy barrier (Ef), entropy of activation (∆Sf), Gibbs free energy barrier (∆Gf), entropy of
reaction (∆Sr) and Gibbs free energy of reaction (∆G r) for the CO2* scission reaction on a clean
Co(211) surface. ........................................................................................................................96

Table 4-7 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch
conditions (500 K, 20 bar, 60% conversion) for the most stable configuration of reaction
intermediates on a clean Cu(111) surface. A more extended summary is given in Appendix B
Table B-10. ................................................................................................................................98

Table 4-8 Energy barriers (Ef), entropy of activation (∆Sf), Gibbs free energy barriers (∆G f), entropy of
reaction (∆Sr) and Gibbs free energy of reaction (∆G r) for CO* and CO2* scission reactions on a
clean Cu(111) surface. The accompanying transition state structures are given in Table B-12 in
Appendix B. ...............................................................................................................................98
XV

Table 4-9 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch
conditions (500 K, 20 bar, 60% conversion) for CO* and CO2* adsorbates on a Cu(111) surface
with CO coverage of 1/3 ML. A more elaborated summary is given in Table B-13 in Appendix B.
................................................................................................................................................ 100

Table 4-10 Energy barrier (Ef), entropy of activation (∆Sf), Gibbs free energy barrier (∆G f), entropy of
reaction (∆Sr) and Gibbs free energy of reaction (∆G r) for the CO2* scission reaction on a Cu(111)
surface with CO coverage of 1/3 ML. ....................................................................................... 101

Table 4-11 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch
conditions (500 K, 20 bar, 60% conversion) for the most stable configuration of reaction
intermediates on a clean Ni(111) surface. A more extended summary is given in Table B-16 in
Appendix B. ............................................................................................................................. 103

Table 4-12 Energy barriers (Ef), entropy of activation (∆Sf), Gibbs free energy barriers (∆Gf), entropy of
reaction (∆Sr) and Gibbs free energy of reaction (∆G r) for CO* and CO2* scission reactions on a
clean Ni(111) surface. The accompanying transition state structures are given in Table B-18 in
Appendix B. ............................................................................................................................. 103

Table 4-13 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch
conditions (500 K, 20 bar, 60% conversion) for CO* and CO 2* adsorbates on a Ni(111) surface
with CO coverage of 1/3 ML. A more elaborated summary is given in Table B-19 in Appendix B.
................................................................................................................................................ 105

Table 4-14 Energy barrier (Ef), entropy of activation (∆Sf), Gibbs free energy barrier (∆G f), entropy of
reaction (∆Sr) and Gibbs free energy of reaction (∆G r) for the CO2* scission reaction on a Ni(111)
surface with CO coverage of 1/3 ML. ....................................................................................... 106

Table 4-15 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch
conditions (500 K, 20 bar, 60% conversion) for the most stable configuration of reaction
intermediates on a clean Au(111) surface. A more extended summary is given in Table B-22 in
Appendix B. ............................................................................................................................. 108

Table 4-16 Energy barriers (Ef), entropy of activation (∆Sf), Gibbs free energy barriers (∆Gf), entropy of
reaction (∆Sr) and Gibbs free energy of reaction (∆Gr) for CO* and CO2* scission reactions on a
clean Au(111) surface. The accompanying transition state structures are given in Table B-24 in
Appendix B. ............................................................................................................................. 108

Table 4-17 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch
conditions (500 K, 20 bar, 60% conversion) for CO* and CO 2* adsorbates on a Au(111) surface
XVI

with CO coverage of 1/3 ML. A more elaborated summary is given in Table B-25 in Appendix B.
................................................................................................................................................ 110

Table 4-18 Energy barrier (Ef), entropy of activation (∆Sf), Gibbs free energy barrier (∆G f), entropy of
reaction (∆S r) and Gibbs free energy of reaction (∆Gr) for the CO2* scission reaction on a
Au(111) surface with CO coverage of 1/3 ML. ......................................................................... 111

Table 5-1 Summary of the reactions taken in account for the activation of CO* on the catalyst surface.
Distinction is made between the direct CO dissociation, the H-assisted and OH-assisted CO
activation pathway. Values are provided by the coach of this thesis. ......................................... 119

Table 5-2 Surface coverages simulated for the surface species at 500 K and 20 bar. .................................. 120

Table 5-3 Dominant reactions involved in the CO insertion mechanism. An extended summary is given in
Appendix C in Table C-1. ........................................................................................................ 125

Table 5-4 Surface coverages simulated for the main surface species at 500 K and 20 bar. .......................... 126

Table 5-5 Rates of CO insertion and C-C coupling (s-1) for the base case (500 K, 20 bar, H2/CO = 2). ....... 126

Table 5-6 Comparison of the net production rate (s-1) for the propagation reactions included in the CO
insertion mechanism. ............................................................................................................... 127

Table 5-7 Comparison of the net production rate (s-1) for the propagation reactions included in the carbide
mechanism. .............................................................................................................................. 127

Table 5-8 Reactions included for the activation of CO2 on Co(111) under Fischer-Tropsch synthesis
conditions. The corresponding DFT calculated activation energies are given in kJ/mol. ............ 133

Table A-1: Cost breakdown for the Distributed Forecourt Production of Hydrogen from Bio-Derived
Renewable Liquids – High Temperature Ethanol Reforming. Plant design capacity is set at 1500
kg/day and a yearly operation of 8500h [1]. .............................................................................. 143

Table A-2: Cost calculation for the Biomass Gasification/Pyrolysis Hydrogen Production. Production
capacity is set at 155 tH2/day. For the current case this requires 2070 t/day of biomass ($75/kg)
and for the future case (2020) this is 2000 t/day of biomass ($63/kg) [1]. .................................. 143

Table A-3: Target for Hydrogen production cost via solar-driven high-temperature thermochemical hydrogen
production. Production capacity is set at 100 000 kgH2/day [1]. ................................................. 144
XVII

Table A-4: Target for Hydrogen production cost via Photoelectrochemical Hydrogen Production:
Photoelectrode system with Solar Concentration. Production capacity is set at 50 000 kg H2/day
[1]. ........................................................................................................................................... 144

Table A-5: Targets for Hydrogen cost via Photolytic Biological Hydrogen Production. Production capacity is
set at 50 000 kgH2/day [1]. ........................................................................................................ 144

Table A-6: Input parameters for H2A Production cases for PEM electrolysis (costs in 2007$ and in 2012$).
[2] ............................................................................................................................................ 145

Table A-7: Hydrogen cost, capital cost, energy efficiency and electricity price for current and future,
distributed production facilities using alkaline electrolysis systems [1]. .................................... 145

Table A-8: Cost contributions to the hydrogen production cost for current and future distributed production
facilities (1500 kgH2/day) using alkaline electrolysis technology. Costs are expressed on a $/kg H2
basis [1]. .................................................................................................................................. 146

Table A-9: Hydrogen cost, capital cost, energy efficiency and electricity price for current and future, central
production facilities (52 300 kgH2/day) using alkaline electrolysis systems [1]. ......................... 146

Table A-10: Cost contributions to the hydrogen production cost for current and future central production
facilities (52 300 kgH2/day) using alkaline electrolysis technology. Costs are expressed on a $/kg H2
basis [1]. .................................................................................................................................. 146

Table A-11: Estimated energy required to remove and recover CO2 from coal-fired power plants using
various technologies.[3] ........................................................................................................... 146

Table A-12: Mass balance for the Aspen simulations performed by Mignard et al. [4]. Values are given in
tons per day. aAdditional emissions from process steam production in case no waste heat is
available. bAdditional CO2 abatement in case the oxygen produced can be valorized. ............... 147

Table A-13: Energy balance for the Aspen simulations performed by Mignard et al. [4]. Values are given in
tons per day.............................................................................................................................. 147

Table A-14: Energy efficiency for the Aspen simulations performed by Mignard et al. [4]. ....................... 147

Table A-15: Mass balance for the Aspen simulations performed by Van-Dal et al. [5]. Values are expressed
in tons per hour. ....................................................................................................................... 147

Table A-16: Energy balance for the Aspen simulations performed by Van-Dal et al. [5]. Values are expressed
in tons per hour. ....................................................................................................................... 148
XVIII

Table A-17: CO2 balance for the Aspen simulations performed by Van-Dal et al. [5]. Values are expressed in
tons per hour. ........................................................................................................................... 148

Table A-18: Energy production cost from multiple sources for 2014. Values obtained from the Annual
Energy Outlook for 2014 provided by the Energy Information Agency.[6] ............................... 148

Table A-19: Price summary for energy (2013 dollars per unit) and prospects up to 2040. 1 barrel has an
approximate energy capacity of 1628 kWh, 106 Btu is equal to 293 kWh and 1 ton of coal
equivalent is equal to 8141 kWh. [7] ........................................................................................ 149

Table A-20 Stream data for the methanol to olefins (MTO) process [8]. .................................................... 150

Table A-21 Continued............................................................................................................................... 150

Table B-1 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch
conditions (500 K, 20 bar, 60% conversion) for reaction intermediates on a clean Co(111) surface.
Accompanying structures are illustrated in Table B-2. .............................................................. 152

Table B-2 Optimized configurations for reaction intermediates on a clean Co(111) surface presented in Table
B-1. ......................................................................................................................................... 153

Table B-3 Transition state geometries for the reactions presented in Table 4-2 and Figure 4-5 on clean
Co(111). .................................................................................................................................. 154

Table B-4 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch
conditions (500 K, 20 bar, 60% conversion) for reaction intermediates on a Co(111) surface with
coverage of 1/3 ML. Accompanying structures are illustrated in Table B-5. .............................. 157

Table B-5 Optimized configurations for reaction intermediates on Co(111) surface with coverage of 1/3 ML
presented in Table B-4. ............................................................................................................ 157

Table B-6 Transition state geometry for CO2* dissociation on Co(111) with coverage of 1/3 ML, presented
in Table 4-4.............................................................................................................................. 157

Table B-7 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch
conditions (500 K, 20 bar, 60% conversion) for reaction intermediates on a clean Co(211) surface.
Accompanying structures are illustrated in Table B-8. .............................................................. 158

Table B-8 Optimized configurations for reaction intermediates on a clean Co(211) surface presented in Table
B-7. ......................................................................................................................................... 158
XIX

Table B-9 Transition state geometry for CO2* dissociation on a clean Co(211) surface presented in Table
4-6. .......................................................................................................................................... 159

Table B-10 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch
conditions (500 K, 20 bar, 60% conversion) for reaction intermediates on a clean Cu(111) surface.
................................................................................................................................................ 160

Table B-11 Optimized configurations for reaction intermediates on a clean Cu(111) surface presented in
Table B-10. .............................................................................................................................. 161

Table B-12 Transition state geometries for the reactions presented in Table 4-8 on clean Cu(111). ............ 162

Table B-13 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch
conditions (500 K, 20 bar, 60% conversion) for reaction intermediates on a Cu(111) surface with
coverage of 1/3 ML. (Extension to Table 4-9.) ......................................................................... 162

Table B-14 Optimized configurations for reaction intermediates on Cu(111) surface with CO coverage of 1/3
ML presented in Table B-13. .................................................................................................... 163

Table B-15 Transition state geometry for CO2* dissociation on Cu(111) with CO coverage of 1/3 ML,
presented in Table 4-10. ........................................................................................................... 163

Table B-16 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch
conditions (500 K, 20 bar, 60% conversion) for reaction intermediates on a clean Ni(111) surface.
................................................................................................................................................ 164

Table B-17 Optimized configurations for reaction intermediates on a clean Ni(111) surface presented in
Table B-16. .............................................................................................................................. 165

Table B-18 Transition state geometries for the reactions presented in Table 4-12 on clean Ni(111)............ 166

Table B-19 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch
conditions (500 K, 20 bar, 60% conversion) for reaction intermediates on a Ni(111) surface with
coverage of 1/3 ML. (Extension to Table 4-13.) ....................................................................... 166

Table B-20 Optimized configurations for reaction intermediates on Ni(111) surface with coverage of 1/3 ML
presented in Table B-19. .......................................................................................................... 167

Table B-21 Transition state geometry for CO2* dissociation on Ni(111) with coverage of 1/3 ML, presented
in Table 4-14. ........................................................................................................................... 167
XX

Table B-22 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch
conditions (500 K, 20 bar, 60% conversion) for reaction intermediates on a clean Au(111) surface.
................................................................................................................................................ 168

Table B-23 Optimized configurations for reaction intermediates on a clean Au(111) surface presented in
Table B-22. .............................................................................................................................. 169

Table B-24 Transition state geometries for the reactions presented in Table 4-16 on clean Au(111). .......... 170

Table B-25 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch
conditions (500 K, 20 bar, 60% conversion) for reaction intermediates on a Au(111) surface with
coverage of 1/3 ML. ................................................................................................................. 170

Table B-26 Optimized configurations for reaction intermediates on Au(111) surface with coverage of 1/3
ML presented in Table B-25. .................................................................................................... 171

Table B-27 Transition state geometry for CO2* dissociation on Au(111) with coverage of 1/3 ML, presented
in Table 4-18. ........................................................................................................................... 171

Table C-1 Summary of the reactions taken in account for chain propagation in FTS on the catalyst surface.
Distinction is made between the CO insertion and carbide mechanism. Values for the CO
insertion mechanism are provided by the coach of this thesis, values for the carbide mechanism
were found in literature [1-4]. ................................................................................................... 172

Table D-1 List of simulations for adsorption on the cobalt catalyst surface ................................................ 175

Table D-2 List of transition state calculations for the cobalt catalyst surface. ............................................. 176

Table D-3 List of simulations for adsorption on the copper catalyst surface. .............................................. 176

Table D-4 List of transition state calculations for the copper catalyst surface. ............................................ 177

Table D-5 List of simulations for adsorption on the nickel catalyst surface. ............................................... 177

Table D-6 List of transition state calculations for the nickel catalyst surface. ............................................. 177

Table D-7 List of simulations for adsorption on the gold catalyst surface. ................................................. 178

Table D-8 List of transition state calculations for the gold catalyst surface. ............................................... 178

Table D-9 CD Rom summary.................................................................................................................... 178


XXI

List of abbreviations and symbols


Abbreviations

ASF Anderson-Schulz-Flory

BOP Balance of Plant

CAGR Compound Annual Growth Rate

CapEx Capital Expenditures

CCS Carbon Capture and Storage/Sequestration

CCU Carbon Capture and Utilization

cNEB climbing image Nudged Elastic Band

CTL Coal To Liquid

DFT Density Functional Theory

DOE United States Department of Energy

ECBM Enhanced Coal-Bed Methane recovery

ECFORM Electrochemical Reduction of CO2 to Formate/Formic Acid

EOR Enhanced Oil Recovery

FCC Fluid Catalytic Cracking

FGD Flue-Gas Desulphurization

FT Fischer-Tropsch

FTS Fischer Tropsch Synthesis

GDE Gas Diffusion Electrodes

GDL Gas Diffusion Layer

gge Gasoline gallon equivalent

gle Gasoline liter equivalent

GTL Gasoline To Liquid

H2A Hydrogen Analysis

HER Hydrogen Evolution Reaction

HHV Higher Heating Value

HMB Hexa-Methyl-Benzene
XXII

HP Hydrocarbon Pool

IEA International Energy Agency

LCA Life Cycle Analysis

LHV Lower Heating Value

LNG Liquid Natural Gas

LPG Liquid Petroleum Gas

MARR Minimum Acceptable Rate of Return

MD Molecular Dynamics

ML monolayer

MOF Metal-Organic Framework

MSDS Material Safety Data Sheet

MTG Methanol To Gasoline

MTHC Methanol To Hydro-Carbons

MTO Methanol-To-Olefin

NASA National Aeronautics and Space Administration

NPV Net Present Value

NREL National Renewable Energy Laboratory

O&M Operating and Maintenance

ODH Oxidative Dehydrogenation

ODE Ordinary Differential Equation

OER Oxygen Evolution Reaction

OpEx Operating Expenditures

PC Product Cost

PEC Photoelectrochemical

PEM Polymer Electrolyte Membrane

PFB Pressurized Fluidized Bed

PFR Plug Flow Reactor

PSA Pressure Swing Adsorption


XXIII

PV Photovoltaic

PVC Poly-Vinyl-Chloride

ROI Return on Investment

RWGS Reversed Water Gas Shift

SDEN Site Density

SNG Substitute Natural Gas

SOE Solid Oxide Electrolysis

STH Solar to H2

TCI Total Capital Investment

TOF Turnover Frequency

TSA Temperature Swing Adsorption

VASP Vienna Ab-Initio Simulation Package

vdW-DF2 Van der Waals Density Functional

WGSR Water Gas Shift Reaction

ZPE Zero Point Energy

Symbols

Roman symbols

𝑠̇𝑘 Net production rate

∆𝐺𝑎𝑑𝑠 Gibbs free energy of adsorption

∆𝐻𝑎𝑑𝑠 Adsorption enthalpy

∆𝑆𝑎𝑑𝑠 Adsorption entropy

𝐴𝑠 Site surface area

𝐶𝐴𝐶 Administration cost

𝐶𝐷 Depreciation cost

𝐶𝐷𝑆𝐶 Distribution and selling cost

𝐶𝑂&𝑀 Operating and maintenance cost

𝐶𝑃𝑂𝐶 Plant overhead cost

𝐶𝑅 Raw material cost


XXIV

𝐶𝑈 Utility cost
°
𝐶𝑝𝑘 Specific heat capacity

𝐸𝑎 Activation energy

𝐻𝑘° Standard enthalpy

𝐼𝐸 Equipment cost

𝐼𝑖 Principal moments of inertia

𝑄𝑖𝑟𝑟 Generated heat during electrolysis

𝑄𝑟𝑒𝑣 Released heat during electrolysis

𝑆𝑘° Standard entropy

𝑆𝑟𝑜𝑡 Rotational entropy correction

𝑆𝑡𝑟𝑎𝑛𝑠 Translational entropy correction

𝑆𝑣𝑖𝑏 Vibrational entropy correction

𝑉𝑟𝑒𝑣 Reversible cell potential

𝑉𝑡ℎ Thermoneutral cell potential

𝑊𝑛 Hydrocarbon weight fraction

𝑒− electrode

𝑘𝐵 Boltzmann constant

𝑞𝑖 Rate of production

𝑟𝑝,𝑛 Propagation rate

𝑟𝑡,𝑛 Termination rate

∆𝐺° Standard Gibbs free energy change

∆𝐻 Enthalpy change

∆𝐻° Standard enthalpy change

∆𝑆 Entropy change

Au Gold

C Carbon

C2H4 Ethylene

CH2 Methylene
XXV

CH3 Methyl

CH3OH Methanol

CH4 Methane

Co Cobalt

CO Carbon monoxide

CO2 Carbon dioxide

COOH Hydrocarboxyl

Cu Copper

H Hydrogen

ℎ Planck’s constant

H2 Hydrogen

H2O Water

HCOH Hydroxyl methylene

HCOO Formate

HCOOH Formic acid

MEA Monoethanolamine

Ni Nickel

O Oxygen

O2 Oxygen

OH Hydroxyl

𝐴 Pre-exponential factor

𝐴𝐵𝐶 Rotational constants

𝐸 Specific energy consumption

𝐹 Faraday’s constant

𝑃 Pressure

𝑃𝐶 Product cost

𝑄 Partition function

𝑅 Universal Gas constant


XXVI

RC-C Rate of C-C coupling in the carbide mechanism

RCO Rate of CO insertion

𝑅𝐹 Ratio factor

𝑇 Temperature

𝑉 Operating voltage

𝑘 Rate coefficient

𝑛 Number of transferred electrons

𝑥 Electrolyzer production capacity

𝑦 Electrolyzer capital cost

Greek symbols

𝜃 Coverage

𝜃𝑟𝑜𝑡 Characteristic temperature of rotation

𝜈𝑖 Vibrational frequencies

𝜐𝑘𝑖 Stoichiometric coefficients

𝜒𝑘 Chemical symbol representation

𝛼 Chain growth probability

𝜀 Electrolysis efficiency

𝜎 External symmetry number

Superscripts

† Transition state

* Surface species

Subscripts

aq aqueous

cw cooling water

f forward

g gaseous

l liquid

r reverse
1

Chapter 1
Introduction
As the concentration of CO2 becomes an environmental problem, the catalytic conversion of CO 2 to
petrochemicals and fuels represents a possible pathway to reduce the CO 2 emissions and create a carbon-
neutral fuel and chemicals cycle. Rapid activation of the very stable CO 2 molecule is however a key
challenge for conversion pathways to be feasible. Using computational catalysis, more effective catalyst
systems for CO2 activation reactions can be sought. Utilization of kinetic modeling can give additional
insight in the reaction mechanism, and indications on some specific catalytic properties.
2 Introduction

1.1. Techno-economic analysis of renewable pathways for CO 2


conversion.
Over the last decades, greenhouse gas emissions are building up to alarming rates and the CO 2 concentration
in our atmosphere becomes an environmental problem. Within approximately 50 years the emissions
increased tenfold and are currently still rising, leading to an increase of the CO 2 concentration, from 270 ppm
to >400 ppm, which nature is not capable to compensate [1, 2]. The general belief is that the increase of CO2
concentration is caused by human activity, and processes involving the consumption of fossil fuels have been
the main contributors to CO2 emissions, requiring attention as more than 80% of the world energy portfolio
relies on non-renewable sources such as fossil fuels [3]. The environmental impact extends beyond global
warming, to e.g. the acidification of oceans [4], and to reduce the CO2 concentration in the atmosphere,
human actions are required. Awareness for the environmental issues caused by the increasing CO2 emissions
has grown worldwide, leading to several international agreements such as the Kyoto protocol [5] in 1997 and
recently the Paris agreement [6] of 2015. Other political incentives, such as government policy and
regulations, will be of major importance for the growth of CO 2 conversion processes. Commitment of
national governments will be necessary for the integration and stimulation of renewable energy via
investments and budgetary support [7].

Figure 1-1 Different pathways for CO2 utilization [8].

There are essentially four approaches for the reduction of CO 2 emissions in our atmosphere (Figure 1-1): (1)
improving energy efficiencies; (2) the use of non-carbon or renewable energy sources such as solar, wind or
nuclear energy; (3) carbon capture and sequestration (CCS); and (4) carbon capture and utilization (CCU)
[9]. Improving energy efficiency has had intense research and industry is pursuing constant optimization.
However, it does not address the non-renewable nature of fossil fuels. Renewable energy sources are
Introduction 3

attractive but are subject to fluctuations in its energy generation. Problems arise concerning transportation
and storage of the generated power. CCS has the prospect of possible reductions of the atmospheric CO 2
concentration, but offers no alternative production route towards fuels or raw chemicals. As it does not
address the non-renewable nature of fossil fuels, it is believed that CCS will merely be a bridging technology
to more sustainable energy supply [3, 10-12]. The fourth option considers CO2 as a feedstock for a variety of
purposes, instead of treating it as a waste product [9]. Three general pathways for CO2 utilization exist: (1)
conversion of CO2 into fuels (storage of renewable energy); (2) production of chemical feedstock; and (3)
non-conversion use of CO2 (e.g. supercritical solvents, EOR, or food and pharmaceutical applications) [8].

Separation and capture of CO2 on an industrial scale is mainly done by liquid-phase absorption, using e.g.
alkanolamine solutions, or by cryogenic distillation. Product streams with high CO 2 concentrations from
power plants or cement factories are used for carbon capture, but the actual processes are costly and energy
intensive. Effective solid absorbents are suggested to be superior alternatives. As ultimately the high
concentration sources of CO2 will disappear, techniques will be necessary to capture carbon from other
sources such as air. This would allow to deal with small and dispersed CO 2 emitters and avoids the
construction of infrastructure for local CO2 collection [13].

Utilization of CO2 has the possibility of securing the future energy supply and to decrease the net CO 2
emissions to the atmosphere. Catalytic conversion of CO2 to petrochemicals and fuels represents a possible
pathway and creates a carbon-neutral fuel and chemicals cycle. Possible products are fuels such as methanol,
or hydrocarbons such as ethylene or propylene. The combination of different utilization technologies has the
potential to reduce CO2 emissions at least 3.7 Gt/y, or approximately 10% of current annual emissions. Fuel
production from biomass, from the (bio-)chemical conversion of CO2, has the advantage of yielding readily
usable chemicals and replacement of 5% of the liquid fossil fuel by biomass-based liquid fuel could avoid
approximately 0.4 Gt/y of CO 2. Conversion to minerals and polymers captures CO2 in a stable structure over
a longer period in time. If 10% of global building material demand was produced from CO 2 recycling, up to
1.6 Gt/y could be avoided. (Electro-)chemical conversion into chemical feedstock and intermediates could
reduce CO2 emissions by 0.3 Gt/y and could be an alternative for production from fossil fuel. Non-
conversion applications has the potential of 1.4 Gt/y reduction of CO 2 emissions [8].

Energy has to be added to the highly stable CO 2 in order to convert it to useful products. This can be done in
the form of (1) high energy electrons for the electrochemical reduction; (2) photons for the photocatalytic
conversion; (3) heat for solar-thermal conversion; or (4) high energy reductants as H2 or CH4 for the catalytic
conversion of CO2. Hence, one can consider catalytic CO2 conversion as an approach to store renewable
energy or the energy invested to produce renewable H2 in a more convenient liquid form. Although it would
be more efficient to use the electrical energy from renewable power sources directly, the variability poses a
problem. Furthermore, distribution infrastructure for hydrocarbon fuels is well available [10, 14]. Examples
for the storage of renewable energy can be found in nature; the photosynthesis process splits H2O to form H2
and O2 and hydrogen is stored by reaction with CO2 to form glucose and other carbohydrates [15].
4 Introduction

Alternative conversion processes for CO2 conversion attempt to provide similar but accelerated pathways,
illustrated in Figure 1-2.

Figure 1-2 Closed reaction cycle for the conversion of CO 2 to fuels and chemicals. Alternative conversion
processes attempt to provide a similar, but accelerated pathway.

Two main driving factors for CO2 recycle exist: (1) the product/process value; and (2) the economics related
to carbon taxes. CO2 is an economic raw material (its price is solely determined by the costs for capture and
separation), and can be used to produce several valuable products [13]. Fine chemicals or polycarbonates
have been reported to be economically feasible without carbon taxes. However, production scale would not
allow for significant reduction of global CO2 emissions. Large capacity compounds such as methanol or light
olefins have the potential of CO2 emission control, but the exploitation will be highly dependent on carbon
taxes [16].
Introduction 5

1.2. Ab-initio calculations


A key challenge for alternative pathways is the rapid activation of the very stable CO 2 molecule. Next to
energy requirements (in the form of high energy electrons, photon, or high energy reductants), catalysts are
necessary to tackle this issue. Breakthrough catalysts for rapid activation of CO 2 by undergoing fast and
thermodynamically allowed surface redox steps are required. Using computational catalyst screening, more
effective catalyst systems can be sought.

Due to advances in density functional theory (DFT) it has become possible to describe catalytic surface
reactions in high detail and with high accuracy, to an extent where comparison with experiments becomes
possible. Hence, by the use of theoretical models the surface reaction mechanism can be analyzed and
catalytic activity of different catalysts can be compared. An example of such an analysis is illustrated in
Figure 1-3, showing a comparison between CO2 dissociation on Co(111) and Cu(111).

Figure 1-3 Comparison of the reaction path for dissociation of CO 2* to O* and CO* adsorbed on a free Co(111)
and Cu(111) surface. The different catalyst morphology and properties affect the adsorption energy and
activation barrier for dissociation; and hence, its capability for activation of CO2.

The ultimate goal of computational catalysis is to be able to determine the material required to catalyze a
certain reaction under a set of specified conditions. Therefore sufficient insight is necessary into the factors
determining the activity of a certain catalyst morphology to be able to tailor the catalysts atom-by-atom [17].
6 Introduction

1.3. Microkinetic modeling


Once an applicable reaction mechanism, including all elementary reactions at microscopic level (without
assuming the existence of a rate-determining step), and the appropriate thermodynamic data is obtained,
together with the desired reactor model; the reaction kinetics can be analyzed via microkinetic modeling.
Specialized software such as Chemkin is then capable of symbolically describing the reaction mechanism
and to solve the reaction and reactor equations for the surface coverages and the gaseous composition
following from the reaction mechanism and the desired reactor model [18-20]. Verification of microkinetics
occurs via comparison with more detailed surface science measurement techniques [21].

Via microkinetic modeling a better insight in the dominance of different reaction pathways can be obtained.
Microkinetic models are also more robust than macro-kinetic models when extrapolating for a wider range of
operating conditions. Insight on the reaction pathways can be related to the catalyst properties, and could
indicate important information on the influence of the catalyst structure on product selectivity or yield and
hence, allows for faster catalyst development [21, 22].

1.4. Thesis structure


In this thesis a techno-economic analysis is performed on the possible activation pathways and catalytic
conversion processes for CO2 with renewable energy (Chapter 2). CO2 is hence evaluated as an energy
carrier for renewable energy. (1) Production of hydrogen from electrolysis of water with renewable energy is
analyzed first; hydrogen then has the potential to serve as high energy reductant for conversion of CO 2. (2)
Production of methanol via conversion of CO2 with hydrogen; (3) ethylene synthesis via electrochemical
reduction of CO2 and water, a modified FTS process for CO2 and H2, and the methanol-to-olefins process;
and (4) the electrochemical reduction of CO2 to formic acid.

The activation of CO2 is further analyzed via DFT calculations (Chapter 4). The Co(111) surface is evaluated
as a reference case for adsorption of various reaction intermediates and calculation of activation energies for
direct dissociation and a hydrogen-assisted pathway. Furthermore, effects of increased coverage and step
sites, i.e. the Co(211) surface, are taken in account. By means of comparison the Cu(111), Ni(111) and
Au(111) are evaluated for their capability of CO2 activation.

An initial kinetic model is composed to describe the CO activation and carbon coupling reactions in Fischer-
Tropsch synthesis (Chapter 5). Preliminary analysis of CO2 activation is done by incorporation of CO 2
activation steps in the reaction mechanism for FTS. To obtain insight in the activation of CO 2, comparison is
made with the CO activation pathway.
Introduction 7

1.5. References

1. Marland, G., et al., Global, regional, and national fossil fuel CO2 emissions. Trends: A
Compendium of Data on Global Change, 2007: p. 37831-6335.

2. Network, G.G.G.R. Recent Monthly Average Mauna Loa CO2. 2016 April 5, 2016 [cited 2016 April
30]; Available from: http://www.esrl.noaa.gov/gmd/ccgg/trends/.

3. Olajire, A.A., Valorization of greenhouse carbon dioxide emissions into value-added products by
catalytic processes. Journal of Co2 Utilization, 2013. 3-4: p. 74-92.

4. Appel, A.M., et al., Frontiers, opportunities, and challenges in biochemical and chemical catalysis
of CO2 fixation. Chemical reviews, 2013. 113(8): p. 6621-6658.

5. Oberthür, S. and H.E. Ott, The Kyoto Protocol: international climate policy for the 21st century.
1999: Springer Science & Business Media.

6. Commission, E., The Paris Protocol – A blueprint for tackling global climate change beyond 2020,
E. Union, Editor. 2015: Paris.

7. Annual Energy Outlook 2015 with projections to 2040, U.S.E.I. Administration, Editor. 2015, U.S.
Department of Energy: Washington.

8. Veritas, D.N., Electrochemical Conversion of CO2–Opportunities and Challenge. 2011, Norway.

9. Agarwal, A.S., et al., The electrochemical reduction of carbon dioxide to formate/formic acid:
engineering and economic feasibility. ChemSusChem, 2011. 4(9): p. 1301-1310.

10. Aresta, M., A. Dibenedetto, and A. Angelini, The changing paradigm in CO2 utilization. Journal of
Co2 Utilization, 2013. 3-4: p. 65-73.

11. Aresta, M., A. Dibenedetto, and A. Angelini, Catalysis for the valorization of exhaust carbon: From
CO2 to chemicals, materials, and fuels. Technological use of CO2. Chemical reviews, 2013. 114(3):
p. 1709-1742.

12. Whipple, D.T. and P.J.A. Kenis, Prospects of CO2 Utilization via Direct Heterogeneous
Electrochemical Reduction. The Journal of Physical Chemistry Letters, 2010. 1(24): p. 3451-3458.

13. Olah, G.A., G.K.S. Prakash, and A. Goeppert, Anthropogenic Chemical Carbon Cycle for a
Sustainable Future. Journal of the American Chemical Society, 2011. 133(33): p. 12881-12898.

14. Centi, G., E.A. Quadrelli, and S. Perathoner, Catalysis for CO2 conversion: a key technology for
rapid introduction of renewable energy in the value chain of chemical industries. Energy &
Environmental Science, 2013. 6(6): p. 1711-1731.

15. Bensaid, S., et al., Towards Artificial Leaves for Solar Hydrogen and Fuels from Carbon Dioxide.
ChemSusChem, 2012. 5(3): p. 500-521.

16. Centi, G., G. Iaquaniello, and S. Perathoner, Can We Afford to Waste Carbon Dioxide? Carbon
Dioxide as a Valuable Source of Carbon for the Production of Light Olefins. ChemSusChem, 2011.
4(9): p. 1265-1273.
8 Introduction

17. Norskov, J.K., et al., Towards the computational design of solid catalysts. Nat Chem, 2009. 1(1): p.
37-46.

18. Kee, R.J., J.A. Miller, and T.H. Jefferson, CHEMKIN: A general-purpose, problem-independent,
transportable, FORTRAN chemical kinetics code package. 1980, Sandia Labs.

19. Getting stared with CHEMKIN, R. Design, Editor. 2015.

20. Storsæter, S., D. Chen, and A. Holmen, Microkinetic modelling of the formation of C 1 and C 2
products in the Fischer–Tropsch synthesis over cobalt catalysts. Surface Science, 2006. 600(10): p.
2051-2063.

21. Hansen, A.G., W.J. van Well, and P. Stoltze, Microkinetic modeling as a tool in catalyst discovery.
Topics in Catalysis, 2007. 45(1): p. 219-222.

22. Metaxas, K., et al., A microkinetic vision on high-throughput catalyst formulation and optimization:
development of an appropriate software tool. Topics in Catalysis, 2010. 53(1-2): p. 64-76.
9

Chapter 2
Techno-economic analysis of renewable
pathways for CO2 conversion.
In this chapter a techno-economic analysis is performed on possible reaction paths for conversion of CO 2 to
methanol, ethylene and formic acid; with renewable energy. Additionally, production of hydrogen from
electrolysis of water with renewable energy is evaluated. Hydrogen is then used as a feedstock for the
catalytic conversion of CO2. First some general issues and challenges for CO2 conversion are discussed. CO2
is one of the main causes for global warming, and its high atmospheric concentration is mainly due to human
activity. Technologic and economic concerns are discussed as well as the benefits of CO2 recycling
(environmentally and economically). Recycling processes for CO2 require that no additional CO2 is emitted
during operation and energy supply is carbon neutral. Renewable energy is thus required for a sustainable
carbon cycle. Sources of renewable energy and utilization are discussed. Utilization of the renewable energy
can be direct, or can be inserted in the energy cycle via water or carbon dioxide electrolysis resulting in
energy carrier compounds such as hydrogen, methanol or formic acid.

Production of hydrogen is briefly evaluated for some emerging technologies and more thoroughly for
electrolysis. The thermodynamics of electrolysis are discussed, followed by the techniques for alkaline and
PEM electrolysis. Subsequently an analysis on the economics of electrolysis is performed. Finally some
general thoughts on hydrogen purification and incentives for utilization as a feedstock for CO 2 conversion
are included.

Multiple highly concentrated emission sources of CO 2 currently exist, such as coal-fired power plants, oil
refineries or cement factories. These sources should initially be exploited for CO 2 feedstock. For future
conversion of CO2, new techniques will be necessary to capture carbon from other sources such as air.
10 Techno-economic analysis

Several reaction paths for CO2 conversion are currently explored, and in this master thesis the conversion of
CO2 to methanol, ethylene, and formic acid are evaluated in depth. An introduction is provided on the
different techniques and an economic evaluation is based on case studies performing simulations of the
process operations. Methanol production is evaluated for CO2 captured from a coal-fired power plant and
hydrogen supplied from water electrolysis. Ethylene synthesis is evaluated for the electrochemical reduction
of CO2 and water, a modified Fischer-Tropsch process (CO2 + H2), and the methanol-to-olefins process.
Analysis of formic acid production is done for the electrochemical reduction of CO 2.

2.1. CO2 conversion: a challenging opportunity


During the last century the anthropogenic emission of CO2 has encountered a sharp increase. Within
approximately 50 years the emissions increased tenfold and are currently still rising, leading to an increase of
the CO2 concentration in our atmosphere, from 270 ppm to >400 ppm, which nature is not capable to
compensate [1, 2]. The general belief is that the high increase of CO 2 concentration is predominantly caused
by human activity, especially the combustion of fossil fuels. The increased CO2 concentration has a lot of
detrimental effects on the environment, with global warming (the increase of the average temperature of the
Earth’s climate and the related effects) as main issue. However, also the acidification of the oceans and other
consequences can be attributed to the increased carbon dioxide emissions, see Figure 2-1 [3-6]. Therefore,
human actions are required.

Figure 2-1 Left: Mean carbon dioxide concentration measured at the Mauna Loa Observatory, Hawaii [7]. The
red line represents the monthly atmospheric CO 2 concentration, the black line represents the yearly average.
Right: schematic representation of the perturbation of the global carbon cycle caused by human activities,
averaged globally for the decade 2004- 2013 [7].

In recent years the awareness for the environmental issues of CO2 emissions has grown and several
international agreements (the Kyoto protocol [8] in 1997 and recently the Paris agreement [9] of 2015) were
Techno-economic analysis 11

settled. Current efforts for reduction of CO2 emissions are mainly focused on the improvement of energy
efficiencies of processes involving fossil fuels (e.g. electric energy production), and increasing the efficiency
in energy utilization. However, this approach will merely contribute a small part to the necessary reduction
of emissions and active CO2 consuming activities seem to be required [10]. In 2013, the industrial utilization
of CO2 for chemical industry was reported to be around 200 Mt/y, which is only a minor part of the CO 2
waste stream generated by human activity (~32 000 Mt/y) [11, 12]. Current CO2 is found in the production of
salicylic acid, the Solvay process for the production of sodium carbonate and urea production. Also in the
food and pharmaceutical industry, in Enhanced Oil Recovery (EOR) and in Enhanced Coal-Bed Methane
recovery (ECBM), carbon dioxide is directly used [4, 5, 11, 13]. A relatively new, and promising technique
is Carbon Capture and Storage (CCS), which is still under research and only small scale processes exist up to
now; uncertainties about the permanence time of storage and the impact of large volumes of CO 2 on natural
systems as well as the high cost and intensive energy requirements limit the advancement in research on CO 2
injection. CCS has the prospect of possible reduction of the atmospheric CO 2 concentration, but offers no
alternative production route towards fuels or raw chemicals. As it does not address the non-renewable nature
of the fossil fuels, it is believed that CCS will merely be a bridging technology to more sustainable energy
supply [10, 11, 13, 14].

Due to the detrimental environmental impact of fossil fuels, the dependence on oil-rich countries and the risk
of depleting resources, a shift towards renewable energy sources is necessary. Solar-, wind-, and hydro-
energy sources become more and more important within our energy economy, but mainly generate electricity
which is difficult to store, and difficulties arise from the discontinuous production of electrical energy.
Furthermore, it can be expected that the energy economy will remain dominated by liquid fuels [15, 16].
Widespread infrastructure for processing and distribution is at hand, it has a high penetration in the current
society, a high volumetric energy density and safe transportation, and has advantageous production prices.
Hence, the electrical energy does not match with the energy needs of our current society. Introduction of
renewable electrical energy within the chemical industry would be a viable option to solve these problems
[11, 17]. Conversion of CO2 towards liquid fuels or chemical raw materials would have the potential to
reduce the fossil fuel dependency, reduce the CO2 emissions and make use of electrical energy generated
from renewable energy sources. Carbon dioxide could in fact be an effective link between the needs and
requirements of renewable energy injection in the current energy infrastructure [17]. It is a freely available
and cheap carbon source, and elaboration of CO2 conversion processes could induce a sustainable industry
with a low carbon economy [13]. A CO2 based energy economy would have the potential of energy security
and global climate change protection, and could be a means to improve the image of the chemical industry
[18]. Hence, an economic as well as environmental driving force for the exploration of CO 2 conversion exists
[4, 17-19].

A lot of research has been ongoing on the possible conversion processes for CO2 towards liquid fuels and
base chemicals. Two general processes can be examined; first the low energy processes in which the
oxidation state of the carbon atoms is preserved, and secondly the high energy processes which require
12 Techno-economic analysis

energy addition to alter the carbon oxidation state. Examples of the first option are the formation of
carboxylates, carbonates, urea’s and derived materials like polycarbonate or polyurethane polymers. In these
processes CO2 is usually converted by reaction with electron-rich molecules such as H2O, OH-, olefins,
alkynes and others, under mild process conditions. Possible products from the high energy processes are
formic acid, carbon monoxide, formaldehyde, methanol, methane and other hydrocarbons. Energy can be
added for example via electrons, hydrogen or thermal energy [4, 11, 18]. Carbon dioxide has the potential to
be converted into storage components for renewable energy (e.g. hydrogen or electricity storage) or can be a
feedstock for the conversion to base chemicals, polymers and fuels. A third, alternative, application of CO 2 is
to serve as a non-toxic alternative for several reactants, e.g. phosgene [4, 5, 11, 13, 18]. Significant
contributions are expected to the reduction of carbon dioxide emissions [20]. However, even production of
the entire ethylene and propylene consumption by CO2 recycling would not consume enough carbon dioxide
to balance the emissions. It is important to notice that contributions from different sectors will be necessary
to come to an effective solution [11, 15, 18].

Before carbon dioxide conversion becomes a feasible reaction path that is competitive with current fuel and
chemical production, a lot of challenges need to be tackled. A first issue concerns the capture of CO 2. The
level of the majority of these systems varies significantly over different industries; certain industries such as
power plants and oil refineries have already done a lot of effort to implement carbon capture at large-scale.
Other industries still require a transition from small-scale demonstration plants to industrial applications [11,
21]. As about 40% of global anthropogenic CO2 comes from thermoelectric plants driven by fossil fuels,
focus should initially be put on these systems for carbon capture for further utilization [21, 22].

Secondly, energy supply from renewable sources should be carbon neutral. If energy sources would emit a
high amount of carbon dioxide, as is the case for power plants or oil refineries, there would be no net
reduction of CO2 emission. To determine the real CO2 emission reduction, different from the amount of CO2
used or recycled in the process, a Life Cycle Analysis (LCA) is required for the entire system [5, 11, 18].

The third issue is the thermodynamic and kinetic properties of the reactions. Good knowledge of the reaction
steps for every process is necessary to analyze the driving force for the reaction and to optimize the reaction
conditions [4, 5, 13]. Due to the high stability of the carbon dioxide molecules, catalysts will be necessary to
activate CO2 and to convert CO2 on a large scale; the catalysts should be simple, stable and preferably
inexpensive [13, 19]. High efficiencies would also be required for CO2 conversion to become competitive
with current production processes [5]. Research towards CO2 conversion processes found in nature can be a
source of inspiration for efficient catalysts for enhanced CO2 conversion [6].

The last and probably most important aspect for CO2 conversion processes to become competitive are the
economics. Prices of renewable fuels from CO2 need to be competitive with fossil fuels, which would require
high efficiencies within the process, high added value of products and reduction of variable and capital costs
[23]. An option that is currently under consideration, but already introduced in the legislation of several
countries (e.g. Finland, Sweden, Switzerland, Ireland and Great-Britain), is a taxation on carbon. This is a tax
on the carbon content of fossil fuels (coal, oil and natural gas) and biofuels, more specifically on the level of
Techno-economic analysis 13

carbon dioxide emission (in ton CO2 equivalent). It would raise the price of fossil fuels and increase the
competitiveness of fuels from renewable energy sources [24, 25]. This would enhance the opportunities for
renewable fuels and chemicals and the transition to a sustainable energy economy [26].

2.2. Shift to renewable energy/Electricity generation


Over the last decades, awareness has raised that the energy system requires a shift to renewable energy
utilization [27]. Significant efforts have already been done, with forefront countries such as Denmark
generating 42% of its electricity demand from wind turbines in 2015 [28]. Prominent countries such as
Germany, the United Kingdom, and the United States of America also have done considerable investments,
creating renewable energy shares of 19%, 10% and 6% respectively [29].

The general belief and predictions forecast a consistent increase in renewable energy utilization and market
share. Current renewable energy share is estimated at 7%, with bioenergy (such as wood, agricultural waste,
sugarcane, etc.) as the main contributor (53%), and wind (7%) and solar (1%) energy steadily growing.
Reports suggest that by 2020, 20% of the total energy consumption will be based on the utilization of
renewable energy, in case coordinated actions between countries are supported [30].

Five primary sources of renewable energy exist: (1) biomass, (2) wind power, (3) solar power, (4)
hydropower, and (5) geothermal power [30]. Wind energy is expected to become the largest contributor to
renewable electricity production in 2038 at an average annual growth rate of 2.4% per year [27], and the
future cost for wind energy is reported to drop to an average 0.04 €/kWh [31]. In case of solar photovoltaic
energy, an annual growth rate of 6.8% per year is predicted [27] and the International Energy Agency (IEA)
stated that solar energy production will cover 20-25% of the electricity market by 2050. The photovoltaic
(PV) power installed is growing substantially due to reduced material and installation costs, and reduced
payback time from the initial 8-11 years to an expected 2 years [10]. Geothermal energy (the use of heat as
energy source) is predicted to grow at an annual rate of 5.5% [27] and could supply 10-20% of the world’s
energy need in 2050 [30]. Biomass, with an annual growth rate of 3.1%, is expected to retain an important
share in future energy resources [32, 33]. Although the solar energy conversion to biomass is relatively low
(1-8%) compared to conversion towards electricity, with efficiencies in a range of 10-30% [34], biomass will
remain important for it is a viable feedstock for production of biofuels which can easily be integrated in the
current fuel distribution infrastructure.

Several incentives exist for a shift towards renewable energy. Self-evident reasons such as the environmental
issue of CO2 emissions causing global climate change and the depleting fossil resources have caused the
growing awareness, but other aspects such as government policy, technology innovation or process
performance, business strategies etc. are important for the continuation of the extension towards a
sustainable renewable energy system. Influences on the renewable energy utilization can be subscribed to
external forces (government policy, environmental concerns, regulations, sustainability, technology
14 Techno-economic analysis

innovation, …) and internal forces (employee training/talent, performance, stability, operations, processes,
business strategy, technology adoption, …) [30].

The main factors to be considered are presumably the technical elaboration of renewable production
processes and the government policies on integration and stimulation of renewable energy. Long-term
industrial development of fossil fuel processing has rendered a high structural advantage for fossil fuel
production over the less mature, sustainable alternatives such as solar and wind power processes [29].
Technological development will be necessary to obtain high capacity and efficient processes to replace the
current fossil fuel based market. Investments can generate technological insights and hence reduce the cost
of renewables, further contributing to the energy transition. To support the grow of renewable energy
generation, the commitment of federal governments will be of vital importance [27]. However, federal
governments are appointed for a limited period in time and are bound to electoral outcomes, making it
difficult for a single government to sustain a transition towards renewable energy. Consistent government
regulatory and budgetary support will nevertheless be a vital external force for the growth and development
of renewable energy innovation technologies, as these are highly sensitive to policy and market uncertainty
[30]. Technological and political exchanges and successes from foreign governments can favor the transition
to renewable energy, due to increased availability and reduced cost of renewable energy, improving its
profitability, public image and political feasibility [29].

To evaluate the replacement of fossil energy by renewable energy, some simplified calculations can be
performed. First, an analysis will be done for complete replacement of fossil energy by solar and wind
energy. Energy requirements and surface area are determined. Secondly, a comparison will be made for the
energy requirements in case the generated electricity from solar and wind energy is converted to liquid fuels
and chemicals.

Current fossil energy consumption reported by BP in their statistical review for 2015 [35] is estimated at 543
EJ per year. In terms of electricity consumption this would account for 1.51·1014 kWh per year. These
numbers are not very significant and one could translate the energy requirement to the amount of equipment
necessary to produce the globally consumed energy.

Average properties for a modern wind turbine are a capacity of 2-3 MW and direct impact area of 3000 m²
per MW produced [36, 37]. Assuming a capacity of 2 MW and a capacity factor of 25% to account for the
actual time the wind turbine operates at optimal conditions, the average annual production capacity of a wind
turbine is estimated at approximately 4·106 kWh per year. Hence, to globally provide renewable electricity
from wind power, approximately 35 million wind turbines would be necessary, occupying up to 200 000 km²
or approximately 0.05% of Earth’s surface. A similar evaluation can be done for solar energy, with modern
solar panels providing approximately 220 W per m² [38]. On a yearly basis, adjusting for the average amount
of sunlight per day, one square meter of solar panels could provide up to 800 kWh per year. This would
require more than 320 000 km² of solar panels (not taking in account provision for transportation,
infrastructure or storage), equal to 0.06% of Earth’s surface.
Techno-economic analysis 15

If conversion of the electricity were to be considered, an average conversion efficiency of 60% would result
in an electricity requirement of 2.52·1014 kWh per year. This would require up to 58 million wind turbines or
a surface area of 330 000 km² in case of wind power generation. For solar power generation, approximately
550 000 km² would be necessary.

Problems arising from the generation of renewable electricity are the high capital investments, depicted in
Table 2-1. Capital costs are considerably higher for wind and solar energy compared to natural gas or coal
conversion, causing the electricity price to be significantly higher [31].

Table 2-1 Construction and electricity generation cost of different energy sources [31].

Construction cost (€/kWe) Electricity generation cost (¢/kWh)


Coal 1000-1500 2.5-5
Natural gas 400-800 3.7-6
Nuclear 1000-2000 2.1-3.1
Wind 1000-2000 3.5-9.5
Solar 3000-10000 15-40

An additional issue concerning electricity generation is storage and transportation [11, 17]. Electricity
generation from renewable sources is sensitive to weather effects and can be fluctuating. Storage of
electricity, which is not obvious, is required to store the excess energy in peak hours and supply energy in off
peak hours. However, e.g. storage of the electrical energy in batteries appears not to be feasible due to low
energy density of batteries. An extended electricity grid would require high capital investments and have
high losses due to the resistance of the electricity grid. Other problems such as noise and light disturbances
can also complicate installation of numerous amounts of wind mills [39].

As the electrical energy does not match with the energy needs of our current society, introduction of this
electrical energy within the chemical industry would be a viable option. This would require feasible
processes to produce chemical compounds that can be added within the chemical production value chain [11,
17]. In this thesis, the conversion of CO2 with renewable energy will be discussed.
16 Techno-economic analysis

2.3. Hydrogen production


Renewable energy plants such as solar, wind or marine 1 energy plants are obtaining an increasing impact on
the energy production worldwide. Electricity produced can be used directly but as most of these plants
produce a rather variable supply of electricity a solution needs to be found for the excess energy production
during off-peak periods [40].

Electricity is difficult to store per se and with renewable energy sources capacities exceeding the gigawatt
range an appropriate storage system is awaited [41]. As a lot of wind and marine energy generating plants are
situated off-shore, reports suggest that for these energy sources it would be more beneficial to produce fuels,
which are more convenient to transport, than to create a subsea electrical network [42]. A viable option is the
production of H2 in electrolyzer plants. But, these relatively small sources of excess energy are distributed all
over the globe and smaller chemical synthesis plants are necessary to process these streams [43]. H2 can then
be compressed and stored [23], e.g. in floating head or pressurized tanks, and further used to produce
hydrocarbons [40]. Due to the growing penetration of renewable fuels and energy sources, H 2 production can
be an aid in the integration of wind, marine and solar energy resources [23].

Nowadays up to 500 billion Nm³/yr of hydrogen is produced, but around 96% from steam reforming of
natural gas or other fossil fuels [41]. To replace the commercial fuels, hydrogen production should be quasi-
independent of fossil fuels and a high increase of total H2 production is necessary. Also, to be competitive,
hydrogen production costs should be reduced. The estimated cost of hydrogen needed to become competitive
has been examined by the Department of Energy of the United States of America and is thought to be around
4 €/kgH2 [44]. Due to the growing concern for climate change a growing need for clean sources of H 2 can be
expected and the cleanliness of these production processes will largely depend on the source of the electricity
used. Besides, hydrogen from renewable energy source enables the possibility of a greater domestic energy
production and hence less dependence on foreign oil imports [23].

Several production routes for hydrogen exist, of which electrolysis is the most important. Other routes are
biomass gasification/pyrolysis, thermolysis via solar driven thermochemistry, photolysis via
photoelectrochemistry and photolysis via photobiology [23]. First the processes which are still in rather early
development stages are discussed. Finally a more extended discussion of electrolysis is given.

2.3.1. Biomass gasification and pyrolysis

Hydrogen production from the reforming of bio-derived liquids, such as ethanol or sugar alcohols, or through
gasification or pyrolysis of biomass feedstocks is a process currently operational. It offers the potential of a
renewable option and low greenhouse gas emissions, especially if this technology is combined with carbon
capture, storage and utilization [45]. However, this technology is far from being able to run at high capacity

1
The energy carried by ocean waves, tides, salinity and ocean temperature differences.
Techno-economic analysis 17

and significant improvements in technology as well as cost-effectiveness are necessary to become


competitive with commercial hydrogen production [44].

Biomass gasification is used to convert organic or fossil-based carbonaceous biomass materials, such as crop
or forest residues, crops grown for energy use, municipal solid waste and others. High temperatures
(>700°C) and a controlled amount of oxygen and/or steam are required to convert the biomass, without
combustion, into carbon monoxide, hydrogen and carbon dioxide. Via the water-gas shift the carbon
monoxide further reacts with water to form additional hydrogen. An example of a simplified reaction scheme
is shown in Eq. (1) and (2) [45].

(biomass gasification) 𝐶6 𝐻12 𝑂6 + 𝑂2 + 𝐻2 𝑂 → 𝐶𝑂 + 𝐶𝑂2 + 𝐻2 + 𝑜𝑡ℎ𝑒𝑟𝑠 (1)

(water-gas shift) 𝐶𝑂 + 𝐻2 𝑂 → 𝐶𝑂2 + 𝐻2 (+ℎ𝑒𝑎𝑡) (2)

Biomass pyrolysis is a similar process as biomass gasification, but operates in the absence of oxygen.
Gasification is somewhat more complicated than in the presence of oxygen, therefore typically a catalyst is
required to yield a clean syngas mixture. As for the biomass gasification, the water-gas shift increases the
hydrogen yield [45].

Biomass availability in the United States, but as well within the European Union, has led to growing research
concerning the potential of biomass. Studies from the European Commission even revealed that it could be
reasonable to assume that biomass could account for two-thirds of the renewable energy target for the
European Union in 2020 [33, 45, 46].

Several challenges are yet to be tackled for biomass gasification to become competitive. As is the case for
electricity, biomass feedstocks are distributed and costs and availability may vary significantly from region
to region. Hence, reduction in capital and operating costs are required for this distributed production option.
Including identification of more durable, low cost reforming catalysts, more efficient separation trains, cost
reduction of the available bio-derived liquids and intensification of the number of process steps and unit
operations. According to the Multi-Year Research, Development, and Demonstration Plan from the NREL,
hydrogen production costs could be reduced to about 2.0 €/kg H2. A cost calculation for hydrogen production
from ethanol reforming and biomass gasification is given in Appendix A in Table A-1 and Table A-2 [44].

2.3.2. Thermolysis via solar driven thermochemistry

Thermolysis is the chemical decomposition of a substance caused by heat. Research is ongoing to use the
high-temperature energy from concentrated solar power to produce hydrogen through water-splitting
thermochemical cycles. It has the potential of a clean, efficient, and sustainable route for hydrogen
production from water [44].

Thermochemical water splitting uses very high temperatures (500 - 2000°C) to split water into its atomic
components hydrogen and oxygen. Several chemicals are added to the process to enhance hydrogen
production and are reused within each cycle. Only water is thus consumed and hydrogen and oxygen are
18 Techno-economic analysis

produced. In solar driven thermochemistry, these high temperatures are obtained by focusing sunlight on a
thermochemical reactor, illustrated in Figure 2-2. Two set-ups are shown to produce temperatures up to
2000°C. A first approach (a) shows a field of heliostat mirrors that concentrates the sunlight onto a central
reactor tower; the second approach (b) shows a dish mirror focusing the sunlight on an attached reactor
module [44, 47].

Figure 2-2 Two mirror-based approaches for focusing sunlight on a thermochemical reactor to produce
temperatures up to 2000°C. (a) A field of heliostat mirrors concentrates sunlight onto a central reactor tower; (b)
dish mirrors focus sunlight onto an attached reactor module. The solar-generated high-temperature heat can be
used to drive thermochemical reactions that produce hydrogen [47].

The technology of solar driven thermolysis is still in early stages of research. Due to the severe operating
conditions, development of durable cycle reaction materials and construction materials is needed. Feasible
reaction cycles are required and the solar to hydrogen efficiency, currently at a maximum of 17%, needs to
be improved. Actual areas necessary to produce 1 ton per day of H 2 are reported to be in the range of 20-40
km² [10, 44, 48, 49].

Competitiveness of this technique will mainly depend on the development of efficient water-splitting
chemical process cycles and materials, and the cost-effectiveness of the coupling between the
thermochemical cycles and the concentrated solar energy technology. Targeted hydrogen production costs
for solar-driven high-temperature thermochemical hydrogen production are added in Appendix A in Table
A-3. Hydrogen production cost is expected to be around 3.20 €/kgH2.

2.3.3. Photolysis via photelectrochemistry and photobiology

2.3.3.1. Photoelectrochemistry
In photoelectrochemical (PEC) water splitting, light energy is used directly to dissociate water using
specialized semiconductors or PEC materials. The semiconductor is immersed in a water-based electrolyte;
sunlight is absorbed by these semiconductors and is used to split water molecules into hydrogen and oxygen
[44, 50, 51]. Two systems are possible for the construction of PEC reactors, illustrated in Figure 2-3. In the
electrode system (a), planar photovoltaic cells are used to perform the water electrolysis. The cells are
positioned in arrays facing the sun and are immersed in a small water reservoir. Oxygen is produced at the
Techno-economic analysis 19

anode and hydrogen at the cathode face. A more detailed explanation is given in section 2.3.4. In the particle
system (b), aqueous reactor beds containing colloidal suspensions of photovoltaic-active nanoparticles are
used. Each of these nanoparticles is composed of the appropriate layered photovoltaic materials to carry out
the electrolysis reaction [50, 52].

Figure 2-3 Two different approaches to PEC solar hydrogen production reactors. (a) Electrode systems similar
to flat-plate photovoltaic panels; (b) particle systems comprised of slurries of PEC semiconductor particles [50].

PEC water splitting is a promising solar-to-hydrogen pathway for intermediate and high production
capacities. It offers the potential of high conversion efficiency at low operating temperatures. PEC
electrolysis is however still in an early stage of development. It requires significant advancements in
materials, material systems, and reactor concept development (development of high-durability PEC
materials, corrosion resistance, higher efficiencies, cost-effective solar water-splitting reactors etc.) [44].
Research of the NREL shows that a hydrogen production cost of 5.70 $/kg H2 should be achievable by 2020
(see Appendix A Table A-4).

2.3.3.2. Photobiology
The photobiological hydrogen production process consists of microorganisms using sunlight to convert
water, and sometimes organic matter, into hydrogen. The photosynthetic organisms carry out an energy-
storing fuel production reaction, which stores solar energy. Normally, these systems reduce CO 2 to
carbohydrates, but by modification of the process conditions it is possible to enhance H 2 formation. Water is
split by the organisms into hydrogen and oxygen ions, which can combine to form hydrogen and oxygen gas.
The most effective photobiological systems are based on microalgae, with efficiencies up to 10% [51, 53].
20 Techno-economic analysis

Photobiological hydrogen production is in early to mid-stage development and has the prospect of a clean
and sustainable route for hydrogen production. It also has the advantage that it does not require high-purity
water and does not generate any toxic or polluting by-products.

Challenges for this technology lay in the bioengineering of microorganisms, capable of hydrogen production
at high rates and high efficiency and an improved continuity of photoproduction. This also requires improved
reactor designs and processes to optimize biological hydrogen production [44]. Studies done by the NREL
reveal that hydrogen cost via photolytic biological hydrogen production should be optimized to 8.0 €/kg H2 by
2020, see also Table A-5 in Appendix A.

2.3.4. Electrolysis

Electrolysis is a technique that uses an electric current to drive a chemical reaction. No direct use of fossil
fuels is necessary, and the cleanliness of the process depends on the origin of the electricity. Compared to
other alternative hydrogen production methods, it is the furthest developed and therefore it will be examined
thoroughly [44].

A block diagram of the hydrogen production system design is shown in Figure 2-4. Water and power is fed
to the electrolyzer stacks, in which water is split into oxygen and hydrogen gas. The product streams can
then be processed further or can be vented to the atmosphere, as is sometimes done with oxygen if no
possible consumer is available [54].

Figure 2-4 Block diagram of the hydrogen production system design. Process water and power are fed to the
electrolyzer stacks in which water is split into oxygen and hydrogen gas [54].

The overall reaction occurring in the electrolyzer stacks is the decomposition of water into hydrogen and
oxygen.
1
𝐻2 𝑂 ⇌ 𝐻2 + 𝑂2 (∆𝐻298𝐾,1𝑏𝑎𝑟 = 286 𝑘𝐽 𝑚𝑜𝑙−1 ) (3)
2

The minimal requirements for the electrolysis of water to produce 1 kg of hydrogen are 8.9 liters of water at
normal conditions (25°C and 1 atm) and 39.7 kWh. However, efficiencies with respect to the lower heating
value (LHV)2, defined in Eq. (4), are limited to 82%. Efficiencies based on the higher heating value (HHV) 3,

2
LHV: HHV subtracted by the latent heat of vaporization of the water vapor formed at combustion.
Techno-economic analysis 21

see (5), can go up to 100%. Current efficiencies based on the HHV are in the range of 52% to 82% [16, 18,
23].

𝐿𝐻𝑉
𝐸𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑐𝑦𝐿𝐻𝑉 = (4)
𝑅𝑒𝑎𝑙 𝐸𝑛𝑒𝑟𝑔𝑦 𝐼𝑛𝑝𝑢𝑡

𝐻𝐻𝑉
𝐸𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑐𝑦𝐻𝐻𝑉 = (5)
𝑅𝑒𝑎𝑙 𝐸𝑛𝑒𝑟𝑔𝑦 𝐼𝑛𝑝𝑢𝑡

2.3.4.1. Thermodynamics of electrolysis


Initiation of water electrolysis requires a minimum cell voltage: the reversible potential 𝑉𝑟𝑒𝑓 . At this
potential no losses occur in the process.

∆𝐺°
𝑉𝑟𝑒𝑣 = − (6)
𝑛∙𝐹

In this definition 𝑛 represents the number of electrons transferred (2 in case of water electrolysis) and 𝐹 the
Faraday’s constant, equal to 96487 C/mol. For conditions at which the cell operates adiabatically, this cell
voltage is called the thermoneutral potential:

∆𝐻°
𝑉𝑡ℎ = − (7)
𝑛∙𝐹

Cell voltages above the thermoneutral potential, called overpotentials, contain an excess amount of energy
and will cause the cell to heat up. Cell voltages lower than the thermoneutral potential will cause the cell to
cool down. Heat transfer during electrolysis can be defined as the amount of heat released 𝑄𝑟𝑒𝑣 and the
amount of heat generated 𝑄𝑖𝑟𝑟 .

𝑄𝑟𝑒𝑣 = −𝑇∆𝑆 = ∆𝐺° − ∆𝐻° = 𝑛𝐹𝑉𝑟𝑒𝑣 − ∆𝐻° (8)

𝑄𝑖𝑟𝑟 = (𝑉 − 𝑉𝑟𝑒𝑓 )𝑛𝐹 (9)

The total heat generated is the sum of the released and generated heat.

𝑄 = 𝑄𝑟𝑒𝑣 + 𝑄𝑖𝑟𝑟 = 𝑛𝐹𝑉 − ∆𝐻° (10)

Efficiencies for electrolysis can be determined from the thermoneutral (𝑉𝑡ℎ ) and operating voltage (𝑉) and is
in fact the same definition as given in Eq. (5), which is the HHV of one mole of the product divided by the
energy consumption.

∆𝐻° 𝑉𝑡ℎ
𝜀= = (11)
∆𝐺° 𝑉

3
HHV: energy release at complete combustion of a unit quantity of fuel.
22 Techno-economic analysis

Increasing the cell voltage above the reversible potential will augment the production by an increased current
density; hence optimization of the production economy will determine the operational potential and thus the
efficiency [55].

Several types of electrolyzers exist, such as the alkaline, polymer electrolyte membrane (PEM) and solid
oxide (SOE) electrolyzer. Alkaline water electrolysis is a well matured technology for hydrogen production
and is capable of producing up to the megawatt range. Worldwide it is the current standard for large-scale
electrolysis. PEM electrolysis is a technology which was developed in the 1960s to overcome the drawbacks
of alkaline electrolyzers. It makes use of a solid electrolyte membrane, which can provide high proton
conductivity, low permeability for gases and a compact design allowing operation at higher pressures. Solid
oxide electrolysis is a promising technology, but still requires a lot of research. SOEC looks promising as it
can convert electrical energy into chemical energy with high efficiencies. It has drawn a lot of attention over
the years with companies and universities, causing an increasing amount of research on this topic. Studies
mainly investigate novel, improved, low cost and highly durable materials for SOECs [41, 55-57].

2.3.4.2. Alkaline Electrolyzer


Alkaline electrolysis is a mature technology, capable of converting energy up to the megawatt range and is
the current standard for large-scale electrolysis [41]. It consists of a liquid alkaline electrolyte, composed of a
solution of 20-30% KOH, in which two electrodes are immersed. A diaphragm is used to separate the
electrodes. A simplified representation is shown in Figure 2-5.

Figure 2-5 Simplified representation of an alkaline electrolysis cell. The cell consists of two electrodes, separated
by a diaphragm, immersed in an alkaline electrolyte liquid composed of a solution of 20-30% KOH [58].

At the cathode, two molecules of water are reduced to one molecule of hydrogen and two hydroxyl ions.
Desorption of the hydrogen ions from the electrode and recombination forms the gaseous H2. The hydroxyl
ions migrate to the anode compartment and recombine at the anode to form water and oxygen gas. The
reaction mechanism is shown in Eq. (12)-(14) [51, 59].
Techno-economic analysis 23


(𝐶𝑎𝑡ℎ𝑜𝑑𝑒) 2𝐻2 𝑂(𝑙) + 2𝑒 − → 2𝑂𝐻(𝑎𝑞) + 𝐻2(𝑔) (12)

− 1
(𝐴𝑛𝑜𝑑𝑒) 2𝑂𝐻(𝑎𝑞) → 𝐻2 𝑂(𝑙) + 2𝑒 − + 𝑂2(𝑔) (13)
2
1
(𝑇𝑜𝑡𝑎𝑙) 2𝐻2 𝑂(𝑙) → 𝐻2(𝑔) + 𝑂2(𝑔) + 𝐻2 𝑂(𝑙) (14)
2

For sake of safety and separation efficiency the diaphragm is added to the electrolysis cell. This diaphragm is
impermeable for the gaseous products but needs to be permeable for the hydroxyl ions and water molecules
[41]. Advantages of the alkaline electrolysis are that it requires only relatively cheap construction materials,
and cost effective electrode materials can be used such as Fe, Ni and Ni-compounds [55, 60]. However, a lot
of disadvantages arise from the alkaline electrolysis technology. A purge gas is necessary to prevent the
system from reaching explosive limits between oxygen and hydrogen gas, potassium hydroxide is very
corrosive (especially at high temperatures) and large system sizes are required due to the inability of
operating at high pressures. Also, efficiencies are limited because the diaphragm is not entirely impermeable
for the product gases and hence the partial load range is quite low due to safety issues on H 2 and O2 mixture
concentrations (safety threshold of 4 vol.% H2 in O2). Due to the high ohmic losses 4 across the liquid
electrolyte and the diaphragm, high current densities cannot be reached which limits the achievable
production rate [41, 55, 60-62].

2.3.4.3. Polymer Electrolyte Membrane (PEM) electrolyzer


Polymer Electrolyte Membrane (PEM) electrolysis was developed to overcome some of the drawbacks of
alkaline electrolysis. Its main advantages include greater energy efficiency, higher production rates, and
more compact design. The original design was done on behalf of the NASA space program to serve as fuel
cells [41, 55]. Commercial products are limited because product development only gained attention during
the last 20 years. According to Mergel [61], only a few products (<30 Nm³/h hydrogen) were available on the
market in 2012.

Figure 2-6 gives a schematic representation of the polymer electrolyte membrane electrolysis cell. The cell
consists of two electrodes separated by a membrane that conducts protons from the anode to the cathode
while insulating the electrodes electrically.

4
Ohmic loss: the release of heat when an electric current passes through a conductor.
24 Techno-economic analysis

Figure 2-6 Schematic representation of the polymer electrolyte membrane electrolysis cell. The cell consists of
two electrodes separated by a membrane that conducts protons from the anode to the cathode while insulating
the electrodes electrically [63].

The reactions taking place at the anode and cathode are respectively named the oxygen evolution reaction
(OER) and hydrogen evolution reaction (HER) and are given in Eq. (15)-(17). At the anode liquid water is
oxidized to gaseous oxygen, protons and electrons. The protons and electrons are then conducted through the
membrane and are combined to form gaseous hydrogen at the cathode [60].
+
(𝐴𝑛𝑜𝑑𝑒) 2𝐻2 𝑂(𝑙) → 𝑂2(𝑔) + 4𝐻(𝑎𝑞) + 4𝑒 − (15)

+
(𝐶𝑎𝑡ℎ𝑜𝑑𝑒) 4𝐻(𝑎𝑞) + 4𝑒 − → 2𝐻2(𝑔) (16)

(𝑇𝑜𝑡𝑎𝑙) 2𝐻2 𝑂(𝑙) → 2𝐻2(𝑔) + 𝑂2(𝑔) (17)

Advantages over the classical (alkaline) technology are first of all that PEM electrolyzers can operate at
higher current densities. This requires membranes with high protonic conductivity and good electric
insulation to prevent short-circuits within the cell. Use of the membrane also allows to work under a wider
partial load range due to the low gas crossover rate and hence the lower safety risk of reaching explosive
limits of H2 and O2 mixture concentrations. No delay on proton transport across the membrane occurs with
power input. Low permeability of the membrane also permits a better gas separation.

In comparison with the liquid electrolyte used in alkaline electrolysis, the solid electrolyte membrane
possesses a higher mechanical and thermal stability. This makes a more compact design possible because the
electrolysis cell will better withstand an increased operational pressure. Pressures in a range of 0.8 to 20 MPa
are possible, offering the advantage of hydrogen production at high pressure and hence less energy is
afterwards required for compression. Gas volume is also reduced which facilitates the product gas removal.
To minimize risky operation it is even possible to work in a differential pressure configuration to avoid
Techno-economic analysis 25

hazards related to pressurized oxygen. Increased operational pressures will also lead to some disadvantages.
Cross-permeation will be enhanced and therefore thicker membranes are required to maintain the critical gas
concentrations within the safety threshold. Due to the corrosive acidic regime generated by the membrane
(pH~2) and the high applied voltages (~2V), distinct materials are necessary that are highly chemically stable
[41, 51, 55, 60].

Catalyst particles used at the anode to accelerate the OER are mainly Ir, Ru and its oxides. At the cathode the
catalyst mainly consists of Pt and its derivatives. The polymer electrolyte membrane widely used is Nafion®,
which meets all the above mentioned requirements [55, 60, 64].

Critical challenges for the PEM electrolysis technology to become commercially competitive are the
reduction of its high capital cost. Highly resistive materials are necessary to perform in the harsh
environment. Either for the construction materials, e.g. titanium based current collectors or separator plates,
as for the electrocatalysts, where non-noble metals such as cobalt or nickel are subject to corrosion [60].
Furthermore, system efficiency needs to be increased beyond the actual range of 67-82% and integration
with renewable energy sources should be improved [55].

2.3.4.4. Economic analysis of electrolysis


To determine the cost of hydrogen produced from electrolysis, analyses are based on the study of two
electrolysis technologies, namely PEM and alkaline, as these are the only two low-temperature electrolyzers
currently on the market. An important source of information is the H 2A project directed by the U.S.
Department of Energy and executed by the National Renewable Energy Laboratory (NREL). Input and
reviews by independent manufacturers are used to optimize these analyses and to assure the relevance and
accuracy of the study. The models make use of a discounted cash flow rate of return technique to determine
the hydrogen production cost for the desired internal rate of return [54, 65]. In these analyses, not only a
comparison of costs is done but also energy and environmental tradeoffs are considered to come to a clean
energy future [65]. Renewable energy studies conducted in the USA claim that there should be enough wind
and solar potential to produce hydrogen for the transportation sector. However, up to now hydrogen
production is not yet competitive with the commercial (fossil) fuels. Too high electricity and capital costs for
electrolyzer systems are the main obstacles for hydrogen to become competitive [23].

The goal of these analyses is to estimate the hydrogen production cost. The final cost of hydrogen, expressed
in €/kgH2, is broken down into capital costs, feedstock costs, and operating and maintenance (O&M) costs
[23].

Capital costs are determined using correlations developed in the Electrolysis Milestone Report [66], with 𝑦
the electrolyzer capital cost (k€) and 𝑥 the production capacity of the unit (kg/h). This correlation is valid
between 0.1 kg/h and 100 kg/h production capacities.

𝑦 = 224.49 𝑥 0.6156 (18)


26 Techno-economic analysis

Capital costs are determined by the electrolyzer stack, the power electronics, gas conditioning and the
balance of plant (collection of other parts or systems that contribute to the cost of the unit). Reports suggest
that the electrolyzer stack and power electronics appear most promising for future optimization concerning
cost-effectiveness, efficiency increase and mass production [23].

During analyses on the production cost of hydrogen distinction can be made between distributed production
facilities and centralized production facilities. As renewable energy resources are distributed over the world
and high transport and delivery costs could be avoided, distributed production facilities, which are located
close to the energy source and produce a limited amount of hydrogen, could be a viable option [18, 23, 42-
44, 54]. However, advantages from an economy of scale and the increasing demand for hydrogen fuel are in
favor of central production facilities, which have a higher capacity and can be located further away from the
energy sources but will process electricity feeds from a collection of the distributed energy sources [23, 44,
54].

In this section a comparison will be made between hydrogen cost calculations for PEM electrolysis and
alkaline electrolysis systems for both distributed and centralized production facilities. Input parameters
necessary to determine hydrogen production costs are given in Table A-6 in Appendix A [54, 65].

a) Economics of PEM electrolysis

Evaluation of PEM electrolyzers takes in account costs for: (a) the engineering system definition, (b) capital
costs, (c) operating costs, (d) variable and fixed expenses, and (e) replacement costs [54]. Assumptions made
for the distributed production are a 1500 kgH2/day production, hydrogen outlet pressure of 31 bar for current
production and 70 bar for future production, and a depreciation period of 7 year. For the central production a
higher capacity of 50 000 kgH2/day is assumed and depreciation is spread over 20 years [44, 54, 65].

Table 2-2 gives an overview of the main contributions to the hydrogen production cost as identified by the
U.S. Department of Energy [54]. Results are shown for current and future production and for distributed and
central facilities. The costs are represented on a €/kg basis. Electricity feedstock is clearly the main cost
driver.
Techno-economic analysis 27

Table 2-2 Hydrogen production cost breakdown for current (2013) and future (2025), distributed (1500
kgH2/day) and central (50 000 kgH2/day) production facilities. Costs are shown on a €/kgH2 basis5 [54].
Current Future
Current Distributed Future Distributed
Central Central
1,500 kg/day 1,500 kg/day
50,000 kg/day 50,000 kg/day
Stack Capital Cost 0.37 0.14 0.42 0.15
BOP Capital Cost 0.54 0.22 0.47 0.23
Indirect Capital Cost
and Replacement 0.28 0.14 0.28 0.09
Cost
Decommissioning 0.02 0.01 0.00 0.00
Fixed operations and
0.37 0.16 0.35 0.18
maintenance (O&M)
Electricity Feedstock 2.94 3.04 2.97 3.04
Variable O&M 0.01 0.01 0.01 0.01
Total H2 Production
4.53 3.72 4.50 3.70
Cost (€/kg H2)

Notice that in this study it is assumed that the electricity feedstock for the distributed production facilities is
cheaper than for the central production facilities. This is caused by the necessity of transportation of the
energy from the distributed sources to the central facilities.

The results shown in Table 2-2, for the hydrogen production cost determined by the U.S. Department of
Energy, will be evaluated by a sensitivity analysis to determine the impact of certain parameters. This can be
done for the different cases of current and future distributed production and current and future central
production. As electricity is the main cost driver the dependence on average electricity price will be
examined, next to the dependence on the capital costs.

Figure 2-7 illustrates the sensitivity analysis on the hydrogen production cost (€/kg H2) for a low, baseline and
high value of the average electricity price over the plant lifetime (c/kWh); and this for current and future
distributed production and current and future central production. The sensitivity limits are taken as ±50% of
the baseline value.

5
U.S. Dollar to Euro exchange rate: $1 = €0.8797
28 Techno-economic analysis

Figure 2-7 Sensitivity analysis on the hydrogen production cost (€/kgH2) for a low, baseline and high value of the
average electricity price over the plant lifetime (c/kWh). Results are shown for the central distributed production
(top left); future distributed production (top right), current central production (bottom left) and future central
production (bottom right). Sensitivity limits are taken as ±50% of the baseline value [54].

The dependence on the average electricity price is immediately visible and is obviously quite large as
electricity feedstock is the main cost driver for the hydrogen production cost. An increase or decrease of 10%
in electricity price results in an in- or decrease of the production cost of about 6 to 8%. From Figure 2-7 it
can also be seen that the dependence on the average electricity price is higher for larger production
capacities. This is because the major cost for these installations is the electricity feedstock and capital costs
will be less dominant. For smaller production units this relation will be different and capital costs will have a
higher contribution to the production costs [23].

Figure 2-8 gives the sensitivity analysis on the hydrogen production cost (€/kgH2) for a low, baseline and
high value of the uninstalled capital cost (€/kW) for the same four cases. The dependence of the production
cost on the capital cost is less pronounced than on the electricity price. An in- or decrease of 10% of the
uninstalled capital cost would lead to an in- or decrease of the production cost in a range of 1-4% for the four
different cases, which is significantly lower than for the electricity price.

Similar conclusions as for the previous cases can be made on the differences in dependence for the
production cost between the distributed and central production facilities. The larger installations will be less
dependent on the capital cost which is reflected in Figure 2-8, but smaller installations will have a higher
dependence because less product is available to distribute the capital costs [23].
Techno-economic analysis 29

Figure 2-8 Sensitivity analysis on the hydrogen production cost (€/kgH2) for a low, baseline and high value of the
uninstalled capital cost (€/kW). Results are shown for the central distributed production (top left); future
distributed production (top right), current central production (bottom left) and future central production
(bottom right) [54].

b) Economics of alkaline electrolysis


Analysis is also done for alkaline electrolysis. Similarly, distinction is made between distributed and central
production facilities and current data as well as future prospects. Hydrogen production cost, capital costs,
energy efficiencies and electricity price for central and distributed production are given in Table 2-3. More
detailed cost data is given in Table A-7 to Table A-10 in Appendix A. Targets for development were set and
require advancement in material selection and design, as well as advantages from increasing production
capacities [23]. Cost determination is based on the same assumptions made for PEM operation, given in
Table A-6 in Appendix A.

Table 2-3 Hydrogen production cost breakdown for current (2015) and future (2020), distributed (1500
kgH2/day) and central (50 000 kgH2/day) production facilities. Costs are shown on a €/kgH2 basis6 [44].
Current Future Current Future
Distributed Distributed Central Central
1,500 kg/day 1,500 kg/day 50,000 kg/day 50,000 kg/day
Capital Cost 0.50 0.50 0.50 0.40
Fixed operations and
0.20 0.20 0.10 0.10
maintenance (O&M)
Energy efficiency (%) 72 75 73 75
Electricity Feedstock 4.80 3.30 4.20 3.10
Energy cost (€/kWh) 0.070 0.037 0.049 0.031
Variable O&M 0.10 <0.10 0.10 0.10
Total H2 Production Cost
5.60 4.00 4.90 3.70
(€/kgH2)

6
U.S. Dollar to Euro exchange rate: $1 = €0.8797
30 Techno-economic analysis

Similar to PEM electrolysis, electricity feedstock is the main cost driver. For alkaline electrolysis it is not
possible to work at high pressures, and an additional cost contribution from compression significantly raises
the hydrogen production cost [41, 44, 51].

Similar conclusions can be made for a sensitivity analysis on alkaline electrolysis as for PEM electrolysis
systems [23]. The electricity feedstock is the main cost driver and hence a fluctuation in the electricity cost
price or usage will have the highest impact on the production cost of hydrogen. Capital costs also have a high
contribution and influence the production costs. Similar as for PEM electrolysis, systems with high
production capacities will have a higher dependence on electricity cost price and usage relative to the capital
costs. This will be different for systems with smaller production capacities, where the capital cost will have a
higher contribution to the production costs although electricity cost price and usage will still dominate.

c) Comparison of alkaline and PEM electrolysis


Comparison of PEM and alkaline electrolysis (see Table 2-4) shows PEM electrolysis to be the advantageous
technology. Especially for distributed production facilities, PEM appears to have a favorable production cost.
PEM electrolysis also allows operation at higher pressure avoiding additional costs for compression.
However, no large scale applications exist, probably because high capital cost are an obstacle [60, 61].
Alkaline electrolysis is already a mature technology and hence dominates high-capacity hydrogen production
from electrolysis. According to the DOE, there is however still room for improvement, mainly in reduction
of capital costs and energy efficiency. Reduction of the electricity feedstock price also has a large effect on
the hydrogen production cost [44].

Table 2-4 Comparison of the cost analyses of alkaline and PEM electrolysis hydrogen production for current and
future, distributed and central production facilities [44, 54].
Future distributed Current central Future central
Current distributed
production (2020- production (2011- production (2020-
Case Units production (2011-
2025, 1500 2013, 52 000 2025, 50 000
2013, 1500 kgH2/day)
kgH2/day) kgH2/day) kgH2/day)
Alkaline
€/kgH2 5.60 4.00 4.90 3.70
electrolysis
PEM
€/kgH2 4.52 3.72 4.50 3.69
electrolysis

Several optimization challenges still need to be tackled to improve the electrolysis operation and to
strengthen the competitive nature of hydrogen production with respect to commercial fuel production.
Electricity cost, which represent a major contribution to the hydrogen production cost, should be reduced by
developing new materials and systems to improve the efficiency of the process. New designs will be
necessary using lower cost materials and advanced manufacturing methods to develop larger systems, reduce
the capital costs of the electrolysis system, and take advantage of economies of scale. Stacks with integral
electrochemical compression schemes should be developed that are capable of producing hydrogen at higher
pressures to reduce compression costs. Next to optimization of the electrolysis system, also the price of the
electricity feedstock should be optimized. Electricity price lower than $0.04/kWh should be strived for, but
Techno-economic analysis 31

coming from near-zero emission sources. Better integration with renewable electricity sources such as wind
and solar energy sources are of high interest [17, 44, 55].

2.3.5. Hydrogen Purification and Enrichment

In the production of hydrogen, attention has to be paid to the quality of the hydrogen produced. Certain
requirements need to be met before usage in other processes is possible. Therefore, purification and
enrichment is of major importance in hydrogen production facilities. This is mainly important for the current
commercial hydrogen production from fossil fuel sources as hydrocarbons are present in the gaseous
mixture. In case of electrolysis applications oxygen is present in the gaseous mixture which needs to be
removed for further processing [44].

Several aspects need to be taken in account during the selection of the appropriate separation technique: the
system needs to be capable of handling the expected process stream, a high recovery and purity of the
hydrogen product is required, and the equipment needs to be highly resistant against possible poisoning by
impurities in the system.

Important techniques for the purification of hydrogen are: (1) separation with hydrogen permeable
membranes; (2) catalytic purification; (3) metal hydride separation; (4) pressure swing adsorption; and (5)
cryogenic separation [67, 68].

The palladium membrane technique is capable of producing the purest product stream, above 99.9999%
purity, and delivers a recovery up to 99%. However, it is only capable of small to medium capacities.
Catalytic purification is a technique with a high purity (99.999%) and high recovery (up to 99%) and can
handle small to large scale capacities. It is mainly used for removal of oxygen, by catalytic reaction with
hydrogen. The metal hydride separation is based on the reversible reaction of hydrogen with metals to form
hydrides and has a high purity for the hydrogen product stream. However, the recovery is in a range between
75 to 95% and is only capable of handling small to medium process streams. Pressure Swing Adsorption
(PSA) systems will selectively adsorb impurities from the gas stream and can handle any hydrogen rich gas
stream, even for large scale capacities. It yields a very high purity stream, up to 99.999%, but has a relatively
poor recovery of hydrogen of about 70 to 85%. Cryogenic separation is mainly used for petrochemical and
refinery off-gases, and large scale capacities. Its purity and recovery of hydrogen goes up to 98% and 95%
respectively [68, 69].

2.3.6. Relevance of hydrogen production footprint

In search of an alternative energy carrier to replace fossil fuels, hydrogen has been established as a possible
solution. As CO2 emissions are an important driving force in the search for alternative fuels, CO 2 emissions
should be minimized in hydrogen production.
32 Techno-economic analysis

Although footprints for hydrogen production from renewable energy sources are quite difficult to predict as
these technologies are still in stage of development, estimations should be taken in account for the selection
of appropriate technologies towards alternative fuels, even if the actual estimations are highly dependent on
specific cases [17].

Current hydrogen production is mainly from steam reforming of methane. Life cycle analyses describe
emission values of about 8.9 kgCO2 per kgH2 [70]. This could significantly be reduced by utilization of
renewable production processes in which on average 5 to 6 kgCO2 per kgH2 is emitted [17]. It has even been
reported that CO2 emission estimations for electrolysis provided with renewable wind electricity could be
reduced to 1 kgCO2 per kgH2 [71]. Emissions for hydro-electric or solar thermal conversion by electrolysis are
reported to be reduced to 2 kgCO2 per kgH2 [17].

2.3.7. Hydrogen as an intermediate step towards energy storage

Hydrogen could be a good transportation fuel and has the advantage of only producing energy and water
when burned. Unfortunately it has some serious setbacks which makes it more interesting to use it as a
feedstock for chemicals or liquid fuels.

A first issue that concerns hydrogen is its low volumetric energy density compared to commercial liquid
fuels which makes it inconvenient for transportation and storage. Due to its high volatility, hydrogen needs
to be compressed at high pressure (350-700 bar) or liquefied using cryogenic temperatures (-253°C) to
increase the volumetric energy density. Storage is also complicated as hydrogen diffuses through a lot of
metals and materials.

Distribution of hydrogen would require a lot of replacements on the existing infrastructure for liquid fuels
and would represent a high cost. Due to the high flammability of hydrogen, safety measures would augment
these costs.

Although hydrogen could be an alternative for fuels, base chemicals cannot be replaced by hydrogen solely.
Chemicals will remain carbon based and hence the dependence on fossil fuels cannot be nullified via
hydrogen production solely [11, 31]. Making use of hydrogen to convert the excessive amount of CO 2 into
liquid fuels or base chemicals would be a possible way to solve the problems concerning hydrogen. In the
following sections some important opportunities for the conversion of H 2 and CO2 towards liquid fuels and
base chemicals will be discussed.
Techno-economic analysis 33

2.4. CO2 capture and sequestration


Recycling of carbon is becoming more and more important in our society; it is an important objective
towards a sustainable future energy market and has beneficial effects on the detrimental CO 2 emissions.
Several techniques exist to convert carbon dioxide, but not every source of carbon dioxide is usable for
exploitation and certain separation or purification steps are necessary [31].

Although carbon capture will become a major issue to guarantee a closed carbon cycle, multiple highly
concentrated CO2 emission sources exist from which carbon can be captured without too much effort. Point
emissions from coal-fired power plants oil refineries or cement factories should allow sufficient CO2 capture
for utilization in initial conversion processes (Table 2-5). Research should therefore be focused first on
developing adequate conversion processes [17]. As these processes will mature and higher amounts of CO 2
will be consumed, the need for high-tech solutions for carbon capture will be of increasing importance. The
focus of this thesis is on the conversion processes for carbon dioxide; therefore the techniques for carbon
capture will only be discussed briefly.

Table 2-5 CO2 emissions from various industrial sectors in Mt CO2 per year [17].

Ethylene oxide 10−15


LNG sweetening 25−30
Ammonia 160
Ethylene (and other petrochemical processes) 155−300
Fermentation >200
Iron and steel ca. 900
Oil refineries 850−900
Cement >1000

Several techniques exist for carbon capture and Figure 2-9 illustrates the main technologies currently
available. Removal and capture of CO2 from gas streams can be achieved by a range of separation techniques
depending on CO2 concentration, pressure, temperature, etc. [31].

Figure 2-9 Most important separation techniques for carbon capture [31].
34 Techno-economic analysis

Current CO2 capture is mainly done from high concentration emission sources such as power plants, but
exploitation could also be possible from natural gas sources or geothermal vents. From these high
concentration sources separation is easier, more cost-effective and less energy demanding. It is mostly done
by using liquid absorbents, such as alkanolamine solutions, or by cryogenic separation. These techniques are
very costly and energy intensive. Therefore research is ongoing towards effective solid absorbents, which
seems to be a promising technique for large scale CO2 capture. Other improvements for new power plants to
enhance the carbon capture are precombustion CO2 capture or oxy-fuel combustion techniques. At this stage
of development, power plants and other facilities still make use of conventional technologies causing the
postcombustion capture of CO2 [31]. Energy requirements for post-combustion CO2 recovery from coal-fired
power plants are given in Table 2-6.

Table 2-6 Estimated energy required to remove and recover CO 2 from coal-fired power plants using various
technologies [31].

Process CO2 removal efficiency (%) kWh(e)/kgCO2 recovered


Improved amine absorption/ stripping
90 0.22
integrated plant
Oxygen/coal-fired power plant 100 0.30
Amine (MEA) absorption/ stripping
90 0.54
nonintegrated plant
Potassium carbonate absorption/ stripping 90 0.64
Molecular sieves adsorption/ stripping 90 0.80
Refrigeration 90 0.80
Seawater absorption 90 1.60
Membrane separation 90 0.72

In cryogenic separation CO2 and N2 are separated from the flue gases prior to cooling. The remaining
mixture is then separated by sufficient cooling, and pressurized to liquefy the CO2. This is a very energy
intensive process and assumes all other components such as NOx, SOx, etc. to be removed prior to cooling.
Therefore it is considered impractical for large process streams. The sole advantage of cryogenic distillation
over other techniques is the liquefied product which can easily be transported [72].

Membrane separation is based on the principles of diffusion. The porous membranes allow only gases of a
certain size to pass through the pores and hence act as a sieve to separate CO 2 from larger molecules.
Membrane separation has the advantage of being a straightforward technique with only a limited amount of
equipment necessary. However, membranes are mainly suited for high concentrations of CO 2 and lower
performance at high temperature causes difficulties to use membranes for flue gases exiting the stack [31,
72].

Absorption with liquid absorbents is currently the most suitable technology for CO 2 extraction from high-
volume flue gas streams. It is a well-known and mature technology with high selectivity for CO2 capture.
The carbon dioxide is extracted from the flue gas by scrubbing with a solvent, usually an amine-based
solution e.g. monoethanolamine (MEA) [22, 40]. Regeneration of the fixed CO2 is done via heating of the
mixture in a regenerator column. Drawbacks are the high energy requirements for the CO2 regeneration step,
limited amine loading due to corrosion, and amine decomposition and degradation by compounds such as fly
Techno-economic analysis 35

ash or other particulates as SOx and NOx, implying a necessary flue gas desulphurization (FGD) step before
absorption is possible [31, 72].

Solid chemical adsorbents (e.g. CaO, MgA, hydrotalcites) also have a high capacity to trap CO 2 and are
widely available at low cost. They require higher temperatures (200-500°C) to reach acceptable adsorption
rates; 50% absorption can be reached within one hour. The adsorbent can be regenerated within 15 minutes
but this requires even higher temperatures than adsorption and buildup of stable minerals on the surface will
decrease the adsorption rate. Improvement of existing systems focuses on new materials that are more easily
regenerated and are less prone to degradation [31, 72].

Solid physical adsorbents (e.g. silica gel, alumina, activated carbon, zeolites) rely on their potential to
reversibly adsorb species into small cracks, pores or their external surface. The main techniques used for
adsorption are pressure swing adsorption (PSA) and temperature swing adsorption (TSA) and depend on the
partial pressures of CO2, temperature, interaction between the sorbent and CO2, and pore size of the
adsorbent. Disadvantages of these techniques are its fast decrease in adsorption capacity with increasing
temperature, limited range of CO2 concentrations (0.04 – 1.5%) and low selectivity for CO2 adsorption.
Metal-organic frameworks (MOF) appear to be a promising technique but require further research before
commercial applications are possible. Other possibilities such as solid amines, or amines chemically bound
to solid surfaces are under investigation to increase the selectivity for CO 2 adsorption of solid adsorbents [31,
72].

As ultimately the high concentration sources will disappear, techniques will be necessary to capture carbon
from other sources such as air, which would eventually provide an almost limitless source of carbon [31].
Half of the CO2 emissions come from small and dispersed sources which vent CO 2 directly into the
atmosphere. Capture of these emissions at the source seems unreasonable and not economically feasible due
to the necessity of extended and costly infrastructure and due to logistic inconvenience. Therefore, when
high concentration sources are depleted, direct carbon capture from the atmosphere will become necessary.

Research towards carbon capture from the atmosphere is currently done on the fixation of CO2 with basic
absorbents such as calcium hydroxide (Ca(OH)2) and potassium hydroxide (KOH) to form their associated
carbonates. CO2 could then be regenerated by desorption through heating, vacuum, electrochemical means or
any other method. Current absorbents however still require too high energy input for the regeneration of
CO2.

Although CO2 capture from the air is already done at small scale in submarines and space ships to keep the
air breathable, high capacity carbon capture from air is not likely to be executed until all major high
concentration CO2 emission sources are depleted, if applicable at all. For this application no transport of CO2
towards the plant is necessary as the atmosphere operates as a cost-free natural CO2 conveyor belt,
transporting CO2 from the dispersed emission sources. Hence the location of the plant is of minor importance
and as CO2 equilibration is relatively rapid, local depletion is not likely to occur. The major drawback though
for direct carbon capture from air is the very low concentration of CO2 (0.039%). Consequently, large
36 Techno-economic analysis

volumes of air need to be processed resulting in large infrastructure requirements and high capital costs. To
treat the carbon dioxide emissions from a 1000 MW coal power plant (~6 million tons of CO2 per year),
structures of 10 m in height and 30 km in length filled with absorbents would be required.

Despite its high capital cost, direct carbon capture from the atmosphere has several advantages. Carbon
capture would be independent of carbon sources and has no restrictions on the location of the plant. It also
allows the possibility to extract more CO2 than is emitted by human activity, thus stabilizing CO2 levels or
even reducing the atmospheric level of carbon dioxide [31, 73, 74].
Techno-economic analysis 37

2.5. Liquid fuel and raw chemical production


To introduce renewable energy into the current energy economy, conversion towards chemical feedstock and
fuels is a viable option. This would match the current energy carriers, which are predominantly fossil fuels.
Conversion to chemical feedstock and fuels would be an opportunity to validated the excess energy from
renewable sources and only minor changes and costs would be necessary to modify the existing
infrastructure. However, a price will have to be paid for the loss of efficiency by converting electricity into
chemical energy [10]. Several vector molecules exist to inject the renewable energy into our energy
economy, e.g. hydrogen, ammonia and methanol [17, 43]. In this thesis, focus is put on the conversion of
carbon dioxide.

In search of alternative reaction routes towards renewable liquid fuels or chemical raw materials, three
important aspects need to be taken in mind according to Tremel et al. [43]. A first concept that needs to be
evaluated is acceptance; acute toxicity and ecotoxicity based on the material safety data sheet (MSDS) need
to be evaluated, especially when contact with people is possible. Potential use of existing infrastructure
increases the acceptance of an alternative process route. Secondly, the technological aspect of the process is
evaluated. Processes making use of milder process conditions and in need of fewer upgrading steps will be
preferred. The third aspect discussed is the economic part. Most important parameter is the ratio of market
price to production costs, but also the production rate compared to the world market volume, capital
expenditures and the ratio of fuel additive price to production costs can be taken in account.

Products that are believed to be viable options for CO 2 conversion are methanol, short-chain olefins
(ethylene and propylene), methane, higher hydrocarbons, carbon monoxide, DME, higher alcohols, and
polycarbonates [10, 17, 43, 75]. In this master thesis the production of methanol, ethylene and formic acid
will be studied.

2.5.1. Methanol synthesis

Although methanol includes a severe risk towards humans, public acceptance is quite high due to its low
ecotoxicity. Indeed, methanol can be added to classical fuels and is believed to be a viable option towards
renewable fuels. It is also a chemical feedstock for various products, e.g. plastics, synthetic fibers, solvents,
paints and others. The moderate operating conditions and straightforward separation of the methanol product
gives it also a technological advantage. Compared to other processes the ratio of market price to production
costs of methanol from renewable sources is reasonable, and further the methanol process requires only a
moderate investment cost which makes it economically attractive. A ready and extended market for methanol
is available with an annual production in excess of 70 million tons [40, 43, 76, 77].

Applications for methanol are numerous and it can be used in its pure form or as an intermediate for
production of more valuable products. It can be used as a transportation fuel, due to its high octane rating and
easy transport, and only a small incremental cost is necessary to modify the current vehicles to run on blends
38 Techno-economic analysis

of methanol fuel. Methanol can also be used as an excellent hydrogen carrier for fuel cell applications which
become more important in our energy economy. Also in the production of biodiesel, methanol is a key
component. Further, methanol is an important feedstock in the production of olefins and other synthetic
products [76].

Industrial production of methanol is currently based on syngas conversion in a catalytic system at


approximately 250°C and 50 bar, but production from a CO2 feedstock is a promising and much debated
technique. Though, before a true methanol economy will be available, major increase in production capacity
of methanol would be required to replace the actual fossil fuel consumption. Global petroleum consumption
is estimated at 30 billion barrels per year; hence a production of about 60 billion barrels of methanol would
be required, taking in account the volumetric energy densities of methanol and gasoline. This would
represent about 160 times the current methanol production [31, 77].

In the conversion of CO2 to methanol, methanol is used as a hydrogen carrier. Methanol and its further
derived fuels can be used in the existing distribution facilities, power generators, vehicles and other
appliances. Both the penetration of renewable power, via hydrogen production, and the reduction of CO 2
emissions can be enhanced significantly on a relatively short timescale [42]. Conversion to methanol will
also augment the volumetric energy content, which is for hydrogen a lot lower than commercial fuels
nowadays.

2.5.1.1. Reaction cycle


The main catalysts used in methanol production are Cu-Zn-oxide based, possibly with promotors such as
ZrO2, Ga2O3 or SiO2 to increase the catalyst dispersion and stability, and enhance the catalyst performance.
For the methanol synthesis from CO2 three possible reaction paths have been identified, visualized in Figure
2-10. The first is the formate pathway; secondly the combination of the reversed water-gas shift (RWGS)
with the syngas-to-methanol process and as third the hydrocarboxyl hydrogenation pathway. Both the
formate and RWGS + syngas-to-methanol pathway have been reported to occur on the catalyst surface,
although different pathways will be dominant at different operating conditions [31, 75]. Other researchers
however found that the reaction barriers for the hydrocarboxyl hydrogenation pathway are lower than both
other mechanisms [6]. Debate is still ongoing on this topic. The total reaction equation for methanol
production is given in Eq. (19).

CO2 + 3H2 ⇌ CH3 OH + H2 O (∆H298K,1bar = −49.16 kJ mol−1 ) (19)

According to the reaction scheme (Eq. (19)), 1 kg of hydrogen would allow the conversion of approximately
7.3 kg CO2 and production of 5.3 kg methanol. Reports suggest that actual catalysts are capable of reaching
equilibrium conversions for CO2 of approximately 20%, and in case recycle streams are taken in account
conversions up to 89% could be obtained [78].
Techno-economic analysis 39

Figure 2-10 Proposed pathways for the Copper-catalyzed hydrogenation of CO2 to methanol [6, 79].

Throughout the reaction several by-products are formed. Especially water is an important by-product as it
has an inhibiting effect on the catalyst and continuous water removal via distillation or membranes would
enhance the reaction productivity. Experimental analyses have shown that an intermediate water removal
step could also greatly enhance the reaction productivity compared to the single-stage process. Other by-
products are CO, HCOOCH3 (methyl formate), hydrocarbons and higher alcohols, which can be extracted for
further processing [17, 22, 75, 80].

2.5.1.2. Economic evaluation of methanol synthesis


Actual production costs of methanol from syngas are 0.08 €/kg and according to Aresta et al. [10] methanol
can be produced from electrolytic hydrogen at a cost of 0.3 €/kg, assuming CO 2 is easily available and
assuming the best case for Capex and Opex. Carbon taxes might compensate for the higher cost of methanol
from electrolytic hydrogen. Actual C-taxes (if present) are found in the range of 30-100 €/ton or 0.03-0.1
€/kg, which would lead to a methanol cost of about 0.16 €/kg and thus reduces the additional cost of
methanol from renewable energy source.

To model the cost for the production of methanol via CO2 conversion with H2 (see Eq. (19)), several studies
have been conducted in which a modeling tool, such as Aspen, is used to simulate the conversion process
[22, 40, 42].
40 Techno-economic analysis

A simplified scheme of the process is represented in Figure 2-11. Carbon dioxide is extracted from the flue
gases from a coal-driven power plant by amine absorption, using monoethanolamine (MEA) in an absorption
column. Other techniques are possible for CO2 extraction and energy requirements are given in Table A-11
in Appendix A. Flue-gases from a coal power plant have a high CO 2 concentration (up to 16% [42]), and are
therefore attractive for carbon capture as the costs for absorption can be confined. By stripping, the absorbed
CO2 is extracted from MEA. Steam for the stripping of the CO2 – MEA mixture will be supplied by the
overproduction of steam in the methanol synthesis and steam from waste heat of the power plant. Hydrogen
is supplied from water electrolysis plants, see above, and the fluctuating availability of excess energy from
renewable sources is taken in account. Distinction is made whether the oxygen produced in the electrolysis
plant can be validated or not [22, 40].

Figure 2-11 Block diagram of the CO2 to methanol conversion process. [22]

A study by Mignard et al. [40] describes four case studies for the production of methanol based on the
previously mentioned assumptions. A coal power plant is designed, producing a total power output of 1000
MWel for distribution. CO2 is captured from the resulting flue-gases and hydrogen is generated by the
electrolysis of water with renewable electricity. The availability of renewable electricity is modeled to be
dependent on the excess supply of renewable energy sources. Three periods are assumed during a 24h day
and renewable electricity is assumed to be available only at off-peak periods. The model values for
electricity supply and cost are given in Table 2-7. In periods of high-energy demand, renewable energy is
expected to be unavailable and the model uses fossil electricity to operate during these periods.
Techno-economic analysis 41

Table 2-7 Simplified schedule for the availability of renewable electricity. Three periods are assumed during a
24h day and renewable electricity is assumed to be available only at off-peak periods. Fossil electricity is used in
periods with no excess supply of renewable energy [40].
Excess supply (MW) Duration (h) Price (€/kWh)7
Period 1 500 4 0.019
Period 2 100 12 0.013
Period 3 0 8 -

The four case studies are:


 Option A: makes use of conventional alkaline electrolysis, with variable excess renewable energy
supply.
 Option B: equal to A but assumes constant excess renewable energy supply.
 Option C: uses pressurized electrolysis to save energy on compression.
 Option D: uses a hydrogen fuel cell plant to reduce the fossil electricity usage.

An additional consideration made is the possibility of validating the oxygen produced during water
electrolysis. Via modeling with a simulation package, e.g. Aspen, the different cases can be evaluated and a
technical as well as economic analysis is possible.

Results for the mass and energy balance of the simulations are given in

Table A-12 and Table A-13 in Appendix A. In these simulations it is taken in account that indirect emissions
will result from the use of electricity for compressors, pumps and other equipment, as well as from the
purification steps of the CO2 feedstock extracted from the power plant’s flue gases, e.g. NOx and SO2
removal. Important to notice from the general mass balance is that net abatement of CO 2 is only possible
through the availability of waste heat from the power plant to generate low pressure steam. If steam
production from fossil energy would be necessary, the average CO2 emission would be comparable to the
usage and abatement would be nullified.

Comparison of the mass balance for the different cases shows that option D has an advantage on CO 2
abatement on a percentage basis. Theoretically, zero emissions result as it makes use of a fuel cell plant
which is fed with hydrogen generated by water electrolysis with renewable electricity. Hence, 100% of the
incoming CO2 is abated, which is significantly higher than the other cases which make use of fossil
electricity to overcome the high-demand periods. Option C, making use of pressurized electrolysis, gives
abatement of approximately 90% and option B and C give slightly higher abatement than 80% of the CO2
recycled to the system. Conversion efficiencies from electrical to chemical power are similar for the different
cases and are in the range of 60-70% (see Table A-14 in Appendix A).

However, more important results are the costs for methanol production following from each case studied.
Table 2-8 gives a cost breakdown of the capital expenditures and a summary of the variable operating costs

7
British pound to euro exchange rate: £1 = €1.2541.
42 Techno-economic analysis

for each option. Capital investments are expected to be similar to a conventional syngas-based methanol
synthesis plant with in addition the equipment necessary for electrolysis [31]. Calculation of the operating
costs is done assuming 8000 working hours per year and unit costs of €0.74/t steam, €0.06/tcw and €0.05/kWh
(fossil power). Fixed operating costs were taken as 5% of the total capital expenditures.

Table 2-8 Cost breakdown for the capital expenditures studies and summary of the variable operating costs for
each option. Calculations of the variable operating costs assume 8000 working hours per year and unit costs of
€0.74/tsteam, €0.06/tcw and €0.05/kWh (fossil power) [40].
Capital Expenditures in €106
Option A Option B Option C Option D
Electrolysis 176.83 43.52 114.00 43.14
Extra de-NOx capacity 0.13 0.07 0.08 0.07
Extra FGD (Wellman–Lord) 3.60 1.72 1.94 1.69
CO2 recovery 11.74 5.59 6.33 5.52
H2 compression and storage 22.07 10.53 11.93 10.38
CO2 compression 5.59 2.67 3.01 2.62
Reactor 3.09 1.47 1.67 1.45
Separation 1.27 0.60 0.68 0.59
Water purification 0.83 0.40 0.45 0.39
Fuel cell plant 0.00 0.00 0.00 68.35
Total 225.15 66.56 140.08 134.21

Variable Operating Costs in €106/year


Option A Option B Option C Option D
Renewable power 17.56 7.52 7.52 7.52
Fossil fuel power 1.38 0.66 0.51 0.00
Cooling Water 1.72 0.88 0.94 0.87
LP steam 4.43 2.11 2.38 2.07
Feed and Chemicals 0.59 0.28 0.32 0.28
Total 25.71 11.45 11.68 10.75

From these cost estimations a minimum selling price can be calculated for a nullified NPV, in this study over
a period of 15 year. Taxes are taken at 35% of the taxable positive profits, and a yearly depreciation is taken
as 15% of the capital expenditures. The lifetime of the plant is fixed at 15 years. Minimum selling prices of
methanol for a nullified NPV over 15 years are presented in Table 2-9.

Table 2-9 Minimum selling price of methanol produced resulting in a nullified NPV over 15 years for each design
option. Option B shows the most advantageous selling price [40].
Option A Option B Option C Option D
Price (€/ton) 594 430 619 670

Option B appears to be the most feasible design option with a minimum methanol selling price, significantly
lower than the other cases. This can be attributed to the fairly low capital costs required for this design option
compared to the other cases. However, validation of the oxygen produced during water electrolysis has not
yet been taken in account.

Before oxygen can be validated, a local customer should be secured which might require careful planning
and the right environment for the plant. Customers could be chemical plants, and from the high purity of the
oxygen from water electrolysis (99.9%) even medical applications would be possible. In case the oxygen
would be validated throughout the process, several additional cost contributions arise. The most important
Techno-economic analysis 43

are compression, cooling and storage, as well as piping to transport the oxygen product. Piping can even
become a limiting factor, as transportation over 200 km would not be economically feasible. Design option B
and C are compared in Table 2-10 for the implementation of an oxygen plant for the valorization of O 2.
Option B was proven to be most economically feasible in case no local customer is available and option C
offers substantial savings in compression. Capital costs are similar for both options due to the compensation
of the extra compressors necessary in option B with the pressurized storage in option C. However, option C
appears to be much more favorable due to its lower operating costs and hence yields a much higher NPV
after 15 years of €253m.

Table 2-10 Comparison of option B and C for the implementation of an oxygen plant for the valorization of O 2.
Oxygen is assumed to be distributed via 20 km pipelines at 35 bar and is sold at €12.54 per 100 Nm³. Option C is
highly favorable, due to its lower operating costs [40].

O2 plant Option B Option C


6
Capital costs (€10 ) 6.27 6.40
6
Operating costs (€10 /year) 0.60 0.33
NPV (15 years, 10% MARR) €106 69.85 253.33

This study performed by Mignard et al. [40] clearly shows that the feasibility of methanol production from
CO2 and renewable energy will be highly dependent on the availability of renewable energy as well as the
market environment.

A comparable study performed by Van-Dal et al. [22] simulates a similar methanol conversion process. A
645 MW water electrolysis plant is however completely fed by renewable electricity and total methanol
production is approximately 500 000 ton per year. Modeling is done with the Aspen software and starts from
the same basic assumptions as described earlier. Mass, energy and CO2 balance are given respectively in
Table A-15, Table A-16 and Table A-17 in Appendix A. In this section a comparison will be made between
the economic parameters that were assumed in the previously described cases and the economic parameters
that represent the reality and future prospects to the best extent.

An estimation of the capital costs for the 645 MW plant in this case has been made using the correlation
given in Eq. (20), with as reference case the capital cost for the 500 MW plant from the previous simulations.
This results in a capital cost of approximately €263m [43].

𝑆2 0.6
𝐶2= 𝐶1 ( ) (20)
𝑆1
Assumptions concerning the utility costs are summarized in Table 2-11. Distinction is made between the
assumptions made by Mignard [40] and actual values found in literature. For simplicity only modifications
of power and methanol price are taken in account. The current price for renewable power is taken as 0.13
€/kWh which is an average value for the renewable energy cost provided by the Energy Information Agency
in the Annual Energy Outlook for 2015 (see Table A-19 in Appendix A). Prospects for energy costs are also
provided in the Annual Energy Outlook for 2015 [27, 40, 81, 82].
44 Techno-economic analysis

Table 2-11 Estimation of utility cost. Distinction is made between values used for the economic analysis by
Mignard and actual values found in literature [40, 81, 82].
Mignard cases Current Prospect (2040)
Steam (€/t) 0.74 0.74 0.74
Cooling water (€/t) 0.06 0.06 0.06
Fossil power (€/kWh) 0.05 0.09 0.11
Renewable power (€/kWh) 0.02 0.13 0.15
Oxygen (€/Nm³) 12.54 12.54 12.54
Methanol (€/t) 265.56 272.77 275.90

Operating costs for each case are given in Table 2-12 and a clear distinction can be made between the
different cases. Due to the high amount of renewable energy necessary in electrolysis, the cost for this
electricity feedstock will have a high influence on the overall operating costs. For the prospective case an
efficiency increase of 15% is assumed for electricity usage and a 10% decrease in capital cost is expected.
An increased methanol price for the prospective case is to be explained by the expected increase of fossil
fuel cost [12].

Table 2-12 Variable operating costs in €106/year for the Aspen simulations performed by Van-Dal with economic
parameters by Mignard and literature for current and prospective cases. Increased energy efficiency of 15% is
assumed for the prospective case [40, 81-83].

Mignard cases Current Prospect (2040)


Renewable power 84.15 647.12 625.80
Fossil fuel power 7.93 14.55 15.17
Cooling Water 0.01 0.01 0.01
LP steam 0.74 0.74 0.74
Feed and Chemicals 1.30 1.30 1.30
Total 94.06 663.42 642.10

Combination of the capital costs and the operating costs leads to a methanol price per ton (€/ton), see Table
2-13. Fixed operating costs are similar to the Mignard cases taken as 5% of the total capital expenditures.
Other assumptions are a lifetime of the plant fixed at 15 years and a yearly depreciation taken as 15% of the
capital expenditures.

Table 2-13 Methanol price per ton (€/ton) for the different cases starting from the simulations from Van-Dal.
Dependence on the electricity feedstock price is visible by comparing results from Mignard and current
literature values.
Mignard cases Current Prospect (2040)
Methanol price per liter (€/ton) 240 1430 1390

From these analyses it can be seen that the price for the renewable electricity feedstock will be a major
concern towards a feasible economy relying on renewable energy. The assumption for renewable energy
price in the Mignard paper appears to be very opportunistic, but relies on the fact that it only uses excess
energy available at cheap price. In cases where renewable energy is continuously used for electrolysis the
assumption is not valid and results in a sharp increase in methanol production price. Current methanol
production price based on continuous usage of renewable energy would result in a methanol production price
Techno-economic analysis 45

estimated at €1400 per ton, which is significantly higher than the actual selling price. Prospects for methanol
prices predict a decrease in methanol production price due to efficiency increase. Although renewable
electricity price is expected to increase in the future [27], energy efficiency optimization will lead to a
reduced production price.

The current methanol price for methanol in Europe (1/01 – 31/03) is set at 275 €/ton [82]. This is a lot lower
than the production price determined for methanol from CO2 and renewable energy. To obtain a methanol
production price that would be competitive with actual commercial methanol production, renewable energy
prices should be reduced to approximately 0.03 €/kWh. Carbon taxes, which are already used in several
countries, are currently set in a range of €30 to €100 per ton of CO 2 equivalent. Taking in account a carbon
tax of €100 per ton of CO2, commercial methanol prices would increase to approximately 500 €/ton and the
price gap between renewable and commercial methanol could be reduced. However, renewable energy prices
would still be too high for renewable methanol to be competitive and a reduction to 0.06 €/kWh would be
required.

2.5.2. Ethylene production

Ethylene is one of the most produced hydrocarbons worldwide (around 200 Mt per year) and is currently
mainly produced from fossil fuels. It is used for synthesis of diverse products such as plastics (polyethylene,
polystyrene, PVC, etc.) and textiles (e.g. polyester). High-grade ethanol is produced via hydration of
ethylene and also gasoline, diesel and jet fuel compounds are produced via polymerization.

Current production of ethylene is derived from steam cracking of long chain hydrocarbons or natural gas
fractions. The desired outcome such as ethylene/propylene ratios and by-products determines the choice of
cracking feed. Steam cracking is one of the most energy-consuming processes in chemical industry (≈ 3 EJ
of primary energy per year) and has a high contribution to CO2 emissions (≈ 200 million tons per year);
about 1.5 to 3 tons of CO2 per ton of ethylene is produced. Furthermore, depleting oil supplies force the
industry to explore alternative production routes for base chemicals [12, 17, 84].

Due to their high energy of formation, short-chain olefins are an excellent opportunity to store renewable
energy and incorporate it in the value chain of chemical production processes. Ethylene is, along with
propylene, the major base chemical in our society and is used in numerous applications. Although the carbon
recycling for ethylene production would not be sufficient to compensate for the imbalance of CO 2 emissions
and natural recycle, it is an interesting compound for research on large scale production processes involving
carbon recycling [11]. Compared to liquid fuels, olefins and other raw materials for petrochemistry also have
the advantage of a higher added-value which increases the cost-competitiveness of olefin production from
CO2 and renewable energy; and would be a possibility to store CO2 over a longer period of time as ethylene
and propylene are mainly used for polymer production [12, 17].
46 Techno-economic analysis

Figure 2-12 illustrates the global ethylene and propylene market (in Mton) for 2010 and prospects for 2020
including a breakdown of the main production methods. Steam cracking and fluid catalytic cracking (FCC)
will probably remain the dominant production methods throughout the coming decades but alternative
methods are expected to grow and have an increasing impact on the ethylene market. Ethylene and propylene
prices, currently in a range of €1000-1200 per ton, are expected to increase following the expected oil price
raise due to depletion of easy available oil sources 8. This will lead to opportunities for alternative production
methods from CO2 to become cost-competitive with commercial ethylene production which will lead to a
moderation of the ethylene cost increase, due to a widening of the number of possible production sources for
base chemicals [12].

Figure 2-12 World market for ethylene and propylene with a breakdown of the main production methods.
Values are given for 2010 and prospect for 2020. Distinction is made between steam cracking, fluid catalytic
cracking (FCC), dehydrogenation from ethane or propane, oxidative dehydrogenation (ODH), production from
syngas and others [12].

2.5.2.1. Electrochemical reduction

The electrochemical reduction of carbon dioxide is a means to convert CO 2 to more reduced chemical
species such as formic acid, ethylene, or carbon monoxide with electricity as feedstock, and uses aqueous
electrolytes as H+ source. The half-reaction for the reduction towards ethylene on the surface is given by:

2CO2 + 8H2 O + 12e− → C2 H4 + 12OH − (21)

and requires a minimal energy addition of 3.6 MJ/mol. CO2 is believed to first undergo reduction towards
CHxO* (x = 0,1,2) bound on the surface. Via intermolecular C-C binding this yields C2HxO2* (x=0-4), which
can further be reduced to ethylene. Up to now, copper-based catalysts give the best results, but selectivity
improvements are still required [85]. The main possibilities for selectivity enhancement are reported to be
the optimization of the local pH at the surface and alterations of the surface morphology (step and edge
sites).

8
US Dollar to Euro exchange rate: $1 = €0.8797.
Techno-economic analysis 47

Application of acidic solutions (pH < 7) is a means to reduce the poisonous deposition of graphitic carbon
and enhance the catalytic activity and efficiency of the process. Hence the rapid decrease of efficiency during
electrolysis is confined, and the catalytic activity is maintained by avoiding the deactivation of the catalyst
by deposition of poisonous compounds [85, 86].

For electrochemical reduction to become feasible, high energy efficiency and high reaction rates are
required. High energy efficiency is achieved through a combination of high selectivity (Faradaic or current
efficiency) and low overpotentials (see section 2.3.4.1). Efficiencies for ethylene production are reported to
be around 65-70% and reaction rates are measured as current densities, determining the reactor size and thus
capital cost of the process [14]. As hydrogen production is also an attractive option for electricity storage,
efficiencies for CO2 reduction should be close to those for water electrolysis. Optimization of the reduction
processes will require the exploration of different catalyst and electrolyte types, and determination of optimal
temperatures as well as reactor design. Current technology still requires high overpotentials for CO 2
reduction, with as rate limiting step the formation of the ·CO 2- radical anion intermediate. One of the main
functions of the catalysts is to stabilize this intermediate in order to reduce the potential necessary to form
the radical. A second issue comprises the competitive hydrogen evolution reaction. Via selection of the
appropriate electrolytes and reaction conditions, the selectivity for CO 2 reduction over hydrogen evolution
should be guaranteed. Other issues which need to be resolved are the poor thermodynamic efficiency of CO 2
reduction due to the requirement of high overpotentials, low current efficiencies, slow kinetics, and poor
stability of the catalysts. An example of progress that already has been made in this field is the utilization of
gas diffusion electrodes (GDEs) to limit the effect of mass transfer of CO2 due to its low solubility in many
electrolytes. GDEs create a three-phase interface between the gaseous reactants, the solid catalyst and the
electrolyte, which facilitates the conversion process [14].

2.5.2.2. Modified Fischer-Tropsch (olefins from syngas: CO + H2)

Producing olefins from syngas via Fischer-Tropsch synthesis is a well-known production process and has
been under extensive study for more than 50 years [87]. Current operation already incorporates a small
amount of CO2 in the conversion process to promote the reaction rate and conversion of pure CO 2 in a
similar process is under investigation, but requires novel or improved catalysts [17].

The olefin synthesis can be presented as the combination of the RWGS reaction and consecutive FTS:

𝐶𝑂2 + 𝐻2 ⇌ 𝐶𝑂 + 𝐻2 𝑂 (22)

𝐶𝑂 + 𝐻2 ⇌ 𝐶𝑛 𝐻2𝑛 + 𝐶𝑛 𝐻2𝑛+2 + 𝐻2 𝑂 + 𝐶𝑂2 (𝑛 = 1,2, … ) (23)

The overall reaction for ethylene formation is 215 kJ/mol endothermic and the main catalysts used in this
process are iron and cobalt based FT-catalysts. Processes involving Fe-based catalysts are reported to
produce more lighter hydrocarbons, where processes involving Co-based catalysts produce higher amounts
of heavier hydrocarbons [12]. According to Xu et al. [88] doping with low amounts of potassium has
beneficial effects on the hydrocarbon yield, CO2 conversion and selectivity for light olefins, and inhibits the
48 Techno-economic analysis

formation of methane. However, researchers have also reported negative effects of the presence of CO 2 in FT
synthesis: catalyst stability is disturbed, lower rates have been reported, as well as lower selectivities for light
olefins [17, 89]. Application of the FT-technology has been limited by low olefin selectivity and high
methane selectivity, as well as severe carbon deposition.

Figure 2-13 Schematic representation of the changes in the reaction rates of key reaction steps as a function of
metal-carbon (M-C) interaction energies [17].

Selective formation of light olefins appears to be impossible without the formation of large amounts of
methane and C4+ hydrocarbons, as is indicated in Figure 2-13. A wide chain length distribution is obtained,
which can be described by an Anderson-Schulz-Flory (ASF) distribution. Selectivities for C 2 to C4 olefins
are predicted not to exceed 58% [87].

For the modified FT synthesis (starting from CO2 + H2 instead of syngas) to become a competitive
production process, the formation of alkanes should be minimized via modification of the catalyst dedicated
to CO2 conversion instead of using existing FT catalysts. Inclusion of water removal to avoid inhibition of
the catalyst will also enhance the process yield [12].

The modified FT synthesis can be summarized as a simpler process than direct CO2 conversion, and is
comprised of processes that are commercially available, but not specified for CO 2 conversion. The
disadvantage is that it results in a broader hydrocarbon distribution [17].

2.5.2.3. Methanol-To-Olefins (MTO)


The methanol to hydrocarbons (MTHC) conversion mechanism over acidic zeolite catalysts was discovered
by Chang and Silvestri in 1977 at Mobil Oil. Mainly the methanol-to-gasoline (MTG) and methanol-to-olefin
(MTO) reactions were highly attractive and were considered to be new routes to produce high-octane
gasoline and chemicals from natural gas and coal reserves. As the demand for light olefins increased
throughout the past decades, and methanol was more beneficial than gasoline (cheaply available from fossil
fuel sources) the research in the field of heterogeneous catalysis focused on the mechanisms of the MTO
Techno-economic analysis 49

process [90]. Compared to the direct conversion of CO2 to olefins via the modified FT route, conversion via
MTO has the advantage of being already partly commercially available and has a more tuned and narrow
product distribution [17].

In a simplified matter, the overall process for the conversion of methanol to hydrocarbons can be
summarized by the following reactions [91]:

𝑝𝑎𝑟𝑎𝑓𝑓𝑖𝑛𝑠
±𝐻2𝑂 −𝐻2 𝑂 𝑙𝑜𝑤𝑒𝑟 ℎ𝑖𝑔ℎ𝑒𝑟 𝑜𝑙𝑒𝑓𝑖𝑛𝑠
2𝐶𝐻3 𝑂𝐻 ↔ 𝐶𝐻3 𝑂𝐶𝐻3 → → (24)
𝑜𝑙𝑒𝑓𝑖𝑛𝑠 𝑎𝑟𝑜𝑚𝑎𝑡𝑖𝑐𝑠
𝑛𝑎𝑝ℎ𝑡𝑒𝑛𝑒𝑠

In the first step an equilibrium composition of methanol, dimethyl ether (DME), and water is formed via
dehydration of methanol. The mixture is further converted to lower olefins and this can be used to produce
alkanes, aromatics, napthenes and higher olefins by consecutive reactions (hydrogen transfer, isomerization,
alkylation, and polycondensation) [90]. The exact nature of the first C-C bond formation is however still
under debate; a consecutive and a parallel mechanism of olefin formation have been proposed, but no
exclusive proof has been found yet.

The consecutive mechanism assumes olefins to be produced via the consecutive attachment of a carbon atom
to the initial hydrocarbon. Surface-adsorbed intermediates react with methanol or DME to form the first C-C
bond. The parallel mechanism describes the formation of C-C bonds via macromolecular compounds, also
referred as the ‘hydrocarbon pool’ (HP) mechanism. Large carbonaceous species (both aromatics and long-
chain alkenes) are proposed to be formed on the surface and can further add reactants and split off products
in the steady-state of the reaction. For aromatics, scission of the side bond can lead to alkenes; for long-chain
alkenes, cracking and polymerization reactions lead to short-chain olefins [90-92].

Figure 2-14 Left: illustration of a consecutive pathway involving methylation, cracking and aromatization
reactions [93]. Right: general illustration of the parallel pathway (with hydrocarbon pool) [91].

In the initial MTO process by Mobil Oil, ZSM-5 zeolites were used as catalyst (Figure 2-15, left).
Combination of the proper reaction conditions (temperatures of about 673 K and methanol partial pressures
of 2-5 bar) resulted in high yields for light olefins. Further research has led to advanced zeolite materials,
such as the SAPO-34 zeolite (Figure 2-15, right) which is currently the main zeolite used in industrial
applications for MTO. By means of its small pore size (about 4 Å) diffusion of heavy and branched
hydrocarbons is restricted, resulting in high selectivity for small, linear olefins. The optimized acidity of the
SAPO-34 molecular sieve compared to other zeolite materials leads to much lower formation of paraffinic
50 Techno-economic analysis

by-products due to hydride transfer reactions [94]. By increased temperature (thermodynamic effect) and
reduced pressure (kinetic effect), the light olefins yield can further be enhanced [17]. Reactor optimization
actions such as water removal via catalytic distillation or inorganic membranes, and optimal COx/H2 ratios to
optimize the selectivity and productivity towards light olefins are other means to enhance the process
operation [12].

Figure 2-15 Framework topologies of HZSM-5 (left) and SAPO-34 (right) zeolites [93].

Nowadays, two proprietary MTO technologies dominate the market, producing up to 11 Mton/year [92]. The
first process, developed by UOP9 in the 1990s in cooperation with INEOS 10, uses a low-pressure fluidized-
bed reactor and SAPO-34 zeolites to convert methanol into ethylene and propylene. Integration of an Olefin
Cracking Process (OCP, Total/UOP) has led to carbon selectivities from methanol to C 2-C3 olefins up to 85-
90% by further cracking of the C4-C6+ by-product [17].

Figure 2-16 Left: UOP/INEOS/Total MTO process with integration of an OCP process [95]. Right:
DICP/SINOPEC/SYN Energy Technology MTO process with recycle of C4+ stream [93].

9
Universal Oil Products.
10
International Ethylene Oxyde Sales
Techno-economic analysis 51

The second commercial MTO process was established by DICP 11 and also uses a fluidized-bed reactor with a
SAPO-34 zeolite catalyst. The difference with the UOP/INEOS/Total process is the treatment of the C 4+ by-
product. Instead of further cracking, the by-product stream is recycled to the MTO reactor for maximization
of the ethylene and propylene yield [93].

As it would be desirable to directly convert CO2 to short-chain olefins, another process involving zeolite
materials is currently under investigation. It involves the direct one-step formation of short-chain olefins
from CO2 and H2 by combination of catalysts for methanol production and zeolites for MTO. However,
actual catalysts are adapted catalysts for syngas (CO/H2) conversion processes, but not specifically
developed for carbon dioxide conversion. Although methanol, MTO, and FT processes are well-established,
new catalysts are required to convert pure CO 2. FT catalysts should be modified to minimize the formation of
alkanes (mainly CH4) and increase the selectivity for C2-C3 olefins [17].

2.5.2.4. Economic evaluation of ethylene production from CO2

a) Methanol to olefins
The methanol-to-olefins (MTO) technology can be an attractive process for future production of valuable
hydrocarbons. Methanol could serve as an alternative feedstock for the depleting oil resources and could also
contribute to the increasing demand for propylene. Propylene/ethylene ratios from steam cracking of oil
fractions or natural gas cannot be raised significantly, and the selective production of propylene via MTO
(with higher flexibility for ethylene/propylene ratios) can hence be a means to make up for the expected
propylene demand. Most importantly in light of this thesis, methanol can be produced from CO2 and
renewable energy and can hence contribute to the reduction of CO2 emissions [94].

However, the competitiveness of MTO plants will highly depend on the price of the methanol feedstock.
Comparisons between a naphtha cracker and a MTO plant can be found in literature [94] which show that the
MTO process requires relatively inexpensive methanol to be competitive, depicted in Figure 2-17. With
current crude oil prices above 35 €/bbl, the MTO process would be feasible for methanol feedstock prices in
the order of 175 €/ton [96].

Figure 2-17 Comparison of the return on investment (ROI) of a naphtha cracker and MTO plant for a range of
methanol feedstock prices [94].

11
Dalian Institute of Chemical Physics
52 Techno-economic analysis

Analysis of the economics of the MTO process can be done by performing a simulation of the process with
specialized software such as Aspen. The simulation is based on the UOP/Hydro MTO process, explained
above, and attempts to give a good representation of the characteristic operating conditions in MTO rather
than giving an exact representation of the UOP/Hydro process [97].

Figure 2-18 illustrates the process flowsheet; it consists of a reaction section for the MTO process and a
separation and purification section. The mass balance and process conditions for the simulations are added in
Appendix A in Table A-20.

Figure 2-18 Flowsheet for the MTO process. [1] represents the reactor, [2] to [9] represents the separation
section, [10] is the cracking section of the UOP/Hydro MTO process [97].

Methanol is fed to the system and ethylene and propylene are obtained, along some minor streams of others.
The mass balance and prices for the main products are given in Table 2-14.

Table 2-14 Mass balance and product prices12 for the main products in the MTO process simulations [97].

Chemical Methanolin Propyleneout Ethyleneout


MTO (kton/y) 1560 386 214
MTO (€/ton) base case 300 1150 1150
MTO (€/ton) renewable 1426 1150 1150

In order to evaluate the MTO process with renewable methanol feedstock, a reference case involving
conventional methanol feedstock is also presented. The conventional feedstock price is fixed at 300 €/ton
and renewable methanol is taken as 1426 €/ton, see section 2.5.1.

Table 2-15 Key economic results12 of the MTO process for the conventional and renewable methanol feedstock
[97].

Feedstock Methanol (conv.) Methanol (renew.)


6
Fixed Capital Investment (10 €) 280 280
Sales (106 €/a) 700 700
Raw Material Cost (106 €/a) 500 2200
Cost of Energy Consumption (106 €/a) 33 33

12
US Dollar to Euro exchange rate: $1 = €0.8797.
Techno-economic analysis 53

The key economic results of the analysis are summarized in Table 2-15. Distinction is made for methanol
feedstock price. The production cost for ethylene and propylene can be estimated taking in account an annual
depreciation cost of 10% of the fixed capital investment. The production cost for ethylene and propylene
from conventional methanol is estimated at 935 €/ton and for renewable feedstock at 3700 €/ton. When
compared to the ethylene and propylene selling price in the order of 1200 €/ton, renewable MTO cannot be
feasible at this moment and requires a reduction of feedstock price to approximately 500 €/ton to be
competitive.

Taking in account the CO2 emissions for the MTO process (Table 2-16), the MTO process based on
renewable feedstock has a lower impact on the environment. Up to 1.35 ton of CO 2 per ton of methanol is
recycled in the production of methanol [22].

Table 2-16 Comparison of the CO2 emissions of the MTO process simulation for the conventional and renewable
methanol feedstock [97].

CO2 emissions MTO (conv.) MTO (renew.)


per year (MMt) 8 6
per ton product (ton/ton) 13.4 10
per ton propylene (ton/ton) 20.9 15.5

Including CO2 taxation could benefit the MTO process from renewable methanol production. However, this
taxation would be unreasonably high for the MTO process to become feasible due to the high feedstock
price.

b) Modified Fischer-Tropsch process


Analyses performed in literature on the production of ethylene via electrochemical reduction identified
syngas production by electrolyzing H 2O and CO2 in solid oxide electrolysis (SOE) cells to be most promising
[34]. Therefore a techno-economic analysis for the modified FT process will be evaluated based on the
representation in Figure 2-19. CO2 is captured from an air stream and is fed to the SOE system. An
appropriate amount of water is added to the electrolyzer to produce syngas with a proper composition for
FTS. The SOE system is assumed to be driven by renewable energy and FTS yields ethylene and propylene.

Figure 2-19 Schematic representation of the modified Fischer-Tropsch process. CO2 is captured and
subsequently dissociated with an appropriate amount of H 2O to produce syngas (CO + H2) with an appropriate
composition for FT synthesis [34].
54 Techno-economic analysis

Assumptions made for the evaluation of the modified FT process are summarized in Table 2-17, depicting
costs for feedstock species, energy consumption and investment costs.

Table 2-17 Assumptions for cost estimation12 of the modified Fischer-Tropsch process simulation [34].

Cost of CO2 capture € 26/tonCO2


Cost of H2O € 1 m³
Cost of fuel synthesis € 1.30/GJ Fischer–Tropsch gasoline from syngas
Cost of dissociation
Operating temperature 850°C
Electrolysis cell stack € 2000/m² investment including financing
Balance of system € 5000/m² investment including financing
Operation and maintenance € 0.5/GJ fuel

An efficiency of 95% is assumed for the SOE stacks [34]. The mass balance (Table 2-18) is limited to the
main components involved in the system: CO2 and H2O going in and ethylene and propylene going out.
Product prices are given, and taken the same for ethylene and propylene as in the case of MTO.

Table 2-18 Mass balance and product prices for the main products in the modified Fischer-Tropsch process
simulations.
Chemical CO2in H2Oin Ethyleneout Propyleneout
Electrolysis system (kton/y) 1935 792
Fischer-Tropsch (kton/y) 214 386
Product price (€/ton) 30 1 1150 1150

The key economic results for the simulation of the modified Fischer-Tropsch process are summarized in
Table 2-19. Taking in account a depreciation cost of 10% of the capital investment and the assumptions
made above allows determination of the production price for ethylene and propylene. No distinction is made
in production of ethylene and propylene, and a cost of approximately 2100 €/ton is obtained.

Table 2-19 Key economic results12 for the modified Fischer-Tropsch process simulations.

Capital Investment (106 €) 750


Raw material cost (10 €/y) 6
55
Sales (106 €/y) 700
Cost for energy consumption (106 €/y) 1100
Production cost (€/ton) 2100

The main cost contribution can be attributed to the cost for energy consumption from electrolysis of CO 2 and
H2O. Also the relatively high capital investment for the electrolyzer stacks, compared to the capital
investment of the MTO process, has a significant contribution to the production cost for ethylene and
propylene. As a consequence the production cost is significantly higher than the actual selling price
(approximately 1200 €/ton). When evaluating the reduction of CO 2 emissions by recycle in the process,
approximately 3 ton CO2 per ton product is recycled. In case CO2 taxation of 100 €/tonCO2 is introduced, the
difference between production price and selling price for renewable and conventional ethylene and propylene
could be reduced by approximately 300 €/ton.
Techno-economic analysis 55

2.5.3. Electrochemical reduction to formic acid

Although the current demand for formic acid is low (global capacity of 700·10 3 ton/y in 2013 [98]), the
global market is expected to grow at a compound annual growth rate (CAGR) of 5.63% over the period
2015-2019 [99]. Formic acid is mainly used in the animal feed market and in leather tanning, and demand is
growing for the pharmaceutical as well as biofuel market due to the increasing production volume and price
reductions (€0.51-0.60/kg for 2014 in Europe [98]) [100]. Major chemical companies such as BASF are
developing new uses for formic acid in hydrogen storage and fuel cells, making it an attractive chemical
[101].

The electrochemical reduction of CO2 to formic acid has several advantages over other pathways for
recycling CO2 and incorporating renewable energy in the chemical value chain: (1) the process can be
performed at mild conditions (i.e. room-temperature and ambient pressure); (2) the overall chemical
consumption can be reduced to wastewater by recycling the supporting electrolytes; (3) the process can be
driven by renewable energy; (4) extensive research has led to high selectivity, low cost, heterogeneous
catalyst; and (5) the process consists of compact, modular, and easy to scale-up applications of the
electrolytic cells [100, 101]. Compared to other products from CO 2 conversion the electrochemical reduction
to formic acid requires only a limited energy input. The specific energy consumption 𝐸 (MWh t-1) for an
electrochemical process [100] is given by:

𝑉 (1 + 𝛼 )𝑛
𝐸= (25)
1.641 𝐹𝐸

With 𝑉 the cell voltage, 𝛼 a factor for the auxiliary energy required in the process, 𝑛 the number of electrons
in the reaction, and 𝐹𝐸 the Faradaic efficiency. As reduction to formic acid only requires 2 e-, it avoids high
energy barriers for the activation of multi-electron transfer [102].

Formic acid (4.35 wt% H2) is a means to store hydrogen and incorporate renewable energy. It can be used
directly in formic acid fueled fuel cells [103], but due to technical challenges for the development of formic
acid fuel cells, it is preferred to use conventional hydrogen fuel cells and use formic acid for the storage of
hydrogen [17].

The general reactions occurring at the cathode and anode in the electrolytic cells are shown in Eq. (26) and
(27).

(Cathode) 2𝐶𝑂2 + 4𝐻 + + 4𝑒 − → 2𝐻𝐶𝑂𝑂𝐻 (26)

(Anode) 2𝐻2 𝑂 → 𝑂2 + 4𝑒 − + 4𝐻 + (27)

The general mechanism of the electrochemical reduction of CO2 is depicted in Figure 2-20. CO2 adsorbs on
the cathode surface and via electronation CO2- ions are obtained. Reaction with H 2O and electronation result
in formate ions, which desorb from the surface [102].
56 Techno-economic analysis

Figure 2-20 Mechanism of CO2 electrochemical reduction on an electrolytic cell cathode [102]. (a) Adsorption
and electronation of CO2 on the cathode surface. (b) Reaction of adsorbed CO 2- and H2O. (c) Electronation of
HCO2- on the surface. (d) Desorption of the HCO2- ion.

To obtain formic acid, the formate ions desorbing from the cathode surface need further acidolysis [104].

𝐻𝐶𝑂2− + 𝐻 + → 𝐻𝐶𝑂𝑂𝐻 (28)

The electrolytic cell, illustrated in Figure 2-21, consists of anode and cathode electrodes, coated with
electrocatalysts. Ion transfer between the cathode and anode is provided by the supporting electrolytes. At
the cathode, CO2 is reduced to the desired products. The accompanying oxidation reaction of water occurs at
the anode [100]. An important issue in the reaction mechanism is the competitive hydrogen evolution (Eq.
(29)).

2𝐻2 𝑂 + 2𝑒 − → 𝐻2 + 2𝑂𝐻 − (29)

By means of appropriate electrolyte selection, with a high overpotential for hydrogen evolution, the
competitive reaction can be eliminated [104].

Figure 2-21 Basic structure of a CO2 electroreduction electrolytic cell [102]. At the cathode CO2 is reduced to the
desired products. The accompanying oxidation reaction occurs at the anode.
Techno-economic analysis 57

To avoid the production of intermediates requiring high overpotentials for further processing and to suppress
the competitive hydrogen evolution reaction (HER), the electrochemical reduction process requires catalysts
[104]. Mainly metal catalysts (Pb, Ru-Pd, Sn, Hg, In and Cd) are used, but also metal-complex catalysts
(hydrides and halides) for their improved operation. Bio-catalysts (e.g. the FDH1 enzyme), of interest for
their high selectivity and low overpotentials, are used to a lesser extent because of vulnerability to oxygen
poisoning [102]. Current catalysts already guarantee high Faradaic efficiencies (>80%) for electrochemical
reduction to formic acid [100].

Production of formic acid can be carried out in aqueous and non-aqueous electrolyte solutions and in acid or
neutral conditions. KHCO3 is reported to be one of the most prominent aqueous electrolytes and provides
stable electroconductivity. Aqueous solutions also have the convenience of concentration adjustment. Non-
aqueous electrolytes, such as KOH/methanol or teraethylammonium perchlorate (TEAP) or MeCN, enhance
the CO2 solubility and suppress the HER. An alternative for (non-)aqueous solutions are ionic liquids, which
are highly electrically conductive but still require high operating temperatures.

Issues concerning the mass transfer of CO2 to the cathode surface were a limiting factor to the development
of electrochemical reduction of CO2. Introduction of gas diffusion electrodes (GDE) provided a solution: a
gas diffusion layer (GDL) provides a pathway for CO2 and electrolyte diffusion [102].

2.5.3.1. Economic evaluation of the electrochemical reduction of CO 2 to


formic acid
As mentioned before, the electrochemical reduction of CO2 to formic acid has several advantages over other
recycle pathways. High selectivity, low cost, heterogeneous catalysts are available, the process can be
operated at room temperature and ambient pressure, it can be driven by renewable energy, and the chemical
consumption can be minimized to wastewater. However, to economically process the reduction to formic
acid, several requirements are necessary [100].

At first, high current densities to obtain high product generation limit the reactor size and ultimately limit the
capital costs. Secondly, the used catalyst should guarantee a high Faradaic efficiency to reduce the specific
electricity consumption of the process. Further, long electrode lifetime can be of importance to reduce the
operational cost and allow for continuous operation. Apart from the energetic requirements, the process
should also provide high product selectivity, avoid additional CO2 generation, and guarantee long-term stable
operation.

To estimate the economics of the electrochemical reduction of CO2 to formic acid a simplified simulation of
the process is modeled, according to the conceptual flowsheet in Figure 2-22. By means of a separation
section and recycle loop to an acidolysis reactor, it is assumed that 100% selectivity for formic acid is
obtained [104].
58 Techno-economic analysis

Figure 2-22 Conceptual flowsheet for the electrochemical reduction of CO 2 process [104]. CO2 is reduced with
water in the electrolytic cell. Produced formic acid is separated and the remaining stream is recycled.
The general assumptions made for the analysis of the electrochemical reduction process are given in Table
2-20. CO2 is assumed to be bought at a price of 30 €/ton, and the electricity price (0.13 €/kWh) is taken as an
average price for renewable energy [27].

Table 2-20 Assumptions for cost estimation of the electrochemical reduction process [100].

Cost of CO2 capture (€/ton) 30[34]


Cost of H2O (€/m³) 1[34]
Electricity price (€/kWh) 0.13[27]
Electrode cost (€/tpd) 2000

The mass balance for the main components in the process is depicted in Table 2-21. CO2 and H2O are
inserted as feedstock to the system; H 2SO4 and NaCl are added as electrolyte; and HCOOH, O2 and H2 are
obtained as product. Product prices for each compound are given. Actual market prices for formic acid are in
the range of 600-700 €/ton and a value of 700 €/ton will be used for the model simulations [98].

Table 2-21 Mass balance and product prices for the main products in the electrochemical reduction process
simulations [100].

Chemical CO2in H2Oin H2SO4in NaClin HCOOHout O2out H2out


Electrochemical
36500 19710 75 75 38000 16500 365
reduction (kton/y)
Product price (€/ton) 30 1 130 130 700[98] 85 2400

The key economic results for the simulation of the modified Fischer-Tropsch process are summarized in
Table 2-22. Sridhar et al. [100] and Oloman et al. [104] report an energy consumption of approximately 7000
kWh/ton, with energy efficiencies in the range of 90-95%. Taking in account a depreciation cost of 10% of
the capital investment and the assumptions made above allows determination of the production price for
formic acid.
Techno-economic analysis 59

Table 2-22 Key economic results12 for the electrochemical reduction process simulation [100].

Capital Investment (106 €) 163


Raw material cost (106 €/y) 1.13
6
Sales (10 €/y) 30
6
Cost for energy consumption (10 €/y) 33
6
Electrode cost (10 €/y) 0.2
Production cost (€/ton) 1385

The modeled production cost for the simulation of the electrochemical reduction of CO2 is approximately
1400 €/ton. This is considerably higher than the actual market value for formic acid. The main cost
contributors for production of formic acid are the electricity and capital cost. Feedstock and electrode cost
constitute only 2.5% of the operational expenditures, and have only a small influence on the production cost.

DNV13 has reported on an optimized process: Electrochemical Reduction of CO 2 to Formate/Formic Acid


process (ECFORM) process, for the reduction of CO2 to formic acid. Simulations have shown that a
significantly smaller amount of energy is necessary in this process: approximately 4000 kWh/t compared to
7000 kWh/ton in the previous case. Taking in mind the current selling price for formic acid, the
electrochemical reduction of CO2 to formic acid with renewable energy would not be feasible, as is reported
by DNV [101] in case of a formic acid market value of 1200 €/ton, but the production cost can be reduced to
approximately 1000 €/ton compared to 1400 €/ton obtained from the above simulation.

In case a carbon credit would be taken in account, the price gap between the renewable and conventional
formic acid price can be reduced. Carbon credits of 100 € per ton CO 2 avoided could favor the production of
renewable formic acid, although at current market values carbon credits up to 600 €/tonCO2 would be
necessary for the electrochemical reduction of CO2 to become feasible.

13
Det Norske Veritas
60 Techno-economic analysis

2.6. Conclusion
In this chapter a techno-economic analysis has been performed for the conversion of CO2 with renewable
energy. More specific, analysis has been done for (1) the production of hydrogen from electrolysis of water
with renewable energy; (2) production of methanol via conversion of CO2 with hydrogen; (3) ethylene
synthesis via electrochemical reduction of CO2 and water, a modified FTS process for CO2 and H2, and the
methanol-to-olefins process; and (4) the electrochemical reduction of CO2 to formic acid.

Hydrogen production has been evaluated for alkaline and PEM electrolysis and future hydrogen production
costs are estimated at approximately 4 €/kgH2, efficiencies are expected to increase up to approximately 75%
for alkaline and PEM electrolysis. The main cost contributor to the production of hydrogen is the electricity
feedstock for electrolysis. Also the high capital costs for electrolysis equipment raise the production cost.

The favorable aspects of methanol production are its compatibility with existing distribution infrastructure
and the ability to use it as a replacement for current liquid fuels, as well as an intermediate in chemical
production processes. Production of methanol was evaluated by the conversion of CO 2 with H2 from water
electrolysis. Electricity to chemical energy efficiencies are found in a range of 60 to 70% and the production
cost for renewable methanol is estimated at 1400 €/ton. This is significantly higher than actual methanol
price of 275 €/ton. The high cost can mainly be attributed to the electricity cost of water electrolysis for
hydrogen production. For renewable methanol production to be feasible, electricity cost should be reduced to
0.03 €/kWh. In case of carbon taxes to be included of 100 €/ton CO2, an electricity cost reduction to 0.06
€/kWh is required.

Ethylene synthesis via MTO has the advantage of a high flexibility and good control over the product
selectivity, with in particular the ability to increase the propylene to ethylene product ratio. Analysis was
based on the UOP/Hydro MTO process and for a renewable methanol feedstock a production cost for
ethylene of approximately 3700 €/ton is determined, compared to a production cost of 935 €/ton in case of
conventional methanol feedstock. The production cost is hence highly dependent on the methanol feedstock
price and the renewable feedstock requires a 500 €/ton reduction for ethylene production to be feasible.
Overall 1.35 ton of CO2 is recycled per ton of product, which favors the production of ethylene via MTO
over conventional production. However, carbon taxes would need to be unreasonably high for the renewable
MTO process to become feasible.

Ethylene via modified FTS has been analyzed for syngas production from electrochemical reduction of
carbon dioxide and water in a solid oxide electrolyzer and consecutive FTS. Efficiencies for SOE are
reportedly 95% or higher and is hence suggested to be the most promising technology. However, the
electricity cost for electrolysis remains the highest cost contributor and capital costs are also significantly
higher than in the MTO process. A production cost of approximately 2100 €/ton was obtained, significantly
higher than conventional ethylene prices around 1200 €/ton. Ethylene production via modified FTS estimates
CO2 recycle at 3 ton per ton product.
Techno-economic analysis 61

Formic acid is mainly considered favorable for use in hydrogen storage and fuel cell applications. It benefits
from its low chemical consumption (possibly solely water), lower energy requirement and mild operating
conditions compared to other CO2 conversion processes. However, current production is twice the demand.
Renewable formic acid production price is estimated at 1400 €/ton and is mainly determined by the
electricity and capital cost. Actual selling prices are in a range of 600 to 700 €/ton, making it economically
very difficult for renewable formic acid production to be feasible, requiring carbon credits up to 600 €/ton.
Reports by DNV suggest an optimized process with higher energy efficiency reducing the electricity cost.
Production price could then be reduced to 1000 €/ton, however still significantly higher than actual prices.
62 Techno-economic analysis

2.7. References

1. Marland, G., et al., Global, regional, and national fossil fuel CO2 emissions. Trends: A
Compendium of Data on Global Change, 2007: p. 37831-6335.

2. Network, G.G.G.R. Recent Monthly Average Mauna Loa CO2. 2016 April 5, 2016 [cited 2016 April
30]; Available from: http://www.esrl.noaa.gov/gmd/ccgg/trends/.

3. Climate change. [cited 2016; Available from: http://www.globalchange.gov/climate-change.

4. Hu, B., C. Guild, and S.L. Suib, Thermal, electrochemical, and photochemical conversion of CO2 to
fuels and value-added products. Journal of Co2 Utilization, 2013. 1: p. 18-27.

5. Aresta, M. and A. Dibenedetto, Utilisation of CO2 as a chemical feedstock: opportunities and


challenges. Dalton Transactions, 2007(28): p. 2975-2992.

6. Appel, A.M., et al., Frontiers, opportunities, and challenges in biochemical and chemical catalysis
of CO2 fixation. Chemical reviews, 2013. 113(8): p. 6621-6658.

7. Le Quéré, C., et al., Global carbon budget 2013. Earth Syst. Sci. Data, 2014. 6(1): p. 235-263.

8. Oberthür, S. and H.E. Ott, The Kyoto Protocol: international climate policy for the 21st century.
1999: Springer Science & Business Media.

9. Commission, E., The Paris Protocol – A blueprint for tackling global climate change beyond 2020,
E. Union, Editor. 2015: Paris.

10. Aresta, M., A. Dibenedetto, and A. Angelini, Catalysis for the valorization of exhaust carbon: From
CO2 to chemicals, materials, and fuels. Technological use of CO2. Chemical reviews, 2013. 114(3):
p. 1709-1742.

11. Aresta, M., A. Dibenedetto, and A. Angelini, The changing paradigm in CO2 utilization. Journal of
Co2 Utilization, 2013. 3-4: p. 65-73.

12. Centi, G., G. Iaquaniello, and S. Perathoner, Can We Afford to Waste Carbon Dioxide? Carbon
Dioxide as a Valuable Source of Carbon for the Production of Light Olefins. ChemSusChem, 2011.
4(9): p. 1265-1273.

13. Olajire, A.A., Valorization of greenhouse carbon dioxide emissions into value-added products by
catalytic processes. Journal of Co2 Utilization, 2013. 3-4: p. 74-92.

14. Whipple, D.T. and P.J.A. Kenis, Prospects of CO2 Utilization via Direct Heterogeneous
Electrochemical Reduction. The Journal of Physical Chemistry Letters, 2010. 1(24): p. 3451-3458.

15. Stechel, E.B. and J.E. Miller, Re-energizing CO2 to fuels with the sun: Issues of efficiency, scale,
and economics. Journal of Co2 Utilization, 2013. 1: p. 28-36.

16. Zuttel, A., et al., Storage of Renewable Energy by Reduction of CO2 with Hydrogen. Chimia, 2015.
69(5): p. 264-268.

17. Centi, G., E.A. Quadrelli, and S. Perathoner, Catalysis for CO2 conversion: a key technology for
rapid introduction of renewable energy in the value chain of chemical industries. Energy &
Environmental Science, 2013. 6(6): p. 1711-1731.
Techno-economic analysis 63

18. El Khamlichi, A. and N. Thybaud, Valorisation chimique du CO2: état des lieux. Quantification des
bénéfices énergétiques et environnementaux et évaluation économique de trois voies chimiques.
2014, ENEA consulting, EReIE, Universite Strasbourg: Strasbourg. p. 148.

19. Yang, N. and R. Wang, Sustainable technologies for the reclamation of greenhouse gas CO2.
Journal of Cleaner Production, 2015. 103: p. 784-792.

20. Hu, Y.H., Advances in CO2 conversion and utilization. 2010: American Chemical Society
Washington, DC.

21. Cuellar-Franca, R.M. and A. Azapagic, Carbon capture, storage and utilisation technologies: A
critical analysis and comparison of their life cycle environmental impacts. Journal of Co2
Utilization, 2015. 9: p. 82-102.

22. Van-Dal, É.S. and C. Bouallou, CO2 abatement through a methanol production process. Chemical
Engineering Transactions, 2012. 29: p. 463-468.

23. Saur, G., Wind-to-hydrogen project: electrolyzer capital cost study. 2008: Citeseer.

24. Where Carbon Is Taxed. 2016 01/03/2016 [cited 2016; Available from:
http://www.carbontax.org/where-carbon-is-taxed/.

25. Lamhauge, N. and A. Cox, Climate and Carbon: Aligning Prices and Policies. 2013, Paris: OECD.

26. Majocchi, A. After the oil price crash, it’s time for a carbon tax. 2015; Available from:
https://www.euractiv.com/section/energy/opinion/after-the-oil-price-crash-it-s-time-for-a-carbon-
tax/.

27. Annual Energy Outlook 2015 with projections to 2040, U.S.E.I. Administration, Editor. 2015, U.S.
Department of Energy: Washington.

28. Norskov, J.K.V., C. New record-breaking year for Danish wind power. 2016 15/01/2016 [cited 2016
20 May]; Available from: http://energinet.dk/EN/El/Nyheder/Sider/Dansk-vindstroem-slaar-igen-
rekord-42-procent.aspx.

29. Dumas, M., J. Rising, and J. Urpelainen, Political competition and renewable energy transitions
over long time horizons: A dynamic approach. Ecological Economics, 2016. 124: p. 175-184.

30. Seetharaman, A., et al., Enterprise framework for renewable energy. Renewable and Sustainable
Energy Reviews, 2016. 54: p. 1368-1381.

31. Olah, G.A., G.K.S. Prakash, and A. Goeppert, Anthropogenic Chemical Carbon Cycle for a
Sustainable Future. Journal of the American Chemical Society, 2011. 133(33): p. 12881-12898.

32. Connolly, D., H. Lund, and B.V. Mathiesen, Smart Energy Europe: The technical and economic
impact of one potential 100% renewable energy scenario for the European Union. Renewable and
Sustainable Energy Reviews, 2016. 60: p. 1634-1653.

33. Wiesenthal, T. and A. Mourelatou, How much bioenergy can Europe produce without harming the
environment? 2006.

34. Graves, C., et al., Sustainable hydrocarbon fuels by recycling CO2 and H2O with renewable or
nuclear energy. Renewable and Sustainable Energy Reviews, 2011. 15(1): p. 1-23.

35. BP, BP Statistical Review of World Energy, B. p.l.c., Editor. 2015: London. p. 44.
64 Techno-economic analysis

36. Mortensen, N.G., et al., Wind Atlas Analysis and Application program (WAsP): Vol. 3: Utility
programs. 1999, Risø National Laboratory.

37. Denholm, P., et al., Land-use requirements of modern wind power plants in the United States. 2009:
National Renewable Energy Laboratory Golden, CO.

38. Kalogirou, S.A., Solar energy engineering: processes and systems. 2013: Academic Press.

39. Wagner, S., R. Bareiss, and G. Guidati, Wind turbine noise. 2012: Springer Science & Business
Media.

40. Mignard, D., et al., Methanol synthesis from flue-gas CO2 and renewable electricity: a feasibility
study. International Journal of Hydrogen Energy, 2003. 28(4): p. 455-464.

41. Carmo, M., et al., A comprehensive review on PEM water electrolysis. International Journal of
Hydrogen Energy, 2013. 38(12): p. 4901-4934.

42. Mignard, D. and C. Pritchard, Processes for the synthesis of liquid fuels from CO 2 and marine
energy. Chemical engineering research and Design, 2006. 84(9): p. 828-836.

43. Tremel, A., et al., Techno-economic analysis for the synthesis of liquid and gaseous fuels based on
hydrogen production via electrolysis. International Journal of Hydrogen Energy, 2015. 40(35): p.
11457-11464.

44. Multi-Year Research, Development, and Demonstration Plan, U.S.D.o. Energy, Editor. 2015. p. 1-
44.

45. Hydrogen production: Biomass gasification. 2016 [cited 2016 29/02]; Available from:
http://energy.gov/eere/fuelcells/hydrogen-production-biomass-gasification.

46. Biomass potential. Agriculture and rural development 2015 [cited 2016 29/02]; Available from:
http://ec.europa.eu/agriculture/bioenergy/potential/index_en.htm.

47. Hydrogen production: Thermochemical water splitting. 2016 [cited 2016 29/02]; Available from:
http://energy.gov/eere/fuelcells/hydrogen-production-thermochemical-water-splitting.

48. May, M. New efficiency record for solar hydrogen production is 14 percent. 2015 [cited 2016
16/02]; Available from: http://phys.org/news/2015-09-efficiency-solar-hydrogen-production-
percent.html.

49. May, M.M., et al., Efficient direct solar-to-hydrogen conversion by in situ interface transformation
of a tandem structure. Nat Commun, 2015. 6.

50. Hydrogen production: Photoelectrochemical water splitting. 2016 [cited 2016 29/02]; Available
from: http://energy.gov/eere/fuelcells/hydrogen-production-photoelectrochemical-water-splitting.

51. Zoulias, E., et al., A review on water electrolysis. TCJST, 2004. 4(2): p. 41-71.

52. James, B.D., et al., Technoeconomic analysis of photoelectrochemical (PEC) hydrogen production.
DOE report, 2009.

53. Hydrogen production: Photobiological. 2015 [cited 2016 29/02]; Available from:
http://energy.gov/eere/fuelcells/hydrogen-production-photobiological.
Techno-economic analysis 65

54. Ainscough, C.P., D.; Miller, E., Hydrogen Production Cost From PEM Electrolysis, D.o. Energy,
Editor. 2014: United States of America. p. 11.

55. Tsiplakides, D., PEM water electrolysis fundamentals. 2012.

56. Laguna-Bercero, M., Recent advances in high temperature electrolysis using solid oxide fuel cells: A
review. Journal of Power Sources, 2012. 203: p. 4-16.

57. Dönitz, W. and E. Erdle, High-temperature electrolysis of water vapor—status of development and
perspectives for application. International Journal of Hydrogen Energy, 1985. 10(5): p. 291-295.

58. Kreuter, W. and H. Hofmann, Electrolysis: The important energy transformer in a world of
sustainable energy. International Journal of Hydrogen Energy, 1998. 23(8): p. 661-666.

59. Abe, I., Alkaline water electrolysis, in Energy carriers and conversion systems with emphasis on
hydrogen. p. 146-166.

60. Gouws, S. Voltammetric Characterization Methods for the PEM Evaluation of Catalysts. 2012
[cited 2016 15/02]; Available from: http://www.intechopen.com/books/electrolysis/voltammetric-
characterization-methods-for-the-pem-evaluation-of-catalysts.

61. Mergel, J. and D. Stolten. Challenges in Water Electrolysis and Its Development Potential as a Key
Technology for Renewable Energies. in Meeting Abstracts. 2012. The Electrochemical Society.

62. von Meier, A., Electric Power Systems: A Conceptual Introduction. 2006: Wiley.

63. Lfritz, D., Schematic of the basic operating principle of a polymer electrolyte membrane electrolysis
cell, in Photoshop, PEMelectrolysis.jpg, Editor. 2013.

64. DuPontTM Nafion® PFSA Membranes, D.F. Cells, Editor. 2009: Wilmington.

65. DOE H2A Analysis. [cited 2016 15/02]; Available from:


https://www.hydrogen.energy.gov/h2a_analysis.html.

66. Spath, P.L. and M.K. Mann, Life Cycle Assessment of Renewable Hydrogen Production Via
Wind/electrolysis: Milestone Completion Report. 2004: National Renewable Energy Laboratory.

67. De, R.A.J., Purification of hydrogen utilizing hydrogen-permeable membranes. 1958, Google
Patents.

68. Grashoff, G., C. Pilkington, and C. Corti, The purification of hydrogen. Platinum Metals Review,
1983. 27(4): p. 157-169.

69. Connor, E. Technology Considerations on Hydrogen Generation Technology. 2016 [cited 2016
06/03]; Available from: http://www.peakscientific.com/learn/articles-and-application-
notes/hydrogen-purification-methods/.

70. Ruether, J., M. Ramezan, and E. Grol, Life-cycle analysis of greenhouse gas emissions for hydrogen
fuel production in the United States from LNG and coal. 2005, DOE/NETL-2006/1227. November.

71. Spath, P.L. and M.K. Mann, Life Cycle Assessment of Renewable Hydrogen Production via
Wind/Electrolysis.
66 Techno-economic analysis

72. Aaron, D. and C. Tsouris, Separation of CO2 from flue gas: a review. Separation Science and
Technology, 2005. 40(1-3): p. 321-348.

73. Socolow, R., et al., Direct air capture of CO2 with chemicals: a technology assessment for the APS
Panel on Public Affairs. 2011, American Physical Society.

74. House, K.Z., et al., Economic and energetic analysis of capturing CO2 from ambient air.
Proceedings of the National Academy of Sciences, 2011. 108(51): p. 20428-20433.

75. Wang, W., et al., Recent advances in catalytic hydrogenation of carbon dioxide. Chemical Society
Reviews, 2011. 40(7): p. 3703-3727.

76. Applications for Methanol. [cited 2016 06/03]; Available from: http://www.methanol.org/Methanol-
Basics/Methanol-Applications.aspx.

77. Kothandaraman, J., et al., Conversion of CO2 from Air into Methanol Using a Polyamine and a
Homogeneous Ruthenium Catalyst. Journal of the American Chemical Society, 2016. 138(3): p. 778-
781.

78. Fornero, E.L., et al., CO2 capture via catalytic hydrogenation to methanol: Thermodynamic limit vs.
‘kinetic limit’. Catalysis Today, 2011. 172(1): p. 158-165.

79. Zhao, Y.-F., et al., Insight into methanol synthesis from CO 2 hydrogenation on Cu (111): complex
reaction network and the effects of H 2 O. Journal of Catalysis, 2011. 281(2): p. 199-211.

80. Joo, O.-S., K.-D. Jung, and J. Yonsoo, CAMERE Process for methanol synthesis from CO 2
hydrogenation. Studies in Surface Science and Catalysis, 2004. 153: p. 67-72.

81. Cost, L., Levelized Avoided Cost of New Generation Resources in the Annual Energy Outlook 2014.
Energy Information Agency, 2014.

82. Methanex posts regional contract methanol prices for North America, Europe and Asia. 2015
7/2/2016 [cited 2016 8/2/2016]; Available from: https://www.methanex.com/our-business/pricing.

83. Steam Flow Rate and kW Rating. 2016 [cited 2016; Available from:
http://www.engineeringtoolbox.com/steam-flow-kw-d_1057.html.

84. Ungerer, J., et al., Sustained photosynthetic conversion of CO 2 to ethylene in recombinant


cyanobacterium Synechocystis 6803. Energy & Environmental Science, 2012. 5(10): p. 8998-9006.

85. Ren, D., et al., Selective Electrochemical Reduction of Carbon Dioxide to Ethylene and Ethanol on
Copper (I) Oxide Catalysts. ACS Catalysis, 2015. 5(5): p. 2814-2821.

86. Ogura, K., H. Yano, and F. Shirai, Selective Conversion of CO2 to Ethylene by the Electrolysis at a
Three-Phase (Gas/Liquid/Solid) Interface in an Acidic Solution Containing Cupric Ions. Fuel
Chemistry Division Preprints, 2003. 48(1): p. 264.

87. Jiao, F., et al., Selective conversion of syngas to light olefins. Science, 2016. 351(6277): p. 1065-
1068.

88. Xu, L., et al., The promotions of MnO and K2O to Fe/silicalite-2 catalyst for the production of light
alkenes from CO2 hydrogenation. Applied Catalysis A: General, 1998. 173(1): p. 19-25.

89. Zhao, Y., et al., The kinetic study of light alkene syntheses by CO2 hydrogenation over Fe-Ni
catalysts. Frontiers of Chemical Engineering in China, 2009. 4(2): p. 153-162.
Techno-economic analysis 67

90. Wang, W., Y. Jiang, and M. Hunger, Mechanistic investigations of the methanol-to-olefin (MTO)
process on acidic zeolite catalysts by in situ solid-state NMR spectroscopy. Catalysis Today, 2006.
113(1–2): p. 102-114.

91. Khadzhiev, S.N., M.V. Magomedova, and E.G. Peresypkina, Mechanism of olefin synthesis from
methanol and dimethyl ether over zeolite catalysts: A review. Petroleum Chemistry, 2014. 54(4): p.
245-269.

92. Sastre, G., Confinement effects in methanol to olefins catalysed by zeolites: A computational review.
Frontiers of Chemical Science and Engineering, 2016. 10(1): p. 76-89.

93. Sun, X., Catalytic Conversion of Methanol to Olefins over HZSM-5 Catalysts. 2013, Universität
München.

94. Chen, J.Q., et al., Recent advancements in ethylene and propylene production using the UOP/Hydro
MTO process. Catalysis Today, 2005. 106(1–4): p. 103-107.

95. Olsbye, U., et al., Conversion of Methanol to Hydrocarbons: How Zeolite Cavity and Pore Size
Controls Product Selectivity. Angewandte Chemie International Edition, 2012. 51(24): p. 5810-
5831.

96. Economics, T. Crude Oil 1946-2016 Data. 2016 [cited 2016 May 2]; Available from:
http://www.tradingeconomics.com/commodity/crude-oil.

97. Jasper, S. and M.M. El-Halwagi, A Techno-Economic Comparison between Two Methanol-to-
Propylene Processes. Processes, 2015. 3(3): p. 684-698.

98. Afshar, A.A.N., Chemical profile: Formic acid. 2014, TranTech Consultants Inc. p. 3.

99. Newswire, P. Global Formic Acid Market 2015-2019. 2015 [cited 2016 26/04]; Available from:
http://www.prnewswire.com/news-releases/global-formic-acid-market-2015-2019-300117131.html.

100. Agarwal, A.S., et al., The electrochemical reduction of carbon dioxide to formate/formic acid:
engineering and economic feasibility. ChemSusChem, 2011. 4(9): p. 1301-1310.

101. Veritas, D.N., Electrochemical Conversion of CO2–Opportunities and Challenge. 2011, Norway.

102. Lu, X., et al., Electrochemical reduction of carbon dioxide to formic acid. ChemElectroChem, 2014.
1(5): p. 836-849.

103. Rice, C., et al., Direct formic acid fuel cells. Journal of Power Sources, 2002. 111(1): p. 83-89.

104. Oloman, C. and H. Li, Electrochemical processing of carbon dioxide. ChemSusChem, 2008. 1(5): p.
385-391.
68

Chapter 3
Methods and models
Ab-initio calculations are performed to analyze the activation of CO 2 on the catalyst surface. The different
catalyst surfaces are described. Adsorption and activation energies are obtained via specialized DFT
calculations and the Gibbs free energies for gas phase and adsorbed species can be determined by
combination of the electronic and zero-point vdW-DF2 energies with enthalpy and entropy corrections from
frequency calculations [1].

By means of kinetic modeling, the reaction mechanism can be described and the reaction and reactor
equations can be solved for the surface coverages and the gaseous composition following from the reaction
mechanism and the desired reactor model [2-4]. Better insight in the different reaction pathways can be
obtained and specific information on the catalytic activity can be retrieved; indicating influences of the
catalyst structure on product selectivity or yield, and hence, allows for faster catalyst development [5, 6].
Methods and models 69

3.1. Density Functional Theory


Density Functional Theory (DFT) is a computational quantum-mechanical modeling method to examine
electronic structures of systems such as atoms, molecules, bulk structures, surfaces etc. via approximate
solutions to the Schrödinger equation [7]. It is not the objective of this thesis to fully understand the
modeling method and for an introduction into DFT, the reader is referred to the Nobel lecture of Walter
Kohn [8]. For a more in-depth understanding, one could read the work of Parr and Yang [9].

3.2. Catalyst models

3.2.1. Cobalt catalyst models

As first model for the cobalt catalyst an fcc Co(111) slab was generated, consisting of 5 layers, using a
p(3x3) unit cell with nine surface atoms per layer. The bottom two layers were fixed at the DFT calculated
bulk positions. Second a Co(211) step surface was generated, using a p(4x4) unit cell. An optimized lattice
constant of 3.56 Å was obtained, close to the experimental value of 3.54 Å [10].

Figure 3-1: Top view of the different unit cells for the Co catalyst: left: p(3x3) Co(111), right: p(4x4) Co(211).
70 Methods and models

3.2.2. Copper catalyst model

Similar as for the cobalt catalyst, an fcc Cu(111) slab was generated, using a p(3x3) unit cell with nine
surface atoms per layer and consisting of 5 layers. Again the bottom two layers were fixed at the DFT
calculated bulk positions. An optimized lattice constant of 3.68 Å was obtained, similar to the experimental
value of 3.61 Å [10].

Figure 3-2: Top view of the unit cell for the Cu catalyst: p(3x3) Cu(111).

3.2.3. Nickel catalyst model

For the Ni catalyst also an fcc Ni(111) slab was generated, using a p(3x3) unit cell with nine surface atoms
per layer and consisting of 5 layers. Again the bottom two layers were fixed at the DFT calculated bulk
positions. An optimized lattice constant of 3.56 Å was obtained, similar to the experimental value of 3.52 Å
[11].

Figure 3-3: Top view of the unit cell for the Ni catalyst: p(3x3) Ni(111).
Methods and models 71

3.2.4. Gold catalyst model

The Au catalyst was also modeled as an fcc Au(111) slab, using a p(3x3) unit cell with nine surface atoms
per layer and consisting of 5 layers. Again the bottom two layers were fixed at the DFT calculated bulk
positions. An optimized lattice constant of 4.21 Å was obtained, similar to the experimental value of 4.08 Å
[12].

Figure 3-4: Top view of the unit cell for the Au catalyst: p(3x3) Au(111).

3.3. VASP calculations


The Vienna Ab-initio Simulation Package (VASP) is a complex software-tool for simulating ab-initio
quantum-mechanical molecular dynamics (MD) at finite temperature. A complete understanding of VASP
can be found in the VASP guide by the University of Wien [13].

The general codes to perform the DFT calculations were provided by the coach of this thesis, and the
objective was to understand the outcome of these simulations rather than the codes itself. In the following
sections however a short overview of the codes will be given for the reader to understand the results and
allowing further work on these results.

3.3.1. Adsorption energy

Adsorption energies were calculated for the catalyst slabs as explained in section 3.2. Spin-polarized Density
Functional Theory was used with the revised functional (rPW86) of Langreth and Lundqvist and including
non-local vdW-DF2 correlations [13, 14]. Similar to previous work in the Saeys group [7] a cut-off kinetic
energy of 450 eV was implemented for the plane-wave basis set and an inter-slab spacing of 15 Å was
defined to minimize the interactions between repeated slabs. Also a k-points mesh was generated, with the
Brillouin zone sampled with a (3x3x1) Monkhorst-Pack grid for both the p(3x3) and p(4x4) unit cells. This
allows for accurate integration of the properties computed in the reciprocal space. The k-points mesh follows
from the Monkhorst-Pack procedure [15].
72 Methods and models

Figure 3-5 Left: possible adsorption sites on a closed pack fcc (111) catalyst surface (T-Top, B-Bridge, F-Hollow-
fcc, H-Hollow-hcp); right: 5 layers p(3x3) model slab in the z-direction with interslab spacing of 15 Å, the top
three layers are relaxed while the bottom two layers are constrained at the DFT optimized bulk positions [7].

Eq. (30) defines the adsorption energy for an adsorbate on the catalyst surface.

𝐸𝑎𝑑𝑠 = 𝐸𝑎𝑑𝑠𝑜𝑟𝑏𝑎𝑡𝑒+𝑠𝑙𝑎𝑏 − (𝐸𝑠𝑙𝑎𝑏 + 𝐸𝑎𝑑𝑠𝑜𝑟𝑏𝑎𝑡𝑒(𝑔) ) (30)

With 𝐸𝑎𝑑𝑠 the adsorption energy of the adsorbate in kJ/mol, 𝐸𝑎𝑑𝑠𝑜𝑟𝑏𝑎𝑡𝑒+𝑠𝑙𝑎𝑏 the total electronic energy of the
adsorbate on the slab, 𝐸𝑠𝑙𝑎𝑏 the electronic energy of the slab and 𝐸𝑎𝑑𝑠𝑜𝑟𝑏𝑎𝑡𝑒(𝑔) the electronic energy of the
adsorbate in vacuum [16]. Typical errors associated with DFT calculations are roughly in the order of 5
kJ/mol [17].

3.3.2. Transition state calculation

To determine the transition state of reactions, two methods are used: the (climbing image) nudged elastic
band (NEB) method and the dimer method. Both methods allow the calculation of saddle points. NEB
further allows calculation of minimum energy paths between known reactants and products [18] and the
dimer method is applicable in case the reaction mechanism is unknown [19]. Hence, both methods are fit for
calculation of transition states. DFT calculated activation energies are reliable in a range of roughly 5 kJ/mol
and results should therefore be considered qualitative in an absolute sense, but reliable in a comparative
sense [20].
Methods and models 73

3.3.2.1. Nudged Elastic Band method


The nudged elastic band method is used in case the reactants and products of a certain reaction are known. A
linear path of images intermediate to the initial and final state of the reaction is generated and a constrained
energy optimization is obtained by the addition of spring forces which maintain equal spacing to neighboring
images.

Figure 3-6 Illustration of the climbing image NEB method (cNEB). A constrained energy optimization is
obtained by the addition of spring forces to maintain equal spacing to neighboring images.

In addition to the NEB method a climbing image can be added [21]. This is a modification to the NEB
method in which the highest energy image is driven up to the saddle point. This is done by inversion of the
true forces acting on the image, to maximize its energy.

3.3.2.2. Dimer method


The dimer method does not require from the user to know the reactants and products in advance; it can
determine a saddle point from any initial configuration. Similar to the climbing image NEB (cNEB), the
forces acting on the species are inverted and these directions are used to maximize the energy content of the
image [19, 21].

The cNEB and dimer method can easily be combined to reach the true transition state for a certain reaction.
Via the nudged elastic band method an estimate is generated for the reaction pathway and with the dimer
method the true transition state can be determined.
74 Methods and models

3.3.3. Gibbs free energy

Calculation of adsorption energies generates interesting molecular information, but does not give direct
thermodynamic information. To evaluate the thermodynamic stability of the adsorbates, the Gibbs free
energy ∆𝐺𝑎𝑑𝑠 (𝑇, 𝑝) needs to be calculated. This also allows determining pressure and temperature
dependencies [22]. The general expression for the Gibbs free energy of adsorption is given in Eq. (31).

1
∆𝐺𝑎𝑑𝑠 (𝑇, 𝑝) = ∆𝐻𝑎𝑑𝑠 (𝑇) − 𝑇 ∙ ∆𝑆𝑎𝑑𝑠 + 𝑅 ∙ 𝑇 ∙ ln( ) (31)
𝑃𝑎𝑑𝑠(𝑔)

Energy calculations from VASP are performed at a temperature of 0 K and pressure of 0 Pa. To obtain the
Gibbs free energy for a specified temperature and pressure, the enthalpy 𝐻 temperature correction, entropy 𝑆,
and the zero-point energy (𝑍𝑃𝐸) are necessary. These can be determined from the vibrational, translational
and rotational partition functions of the adsorbates. This requires the knowledge of the vibrational
frequencies of the adsorbates on the surface.

Figure 3-7 Possible vibrational modes for an adsorbed CO molecule on the catalyst surface: 1) C-O stretching, 2)
Co-C stretching, 3) frustrated rotation, 4) frustrated translation [7].

The zero-point energy, denoting the energy of the ground state of the adsorbate, is calculated as follows;
3𝑁−6
1
𝑍𝑃𝐸 = ∑ 𝜈𝑖 (32)
2
𝑖=1

with 𝜈𝑖 the vibrational frequencies and 𝑁 the number of atoms of the adsorbate species. The enthalpy 𝐻
correction and the entropy 𝑆 can be split into three contributions: the vibrational, translational and rotational
𝑙𝑖𝑛
part. The partition functions for the entropy are given by 𝑆𝑣𝑖𝑏 , 𝑆𝑡𝑟𝑎𝑛𝑠 and 𝑆𝑟𝑜𝑡 (𝑆𝑟𝑜𝑡 can be used in case of
linear species) in J/molK in Eq. (33) - (36).
−ℎ𝜈𝑖
−ℎ𝜈𝑖 ℎ𝜈𝑖 𝑒 𝑘𝐵 𝑇
𝑆𝑣𝑖𝑏 = −𝑅 ∑ ln (1 − 𝑒 𝑘𝐵 𝑇 ) + 𝑅∑ −ℎ𝜈𝑖 (33)
𝑘𝐵 𝑇
𝑖 𝑖 (1 − 𝑒 𝑘𝐵 𝑇 )
Methods and models 75

3 2𝜋𝑚 5 5
𝑆𝑡𝑟𝑎𝑛𝑠 = 𝑅[ ln( 2 ) + ln(𝑘𝐵 𝑇) − ln(𝑝 ) + ] (34)
2 ℎ 2 2

8𝜋 2 3 2𝜋𝑘𝐵 𝑇 1 3
𝑆𝑟𝑜𝑡 = 𝑅 [ln ( ) + ln ( 2
) + ln(𝐼𝐴 𝐼𝐵 𝐼𝐶 ) + ]
𝜎 2 ℎ 2 2
(35)
3 𝑘𝐵 𝑇 1 𝐴𝐵𝐶 3
= 𝑅 [ ln ( ) − ln ( ) − ln(𝜎) + ]
2 ℎ 2 𝜋 2

𝑙𝑖𝑛
8𝜋 2 𝐼𝑘𝐵 𝑇 𝑘𝐵 𝑇
𝑆𝑟𝑜𝑡 = 𝑅 [ln ( 2
) + 1] = 𝑅 [ln ( ) + 1] (36)
𝜎ℎ 𝜎ℎ𝐵

The partition functions for the enthalpy are given in Eq. (37) – (39).
−ℎ𝜈𝑖
ℎ𝜈𝑖 𝑒 𝑘𝐵 𝑇
[𝐻(𝑇) − 𝐻(0)]𝑣𝑖𝑏 = 𝑅𝑇 ∑ −ℎ𝜈𝑖 (37)
𝑘𝐵 𝑇
𝑖 (1 − 𝑒 𝑘𝐵𝑇 )

5
[𝐻(𝑇) − 𝐻(0)]𝑡𝑟𝑎𝑛𝑠 = 𝑅𝑇 (38)
2

[𝐻(𝑇) − 𝐻(0)]𝑟𝑜𝑡 = 𝑅𝑇 (39)

Where, 𝑘𝐵 = 1.38066 x 1023 J/K is the Boltzmann constant, ℎ = 6.626 x 10-34 Js the Planck constant, 𝑅 =
8.314 J/molK the universal gas constant, 𝜈𝑖 the vibrational frequencies of the adsorbed species in s -1, 𝜎 the
external symmetry number, 𝐼𝑖 the principal moments of inertia in gcm² and 𝐴𝐵𝐶 the rotational constants
[22].
76 Methods and models

3.4. Chemkin
Chemkin is a software program allowing the kinetic modeling of a large set of reactions, including gas-phase
and surface reactions. It includes a large choice of reactor models that addresses a wide range of industry-
specific reacting-flow conditions. The Chemkin software is a reliable tool to simulate reaction processes and
reduce the necessity of difficult and expensive experimental testing.

Chemkin is constructed of a package of FORTRAN programs designed to model chemical kinetics. Once an
applicable reaction mechanism, including rate constants, and the appropriate thermodynamic data is
implemented in Chemkin, together with the desired reactor model, it is capable of describing symbolically
the reaction mechanism and to solve the reaction and reactor equations for the surface coverages and gaseous
composition following from the reaction mechanism and the desired reactor model [2-4].

Several solvers exist for the solution of a set of ordinary differential equations (ODE) representing the
reaction mechanism and reactor model. Frequently used solvers are the ODEPACK solver, the LSODA
solver and the DASPK solver, which is used in Chemkin. The LSODA solver is capable of automatically
selecting between stiff (Adams multistep) or non-stiff (BDF) methods, and uses a dense or banded Jacobian
in case of a stiff problem. Initially non-stiff methods are used, but via dynamically monitoring of data the
appropriate method is selected [23]. ODEPACK is a systemized collection of initial value ODE solvers,
consisting of LSODE and its variants, and is capable of handling stiff as well as non-stiff problems. The
ODEPACK solver also makes use of Adams multistep methods for solving non-stiff problems, and backward
differentiation formulas for stiff problems [24]. Chemkin employs a modified version of the DASPK solver
for transient simulations. In this thesis simulations are done applying a plug-flow reactor (PFR) in steady-
state and as modeling of the PFR is included in the DASPK solver, this solver is used with the assumption of
steady-state. The DASPK solver makes use of variable-order variable-stepsize backward differentiation
formulas for integration of the system of ordinary differential equations. In case of large-scale systems, the
iterative method can require a suitable preconditioner [25, 26].

Chemkin will be used as a tool to kinetically model the CO insertion mechanism in the Fischer-Tropsch
reaction process and to model CO2 activation reactions, assuming steady-state conditions in a plug-flow
reactor.
Methods and models 77

3.5. Implementation of a Chemkin project


For this master thesis, Chemkin will be used to model the Fischer-Tropsch process and additionally the
catalytic activation of CO2, and therefore will include solely surface reactions. Therefore, getting started with
Chemkin requires three files. The thermodynamic data file describes the chemical species that are present in
the system. The gas-phase kinetics input file provides a symbolic description of the elementary reactions
included in the mechanism. However, as no gas-phase reactions occur for the Fischer-Tropsch process this
file will not include any reaction. Finally, the surface kinetics input file will, similar as the gas-phase kinetics
input file, describe the elementary reactions in the proposed mechanism occurring on the surface.

3.5.1. Thermodynamic data file

In the thermodynamic data file all desired chemical species, appearing in the mechanism, are declared. A
default format (see Table 3-1) is required for the implementation in Chemkin. The first line of the file
specifies the following lines to be a set of thermodynamic data and line two gives three temperatures (low,
break and high value) that are used in the fitting procedure. Further, for every species four lines are required.
The first line describes the name and elements of the chemical component (e.g. OH consists of 1 O and 1 H),
the aggregation state (e.g. gas phase G for OH) and a temperature interval over which integration will take
place for the calculation of the specific heat capacity and the standard enthalpy and entropy defined in Eq.
(40) to (42). Calculation of these thermodynamic properties occurs via the polynomial coefficients given in
line two to four of each component. To finalize the set of thermodynamic data an END line is required.

Table 3-1 Default representation of a chemical species in the thermodynamic data file in Chemkin.

THERMO ALL
200.000 1000.000 6000.000
OH RUS 78O 1H 1 G 200.000 3500.000 1000.000 1
3.09288767E+00 5.48429716E-04 1.26505228E-07 -8.79461556E-11 1.17412376E-14 2
3.85865700E+03 4.47669610E+00 3.99201543E+00 -2.40131752E-03 4.61793841E-06 3
-3.88113333E-09 1.36411470E-12 3.61508056E+03 -1.03925458E-01 4
CO TPIS79C 1O 1 G 200.000 3500.000 1000.000 1
2.71518561E+00 2.06252743E-03 -9.98825771E-07 2.30053008E-10 -2.03647716E-14 2
-1.41518724E+04 7.81868772E+00 3.57953347E+00 -6.10353680E-04 1.01681433E-06 3
9.07005884E-10 -9.04424499E-13 -1.43440860E+04 3.50840928E+00 4
END

Line two to four of each species gives the polynomial coefficients necessary for the calculation of the
thermodynamic properties. Line one and the first two coefficients from line two represent the upper
temperature interval, the other coefficients represent the lower temperature interval [27].
78 Methods and models

°
The specific heat capacity 𝐶𝑝𝑘 of chemical component 𝑘 at standard conditions (ideal gas at 1 atmosphere) is
calculated as follows [26]:

° 𝑀
𝐶𝑝𝑘
= ∑ 𝑎𝑚𝑘 𝑇𝑘𝑚−1 (40)
𝑅
𝑚=1

Standard enthalpy 𝐻𝑘° of component 𝑘 is given by


𝑀
𝐻𝑘° 𝑎𝑚𝑘 𝑇𝑘𝑚−1 𝑎𝑀+1,𝑘
= ∑ + (41)
𝑅𝑇𝑘 𝑚 𝑇𝑘
𝑚=1

and the standard entropy 𝑆𝑘° of component 𝑘 is given by


𝑀
𝑆𝑘° 𝑎𝑚𝑘 𝑇𝑘𝑚−1
= 𝑎1𝑘 ln(𝑇𝑘 ) + ∑ + 𝑎𝑀+2,𝑘 (42)
𝑅 𝑚−1
𝑚=1

3.5.2. Gas phase kinetics file

The gas phase kinetics file provides a symbolic description of the elementary reactions in the mechanism. It
includes information on elements, species, thermodynamic data, and the reactions occurring in the gas phase.
For the research done in this Master thesis, only surface reactions are being considered, hence no reactions
need to be included in the gas phase kinetics file [27]. The structure of the gas phase kinetics file is
illustrated in Table 3-2 with an example. Declarations start with the possible elements in the system, next the
different species are denoted, then the thermodynamic properties of the species are added (in case this has
not been implemented in the thermodynamics data file), and finally the reactions are specified. Each section
should be finalized by an END statement [27].
Table 3-2 Structure of the gas phase kinetics file in Chemkin on the basis of an example for a reaction
mechanism including solely surface reactions: conversion of carbon monoxide towards hydrocarbons and
oxygenates.

ELEMENTS C H O Co
END

SPECIES
CO CO2 H2 H2O
CH4 C2H6 CH3HCO C2H4 CH2O
END

THERMO
Declaration of thermodynamics of gas phase species if not implemented in thermodynamics data file.
END

REACTIONS
END
Methods and models 79

3.5.3. Surface kinetics file

The surface kinetics file provides a symbolic description of the surface reaction mechanism. It includes
information on the surface sites (phases), surface species, bulk phases (optional), bulk species (optional),
thermodynamic data, and the reaction mechanism. The structure of the surface kinetics file is illustrated in
Table 3-3. Different surface reaction mechanisms can be declared in one file, but each set is initialized
equally by specification of a new ‘MATERIAL’. For the surface reaction occur on sites, the properties of the
sites and the number of sites of the catalyst need to be implemented in Chemkin. The type and site density
(SDEN), expressed in mole per cm², are necessary parameters. Next, the surface species need to be
identified, thermodynamic properties should be added if necessary and the reaction mechanism can be
inserted. Each data set should be finalized by an END statement [27].
Table 3-3 Structure of the surface kinetics file in Chemkin on the basis of an example: CO and H 2 adsorption on
the catalyst surface. For every reaction respectively the pre-exponential factor A, the temperature dependency β
and the activation energy Ea need to be inserted for determination of the rate coefficients from the Arrhenius
equation.

MATERIAL KATALYSATOR
SITE/COBALT/ SDEN/2.42E-9
Co(S) H(S) CO(S) O(S) OH(S) C(S) CH(S) CH2(S) CH3(S)
HCO(S) COH(S) CHOH(S) CH2O(S)
END

THERMO
Declaration of thermodynamics of surface species if not implemented in thermodynamics data file.
END

REACTIONS BAR SITE KJOULES/MOLE


CO + Co(S) => CO(S) 1.70E+03 0.0 0.0
CO(S) => CO + Co(S) 1.22E+16 0.0 129.87

H2 + 2Co(S) => 2H(S) 5.72E+09 0.0 0.0


2H(S) => H2 + 2Co(S) 5.90E+12 0.0 55.43
END
The reaction mechanism can consist of every possible chemical reaction including the surface species named
at the top of the file. The reactions may be reversible or irreversible and rate coefficients are determined by
input of the variables appearing in the Arrhenius equation. The formula for the Arrhenius equation is given
by
𝐸𝑎
𝑘 = 𝐴 ∙ 𝑇𝛽 ∙ exp(− ) (43)
𝑅𝑇
The parameters A, β and 𝐸𝑎 are required for each reaction included in the surface kinetics file. The units of
these parameters can be specified by auxiliary reaction keywords, inserted next to the ‘REACTIONS’ term.
For the example in Table 3-3, the pre-exponential factor A is expressed in (bar s)-1, β is dimensionless, and
𝐸𝑎 is expressed in kJ mol-1.
80 Methods and models

3.6. Calculation of kinetic parameters


Calculation of the forward and backward rate coefficients for the elementary reactions is done with the
transition state theory, by combination of the activation energies and corresponding vibrational frequencies,
used to calculate the partition functions for the initial, transition, and final states [28, 29]. The general
expression for the rate coefficient of each elementary surface reaction step is given by the Eyring equation.

𝑘𝐵 𝑇 𝑄† −∆𝐸𝑎
𝑘= exp ( ) (44)
ℎ 𝑄 𝑘𝐵 𝑇

𝑘 represents the rate constant in mol/s, 𝑘𝐵 is the Boltzmann constant, 𝑇 the temperature in K, ℎ is Planck’s
constant, 𝑄† the partition function of the transition state complex, 𝑄 the partition function of the reactants
(initial state), and 𝐸𝑎 is the activation energy of the reaction. The partition functions are the product of the
translational, rotational and vibrational partition functions, given respectively in Eq. (45) – (48). 𝐼𝑖 represents
the moments of inertia, 𝜎 the external symmetry number, and 𝜈𝑖 the fundamental vibrational frequencies.

𝑞𝑡𝑟𝑎𝑛𝑠 2𝜋𝑚𝑘𝐵 𝑇 3/2


=( ) (45)
𝑉 ℎ2

1 8𝜋 2 𝐼𝑘𝐵 𝑇
𝑞𝑟𝑜𝑡 = ( ) (𝑙𝑖𝑛𝑒𝑎𝑟 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒) (46)
𝜎 ℎ2

3/2
𝜋 1/2 8𝜋 2 𝐼𝑘𝐵 𝑇
𝑞𝑟𝑜𝑡 = ( ) (𝐼𝐴 𝐼𝐵 𝐼𝐶 )1/2 (𝑛𝑜𝑛 − 𝑙𝑖𝑛𝑒𝑎𝑟 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒) (47)
𝜎 ℎ2

1
𝑞𝑣𝑖𝑏 = ∏𝑠𝑖=1 −ℎ𝜈𝑖 (48)
1−𝑒 𝑘𝐵𝑇

Adsorption and desorption rate coefficients are calculated from the collision theory [28]. The rate constant
for adsorption reactions is given in Eq. (49), with 𝑃 the pressure in bar, 𝐴𝑠 the surface area of one site, equal
to 6.35e-20 m2, and 𝑆 the sticking coefficient, in this project taken as 1.

𝑃𝐴𝑠
𝑘𝑎𝑑𝑠 = 𝑆 (49)
√2𝜋𝑚𝑘𝐵 𝑇

The rate constant for a desorption reaction is given in Eq. (50), with ℏ = ℎ /2 𝜋 and 𝜃𝑟𝑜𝑡 the characteristic
temperature of rotation.

𝑘𝐵 𝑇 3 𝐴(2𝜋𝑚𝑘𝐵 ) −∆𝐸𝑑𝑒𝑠 (50)


𝑘𝑑𝑒𝑠 = 3
exp ( )
ℎ 𝜎𝜃𝑟𝑜𝑡 𝑘𝐵 𝑇

ℏ2
𝜃𝑟𝑜𝑡 = (51)
2𝑘𝐵 𝐼
Methods and models 81

3.7. Chemical reaction rate expressions


A general expression (Eq. (52)) can be used to describe every elementary reaction in the reaction
mechanism, involving I (ir-)reversible surface reactions and K chemical species.
𝐾 𝐾
′ ′′ (52)
∑ 𝜐𝑘𝑖 𝜒𝑘 ⟺ ∑ 𝜐𝑘𝑖 𝜒𝑘 (𝑖 = 1 … 𝐼)
𝑘=1 𝑘=1

𝜐𝑘𝑖 are the stoichiometric coefficients for the elementary reactions and 𝜒𝑘 the chemical symbol for the 𝑘th
chemical species. For every reaction the rate of production 𝑞𝑖 can be defined as the difference of the forward
and reverse reaction rate:
𝐾 𝐾
′ ′′
𝑞𝑖 = 𝑘𝑓𝑖 ∏[𝑋𝑘 ]𝜐𝑘𝑖 − 𝑘𝑟𝑖 ∏[𝑋𝑘 ]𝜐𝑘𝑖 (53)
𝑘=1 𝑘=1

Calculation of the rate constants 𝑘𝑓𝑖 and 𝑘𝑟𝑖 is described in Section 3.6. Combination of the rates of
production of the different reactions involving the 𝑘th chemical species yields the net production rate 𝑠̇𝑘 ,
′′ ′
with 𝜐𝑘𝑖 = (𝜐𝑘𝑖 - 𝜐𝑘𝑖 ) [30].
𝐼

𝑠̇𝑘 = ∑ 𝜐𝑘𝑖 𝑞𝑖 (𝑘 = 1 … 𝐾) (54)


𝑖=1

3.8. Chemkin simulation


The simulations in Chemkin are performed under steady state and isothermal conditions for a plug flow
reactor (PFR) model, and Table 3-4 illustrates the reaction and inlet conditions for the base case simulations.
Operating conditions are taken similar to Fischer-Tropsch synthesis [31].

Table 3-4 Reaction and inlet conditions for the base case simulations performed in Chemkin. Reaction conditions
are taken similar as for Fischer-Tropsch synthesis [31].

T (K) 500
P (bar) 20
Reactor Length (m) 1
Diameter (m) 0.012
Int. surf. area (m²/m) 3600[32]
Dispersion (molact sites/molcatalyst) 523
Number of active sites per gram of catalyst (g-1) 2.17e+22
Space velocity (mol/kgcats) 62
Feed (kg/s) 6.25e-4
Hydrogen inlet fraction 2/3
CO inlet fraction 1/3
82 Methods and models

The simulations are run for a temperature of 500 K and at pressure of 20 bar. The reactor is taken to be 1 m
long and has a diameter of 0.012 m. A cobalt catalyst is used for the simulation of the FTS and typical values
are taken for the catalytic properties [7]. The catalyst internal surface area per unit length is taken as 3600
m2/m [32], which is equal to a dispersion of 523 molact sites/molcatalyst.. The feed flow has a mass inlet flow rate
of 6.25E-04 kg/s, which agrees to a space velocity of 62 mol/kgcats for a feed of 1/3 CO and 2/3 H2.

3.9. Turnover frequency and selectivity


The turnover frequency (TOF) is defined as the number of molecules converted per active site and per unit of
time [29]. The TOF is a means to determine the overall rate of production, and can be used for comparison
with experimentally obtained values.

The mass balance for CO adsorbed on the surface is given by:

𝑑𝜃𝐶𝑂
= 𝑟𝑎𝑡𝑒 𝑜𝑓 𝐶𝑂 𝑎𝑑𝑠𝑜𝑟𝑝𝑡𝑖𝑜𝑛 − 𝑟𝑎𝑡𝑒 𝑜𝑓 𝐶𝑂 𝑑𝑒𝑠𝑜𝑟𝑝𝑡𝑖𝑜𝑛 − 𝑟𝑎𝑡𝑒 𝑜𝑓 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝐶𝑂 (55)
𝑑𝑡
and at steady state the TOFCO, which is the rate of reaction of CO on the surface, can be expressed as the rate
of adsorption minus the rate of desorption.

𝑇𝑂𝐹𝐶𝑂 = 𝑘𝑎𝑑𝑠 𝑝𝐶𝑂 𝜃 ∗ − 𝑘𝑑𝑒𝑠 𝜃𝐶𝑂 (56)


Selectivity calculations are a means to represent the product distribution obtained from a certain reaction
mechanism under specific operating conditions [29]. The selectivity for a product P is given in Eq. (57), with
in this case CO or CO2 as reactant R.

𝑚𝑜𝑙𝑒𝑠 𝑜𝑓 𝑃 𝑝𝑟𝑜𝑑𝑢𝑐𝑒𝑑
𝑆= (57)
𝑚𝑜𝑙𝑒𝑠 𝑜𝑓 𝑅 𝑟𝑒𝑎𝑐𝑡𝑒𝑑

3.10. Sensitivity analysis


According to Saltelli et al. [33], a sensitivity analysis is “the study of how uncertainty in the output of a
model (numerical or otherwise) can be apportioned to different sources of uncertainty in the model input”.
This means it is a tool to quantitatively understand how the solution of a problem depends on the parameters
contained in the model [26].

Chemkin allows calculating the first-order sensitivity coefficients with respect to the rate coefficients of the
reaction for several parameters. For this thesis, the parameters of interest are the species fractions and
gaseous composition involved in the reaction mechanism. The calculations are further facilitated, because the
Jacobian for the steady-state solution is already determined in solving the nonlinear reaction equations.

Calculation of the sensitivity coefficients in Chemkin starts from the vector of governing equations 𝐹, in this
case representing the reaction equations for the reaction mechanism.
Methods and models 83

𝐹 (𝜙(𝛼 ); 𝛼 ) = 0 (58)

With 𝜙 the species fractions and 𝛼 the pre-exponential factors. In case of a sensitivity analysis for the species
fractions, the pre-exponential factors 𝛼 in the Arrhenius expressions of the reaction rates are modified to
evaluate the reaction rate sensitivity. By differentiation of Eq. (58) an expression comprising the sensitivity
coefficients is obtained in matrix notation.

𝜕𝐹 𝜕𝜙 𝜕𝐹
+ =0 (59)
𝜕𝜙 𝜕𝛼 𝜕𝛼

𝜕𝐹 𝜕𝐹
𝜕𝜙
represents the Jacobian of the original system, 𝜕𝛼 the matrix of partial derivatives of F with respect to the
𝜕𝜙
parameters and 𝜕𝛼
the sensitivity coefficients, containing quantitative information on how each reaction rate

coefficient 𝛼 affects the species fractions 𝜙. Each column of the matrix of sensitivity coefficients contains
the dependence of the solution vector on a particular chemical reaction. Determination of the sensitivity
coefficients is done by solving Eq. (59) [26].

3.11. Challenges involving the set-up of a Chemkin project


In this section the challenges faced during the first implementation of a Chemkin project will be discussed.
Although at first sight implementation of a Chemkin project seems straightforward, care has to be taken
during preparation of the input values and declaration in the system.

Chemkin requires three files for calculation: the thermodynamic data file, the gas phase kinetics file, and the
surface kinetics file. The thermodynamic data file contains the declaration of the chemical species contained
in the system with a specified format illustrated in Table 3-1. Indentation is of major importance in this
format and calculations will fail in case it is not respected. The polynomials in the declaration of each species
are used to calculate the specific heat capacity and the standard enthalpy and entropy. These are of
importance in case solely forward reaction rates are implemented. Chemkin will then calculate reverse
reaction rates from thermodynamic consistency. However, in case both forward and reverse reactions are
defined, thermodynamic consistency is overwritten and the polynomials could be set to zero as they will not
influence the outcome of the calculations. Note that care should be taken on the indentation of the
thermodynamic data file. For the modeling occurring in this master thesis the gas phase kinetics file is
limited as only surface reactions are involved. Care should be taken that all gas species are incorporated in
the gas phase kinetics file, but other remarks are similar to the surface kinetics file. In construction of the
surface kinetics file one should first make sure that the solid phase is well defined. This means that clear
distinction should be made whether the solid phase contains a difference in bulk and surface species or not.
In this master thesis, a uniform catalyst surface is used, hence declaration of a bulk phase should be avoided.
Presence of a bulk phase would affect the outcome of the calculations. Secondly, similar considerations
should be taken in mind as for the thermodynamic data file in case additional species are declared. When
84 Methods and models

defining the reaction equations, the nomenclature used for the species should be followed strictly and the site
balance needs to be fulfilled for every reaction. For every reaction the kinetic parameters for the Arrhenius
equation need to be given, but indentation is not important. The units of the parameters need to be consistent,
but as explained in section 3.5.3 the units can be modified to the user’s preference. Some numerical
limitations exist for the kinetic parameters. Pre-exponential factors above 1020 cannot be processed by
Chemkin and in case of declaration of both forward and reverse reaction rates, the equilibrium coefficient is
limited in magnitude for the calculations to run. Likewise, the activation energies for forward and reverse
reaction should be consistent.

Drawbacks for the use of Chemkin as a modeling tool are its incapability of implementing neither
temperature dependency nor the coverage dependency of kinetic parameters.
Methods and models 85

3.12. References

1. Gunasooriya, G.K.K., et al., Key role of surface hydroxyl groups in CO activation during Fischer-
Tropsch synthesis. ACS Catalysis, 2016.

2. Kee, R.J., J.A. Miller, and T.H. Jefferson, CHEMKIN: A general-purpose, problem-independent,
transportable, FORTRAN chemical kinetics code package. 1980, Sandia Labs.

3. Getting stared with CHEMKIN, R. Design, Editor. 2015.

4. Storsæter, S., D. Chen, and A. Holmen, Microkinetic modelling of the formation of C 1 and C 2
products in the Fischer–Tropsch synthesis over cobalt catalysts. Surface Science, 2006. 600(10): p.
2051-2063.

5. Hansen, A.G., W.J. van Well, and P. Stoltze, Microkinetic modeling as a tool in catalyst discovery.
Topics in Catalysis, 2007. 45(1): p. 219-222.

6. Metaxas, K., et al., A microkinetic vision on high-throughput catalyst formulation and optimization:
development of an appropriate software tool. Topics in Catalysis, 2010. 53(1-2): p. 64-76.

7. GUNASOORIYA, G.T.K.K., Combined Theoretical and Experimental Study of CO Adsorption and


Reactivity Over Platinum and Cobalt. 2014.

8. Kohn, W., Nobel Lecture: Electronic structure of matter—wave functions and density functionals.
Reviews of Modern Physics, 1999. 71(5): p. 1253.

9. Calais, J.-L., Density-functional theory of atoms and molecules. R.G. Parr and W. Yang, Oxford
University Press, New York, Oxford, 1989. IX + 333 pp. Price £45.00. International Journal of
Quantum Chemistry, 1993. 47(1): p. 101-101.

10. Morrow, P.-S., Contact Magnetoresistance of Multilayered Cobalt/copper Nanostructures Measured


by Scanning Tunneling Microscope. 2008: ProQuest.

11. Yoder, B.L., Steric Effects in the Chemisorption of Vibrationally Excited Methane on Nickel. 2012:
Springer Science & Business Media.

12. Sun, Y. and Y. Xia, Triangular nanoplates of silver: synthesis, characterization, and use as
sacrificial templates for generating triangular nanorings of gold. Advanced Materials, 2003. 15(9):
p. 695-699.

13. Kresse, G.M., Martijn; Furthmüller, Jürgen. VASP the Guide. 2015 [cited 2015; Available from:
http://cms.mpi.univie.ac.at/vasp/vasp/vasp.html.

14. Lee, K., et al., Higher-accuracy van der Waals density functional. Physical Review B, 2010. 82(8):
p. 081101.

15. Monkhorst, H.J. and J.D. Pack, Special points for Brillouin-zone integrations. Physical Review B,
1976. 13(12): p. 5188.

16. Zhao, Y.-F., et al., Insight into methanol synthesis from CO 2 hydrogenation on Cu (111): complex
reaction network and the effects of H 2 O. Journal of Catalysis, 2011. 281(2): p. 199-211.

17. Sholl, D. and J.A. Steckel, Density functional theory: a practical introduction. 2011: John Wiley &
Sons.
86 Methods and models

18. Nudged elastic band. [cited 2016; Available from: http://theory.cm.utexas.edu/vtsttools/neb.html.

19. The dimer method. [cited 2016; Available from:


http://theory.cm.utexas.edu/vtsttools/dimer.html#dimer.

20. van Santen, R.A., M. Ghouri, and E.M. Hensen, Microkinetics of oxygenate formation in the
Fischer–Tropsch reaction. Physical Chemistry Chemical Physics, 2014. 16(21): p. 10041-10058.

21. Henkelman, G., B.P. Uberuaga, and H. Jónsson, A climbing image nudged elastic band method for
finding saddle points and minimum energy paths. The Journal of chemical physics, 2000. 113(22): p.
9901-9904.

22. Irikura, K.K. Essential statistical thermodynamics. in ACS Symposium Series. 1998. American
Chemical Society.

23. Petzold, L., Automatic selection of methods for solving stiff and nonstiff systems of ordinary
differential equations. SIAM journal on scientific and statistical computing, 1983. 4(1): p. 136-148.

24. Hindmarsh, A.C., ODEPACK, A Systematized Collection of ODE Solvers, RS Stepleman et al.(eds.),
North-Holland, Amsterdam,(vol. 1 of), pp. 55-64. IMACS transactions on scientific computation,
1983. 1: p. 55-64.

25. Li, S. and L. Petzold, Design of new DASPK for sensitivity analysis. University of California at
Santa Barbara, Santa Barbara, CA, 1999.

26. Design, R., CHEMKIN Theory Manual. San Diego, CA, 2007.

27. Design, R., CHEMKIN Software Input Manual. San Diego, CA, 2007.

28. Chemistry, R.S.o., Process and Chemical Engineering. 1996: Royal Society of Chemistry.

29. Chorkendorff, I. and J.W. Niemantsverdriet, Concepts of modern catalysis and kinetics. 2006: John
Wiley & Sons.

30. Kee, R., et al., SURFACE CHEMKIN: A Software Package for the Analysis of Heterogeneous
Chemical Kinetics at a Solid-Surface–Gas-Phase Interface. CHEMKIN Collection Release. 3.

31. Lox, E., B. Engler, and E. Koberstein, Diesel emission control. Studies in Surface Science and
Catalysis, 1991. 71: p. 291-321.

32. Jacobs, G., et al., Fischer–Tropsch synthesis: support, loading, and promoter effects on the
reducibility of cobalt catalysts. Applied Catalysis A: General, 2002. 233(1–2): p. 263-281.

33. Saltelli, A., et al., Global sensitivity analysis: the primer. 2008: John Wiley & Sons.
87

Chapter 4
Ab-initio calculations
Activation of the thermodynamically stable CO 2 molecule is a key challenge for conversion pathways to deal
with. It requires a high amount of energy for its conversion to a higher energy state, and minimization of the
energy requirement should be strived for. Therefore, appropriate catalyst selection is of high importance [1].
By using computational catalyst screening, more effective catalyst systems can be sought. Furthermore,
advanced DFT calculations allow describing surface reactions in high detail and with high accuracy. The
ultimate goal of computational catalysis would be to determine the material required to catalyze a certain
reaction under a set of specified conditions [2].

In this chapter a first activation of CO2 on the Co(111) surface is evaluated. Next, the effect of coverage and
presence of step sites, i.e. CO2 activation on Co(211), is studied. Finally, the study is extended to other
catalysts such as Cu, Ni and Au.
88 Ab-initio calculations

4.1. Analysis of the cobalt catalytic surface


Cobalt catalysts are widely used for FTS performed at relatively high pressures and low temperature (20 bar,
500 K). The FTS turnover frequency (TOF) is reported to be independent of particle size above 10 nm [3],
and the relevant reaction steps can be concluded to occur on terraces. Therefore, CO 2 activation is initially
analyzed for the Co(111) surface as a reference case. Under realistic reaction conditions (20 bar, 500 K) CO
coverage is high and has a significant effect on the reaction kinetics and selectivity [4]. Therefore the effect
of CO coverage is evaluated to obtain a more realistic model of the Co surface under FTS. Moreover,
significantly lower activation barriers have been calculated for CO dissociation on step sites [5], therefore
the impact of step sites on the Co(211) surface is examined for the CO 2 activation.

4.1.1. Adsorption of reaction intermediates on Co(111)

The first step in the activation of CO2 is its adsorption on the catalyst surface. DFT calculations render an
adsorption energy for CO2 of -15 kJ/mol for top and hollow sites and -16 kJ/mol for bridge sites on a clean
Co(111) surface. This is in good agreement with values from literature, denoting adsorption energies in a
range of -16 to -28 kJ/mol [5, 6]. CO2 adsorbs preferentially on the bridge sites, although the difference for
top and hollow sites is only 1 kJ/mol. For the adsorbed CO2 molecule a fairly high entropy of 120 J/molK is
calculated. Calculation of the Gibbs free energy however indicates that, although the adsorption energy is
slightly negative, adsorption of CO2 is thermodynamically unfavorable with a Gibbs free energy of
adsorption of 65 kJ/mol at 500 K and 1 bar. Table 4-1 summarizes the DFT calculated adsorption energies,
entropies and Gibbs free energies of adsorption, accompanied by the optimized structures for CO and CO2 in
Figure 4-1.

(a) (b)

Figure 4-1 Optimized adsorption configuration for (a) CO* and (b) CO2* on a clean Co(111) surface.

To further evaluate the competitiveness between CO activation and CO2 activation under FTS, adsorption of
CO is also analyzed. Adsorption energies for CO are represented in Table 4-1 and CO was found to adsorb
preferentially on top sites, with an adsorption energy of -121 kJ/mol; in good agreement with the
experimentally determined low coverage CO adsorption enthalpy of -128 kJ/mol [7]. Gibbs free energies of
adsorption for CO are calculated to be -55 kJ/mol and hence, adsorption of CO is favorable compared to the
thermodynamically unstable CO2 adsorption.
Ab-initio calculations 89

Table 4-1 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch conditions
(500 K, 20 bar, 60% conversion) for the most stable configuration of reaction intermediates on a clean Co(111)
surface. An extensive summary is given in Table B-1 in Appendix B.

Reaction Adsorption energy Entropy of adsorbed species Gibbs free adsorption energy
intermediate (kJ/mol) (J/mol K) (kJ/mol)

C* -625 18 53

O* -543 24 -16

CO* -121 75 -55

CO2* -15 134 43

H* -259 7 -2

OH* -374 41 5

HCO* -170 64 68

COOH* -203 97 121

HCOO* -298 111 37

HCOOH* -43 124 115


The relative stability of the reaction intermediates is evaluated relative to a CO, CO 2, H2 and H2O gas phase
reservoir from the reaction Gibbs free energy equations under FT conditions (500 K, 20 bar, 60% conversion);
and includes the effects of pressure, composition, and temperature. The Gibbs free energy reactions:
𝑪𝑶(𝒈) + 𝑯𝟐(𝒈) ⇌ 𝑪∗ + 𝑯𝟐 𝑶(𝒈) (60)
𝒙
𝑯𝟐 𝑶(𝒈) + ∗ ⇌ 𝑶𝑯∗𝒙 + (𝟏 − )𝑯𝟐(𝒈) (61)
𝟐
𝒑𝑯𝟐 𝑶
∆𝑮𝑪(𝑻, 𝒑) = ∆𝑮𝟎 (𝟓𝟎𝟎 𝑲) + 𝑹𝑻𝒍𝒏( ) (62)
𝒑𝑪𝑶 𝒑𝑯𝟐
𝒑𝑯 𝟏−𝒙/𝟐
∆𝑮𝑶𝑯𝒙 (𝑻, 𝒑) = ∆𝑮𝟎 (𝟓𝟎𝟎 𝑲) + 𝑹𝑻𝒍𝒏( 𝟐 ) (63)
𝒑𝑯𝟐 𝑶
The Gibbs free energies for gas phase and adsorbed species were obtained by combining the electronic and zero-
point vdW-DF2 energies with enthalpy and entropy corrections from frequency calculations for the full
structure [5].

Adsorption of hydrogenating species such as H* and OH* and several reaction intermediate were evaluated
next. Adsorption energies, entropies and Gibbs free adsorption energies are summarized in Table 4-1, and
the accompanying optimized structures for H* and OH* in Figure 4-2. The reaction intermediates are
evaluated relative to CO, CO2, H2 and H2O in the gas phase for FT conditions (500 K, 20 bar, 60%
conversion). Calculations for H* and OH* show relatively high adsorption energies of -259 kJ/mol and -373
kJ/mol respectively, close to values found in literature [5, 6, 8]. However, calculation of the Gibbs free
energies of adsorption indicate that adsorption of hydrogen is only slightly thermodynamically favorable
(∆Gads = -2 kJ/mol), and adsorption of OH* is unfavorable (∆G ads = 5 kJ/mol). This would imply that only
few hydroxyl groups will be present on the surface and hydrogenation reactions will mainly occur via H*
species. Reaction intermediates such as HCOO*, COOH*, HCO* and HCOOH* have reasonable adsorption
energies, and entropy values are quite high. The intermediates are however thermodynamically unstable,
illustrated by the highly positive Gibbs free energies of adsorption relative to CO, CO 2, H2 and H2O in the
90 Ab-initio calculations

gas phase. Calculation of the thermodynamic properties of the reaction intermediates will further be used for
analysis of the reaction path for CO2 activation.

(a) (b)

Figure 4-2 Optimized adsorption configuration for (a) H* (fcc) and (b) OH* (fcc) on a clean Co(111) surface.

4.1.2. Reaction mechanism for activation of CO2 on Co(111)

Activation of CO2 has first been examined for direct dissociation on the clean Co(111) surface.
Determination of the reaction barriers is done via the climbing image nudged elastic band (cNEB) and
refined with the dimer method, as explained in Chapter 3. Energy, entropy and Gibbs free energy barriers are
determined, along with the entropy and Gibbs free energy change of the reaction, summarized in Table 4-2.
An activation energy for CO2 dissociation of 60 kJ/mol is obtained, illustrated in Figure 4-3. This calculated
value is significantly lower than the previously reported activation barrier of 261 kJ/mol by Iglesia et al. [9].
The Gibbs free reaction energy is determined at -72 kJ/mol and hence, dissociation of the CO2 molecule on
Co(111) is thermodynamically feasible and has a reasonably low activation barrier. However, as has been
discussed in Section 4.1.1, adsorption of CO2 is unfavorable.

Figure 4-3 Energy profile for the dissociation of carbon dioxide on a clean Co(111) surface (CO 2* + *  CO* +
O*). An energy barrier of 60 kJ/mol and Gibbs free energy of reaction of -72 kJ/mol is obtained. Detailed
visualization of the transition state is given in Table B-3 in Appendix B.

Comparison can again be made with the dissociation of CO, illustrated in Figure 4-4. An activation energy of
243 kJ/mol has been calculated, which is in agreement with values found in literature [5, 6, 10]. From
calculation of the Gibbs free energy of reaction (94 kJ/mol), it is also clear that the dissociation of CO* on
the Co(111) surface is thermodynamically unfavorable, and one could conclude that the dissociation of CO 2 *
Ab-initio calculations 91

would be the preferred reaction. However, it is also important to take in account the coverages of the species
on the surface; and as the adsorption of CO is highly favorable over CO2 it is not straightforward to predict
which reaction to be dominant.

Figure 4-4 Energy profile for the dissociation of carbon monoxide on a clean Co(111) surface (CO* + *  C* +
O*). An energy barrier of 243 kJ/mol and Gibbs free energy of reaction of 94 kJ/mol is obtained. Detailed
visualization of the transition state is given in Table B-3 in Appendix B.

Pichler and Schulz originally proposed the hydrogen-assisted pathway for CO activation, suggesting CO* is
hydrogenated to HCO* and HCOH* species and in a second step scission of the C-O bond occurs to obtain
CH* species on the surface. Literature reports have shown that the hydrogen-assisted pathway provides a
more favorable CO activation route on Co terraces. Hydrogen indeed weakens the C-O bond and facilitates
the scission reaction [10, 11]. Similarly, studies have been done on the activation of CO2 via hydrogen-
assisted pathways [12]. Therefore, hydrogenating species, in this case H*, are included to evaluate a
hydrogen-assisted activation pathway for CO2 on a clean Co(111) surface. The reactions taken in account are
represented in Table 4-2: (1) the hydrogenation of CO2 to form carboxyl or formate species; (2) dissociation
of the carboxyl to carbon monoxide and hydroxyl groups; (3) dissociation of the formate species to form
HCO* species and oxygen on the surface; (4) hydrogenation of the hydrocarboxyl and formate groups to
formic acid compounds; and (5) dissociation of the formic acid compounds to form HCO* species and
hydroxyl on the surface.
92 Ab-initio calculations

Table 4-2 Energy barriers (Ef), entropy of activation (∆Sf), Gibbs free energy barriers (∆Gf), entropy of reaction
(∆Sr) and Gibbs free energy of reaction (∆Gr) for CO* and CO2* scission reactions and CO2* activation reactions
on a clean Co(111) surface. The accompanying transition state structures are given in Table B-3 in Appendix B.

Surface reactions Energy Ef Entropy ∆Sf Gibbs Free Entropy ∆Sr Gibbs Free
(kJ/mol) (J/mol K) Activation (J/mol K) Reaction
Energy ∆Gf Energy ∆Gr
(kJ/mol) (kJ/mol)

CO* + *  C* + O* 243 -23 251 -33 94

CO2* + *  CO* + O* 60 -33 53 -29 -72

CO2* + H*  COOH* + * 154 -24 162 -38 70

CO2* + H*  HCOO* + * 94 -19 103 -24 -13

COOH* + H*  HCOOH* + * 61 4 71 21 -8

COOH* + *  CO* + OH* 63 -77 67 17 -127

HCOO* + H*  HCOOH* + * 110 16 116 6 75

HCOO* + *  HCO* + O* 107 -21 112 -23 62

HCOOH* + *  HCO* + OH* 186 26 162 -22 2

Figure 4-5 shows the electronic energy profile for the activation pathways for CO 2, calculated with the DFT-
vdW-DF2 functional. Direct dissociation of CO2 appears to be the most favorable step, with lowest
activation energy as well as the most favorable Gibbs free energy of reaction. Further conversion can then be
related to conversion of carbon monoxide, and effective barriers as well as competition with CO adsorbed
from the gas phase needs to be further examined.

For the hydrogen-assisted pathway, hydrogenation to formate species has a significant advantage over the
hydrocarboxyl pathway with activation energies of 91 kJ/mol and 154 kJ/mol respectively. The activation
energies for the hydrocarboxyl groups are significantly lower than for formate species favoring the
hydrocarboxyl pathway. However, the Gibbs free energy of reaction of -13 kJ/mol for the hydrogenation of
CO2 to formate indicates that it is thermodynamically favorable, in comparison with the hydrocarboxyl
intermediate which is 70 kJ/mol endothermic. The effective barrier for the formate pathway will hence be
lower. The dissociation reaction for the formate group towards HCO* and oxygen on the surface also favors
the formate pathway as it has the lowest activation barrier for formation of HCO* compared to HCOOH*
dissociation. Hence, one could conclude that the formate pathway will be dominant.
Ab-initio calculations 93

Figure 4-5 Electronic energy profile calculated with the DFT-vdW-DF2 functional for the activation of CO2 via
(a) direct dissociation; (b) hydrogen-assisted pathway via HCOO*; and (c) hydrogen-assisted pathway via
COOH*. The activation barriers are indicated and selected transition state structures are shown, a complete
overview of the transition state structures is given in Appendix B Table B-3.

Further analysis of the reaction pathways is however required to obtain thorough insight in the activation of
CO2. Additional hydrogenation reactions could be included, with additional intermediates such as COHOH*,
HCOHOH*, H2COO* and H2CO*, as has been studied by Zhao et al. [13] for conversion of CO2 to methanol
on the Cu(111) surface. Additional research might lead to alternative pathways to the ones presented in
Figure 4-5, and more favorable pathways could be detected. The species from the activation of CO 2 (CO*
and HCO*) can further be incorporated in the reaction pathways for CO activation, explained further in this
thesis.

4.1.3. Coverage effects on Co(111)

Under FT conditions, carbon monoxide coverages are typically in a range of 0.33 to 0.50 [5, 14]. To assess the
the impact of coverage effects on the activation of CO 2, adsorption and activation energies are calculated for a
a Co(111) surface with CO coverage of 1/3 ML. Table 4-3 summarizes the DFT calculated adsorption energies,
entropies of adsorbed species and Gibbs free energy of adsorption for CO* and CO2* under FT conditions;

Figure 4-6 illustrates the optimized structures.

Table 4-3 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch conditions
(500 K, 20 bar, 60% conversion) for CO* and CO2* adsorbates on a Co(111) surface with CO coverage of 1/3
ML. A more elaborated summary is given in Table B-4 in Appendix B.

Reaction Adsorption energy Entropy of adsorbed species Gibbs free adsorption energy
intermediate (kJ/mol) (J/mol K) (kJ/mol)

CO* -122 71 -53

CO2* -23 142 34


94 Ab-initio calculations

(a) (b)

Figure 4-6 Optimized adsorption configuration for (a) CO* and (b) CO2* on a Co(111) surface with CO coverage
of 1/3 ML.

Comparison with the clean Co(111) surface, see Table 4-1, indicates that the higher coverage has a
stabilizing effect and the adsorption becomes stronger by 8 kJ/mol. The entropy of the adsorbates is also
slightly increased, hence the favorability of the adsorption of CO 2 has been enhanced. However, Gibbs free
adsorption energy still indicates the adsorption to be thermodynamically unstable. For CO adsorption, it has
been observed that for increased CO coverage of 1/3 ML on a Co(0001) p(3x3) unit slab, the adsorption
energy also tends to increase due to reduced Pauli repulsion between the partially filled Co dz 2 state and the
CO LP NBO, and reduced back-donation to the 2π* CO orbitals [14]. A similar cause is expected for the
trends observed in CO2 adsorption.

Figure 4-7 Energy profile for the dissociation of carbon dioxide on a Co(111) surface with CO coverage of 1/3
ML (CO2* + *  CO* + O*). An energy barrier of 81 kJ/mol and Gibbs free energy of reaction of -55 kJ/mol is
obtained. Detailed visualization of the transition state is given in Table B-6 in Appendix B.

Looking at the dissociation reaction of CO2 (Figure 4-7 and Table 4-4) shows that, due to the higher stability
of the CO2 molecule and the presence of CO on the surface, the activation energy increases by 21 kJ/mol.
This means that under FT conditions, the dissociation of CO 2 would be less favorable. Also note that under
real FT conditions the surface is likely to be more complex and the presence of reaction intermediates might
further affect the stability of the CO2 adsorbates [11].
Ab-initio calculations 95

Table 4-4 Energy barrier (Ef), entropy of activation (∆Sf), Gibbs free energy barrier (∆Gf), entropy of reaction
(∆Sr) and Gibbs free energy of reaction (∆Gr) for the CO2* scission reaction on a Co(111) surface with CO
coverage of 1/3 ML.

Surface reactions Energy Ef Entropy ∆Sf Gibbs Free Entropy ∆Sr Gibbs Free
(kJ/mol) (J/mol K) Activation (J/mol K) Reaction
Energy ∆Gf Energy ∆Gr
(kJ/mol) (kJ/mol)

CO2* + *  CO* + O* 81 -12 77 -29 -55

4.1.4. Effect of step sites on Co(211)

Significantly lower activation barriers have been reported for the CO dissociation reaction on step sites [15],
e.g. Ge and Neurock [16] reported a dissociation barrier of 200 kJ/mol on a clean Co(211) surface, which is
approximately 40 kJ/mol lower than for the Co(111) surface. In search of an appropriate catalyst model for
the activation of CO2, the Co(211) surface is evaluated. An important remark to model a realistic surface is
however that due to their low thermodynamic stability and high affinity for reaction intermediates and
carbon, it is expected that step sites might not be available under FTS conditions [5].

Analysis of adsorption energies and Gibbs free energies of adsorption shows that step sites are beneficial for
the adsorption of CO species as well as CO2 species (see Table 4-5), compared to the Co(111) surface. CO is
calculated to preferentially bind on a top site (Figure 4-8), with an adsorption energy of -204 kJ/mol,
comparable to reported values of -160 kJ/mol in literature [16], and Gibbs free energy of adsorption of -132
kJ/mol. Also adsorption of CO 2 is calculated to be thermodynamically favorable, with an adsorption energy
of -99 kJ/mol and Gibbs free adsorption energy of -27 kJ/mol for top sites.

Table 4-5 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch conditions
(500 K, 20 bar, 60% conversion) for CO* and CO2* adsorbates on a clean Co(211) surface. Accompanying
structures are illustrated in Figure 4-8. A more elaborated summary is given in Table B-4 in Appendix B.

Reaction Adsorption energy Entropy of adsorbed species Gibbs free adsorption energy
intermediate (kJ/mol) (J/mol K) (kJ/mol)

CO* -204 67 -132

CO2* -99 102 -27


96 Ab-initio calculations

(a) (b)

Figure 4-8 Optimized adsorption configuration for (a) CO* and (b) CO2* on a clean Co(211) surface.

The thermodynamic feasibility of the adsorption of CO2 on step sites on the Co(211) surface indicates that
step sites are potentially good catalyst structures for conversion of CO2.

Figure 4-9 Energy profile for the dissociation of carbon dioxide on a clean Co(211) surface (CO 2* + *  CO* +
O*). An energy barrier of 54 kJ/mol and Gibbs free energy of reaction of -121 kJ/mol is obtained. Detailed
visualization of the transition state is given in Table B-9 in Appendix B.

Figure 4-9 illustrates the energy profile for the dissociation of carbon dioxide on a clean Co(211) surface. An
activation energy of 54 kJ/mol is calculated, which is in the same order of magnitude as for dissociation on
the Co(111) surface and is a reasonably low value when compared to the activation energy for CO*
dissociation. The Gibbs free energy of reaction (see Table 4-6) is determined at -121 kJ/mol. Hence,
dissociation of CO2 is also thermodynamically favored and step sites appear to have high potential for the
conversion of CO2.

Table 4-6 Energy barrier (Ef), entropy of activation (∆Sf), Gibbs free energy barrier (∆Gf), entropy of reaction
(∆Sr) and Gibbs free energy of reaction (∆G r) for the CO2* scission reaction on a clean Co(211) surface.

Surface reactions Energy Ef Entropy ∆Sf Gibbs Free Entropy ∆Sr Gibbs Free
(kJ/mol) (J/mol K) Activation (J/mol K) Reaction
Energy ∆Gf Energy ∆Gr
(kJ/mol) (kJ/mol)

CO2* + *  CO* + O* 54 -6 80 -8 -121


Ab-initio calculations 97

4.2. Analysis of the Cu(111) surface


Copper catalysts have been identified as an active phase for methanol synthesis via CO 2 hydrogenation and
already extensive studies have been performed for Cu catalysts via experimental and theoretical methods
[13]. By means of comparison with the previously discussed cobalt surface, CO 2 activation is also evaluated
on the Cu(111) surface.

4.2.1. Adsorption of reaction intermediates on Cu(111)

Figure 4-10 shows the DFT calculated adsorption configurations for CO* and CO2* on the clean Cu(111)
surface. CO and CO2 were both found to preferentially adsorb on a top site position and respective
adsorption energies of -32 kJ/mol and -15 kJ/mol are calculated, close to values found in literature [13, 17,
18].

(a) (b)

Figure 4-10 Optimized adsorption configuration for (a) CO* and (b) CO2* on a clean Cu(111) surface.

However, neither CO nor CO2 adsorption appears to be thermodynamically favorable. Gibbs free energy of
adsorption calculations show values of 73 kJ/mol for CO adsorption and 100 kJ/mol for CO2 adsorption. A
summary of the DFT calculated adsorption energies, entropies and Gibbs free energies of adsorption for the
most stable configurations of intermediates on the clean Cu(111) surface is given in Table 4-7. The positive
values obtained for the Gibbs free energies of adsorption indicate that the intermediates will be
thermodynamically unstable.
98 Ab-initio calculations

Table 4-7 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch conditions
(500 K, 20 bar, 60% conversion) for the most stable configuration of reaction intermediates on a clean Cu(111)
surface. A more extended summary is given in Appendix B Table B-10.

Reaction Adsorption energy Entropy of adsorbed species Gibbs free adsorption energy
intermediate (kJ/mol) (J/mol K) (kJ/mol)

C* -414 21 246

O* -455 24 73

CO* -32 36 73

CO2* -15 64 100

H* -235 23 6

OH* -329 44 48

COOH* -145 103 175

HCOO* -261 91 85
The relative stability of the reaction intermediates is evaluated relative to a CO, CO 2, H2 and H2O gas phase
reservoir from the reaction Gibbs free energy equations under FT conditions (500 K, 20 bar, 60% conversion);
and includes the effects of pressure, composition, and temperature. The Gibbs free energy reactions:
𝑪𝑶(𝒈) + 𝑯𝟐(𝒈) ⇌ 𝑪∗ + 𝑯𝟐 𝑶(𝒈) (60)
𝒙
𝑯𝟐 𝑶(𝒈) + ∗ ⇌ 𝑶𝑯∗𝒙 + (𝟏 − )𝑯𝟐(𝒈) (61)
𝟐
𝒑𝑯𝟐 𝑶
∆𝑮𝑪(𝑻, 𝒑) = ∆𝑮𝟎 (𝟓𝟎𝟎 𝑲) + 𝑹𝑻𝒍𝒏( ) (62)
𝒑𝑪𝑶 𝒑𝑯𝟐
𝒑𝑯 𝟏−𝒙/𝟐
∆𝑮𝑶𝑯𝒙 (𝑻, 𝒑) = ∆𝑮𝟎 (𝟓𝟎𝟎 𝑲) + 𝑹𝑻𝒍𝒏( 𝟐 ) (63)
𝒑𝑯𝟐 𝑶
The Gibbs free energies for gas phase and adsorbed species were obtained by combining the electronic and zero-
point vdW-DF2 energies with enthalpy and entropy corrections from frequency calculations for the full
structure [5].

Activation of the CO2* species on the surface is evaluated for direct dissociation. Similar to the analysis on
the cobalt surface, reaction barriers are determined via the cNEB and refined with the dimer method, and
results are summarized in Table 4-8.

Table 4-8 Energy barriers (Ef), entropy of activation (∆Sf), Gibbs free energy barriers (∆Gf), entropy of reaction
(∆Sr) and Gibbs free energy of reaction (∆Gr) for CO* and CO2* scission reactions on a clean Cu(111) surface.
The accompanying transition state structures are given in Table B-12 in Appendix B.

Surface reactions Energy Ef Entropy ∆Sf Gibbs Free Entropy ∆Sr Gibbs Free
(kJ/mol) (J/mol K) Activation (J/mol K) Reaction
Energy ∆Gf Energy ∆Gr
(kJ/mol) (kJ/mol)

CO* + *  C* + O* 546 10 534 -28 283

CO2* + *  CO* + O* 159 -16 153 -20 100

An activation energy of 159 kJ/mol is calculated for the dissociation reaction of CO2, close to the value of
200 kJ/mol reported in literature [18], and, as illustrated in Figure 4-11, the reaction is 100 kJ/mol
Ab-initio calculations 99

endothermic. The activation barrier is significantly higher (100 kJ/mol) than for the dissociation reaction on
Co(111) and the Gibbs free reaction energy of 100 kJ/mol for dissociation on Cu(111) shows that it is
thermodynamically less favorable than for the Co(111) surface.

Figure 4-11 Energy profile for the dissociation of carbon dioxide on a clean Cu(111) surface (CO2* + *  CO* +
O*). An energy barrier of 159 kJ/mol and Gibbs free energy of reaction of 100 kJ/mol is obtained. Detailed
visualization of the transition state is given in Table B-12 in Appendix B.

The dissociation of CO on Cu(111) is illustrated in Figure 4-12, and an activation energy of 546 kJ/mol is
calculated. This is significantly higher than the dissociation barrier on the Co(111) surface. It can hence be
concluded that for the activation of CO and CO2 by direct dissociation, the cobalt catalyst will be preferred
over the cupper catalyst.

Figure 4-12 Energy profile for the dissociation of carbon monoxide on a clean Cu(111) surface (CO* + *  C* +
O*). An energy barrier of 546 kJ/mol and Gibbs free energy of reaction of 283 kJ/mol is obtained. Detailed
visualization of the transition state is given in Table B-12 in Appendix B.

In Figure 4-12 the Cu surface atoms is seen to shift upwards significantly for the adsorption of CO on a top
site, requiring an energy addition of 11 kJ/mol.
100 Ab-initio calculations

4.2.2. Coverage effects on Cu(111)

Similar to the analysis for the Co(111) surface, the influence of increased CO coverage on the Cu(111) is
examined. Table 4-9 summarizes the DFT calculated adsorption energies, entropies of adsorbed species and
Gibbs free energy of adsorption for CO* and CO2* under FT conditions; Figure 4-13 illustrates the DFT
calculated structures.

Table 4-9 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch conditions
(500 K, 20 bar, 60% conversion) for CO* and CO2* adsorbates on a Cu(111) surface with CO coverage of 1/3
ML. A more elaborated summary is given in Table B-13 in Appendix B.

Reaction Adsorption energy Entropy of adsorbed species Gibbs free adsorption energy
intermediate (kJ/mol) (J/mol K) (kJ/mol)

CO* -34 100 20

CO2* -28 99 51

(a) (b)

Figure 4-13 Optimized adsorption configuration for (a) CO* and (b) CO2* on a Cu(111) surface with CO
coverage of 1/3 ML.

Comparison with the clean Cu(111) surface, see Table 4-7, shows that the higher coverage has a stabilizing
effect, and the CO2 and CO adsorption become respectively 13 kJ/mol and 2 kJ/mol stronger. Also the
entropy increase contributes to the reduced thermodynamic instability, following from reduced Gibbs free
energies of adsorption. The explanation for this effect is assumed to be similar as for the Co(111) surface. In
comparison with the Co(111) surface, CO adsorption appears to be significantly less stable with a difference
of 88 kJ/mol. CO2 adsorbs stronger on the Cu(111) with a difference of 5 kJ/mol, but, calculation of the
Gibbs free energy indicates the adsorption reaction is thermodynamically less favorable.
Ab-initio calculations 101

Figure 4-14 Energy profile for the dissociation of carbon dioxide on a Cu(111) surface with CO coverage of 1/3
ML (CO2* + *  CO* + O*). An energy barrier of 156 kJ/mol and Gibbs free energy of reaction of 89 kJ/mol is
obtained. Detailed visualization of the transition state is given in Table B-15 in Appendix B.

The dissociation reaction of CO2, illustrated in Figure 4-14 and summarized in Table 4-10, is shown to have
a slightly lower activation barrier under conditions of 1/3 ML CO coverage. Hence, under FT conditions the
CO2 dissociation is slightly enhanced. However note that under real FT conditions the surface is likely to be
more complex and the presence of reaction intermediates might further affect the stability of the CO 2
adsorbate [11]. When compared to the dissociation reaction on Co(111), CO2 dissociation barriers are
significantly higher and dissociation is endothermic, whereas for Co(111) the reaction is exothermic. One
can conclude that for the activation of CO2 by direct dissociation, cobalt catalysts should be preferred over
copper catalysts.

Table 4-10 Energy barrier (Ef), entropy of activation (∆Sf), Gibbs free energy barrier (∆Gf), entropy of reaction
(∆Sr) and Gibbs free energy of reaction (∆G r) for the CO2* scission reaction on a Cu(111) surface with CO
coverage of 1/3 ML.

Surface reactions Energy Ef Entropy ∆Sf Gibbs Free Entropy ∆Sr Gibbs Free
(kJ/mol) (J/mol K) Activation (J/mol K) Reaction
Energy ∆Gf Energy ∆Gr
(kJ/mol) (kJ/mol)

CO2* + *  CO* + O* 156 -29 164 -1 89


102 Ab-initio calculations

4.3. Analysis of the Ni(111) surface


Nickel catalysts have reportedly shown good results for methanation reactions of CO 2 [19, 20]. Nickel
catalysts can have highly porous structures with high specific areas making them highly suitable as catalyst
for hydrogenation reactions at low temperature [21]. Therefore, CO2 activation is also evaluated on the
Ni(111) surface.

4.3.1. Adsorption of reaction intermediates on Ni(111)

Figure 4-15 shows the DFT calculated adsorption configurations for CO* and CO2* on the clean Ni(111)
surface. CO is calculated to preferentially adsorb on a hollow site position with adsorption energy of -122
kJ/mol and CO2 adsorbs preferentially to a bridge site position with an adsorption energy of -16 kJ/mol. The
DFT calculations are in agreement with values found in literature [17, 19].

(a) (b)

Figure 4-15 Optimized adsorption configuration for (a) CO* and (b) CO2* on a clean Ni(111) surface.

However, neither CO nor CO2 adsorption appears to be thermodynamically favorable. Gibbs free energies of
adsorption calculations show values of 70 kJ/mol for CO adsorption and 125 kJ/mol for CO 2 adsorption,
analogously to adsorption on the Cu(111) surface. A summary of DFT calculated adsorption energies,
entropies and Gibbs free energies of adsorption for the most stable configurations of intermediates on the
clean Ni(111) surface is given in Table 4-11. It can be remarked that solely hydrogen adsorption is
thermodynamically stable, indicated by the positive Gibbs free energies of adsorption for the remaining
adsorbates.
Ab-initio calculations 103

Table 4-11 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch conditions
(500 K, 20 bar, 60% conversion) for the most stable configuration of reaction intermediates on a clean Ni(111)
surface. A more extended summary is given in Table B-16 in Appendix B.

Reaction Adsorption energy Entropy of adsorbed species Gibbs free adsorption energy
intermediate (kJ/mol) (J/mol K) (kJ/mol)

C* -596 17 66

O* -514 23 14

CO* -122 -54 70

CO2* -16 50 125

H* -262 7 -5

OH* -352 39 28
The relative stability of the reaction intermediates is evaluated relative to a CO, CO 2, H2 and H2O gas phase
reservoir from the reaction Gibbs free energy equations under FT conditions (500 K, 20 bar, 60% conversion);
and includes the effects of pressure, composition, and temperature. The Gibbs free energy reactions:
𝑪𝑶(𝒈) + 𝑯𝟐(𝒈) ⇌ 𝑪∗ + 𝑯𝟐 𝑶(𝒈) (60)
𝒙
𝑯𝟐 𝑶(𝒈) + ∗ ⇌ 𝑶𝑯∗𝒙 + (𝟏 − )𝑯𝟐(𝒈) (61)
𝟐
𝒑𝑯𝟐 𝑶
∆𝑮𝑪(𝑻, 𝒑) = ∆𝑮𝟎 (𝟓𝟎𝟎 𝑲) + 𝑹𝑻𝒍𝒏( ) (62)
𝒑𝑪𝑶 𝒑𝑯𝟐
𝒑𝑯 𝟏−𝒙/𝟐
∆𝑮𝑶𝑯𝒙 (𝑻, 𝒑) = ∆𝑮𝟎 (𝟓𝟎𝟎 𝑲) + 𝑹𝑻𝒍𝒏( 𝟐 ) (63)
𝒑𝑯𝟐 𝑶
The Gibbs free energies for gas phase and adsorbed species were obtained by combining the electronic and zero-
point vdW-DF2 energies with enthalpy and entropy corrections from frequency calculations for the full
structure [5].

Activation of the CO2* species on the surface is evaluated for the direct dissociation. Similar to the analysis
on the cobalt surface, reaction barriers are determined via the cNEB and dimer method and results are
summarized in Table 4-12.

Table 4-12 Energy barriers (Ef), entropy of activation (∆Sf), Gibbs free energy barriers (∆Gf), entropy of reaction
(∆Sr) and Gibbs free energy of reaction (∆Gr) for CO* and CO2* scission reactions on a clean Ni(111) surface.
The accompanying transition state structures are given in Table B-18 in Appendix B.

Surface reactions Energy Ef Entropy ∆Sf Gibbs Free Entropy ∆Sr Gibbs Free
(kJ/mol) (J/mol K) Activation (J/mol K) Reaction
Energy ∆Gf Energy ∆Gr
(kJ/mol) (kJ/mol)

CO* + *  C* + O* 289 -21 295 -30 140

CO2* + *  CO* + O* 79 -39 73 -32 -43


104 Ab-initio calculations

An activation energy of 79 kJ/mol is calculated for the dissociation reaction of CO2, rather low compared to
the value of 121 kJ/mol reported by Choe et al. [19], and, as illustrated in Figure 4-16, the reaction is
calculated to be 43 kJ/mol exothermic. The DFT calculated activation barrier is higher than for Co(111) but
is found to be in the same order of magnitude. Comparison of the Gibbs free energies of reaction shows that
dissociation on Ni(111) is less favorable than on Co(111), with a difference of 29 kJ/mol.

Figure 4-16 Energy profile for the dissociation of carbon dioxide on a clean Ni(111) surface (CO2* + *  CO* +
O*). An energy barrier of 79 kJ/mol and Gibbs free energy of reaction of -43 kJ/mol is obtained. Detailed
visualization of the transition state is given in Table B-18 in Appendix B.

The dissociation of CO on Ni(111) is illustrated in Figure 4-17, and an activation energy of 288 kJ/mol is
calculated. This is comparable to the dissociation barrier on the Co(111) surface, but still slightly higher. It
can hence be concluded that for the activation of CO and CO 2 by direct dissociation, the nickel and cobalt
catalyst will have a similar operation, although cobalt catalysts are preferred.

Figure 4-17 Energy profile for the dissociation of carbon monoxide on a clean Ni(111) surface (CO* + *  C* +
O*). An energy barrier of 289 kJ/mol and Gibbs free energy of reaction of 140 kJ/mol is obtained. Detailed
visualization of the transition state is given in Table B-18 in Appendix B.
Ab-initio calculations 105

4.3.2. Coverage effects on Ni(111)

Similar to the analysis for the Co(111) and Cu(111) surface, the influence of increased CO coverage on the
Ni(111) surface is examined. Table 4-13 summarizes the DFT calculated adsorption energies, entropies of
adsorbed species and Gibbs free energy of adsorption for CO* and CO 2* under FT conditions; Figure 4-18
illustrates the DFT calculated configurations.

Table 4-13 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch conditions
(500 K, 20 bar, 60% conversion) for CO* and CO2* adsorbates on a Ni(111) surface with CO coverage of 1/3
ML. A more elaborated summary is given in Table B-19 in Appendix B.

Reaction Adsorption energy Entropy of adsorbed species Gibbs free adsorption energy
intermediate (kJ/mol) (J/mol K) (kJ/mol)

CO* -115 79 -51

CO2* -24 141 33

(a) (b)

Figure 4-18 Optimized adsorption configuration for (a) CO* and (b) CO2* on a Ni(111) surface with CO
coverage of 1/3 ML.

Comparison with the clean Ni(111) surface, see Table 4-11, shows that the higher coverage has a
destabilizing effect on the CO adsorption, with adsorption on the clean surface being 7 kJ/mol stronger; but
has a stabilizing effect for CO2 adsorption and adsorption for increased coverage is 8 kJ/mol stronger. The
destabilization of the CO adsorption can be interpreted by repulsive lateral interactions, causing the bond
between the CO adsorbate and the catalyst surface to be impaired. For the interpretation of the stabilizing
effect on CO2 adsorption the reader is referred to Section 4.1.3 [22]. In comparison with the Co(111) surface,
CO adsorption appears to be only slightly less favorable, with a difference of 7 kJ/mol. Also the CO 2
adsorption energies are similar, with adsorption on the Ni(111) surface 1 kJ/mol stronger. However,
calculation of the Gibbs free reaction energy indicates that the adsorption reactions on Co(111) are preferred
above adsorption on Ni(111).
106 Ab-initio calculations

Figure 4-19 Energy profile for the dissociation of carbon dioxide on a Ni(111) surface with CO coverage of 1/3
ML (CO2* + *  CO* + O*). An energy barrier of 95 kJ/mol and Gibbs free energy of reaction of -18 kJ/mol is
obtained. Detailed visualization of the transition state is given in Table B-21 in Appendix B.

The dissociation reaction of CO2, illustrated in Figure 4-19 and summarized in Table 4-14, is shown to have
an activation barrier 16 kJ/mol higher under conditions of 1/3 ML CO coverage. The lateral interactions
hence slightly disfavor the dissociation reaction on Ni(111). When compared to the dissociation reaction on
Co(111), the CO2 dissociation barrier is 14 kJ/mol higher. The reaction is 18 kJ/mol exothermic, and is less
favorable than the dissociation reaction on Co(111). One can conclude that for the activation of CO 2 by
direct dissociation, cobalt catalysts will be preferred over nickel catalysts.

Table 4-14 Energy barrier (Ef), entropy of activation (∆Sf), Gibbs free energy barrier (∆Gf), entropy of reaction
(∆Sr) and Gibbs free energy of reaction (∆Gr) for the CO2* scission reaction on a Ni(111) surface with CO
coverage of 1/3 ML.

Surface reactions Energy Ef Entropy ∆Sf Gibbs Free Entropy ∆Sr Gibbs Free
(kJ/mol) (J/mol K) Activation (J/mol K) Reaction
Energy ∆Gf Energy ∆Gr
(kJ/mol) (kJ/mol)

CO2* + *  CO* + O* 95 -47 89 -39 -18


Ab-initio calculations 107

4.4. Analysis of the Au(111) surface


Reports on gold catalysts demonstrate that it has unique properties as a catalyst. It has no stable oxide, has a
high electronegativity, and its chemical properties are in many ways comparable to the properties of
mercury. Although it is reported to have rather low adsorption qualities, gold is an excellent compound for
construction of complex chemical architectures and has the ability to easily activate C=C double bonds [23,
24]. Therefore also the activation of CO2 on the Au(111) surface is evaluated.

4.4.1. Adsorption of reaction intermediates on Au(111)

Figure 4-20 shows the DFT calculated adsorption configurations for CO* and CO2* on the clean Au(111)
surface. CO is calculated to preferentially adsorb on a top site position with adsorption energy of -12 kJ/mol,
and CO2 adsorbs preferentially to a bridge site position with an adsorption energy of -16 kJ/mol, somewhat
less stable than values found in literature [25].

(a) (b)

Figure 4-20 Optimized adsorption configuration for (a) CO and (b) CO2 on a clean Au(111) surface.

However, neither CO nor CO2 adsorption appears to be thermodynamically favorable. Gibbs free energy of
adsorption calculations show values of 33 kJ/mol for CO adsorption and 43 kJ/mol for CO 2 adsorption. A
summary of DFT calculated adsorption energies, entropies and Gibbs free energies of adsorption for the
most stable configurations of intermediates on the clean Au(111) surface is given in Table 4-15. The positive
values obtained for the Gibbs free energies of adsorption indicate that the intermediates are
thermodynamically unstable.
108 Ab-initio calculations

Table 4-15 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch conditions
(500 K, 20 bar, 60% conversion) for the most stable configuration of reaction intermediates on a clean Au(111)
surface. A more extended summary is given in Table B-22 in Appendix B.

Reaction Adsorption energy Entropy of adsorbed species Gibbs free adsorption energy
intermediate (kJ/mol) (J/mol K) (kJ/mol)

C* -417 243 21

O* -321 29 209

CO* -12 119 33

CO2* -16 137 43

H* -204 10 49

OH* -754 45 134


The relative stability of the reaction intermediates is evaluated relative to a CO, CO 2, H2 and H2O gas phase
reservoir from the reaction Gibbs free energy equations under FT conditions (500 K, 20 bar, 60% conversion);
and includes the effects of pressure, composition, and temperature. The Gibbs free energy reactions:
𝑪𝑶(𝒈) + 𝑯𝟐(𝒈) ⇌ 𝑪∗ + 𝑯𝟐 𝑶(𝒈) (60)
𝒙
𝑯𝟐 𝑶(𝒈) + ∗ ⇌ 𝑶𝑯∗𝒙 + (𝟏 − )𝑯𝟐(𝒈) (61)
𝟐
𝒑𝑯𝟐 𝑶
∆𝑮𝑪(𝑻, 𝒑) = ∆𝑮𝟎 (𝟓𝟎𝟎 𝑲) + 𝑹𝑻𝒍𝒏( ) (62)
𝒑𝑪𝑶 𝒑𝑯𝟐
𝒑𝑯 𝟏−𝒙/𝟐
∆𝑮𝑶𝑯𝒙 (𝑻, 𝒑) = ∆𝑮𝟎 (𝟓𝟎𝟎 𝑲) + 𝑹𝑻𝒍𝒏( 𝟐 ) (63)
𝒑𝑯𝟐 𝑶
The Gibbs free energies for gas phase and adsorbed species were obtained by combining the electronic and zero-
point vdW-DF2 energies with enthalpy and entropy corrections from frequency calculations for the full
structure [5].

Activation of the CO2* species on the surface is evaluated for the direct dissociation reaction. Similar to the
analysis on the cobalt surface, reaction barriers are determined via the cNEB and refined with the dimer
method, and results are summarized in Table 4-16.

Table 4-16 Energy barriers (Ef), entropy of activation (∆Sf), Gibbs free energy barriers (∆Gf), entropy of reaction
(∆Sr) and Gibbs free energy of reaction (∆Gr) for CO* and CO2* scission reactions on a clean Au(111) surface.
The accompanying transition state structures are given in Table B-24 in Appendix B.

Surface reactions Energy Ef Entropy ∆Sf Gibbs Free Entropy ∆Sr Gibbs Free
(kJ/mol) (J/mol K) Activation (J/mol K) Reaction
Energy ∆Gf Energy ∆Gr
(kJ/mol) (kJ/mol)

CO* + *  C* + O* 447 -67 475 -70 414

CO2* + *  CO* + O* 310 -1 304 9 240


Ab-initio calculations 109

An activation energy of 310 kJ/mol is calculated for the dissociation reaction of CO2 and, as illustrated in
Figure 4-21, the reaction is calculated to be 240 kJ/mol endothermic. The calculated activation barrier is
significantly higher than for Co(111) and CO 2 dissociation will probably be unlikely on the gold catalyst
surface. Comparison of the Gibbs free energies of reaction shows that dissociation on Au(111) is highly
unfavorable compared to dissociation on Co(111).

Figure 4-21 Energy profile for the dissociation of carbon dioxide on a clean Au(111) surface (CO 2* + *  CO* +
O*). An energy barrier of 310 kJ/mol and Gibbs free energy of reaction of 240 kJ/mol is obtained. Detailed
visualization of the transition state is given in Table B-24 in Appendix B.

The dissociation of CO on Au(111) is illustrated in Figure 4-22, and an activation energy of 447 kJ/mol is
calculated. This is significantly higher than the dissociation barrier on the Co(111) surface. The gold catalyst
surface appears to be incapable of efficiently catalyzing the dissociation reactions of CO and CO2 and cobalt
catalysts should be preferred over gold catalysts.

Figure 4-22 Energy profile for the dissociation of carbon monoxide on a clean Au(111) surface (CO* + *  C* +
O*). An energy barrier of 447 kJ/mol and Gibbs free energy of reaction of 414 kJ/mol is obtained. Detailed
visualization of the transition state is given in Table B-24 in Appendix B.
110 Ab-initio calculations

4.4.2. Coverage effects on Au(111)

Similar to the analysis for the previous catalyst surfaces, the influence of increased CO coverage on the
Au(111) surface is examined. Table 4-17 summarizes the DFT calculated adsorption energies, entropies of
adsorbed species and Gibbs free energy of adsorption for CO* and CO 2* under FT conditions; Figure 4-23
illustrates the DFT calculated configurations.

Table 4-17 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch conditions
(500 K, 20 bar, 60% conversion) for CO* and CO2* adsorbates on a Au(111) surface with CO coverage of 1/3
ML. A more elaborated summary is given in Table B-25 in Appendix B.

Reaction Adsorption energy Entropy of adsorbed species Gibbs free adsorption energy
intermediate (kJ/mol) (J/mol K) (kJ/mol)

CO* -15 117 31

CO2* -27 120 41

(a) (b)

Figure 4-23 Optimized adsorption configuration for (a) CO* and (b) CO2* on a Au(111) surface with CO
coverage of 1/3 ML.

Comparison with the clean Au(111) surface, see Table 4-15, shows that the higher coverage has a stabilizing
effect on the adsorption of CO and CO2, with the bonds being respectively 3 kJ/mol and 11 kJ/mol stronger.
For the interpretation of the increase in adsorption strength, the reader is referred to Section 4.1.3 [22]. In
comparison with the Co(111) surface, CO adsorption is significantly less favorable, with a difference of 107
kJ/mol. The CO2 adsorption energies are similar, with adsorption on the Au(111) surface being 4 kJ/mol
stronger. However, calculation of the Gibbs free energy indicates that the adsorption reactions on Co(111)
are preferred above adsorption on Au(111).
Ab-initio calculations 111

Figure 4-24 Energy profile for the dissociation of carbon dioxide on a Au(111) surface with CO coverage of 1/3
ML (CO2* + *  CO* + O*). An energy barrier of 314 kJ/mol and Gibbs free energy of reaction of 234 kJ/mol is
obtained. Detailed visualization of the transition state is given in Appendix B.

The dissociation reaction of CO2, illustrated in Figure 4-24 and summarized in Table 4-18, is shown to have
an activation barrier 4 kJ/mol higher under conditions of 1/3 ML CO coverage. The lateral interactions hence
slightly disfavor the dissociation reaction on Au(111). When compared to the dissociation reaction on
Co(111), the CO2 dissociation barrier is significantly higher. The reaction is 234 kJ/mol endothermic, and
hence the Au(111) surface is not capable of efficiently catalyzing the dissociation of CO2. One can conclude
that for the activation of CO2 by direct dissociation, cobalt catalysts will be preferred over gold catalysts.

Table 4-18 Energy barrier (Ef), entropy of activation (∆Sf), Gibbs free energy barrier (∆Gf), entropy of reaction
(∆Sr) and Gibbs free energy of reaction (∆G r) for the CO2* scission reaction on a Au(111) surface with CO
coverage of 1/3 ML.

Surface reactions Energy Ef Entropy ∆Sf Gibbs Free Entropy ∆Sr Gibbs Free
(kJ/mol) (J/mol K) Activation (J/mol K) Reaction
Energy ∆Gf Energy ∆Gr
(kJ/mol) (kJ/mol)

CO2* + *  CO* + O* 314 -14 308 26 234


112 Ab-initio calculations

4.5. Conclusion
In this chapter a first evaluation has been done for the activation of CO 2 on the Co(111) surface via DFT
calculations. CO2 was found to adsorb preferentially on a top site position with adsorption energy of -16
kJ/mol, but DFT calculations show that the Gibbs free energy of adsorption is 65 kJ/mol at 500 K and 1 bar,
and adsorption is thermodynamically unstable. The dissociation reaction on Co(111) is shown to be feasible,
with a relatively low activation barrier of 60 kJ/mol and a Gibbs free energy of reaction of -72 kJ/mol. Next,
the adsorption of hydrogenating species and several reaction intermediates was evaluated. Hydrogen
adsorption appears to be thermodynamically stable, and a high adsorption energy of -259 kJ/mol was
obtained. Hydrogen is hence expected to be the main species involved in hydrogenation reactions. Activation
of CO2 by a hydrogen-assisted pathway was evaluated and the formate pathway is shown to be preferential,
due to its lower activation barriers and lowest endothermicity. To obtain a more realistic catalyst model, the
coverage effect on CO2 activation was evaluated on Co(111). DFT calculations show that in situations of
high coverage, CO2 has a higher adsorption energy and dissociation barriers increase. However note that
under real FT conditions the surface is likely to be more complex and the presence of reaction intermediates
might further affect the stability of the CO2 adsorbate. Because significantly lower activation barriers have
been reported for CO dissociation reactions on step sites, CO2 activation was also evaluated on the step sites
of the Co(211) surface. Adsorption is calculated to be thermodynamically stable on top sites, with an
adsorption energy of -99 kJ/mol and Gibbs free energy of adsorption of -27 kJ/mol. The dissociation reaction
has an energy barrier of 54 kJ/mol, close to the DFT calculated value for terraces, and Gibbs free energy of
reaction of -121 kJ/mol. Hence, dissociation of CO2 on step sites is found to be thermodynamically favorable
and step sites appear to have high potential for the conversion of CO 2. However, step sites are expected to be
unavailable for CO2 molecules, due to their high affinity for reaction intermediates and low thermodynamic
stability under FTS conditions.

For reasons of comparison, evaluation was done for the Cu(111), Ni(111) and Au(111) surface with
inclusion of coverage effects. Adsorption energies for CO2 on Cu(111) are close to the values found for
Co(111), but adsorption is similarly thermodynamically unstable. DFT calculations also show higher energy
barriers for dissociation on clean Cu(111) as well as for conditions of increased coverage. The nickel surface
appears to have a similar capability of CO2 activation as Co(111), although the CO2 adsorbates are
thermodynamically less stable. An activation energy of 79 kJ/mol was calculated, which is closest to the
Co(111) values of the alternative surface examined. However, both for a clean Ni(111) surface and in case of
increased coverage the dissociation barriers are higher on Ni(111). The gold catalyst surface is shown to be
incapable of efficiently catalyzing the CO2 activation reactions. DFT calculations show high dissociation
barriers for the clean surface and in case of increased coverage. Overall, the cobalt catalyst appears to be the
preferred surface over the Cu(111), Ni(111), and Au(111) surfaces.
Ab-initio calculations 113

4.6. References

1. Agarwal, A.S., et al., The electrochemical reduction of carbon dioxide to formate/formic acid:
engineering and economic feasibility. ChemSusChem, 2011. 4(9): p. 1301-1310.

2. Norskov, J.K., et al., Towards the computational design of solid catalysts. Nat Chem, 2009. 1(1): p.
37-46.

3. Bezemer, G.L., et al., Cobalt particle size effects in the Fischer-Tropsch reaction studied with
carbon nanofiber supported catalysts. Journal of the American Chemical Society, 2006. 128(12): p.
3956-3964.

4. Den Breejen, J., et al., On the origin of the cobalt particle size effects in Fischer− Tropsch catalysis.
Journal of the American Chemical Society, 2009. 131(20): p. 7197-7203.

5. Gunasooriya, G.K.K., et al., Key role of surface hydroxyl groups in CO activation during Fischer-
Tropsch synthesis. ACS Catalysis, 2016.

6. Storsæter, S., D. Chen, and A. Holmen, Microkinetic modelling of the formation of C 1 and C 2
products in the Fischer–Tropsch synthesis over cobalt catalysts. Surface Science, 2006. 600(10): p.
2051-2063.

7. Papp, H., The chemisorption of carbon monoxide on a Co (0001) single crystal surface; studied by
LEED, UPS, EELS, AES and work function measurements. Surface Science, 1983. 129(1): p. 205-
218.

8. Balakrishnan, N., Theoretical Studies of Co Based Catalysts on CO Hydrogenation and Oxidation.


2013.

9. Iglesia, E., Design, synthesis, and use of cobalt-based Fischer-Tropsch synthesis catalysts. Applied
Catalysis A: General, 1997. 161(1–2): p. 59-78.

10. Ojeda, M., et al., CO activation pathways and the mechanism of Fischer–Tropsch synthesis. Journal
of Catalysis, 2010. 272(2): p. 287-297.

11. Zhuo, M., A. Borgna, and M. Saeys, Effect of the CO coverage on the Fischer–Tropsch synthesis
mechanism on cobalt catalysts. Journal of Catalysis, 2013. 297: p. 217-226.

12. Wambach, J., G. Illing, and H.J. Freund, CO2 activation and reaction with hydrogen on Ni(110):
formate formation. Chemical Physics Letters, 1991. 184(1): p. 239-244.

13. Zhao, Y.-F., et al., Insight into methanol synthesis from CO 2 hydrogenation on Cu (111): complex
reaction network and the effects of H 2 O. Journal of Catalysis, 2011. 281(2): p. 199-211.

14. Gunasooriya, G.K.K., et al., CO adsorption on cobalt: Prediction of stable surface phases. Surface
Science, 2015. 642: p. L6-L10.

15. Shetty, S. and R.A. van Santen, CO dissociation on Ru and Co surfaces: the initial step in the
Fischer–Tropsch synthesis. Catalysis today, 2011. 171(1): p. 168-173.

16. Ge, Q.F. and M. Neurock, Adsorption and activation of CO over flat and stepped Co surfaces: A
first principles analysis. Journal of Physical Chemistry B, 2006. 110(31): p. 15368-15380.
114 Ab-initio calculations

17. Toyoshima, I. and G. Somorjai, Heats of chemisorption of O2, H2, CO, CO2, and N2 on
polycrystalline and single crystal transition metal surfaces. Catalysis Reviews Science and
Engineering, 1979. 19(1): p. 105-159.

18. Fergusson, A.I., Identifying CO₂ dissociation pathways on stepped and kinked copper surfaces using
first principles calculations. 2012.

19. Choe, S.-J., et al., Adsorbed carbon formation and carbon hydrogenation for CO 2 methanation on
the Ni (111) surface: ASED-MO study. Bulletin of the Korean Chemical Society, 2005. 26(11): p.
1682-1688.

20. Choe, S.J., et al., Adsorption and dissociation reaction of carbon dioxide on Ni(1 1 1) surface:
molecular orbital study. Applied Surface Science, 2001. 181(3–4): p. 265-276.

21. Pavlic, A.A. and H. Adkins, Preparation of a Raney Nickel Catalyst. Journal of the American
Chemical Society, 1946. 68(8): p. 1471-1471.

22. GUNASOORIYA, G.T.K.K., Combined Theoretical and Experimental Study of CO Adsorption and
Reactivity Over Platinum and Cobalt. 2014.

23. Bond, G.C., Gold: a relatively new catalyst. Catalysis Today, 2002. 72(1–2): p. 5-9.

24. Sigma-Aldrich. Gold Catalyst. Technology Spotlights 2016 [cited 2016 16 May]; Available from:
http://www.sigmaaldrich.com/chemistry/chemical-synthesis/technology-spotlights/gold-rush.html.

25. Piccolo, L., et al., The adsorption of CO on Au(1 1 1) at elevated pressures studied by STM, RAIRS
and DFT calculations. Surface Science, 2004. 566–568, Part 2: p. 995-1000.
115

Chapter 5
Microkinetic modeling
Microkinetic modeling is defined by M. Neurock [1] as “an examination of catalytic reactions in terms of
elementary chemical reactions that occur on the catalytic surface and their relation with each other and with
the surface during a catalytic cycle”. A reaction mechanism is drafted, by inclusion of all elementary
reactions at microscopic level without assuming a rate-determining step, and a reactor model is selected.
Specialized software, such as Chemkin, Acuchem or Chemical Workbench, is then capable of
mathematically describing the reaction mechanism by integration of the set of reaction and reactor equations
for the surface coverages and the gaseous composition [2-4]. Microkinetic modeling allows better insight in
the dominance of different reaction pathways and can be used to acquire information on catalyst properties
or indications on the influence of the catalyst structure on product selectivities or yield. Hence, it allows for
faster catalyst development [5, 6]. In this Master thesis microkinetic modeling is performed for a plug flow
reactor model in steady state, under typical Fischer-Tropsch conditions (500 K, 20 bar) using the Chemkin
software. An initial microkinetic model is composed of CO activation and chain propagation reactions in
Fischer-Tropsch synthesis up to C2 product fractions. Preliminary analysis of CO2 activation is done by
incorporation of CO2 activation steps in the reaction mechanism for FTS.
116 Microkinetic modeling

5.1. Fischer-Tropsch synthesis


The Fischer-Tropsch synthesis (FTS) is a surface polymerization reaction to convert syngas (CO + H 2) to
long-chain alkanes and mono-alkenes, and small amounts of oxygenates [7]. The process was developed by
Fischer and Tropsch in 1923 and ever since debate is ongoing on the actual reaction mechanism of initiation
and chain growth. Fischer and Tropsch initially proposed a carbide mechanism [8]. This mechanism suggests
that C-O activation occurs via direct CO dissociation to form carbide species and surface oxygen.
Hydrogenation of the carbide species then yields CHx groups, which serve as monomers in the chain growing
steps. A schematic representation of the carbide mechanism is shown in Figure 5-1.

Figure 5-1 Reaction scheme of Fischer-Tropsch synthesis according to the carbide mechanism [9]. CO is
adsorbed and dissociates to form carbide species on the surface. Hydrogenation yields methane and CH x*,
proposedly the monomer species for chain growth.

Experimental studies, by Brady and Pettit [10], have demonstrated the coupling of methylene groups on FTS
catalysts, and theoretical studies indicated that the most likely chain growth reactions are the coupling of
RCH + CH2 and RCH2 + C. However, in order to ensure fast chain growth high surface coverages of CHx
should be present, which requires a sufficiently fast CO dissociation. Calculated C-O dissociation barriers on
terraces however are found to be approximately 200 kJ/mol, suggesting a low CHx coverage [7]. In the
carbide mechanism this is explained by the necessity for a higher operating temperature in FTS than
necessary for the chain-growth reaction; the high temperature favors the dissociation of CO and prevents CO
from blocking the surface [9].

An alternative chain-growth pathway through CO insertion has been originally proposed by Pichler and
Schulz [11]. Chain-growth is suggested to occur in two steps. First a CO molecule is coupled to a RCH 2
group and afterwards dissociation of the C-O bond occurs, possibly after hydrogenation. Initiation of the
chain-growth is suggested to go via a hydrogen assisted pathway, providing an alternative route to surface
methylene groups [7, 9]. Hence, this mechanism does not require a high surface coverage of CH 2 groups for
fast chain growth relative to chain termination. High pressures needed for FTS and the consequent high CO
coverage favors the CO insertion mechanism. The reaction mechanism for chain-growth proposed by Saeys
et al. [12] is illustrated in Figure 5-2.
Microkinetic modeling 117

Figure 5-2 Proposed propagation reaction paths for the CO insertion mechanism by Saeys et al. [12]. The full
arrows indicate the dominant reaction path and the dotted arrows the minor reaction paths. R represents
hydrogen or an alkyl group.

Recently, an alternative pathway for CO activation has been proposed by Saeys et al. [13], suggesting an
OH* assisted pathway via hydrogenation by a surface hydroxyl group. DFT calculations show a favorable
free-energy barrier of 94 kJ/mol for this pathway under FTS conditions, due to its very small entropy
penalty. COH* species are reported to further hydrogenate to form hydroxyl methylene (HCOH*), which
undergoes C-O scission with a low barrier. Advantages for this pathway are its low barriers and avoidance of
formation of formaldehyde.

The FT product mixture contains numerous organic compounds, such as paraffins, olefins and oxygenates,
and determination of selectivities is an important issue in modeling of FTS. As FTS is often used as an
alternative pathway for production of liquid fuel, the selectivity for C 5+ hydrocarbons is of high interest. A
good parameter to evaluate the selectivity of FTS is the chain growth probability 𝛼, the probability that a
molecule will continue reacting to form a longer chain, defined in Eq. (64).
𝑟𝑝,𝑛
𝛼= (64)
𝑟𝑝,𝑛 + 𝑟𝑡,𝑛

With 𝑛 denoting the chain length, 𝑟𝑝,𝑛 the propagation rate, and 𝑟𝑡,𝑛 the species termination rate. Species are
formed via stepwise growth and the product distribution can be described by an Anderson-Schulz-Flory
distribution, expressed in Eq. (65).

𝑊𝑛
= (1 − 𝛼)2 𝛼 𝑛−1 (65)
𝑛

With 𝑊𝑛 the weight fraction of hydrocarbons with chain length 𝑛. Experimental studies demonstrated the
dependence of the chain growth probability on chain length [14]. However, at high chain lengths (typically
>3) the 𝛼 value is reportedly independent to chain length [15].
118 Microkinetic modeling

5.1.1. CO activation

A first evaluation of CO conversion under FTS is performed by examining the activation of CO on the
catalyst surface. This is done by taking in account the direct dissociation of CO, the hydrogen-assisted
pathway, and the hydroxyl-assisted pathway; Table 5-1 summarizes the reactions included for the different
initiation mechanisms. Rate coefficients for adsorption reactions are calculated from the collision theory and
surface reaction rate coefficients from the transition state theory. The rate coefficient are coverage
independent. However, as initially the simulations yielded very high coverages for CO* of approximately
0.9, the CO adsorption coefficient is modified to obtain realistic coverages for FTS, in this case a coverage of
0.6 [16]. It is noticed that the hydrogen adsorption is highly dependent on the CO* coverage. For the initial
case of high CO* coverage (0.9) a hydrogen coverage of approximately 0.1 was obtained. After modification
of the adsorption coefficient for CO, hydrogen is observed to have a coverage of approximately 0.4. The
model is analyzed for its capability of CH* formation, and by implementation of low hydrogenation barriers
the CH* species are removed from the surface in the form of methane (CH4). Blocking of the surface by
oxygen is avoided by O* removal from the surface through reaction with H* and CO* to form water and
carbon dioxide. In case of the hydrogen- and hydroxyl-assisted pathway also formaldehyde is considered as a
possible product. The reaction and inlet conditions are described in section 3.8, and the calculations for a
reference case consider a temperature of 500 K and a pure feed stream of CO (0.33) and H2 (0.66),
representative for FT conditions [16, 17]. In the carbide mechanism, CO activation proposedly occurs via
direct CO dissociation. However, DFT calculations show that the activation barrier for CO dissociation on
Co(111) surfaces are in the order of 200 kJ/mol for terraces and CO dissociation might not be sufficiently
fast to obtain high propagation rates for FTS [7]. In this study a CO dissociation barrier of 230 kJ/mol is
calculated and expectations are that due to this high barrier, the direct dissociation path will only have a
limited contribution to the activation of CO. For the hydrogen assisted pathway, hydrogenation to HCO* and
COH* species, and subsequently to HCOH*, are first taken in account. HCOH* is then expected to
dissociate and yield CH* on the surface, although also HCO* cleavage has a reasonable dissociation barrier.
DFT calculations have shown that the hydrogen-assisted pathway has a significantly lower activation barrier
than the direct pathway, and from the respective activation energies for hydrogenation of 139 kJ/mol and 188
kJ/mol, the hydrogen-assisted pathway is expected to go through HCO* formation rather than COH*; a
pathway that is also considered in literature [9, 12, 18]. Formation of formaldehyde is found to have a
relatively low barrier and might be an important reaction to take in account. Coverage effects however need
to be taken in account to properly evaluate the reaction mechanism. The OH-assisted pathway also suggests
that CO activation first goes through hydrogenation to form HCO* and COH*, with a preference for COH*
[13], and subsequent hydrogenation to HCOH*, which dissociates to obtain CH* on the surface. However,
questions rise on the presence of OH* groups on the surface.
Microkinetic modeling 119

Table 5-1 Summary of the reactions taken in account for the activation of CO* on the catalyst surface.
Distinction is made between the direct CO dissociation, the H-assisted and OH-assisted CO activation pathway.
Values are provided by the coach of this thesis.
Forward Reverse
Forward pre- Reverse pre-
activation activation
Reactions exponential exponential
energy Eaf energy Ear
factor Af (s-1) factor Ar (s-1)
(kJ/mol) (kJ/mol)
Adsorption
CO(g) + * <=> CO* 5.00E+04 0.00 9.22E+16 130.00
H2(g) + 2* <=> 2 H* 5.72E+09 0.00 5.90E+12 55.00
H2(g) + O* + * <=> OH* + H* 2.91E+14 76.00 3.73E+08 82.00
Direct CO dissociation
CO* + * <=> C* + O* 9.97E+11 230.00 1.09E+14 155.00
Alfa hydride addition
C* + H* <=> CH* + * 8.11E+15 20.00 1.20E+14 2.00
CH* + H* <=> CH2* + * 5.44E+12 20.00 2.40E+14 2.00
CH2* + H* <=> CH3* + * 4.24E+12 20.00 3.60E+14 2.00
O removal – water formation
O* + H* <=> OH* + * 1.08E+13 113.00 1.01E+13 91.00
OH* + H* <=> H2O(g) + 2* 1.16E+14 139.00 9.32E+11 61.00
2 OH* <=> H2O(g) + O* + * 8.91E+12 65.00 1.22E+13 9.00
O removal – CO2 formation
CO* + O* <=> CO2(g) + 2* 9.22E+14 138.00 1.20E+14 81.00
H-assisted CO activation
CO* + H* <=> HCO* + * 6.01E+12 139.00 1.81E+13 13.00
CO* + H* <=> COH* + * 2.76E+12 188.00 3.94E+13 96.00
HCO* + H* <=> CH2O* + * 8.02E+12 42.00 1.35E+13 5.00
HCO* + H* <=> HCOH* + * 1.55E+13 90.00 7.01E+12 86.00
COH* + H* <=> HCOH* + * 1.85E+12 83.00 5.87E+13 43.00
OH-assisted CO activation
CO* + OH* <=> HCO* + O* 1.54E+12 159.00 7.05E+13 54.00
CO* + OH* <=> COH* + O* 1.84E+11 85.00 5.90E+14 15.00
HCO* + OH* <=> CH2O* + O* 1.37E+12 108.00 7.94E+13 58.00
HCO* + OH* <=> HCOH* + O* 4.72E+11 14.00 2.30E+14 31.00
COH* + OH* <=> HCOH* + O* 9.46E+11 140.00 1.15E+14 122.00
C-O cleavage
HCO* + * <=> CH* + O* 1.00E+12 55.00 1.08E+14 144.00
COH* + * <=> C* + OH* 2.07E+11 140.00 5.23E+14 143.00
HCOH* + * <=> CH* + OH* 3.48E+13 37.00 3.12E+12 109.00
CH2O* + * <=> CH2* + O* 2.82E+12 37.00 3.84E+13 113.00
Methane desorption
CH3* + H* <=> CH4(g) + 2* 9.13E+11 20.00 1.55E+09 2.00
Formaldehyde desorption
CH2O* <=> CH2O(g) + * 9.22E+16 36.00 3.14E+08 0.00
120 Microkinetic modeling

The pathways for CO* activation are represented in Figure 5-3, and the OH-assisted activation is found to be
the dominant pathway for an inlet of pure hydrogen and carbon monoxide. The direct dissociation has, as
expected, a very low contribution to the activation of CO* due to its high activation barrier. The hydrogen-
assisted pathway has a reasonable contribution, although a factor 2 smaller than the OH-assisted pathway.
CO* hydrogenation with OH* is dominant, as it has the lowest activation barrier (85 kJ/mol). Further
hydrogenation to HCOH* occurs via H* as this reaction has a lower activation energy (90 kJ/mol) than the
hydrogenation with OH* (140 kJ/mol). The activation of CO* will hence be driven by a combination of the
H- and OH- assisted pathway.

COH*

CO(g) CO* HCOH*

HCO* CH*

CH2O* CH2*

Figure 5-3 Reaction paths for the activation of CO* on the cobalt catalytic surface; OH-assisted activation is the
dominant pathway and CH* formation preferably goes through COH* and HCOH* formation. The dominant
pathway is indicated by the bold lines.

The surface coverages of the different species at 500 K are represented in Table 5-2. The surface is
dominated by carbon monoxide and hydrogen, with respective coverages of 0.39 and 0.59. Furthermore,
simulation at 500 K shows a reasonable coverage of OH*, explaining the activation via the OH*-assisted
pathway, but rather low coverages for the other reaction intermediates.

Table 5-2 Surface coverages simulated for the surface species at 500 K and 20 bar.

Surface species Surface coverage Surface species Surface coverage

C* 5.58E-07 CH2* 1.24E-06


H* 3.99E-01 CH3* 1.75E-08
O* 2.32E-07 HCO* 1.02E-12
CO* 5.95E-01 COH* 7.97E-08
OH* 1.65E-03 HCOH* 4.73E-12
CH* 4.54E-05 CH2O* 8.51E-15
Microkinetic modeling 121

5.1.1.1. Effect of H2O partial pressure on the activation mechanism


The availability of OH*-species is a key element in the evaluation of the different activation mechanisms.
Addition of water in the feed stream has been proposed as a possible means to increase the OH* coverage
[13, 17], and therefore water addition to the feed is examined for its effect on the CO activation. Figure 5-4
illustrates the surface coverages; a sharp increase for the OH* and O* coverage is observed at low partial
pressures of water, and for high partial pressures the OH* species tend to dominate the catalyst surface.

H2O Partial pressure (-)


0 0.02 0.04 0.06 0.08 0.1
1.E+00 0.8

Coverage (-)
1.E-02 0.6
log(Coverage) (-)

1.E-04
0.4
1.E-06
0.2
1.E-08
0.0
1.E-10
0 0.05 0.1
1.E-12
pH2O (-)
1.E-14
Free sites H* coverage CO* coverage O* coverage CH* coverage
OH* coverage HCO* coverage COH* coverage HCOH* coverage

Figure 5-4 Surface coverages of the reaction intermediates in function of the partial pressure of water at 500 K
and 20 bar, on a logarithmic scale and detailed linear scale for C*, H*, OH*, and O*. CO*, H* and OH* are the
main surface species. A sharp increase for the OH* and O* coverage is observed already at low water partial
pressures.

Simulations show that water addition has a beneficial effect on the CO consumption rate, as is shown in
Figure 5-5 (a). The turnover frequency is found to sharply increase at low partial pressures up to 0.01, but
encounters stagnation around 0.01 to 0.02 and tends to decrease at higher H 2O partial pressures. This is a
trend that is consistent with the work from Iglesia et al. [17], for the analysis of FTS on ruthenium catalysts.
Figure 5-5 (b) shows their results for the effect of water addition on the CO consumption turnover rate (h -1)
and a similar trend is observed in the CO activation model.

(a) (a) (b)


2.0E-03

1.5E-03
TOFCO (s-1)

1.0E-03

5.0E-04

0.0E+00
0 0.02
pH2O (-)

Figure 5-5 (a) Simulated CO consumption rate (s-1) as a function of the partial pressure of H 2O at 500 K and 20
bar. Increased partial pressure initially raises the TOF, but stagnates around 0.01-0.02; maximum TOF is
observed at a partial pressure of approximately 0.01. (b) CO consumption turnover rate (h -1) in function of H2O
pressure for FTS on a Ru catalyst at 463 K and 2.9 MPa [17].
122 Microkinetic modeling

Reaction path analysis shows that the formation of HCO* occurs mainly through hydrogenation with H* and
COH* formation through hydrogenation with OH*. Figure 5-6 shows the dependency of the rate of
production for the reactions going through COH* and HCO* on the H2O partial pressure. CO activation is
observed to preferably go through HCO* formation, rather than COH* formation. Oxygen on the surface
will block the formation of COH* formation via the OH-assisted path due to its low reverse barrier, shown in
Table 5-1. The HCO* formed is observed to quickly hydrogenate with OH* to HCOH*, due to its low
activation barrier and the high OH* coverage. Hence, simulations show that the dominant pathway for CO
activation is through H-assisted hydrogenation of CO* to HCO*, OH-assisted hydrogenation to HCOH* and
subsequent dissociation; similar to what is proposed by the Saeys group [13]. The decreased rate of
production for higher H 2O partial pressures is explained from the reduced CO* and H* coverage, and
poisoning of the surface by OH* species, visualized in Figure 5-4.

3E-12
Rate of production (mol/cm²s)

2.5E-12
2E-12
1.5E-12
1E-12
5E-13
0
-5E-13
-1E-12
-1.5E-12
0 0.02 0.04 0.06 0.08
H2O Partial pressure (-)
Figure 5-6 H2O partial pressure dependency for the rate of production (mol/cm²s) of the COH* species (blue
line), HCO* species (purple line), COH* hydrogenation to HCOH* (red dashed line), and HCO* hydrogenation
to HCOH* (yellow dashed line). The pathway through HCO* is observed to be dominant, as the COH*
formation is blocked due to the high O* coverage.

5.1.1.2. Temperature effect on the activation mechanism


To evaluate the effect of temperature on the activation mechanism, simulations are performed over a range of
200 K (taking into account the partition functions at 500 K). Figure 5-7 illustrates the surface species
coverages in function of temperature. CO* and H* species are observed to dominate the surface over the
whole temperature range, with CO* the main species at low temperature and H* at higher temperature. This
effect is explained by the higher activation energy for CO* desorption and hence higher temperature
dependency than H* desorption. Simulations show that the OH* coverage increases up to 540 K, but is
steadily reduced at higher temperature.
Microkinetic modeling 123

Temperature (K)
450 500 550 600 650 1.00

1.E-01
0.00
1.E-03 400 500 600
log(Coverage) (-)

0.02
1.E-05

1.E-07 0.00
400 500 600
1.E-09 1.E-07 4.E-11
5.E-08 2.E-11
1.E-11
0.E+00 0.E+00
1.E-13 400 500 600
Free sites H* coverage CO* coverage O* coverage CH* coverage
OH* coverage HCO* coverage COH* coverage HCOH* coverage

Figure 5-7 Temperature dependency for the coverage of the surface species over a range of 200 K, represented
on a logarithmic scale with inclusion of detailed linear scale for the coverages of the most important
intermediates. At low temperatures the surface is dominated by CO* species, but at approximately 510 K, H*
becomes the main surface species. OH* coverage increases up to 520 K, but is reduced at higher temperatures.

Analysis of the CO consumption turnover frequency, see Figure 5-8 (a), shows a peak for the TOF CO at
approximately 540 K. Similar trends for the formation of hydrocarbons from CO and H 2 conversion have
been reported by the Van Santen group [19], and shows the relevance of the CO activation model. As
explained for water addition, reaction path analysis shows that the prominent pathway for HCO* formation
is through hydrogenation of CO* with H*, and COH* formation is mainly from hydrogenation with OH*.
The sharp increase in TOFCO is observed to be mainly due to the increased production rate of HCO* species
and the subsequent hydrogenation with OH*, illustrated in Figure 5-8 (b). This increased rate is linked to the
increasing OH* coverage with its maximum at 540 K. Due to this shift in coverage the dominant pathway for
CO activation is observed to be altered. At low temperature CO activation tends to go through COH* species
and further hydrogenation occurs through hydrogenation with H*, as this hydrogenation reaction has the
lowest barrier. For increased temperature a shift is observed towards the HCO* species and further
hydrogenation with OH*. Around 510 K, hydrogen becomes the main surface species, benefitting the HCO*
formation and a maximum is found at approximately 540 K, where the OH* coverage reaches its maximum.
124 Microkinetic modeling

(a) (b)
0.0006 2.0E-12

Rate of production (mol/cm²s)


0.0005 1.5E-12

1.0E-12
0.0004
TOFCO (s-1)

5.0E-13
0.0003
0.0E+00
0.0002
-5.0E-13
0.0001
-1.0E-12
0 -1.5E-12
450 500 550 600 650 450 500 550 600 650
Temperature (K) Temperature (K)
Figure 5-8 (a) CO consumption turnover frequency (s-1) in function of temperature. (b) Temperature
dependency of the rate of production (mol/cm²s) of the COH* species (blue line), HCO* species (purple line),
COH* hydrogenation to HCOH* (red dashed line), and HCO* hydrogenation to HCOH* (yellow dashed line).
Both the COH* formation and its further hydrogenation, and the HCO* formation and its further
hydrogenation are highly dependent. A shift is observed around 520 K from COH* formation to HCO*
formation.

For increasing temperatures adsorption of CO is less favorable, with a desorption temperature of 740 K. The
CO* coverage is indeed observed to decrease for increasing temperatures and is nearly entirely removed
from the surface at 640 K, see Figure 5-7. This leads to a reduction of the TOF CO and production rate
observed in Figure 5-8.

5.1.2. Chain propagation reactions

Subsequent to CO activation, chain propagation in FTS is evaluated up to C 2 species. Therefore, C-C


coupling reactions are taken in account as have been proposed for the CO insertion and carbide mechanism.
The reactions from the activation mechanism are adopted, and extended with the reactions summarized in
Table C-1. Similarly, rate coefficients for adsorption reactions are calculated from the collision theory and
surface reaction rate coefficients from the transition state theory. The rate coefficients are coverage
independent. The CO adsorption rate coefficient is again modified to obtain a realistic CO coverage under
FTS of 0.6. Analogously to the activation, first a base case (500 K, 20 bar, H 2/CO = 2) is discussed.

The main reactions for the CO insertion mechanism are summarized in Table 5-3. Chain propagation is
suggested to occurs via the consecutive coupling of RCHx* species with CO*, and subsequent (H- or OH-
assisted) C-O bond scission [13]. DFT calculations show that CH2* - CO* coupling has the lowest barrier for
propagation in the CO insertion mechanism, and hydrogenation is suggested to facilitate the C-O scission
reaction [7], although recently a pathway via hydrogenation with OH* is proposed [13]. The obtained CCH3 *
species can further couple with CO* to obtain C 3+ species. However, in this thesis analysis is done up to C2
Microkinetic modeling 125

species and the CCH3* species will be removed from the surface via hydrogenation to obtain ethane and
ethylene.

Table 5-3 Dominant reactions involved in the CO insertion mechanism. An extended summary is given in
Appendix C in Table C-1.
Forward Reverse
Forward pre- Reverse pre-
activation activation
Reactions exponential exponential
energy Eaf energy Ear
factor Af (s-1) factor Ar (s-1)
(kJ/mol) (kJ/mol)
C-C coupling
CH2* + CO* <=> COCH2* + * 1.00E+13 64.00 1.00E+13 85.00
Hydrogenation reactions
COCH2* + H* <=> COCH3* + * 1.00E+13 75.00 1.00E+13 77.00
COCH3* + OH* <=> COHCH3* + O* 1.85E+10 23.00 6.54E+10 14.00
C-O scission reactions
COCH3* + * <=> CCH3* + O* 1.78E+13 51.00 6.50E+13 124.00
COHCH3* + * <=> CCH3* + OH* 6.72E+12 39.00 4.84E+12 118.00

The dominant pathway for the carbide mechanism goes proposedly via coupling of RC* species with CH*,
as illustrated in Figure 5-9. The CHCR* surface species are further hydrogenated to obtain CHCHR*
species, which first dehydrogenate at the α-C atom, and further hydrogenate at the β-CH group to obtain
CCH2R* species which can again couple with a CH*.

Figure 5-9 Reaction path for chain propagation in FTS on Ru(11𝟐 ̅1) (T = 500 K, p = 20 bar, H2/CO = 2),
proposed by Van Santen et al. [19]. The dominant pathway goes through coupling with CH* species.

The simulated surface coverages are summarized in Table 5-4 for the main surface species in the model.
Simulations show a high concentration of CO* and H* present on the surface, and a low coverage for CH*
and CH2*. This favors the CO insertion mechanism, for which a high CH* or CH 2* coverage is not required,
and the propagation rate benefits from the high CO* coverage. The carbide mechanism does require a high
CH* coverage, as the propagation reaction is proposed to go via coupling with CH*. Hence, it can be
expected that CO insertion will be the dominant propagation pathway.
126 Microkinetic modeling

Table 5-4 Surface coverages simulated for the main surface species at 500 K and 20 bar.

Surface species Surface coverage Surface species Surface coverage

C* 1.92E-12 CH2* 9.53E-11


H* 3.96E-01 COCH* 3.38E-07
O* 4.16E-07 COCH2* 2.26E-09
CO* 5.96E-01 COCH3* 2.30E-10
OH* 3.67E-03 COHCH3* 2.01E-11
CH* 2.67E-08 CHCH* 3.41E-07

The dominant pathway obtained for the base case simulation is shown in Figure 5-10. The initiation
mechanism is observed to follow the same path as for the CO activation mechanism, and for propagation the
CO insertion mechanism was observed to be the main pathway. Calculations show that as propagation from
the CO insertion is much faster, the reactions included in the carbide mechanism are reversed.

Figure 5-10 Representation of the pathways for propagation in FTS up to C2 species, simulated with the
reactions presented in Table C-1 (500 K, 20 bar and H2/CO = 2). The dominant pathway forms C2H4 and is
indicated by the bold lines.

Simulations yield a high selectivity for ethylene (approximately 99%) and a CO consumption turnover
frequency of 5E-4 s-1. A chain growth probability α of 0.50 was calculated, reasonable for FTS [20].

5.1.2.1. Relative importance of C-C coupling and CO insertion


Simulations allow for quantitative analysis of the rate of the elementary reactions, and to account for the
relative importance of the carbide and CO insertion mechanism, evaluation is done for the relative rates of
CO insertion and C-C coupling in the base case (500 K, 20 bar, H2/CO = 2).

Table 5-5 Rates of CO insertion and C-C coupling (s-1) for the base case (500 K, 20 bar, H2/CO = 2).

Rate of CO insertion (s-1) Rate of C-C coupling (s-1) RC-C/RCO

1.31E-4 -1.85E-6 1.41E-2

The rate of CO insertion is found to be dominant over the C-C coupling mechanism with a negative relative
rate of the carbide mechanism to CO insertion (Table 5-5). This was also obtained in a thesis by Terry [21]
for FTS on Co(0001). The rate of the CO insertion mechanism is found to be significantly higher than for the
carbide mechanism.
Microkinetic modeling 127

Table 5-6 Comparison of the net production rate (s-1) for the propagation reactions included in the CO insertion
mechanism.

CH* + CO* CH2* + CO* CH3* + CO*

Net production rate (s-1) 1.43E-5 1.17E-4 -2.86E-10

CO insertion mainly goes via CH2* and CO* coupling (see Table 5-6), which is expected considering the
barrier for CO*-CH2* coupling (64 kJ/mol) to be lower than for coupling of CH*-CO* (96 kJ/mol) and
CH3*-CO* (185 kJ/mol). Furthermore, CH*-CO* coupling requires an additional hydrogenation step which
has a high barrier (105 kJ/mol); CHCO* will not undergo hydrogenation to form CHOCH* as this barrier
(139 kJ/mol) is significantly higher.

Table 5-7 Comparison of the net production rate (s-1) for the propagation reactions included in the carbide
mechanism.

C* + C* C* + CH* C* + CH3* CH* + CH* CH2* + CH2*

Net production rate (s-1) -9.01E-15 -5.90E-9 -1.74E-10 -1.80E-6 -4.52E-8

The dominant pathway for C-C coupling is found to be CH*-CH* coupling, see Table 5-7. Although the
CH2*-CH2* coupling reaction has a lower barrier (54 kJ/mol) than CH*-CH* coupling (86 kJ/mol), the
determined CH* coverage is higher than the CH2* coverage. Simulations however show that the carbide
mechanism has a negative reaction rate relative to CO insertion. The increased coverage of C2 species
resulting from the propagation in the CO insertion mechanism will cause the carbide mechanism to be
reversed. Note however that the activation energies for the carbide mechanism are taken from literature [19]
and therefore might not be consistent with the DFT calculated activation energies for the other reactions.
Calculations using the same functionals should be performed to obtain better qualitative insight in the
mechanism.

5.1.2.2. Effect of CH* coverage on the propagation mechanism


Via modification of the barriers for the activation mechanism the coverage of CH* can be altered, which
allows for evaluation of the dependence of the reaction mechanism on the CH* coverage. The effect of CH*
coverage is evaluated for the relative importance of the carbide and CO insertion mechanism, and for the
product selectivity.

In section 5.1.2.1 it is discussed that for the base case the CO insertion mechanism has a clear advantage
over the carbide mechanism. However, the dominant pathway in the carbide mechanism was found to be
CH*-CH* coupling and one could expect that the carbide mechanism would benefit from increased CH*
coverage. It is indeed observed from simulations that for CH* coverages above 1.0E-4 the reaction rate
sharply increases and even overtakes the CO insertion mechanism, illustrated in Figure 5-11. The CO
insertion mechanism is observed to go through a maximum, as for increased CH* coverage the CO*
coverage is reduced and propagation is retarded.
128 Microkinetic modeling

0.025

Rate of production (s-1)


0.02
0.015
0.01
0.005
0
-0.005
1.E-09 1.E-07 1.E-05 1.E-03
log(CH* coverage) (-)
Figure 5-11 Rate of production for the propagation reactions in the CO insertion mechanism (red line) and
carbide mechanism (blue line) in function of the CH* coverage on a logarithmic scale.

These results validate the concept that for the carbide mechanism to be dominant a high coverage of CH* is
necessary, which is not required for CO insertion [7]. Simulation of FTS found in literature also predicts a
low CH* coverage [4]. Substitution to this model suggests that the CO insertion model would be the
dominant pathway for chain propagation as is also found for the base case.

0.18 0.005
0.16
0.004
Selectivity ratio (-)

Selectivity ratio (-)


0.14
0.12 0.003
0.1
0.002
0.08
0.06 0.001
0.04
0
0.02
0 -0.001
0 0.0005 0.001 0.0015
log(CH* coverage) (-)
Figure 5-12 Comparison of the selectivity ratios CH4/C2H4 (orange line), C2H6/C2H4 (dark blue line),
CH3HCO/C2H4 (purple line), and C2H5OH/C2H4 (light blue line) in function of CH* coverage on log scale.

Simulations yield ethylene as the main product and therefore analysis is done for the relative selectivity of
the different products compared to ethylene, illustrated in Figure 5-12. Increased CH* coverage has a
significant influence on the CH4/C2H4 selectivity ratio. Methane is obtained from the consecutive
hydrogenation of CH* and CH2* species and an increased CH* coverage can easily be understood to be
beneficial for the methane production. Simulations show that acetaldehyde selectivity has a maximum value
at CH* coverage of approximately 0.001, which coincides with a minimum for the propagation rate in the
CO insertion mechanism and increased rate for the carbide mechanism. However, reaction path analysis
shows that acetaldehyde formation benefits from the general increase of chain propagation. In case the CH*
coverage is further increased, CO* coverages are reduced and the acetaldehyde selectivity will decrease.
Ethane and ethanol selectivity ratios to ethylene are observed to encounter little influence by the CH*
coverage variation.
Microkinetic modeling 129

5.1.2.3. Temperature effect on the propagation mechanism


The effect of temperature on the propagation mechanism is analyzed over a temperature range from 450 K to
1000 K (taking in account the partition functions for 500 K). Figure 5-13 illustrates the surface species
coverages in function of temperature. Initially CO* is found to be the dominant surface species, but as
temperature is increased the coverage is observed to sharply decrease until CO* is nearly entirely removed
from the surface. At high temperatures hydrogen is observed to be the main surface species. OH* is found to
be dominant at intermediate temperatures with a maximum around 650 K. The temperature dependency of
the surface species coverages is found to be similar to the dependency found in the activation mechanism.
Temperature (K)
450 500 550 600 650 700 750 800 850 900 950 1000
1.00E+00

1.00E-02 1.00
0.80
log(Coverage) (-)

1.00E-04

Coverage (-)
0.60
1.00E-06
0.40
1.00E-08 0.20
0.00
1.00E-10
450 600 750 900
1.00E-12 Temperature (K)

1.00E-14
Free sites H* coverage CO* coverage O* coverage
OH* coverage CH* coverage COCH* coverage COCH2* coverage
COCH3* coverage COHCH3* coverage CHCH* coverage
Figure 5-13 Temperature dependency of the coverage of the surface species, represented on a logarithmic scale
with inclusion of detailed linear scale for the coverages of the most important intermediates (20 bar; H 2/CO = 2).
Initially CO* is the dominant surface species but a fast decrease of coverage is observed for increased
temperature. At higher temperatures hydrogen is found to be dominant. For intermediate temperatures, OH* is
observed to go through a maximum, as also observed for the activation mechanism.

Analysis of the CO consumption TOF CO (s-1) and CH4 formation (mol/s) is illustrated in Figure 5-14 (a). A
maximum TOFCO is found at approximately 650 K, which coincides with the maximal OH* coverage
predicted by the simulations. Falling back to the activation mechanism, CO activation was observed to be
optimal at maximal OH* coverage. Hence, maximal OH* coverage leads to an increased propagation rate.
TOFCO values are calculated in a range of 0.001 to 0.2 s-1 and are in the same order of magnitude as the
experimental value of 0.01 s-1 [13]. CH4 formation is found to be maximal at slightly higher temperatures
than for the maximum TOFCO. This can be related to the increasing hydrogen coverage which will benefit the
hydrogenation reactions, causing increased CH4 yield for the conversion process at higher temperature. At
high temperatures, the CO* coverage is reduced and hence, leading to a reduction of the CO consumption
rate. Evaluation of the different propagation mechanisms shows that the CO insertion mechanism is the
dominant pathway, with reactions included in the carbide mechanism even reversed, illustrated in Figure
5-14 (b).
130 Microkinetic modeling

(a) (b)
0.18 2.5E-05 6E-10
0.16

CH4 formation (mol/s)


4E-10

Rate of production
0.14 2.0E-05
2E-10
TOFCO (s-1)

0.12

(mol/cm²s)
0.10 1.5E-05 0
0.08 -2E-10
1.0E-05
0.06
-4E-10
0.04 5.0E-06
0.02 -6E-10
0.00 0.0E+00 -8E-10
500 600 700 800 900 1000 500 600 700 800 900 1000
Temperature (K) Temperature (K)
Figure 5-14 (a) CO consumption TOFCO (s-1) and CH4 formation (mol/s) in function of temperature for the chain
propagation mechanism (20 bar; H2/CO = 2). (b) Rate of production for CH*-CH* coupling (carbide
mechanism, blue line) and CH2*-CO* coupling (CO insertion, red line) in the chain propagation mechanism.

The simulated selectivities for the propagation mechanism are given in Figure 5-15. Ethylene is found to be
the main product with a selectivity of approximatly 99%. At increased temperature an increased selectivity
for methane and ethane is observed, which can be related to the high H* and low CO* coverage enhancing
the hydrogenation reactions. Also formaldehyde and acetaldehyde formation is found to be more pronounced
at increased temperature. Similarly, the increased hydrogen coverage will favor the hydrogenation reactions.
Besides, the increased temperature will reduce the stability of the surface species and hence, favoring
desorption reactions. Calculations also show that acetaldehyde formation reactions have higher activation
barriers and hence will be more temperature dependent.

Temperature (K)
450 500 550 600 650 700 750 800 850 900 950 1000
1.E+00 1.E+00
1.E-01 1.E-01
1.E-02 1.E-02
Selectivity (-)

1.E-03 1.E-03
1.E-04 1.E-04
1.E-05 1.E-05
1.E-06 1.E-06
1.E-07 1.E-07
1.E-08 1.E-08
Selectivity CO2 Selectivity CH4 Selectivity CH2O
Selectivity C2H6 Selectivity CH3CHO Selectivity C2H4
Selectivity C2H5OH
Figure 5-15 Simulated selectivities for the product species in the chain propagation mechanism in function of
temperature (20 bar; H2/CO = 2). Ethylene is found to be the main product, but for increased temperature an
increased selectivity for methane and ethane is observed, also formaldehyde and acetaldehyde formation is more
pronounced at higher temperatures.
Microkinetic modeling 131

5.1.2.4. Effect of H2/CO ratio on the propagation mechanism

Besides the temperature dependency, also the effect of variation of the ratio of H 2 and CO partial pressures is
investigated; the H2/CO ratio is variated from 1.2 to 6. Figure 5-16 shows the effect on the surface coverages.
At low pressure, CO* is found to be the dominant species due to its higher adsorption energy compared to
hydrogen. For ratios above 3.5, H* is found to be the main surface species due to the increased partial
pressure. Variation of the H 2/CO has a limited effect on the other surface species.

H2/CO ratio (-)

Coverage (-)
1 2 3 4 5 6 1.00
1.E+00
0.50
1.E-02
0.00
log(Coverage) (-)

1.E-04 1 2 3 4 5 6
H2/CO ratio (-)
1.E-06

Coverage (-)
1.E-06
1.E-08 5.E-07
1.E-10 0.E+00
1 2 3 4 5 6
1.E-12 H2/CO ratio (-)
Free sites H* coverage CO* coverage O* coverage
OH* coverage CH* coverage COCH* coverage COCH2* coverage
COCH3* coverage COHCH3* coverage CHCH* coverage
Figure 5-16 Surface species coverages in function of the H2/CO partial pressure ratio, represented on a
logarithmic scale with inclusion of detailed linear scale for the coverages of the most important intermediates
(500 K; 20 bar). CO* is the dominant surface species at lower ratios, with a shift towards H* for ratios above 3.5.

The effect of the H2/CO ratio on the TOFCO (s-1) and CH4 formation (mol/s) is illustrated in Figure 5-17 (a).
A maximum TOFCO is observed at ratios from 3 to 4, which can be related to the coverages of CO* and H*.
At low ratios the CO* species will block the surface and a shortage of H* exists. Similarly, for high ratios
H* species will block the surface and a shortage of CO* exists. For intermediate ratios both CO* and H*
coverage is high and TOFCO is maximized. CH4 formation is observed to steadily increase for increasing
H2/CO ratios. This is explained by the increased hydrogen coverage, which favors the hydrogenation
reactions. Figure 5-17 (b) illustrates the rate of production for CH*-CH* coupling (carbide mechanism) and
CH2*-CO* coupling (CO insertion). The CO insertion is again observed to be the main propagation pathway,
with the carbide mechanism even reversed.
132 Microkinetic modeling

(a) (b)

CH4 formation (mol/s)


0.0006 7E-08 3.5E-13

Rate of production
6E-08 3E-13
TOFCO (s-1)

0.0005

(mol/cm²s)
5E-08 2.5E-13
0.0004 2E-13
4E-08
0.0003 1.5E-13
3E-08 1E-13
0.0002
2E-08 5E-14
0.0001 1E-08 0
0 0 -5E-14
2 3 4 51 6 1 2 3 4 5 6
H2/CO ratio (-) H2 /CO ratio (-)
-1
Figure 5-17 (a) TOFCO (s ) and CH4 formation (mol/s) in function of H2/CO ratio for the chain propagation
mechanism (500 K; 20 bar). Maximum TOFCO is observed at H2/CO = 3; methane production increases with
increasing H2/CO ratio. (b) Rate of production for CH*-CH* coupling (carbide mechanism, blue line) and CH2*-
CO* coupling (CO insertion, red line) in the chain propagation mechanism.

The simulated selectivities for the propagation mechanism are represented in Figure 5-18. Ethylene is again
observed to be the main reaction product with a selectivity of nearly 99%. For increasing H 2/CO ratios an
increased CH4 selectivity is observed, explained by the increased hydrogen coverage enhancing the
hydrogenation reactions. Further, variation of the H 2/CO ratio has only a limited effect on the product
selectivity.

H2/CO ratio (-)


1 2 3 4 5 6
1.E+00 1.E+00

1.E-01 1.E-01
log(Selectivity) (-)

1.E-02 1.E-02

1.E-03 1.E-03

1.E-04 1.E-04

1.E-05 1.E-05

1.E-06 1.E-06

1.E-07 1.E-07
Selectivity CO2 Selectivity CH4 Selectivity CH2O Selectivity C2H6
Selectivity CH3CHO Selectivity C2H4 Selectivity C2H5OH

Figure 5-18 Simulated selectivities of the product species in the chain propagation mechanism in function of
H2/CO ratio (500 K; 20 bar. Increased ratios has a limited effect on the selectivity. Ethylene is found to be the
main product, and methane selectivity increases for increased H2/CO ratio.
Microkinetic modeling 133

5.2. CO2 activation


For reasons of comparison, analysis of CO2 activation is done by incorporation of CO2 activation steps in the
initiation mechanism of FTS, given in Table 5-8. The reaction steps include direct dissociation of CO2,
hydrogen assisted dissociation, and further hydrogenation before dissociation to obtain hydrogenated species.
Several limitations exist in the model and only a preliminary analysis is done; additional research is required
to allow for proper modeling of the CO2 activation mechanism.

Table 5-8 Reactions included for the activation of CO 2 on Co(111) under Fischer-Tropsch synthesis conditions.
The corresponding DFT calculated activation energies are given in kJ/mol.
Reactions Forward pre- Forward Reverse pre- Reverse
exponential activation energy exponential activation energy
factor Af (s-1) Eaf (kJ/mol) factor Ar (s-1) Ear (kJ/mol)
H-assisted CO2 activation
CO2(g) + H* <=> COOH* 1.00E+13 154.00 1.00E+13 84.00
COOH* + H* <=> HCOOH* + * 1.00E+13 61.00 1.00E+13 69.00
CO2(g) + H* <=> HCOO* 1.00E+13 94.00 1.00E+13 107.00
HCOO* + H* <=> HCOOH* + * 1.00E+13 110.00 1.00E+13 35.00
C-O scission reactions
COOH* + * <=> CO* + OH* 1.00E+13 63.00 1.00E+13 190.00
HCOO* + * <=> HCO* + O* 1.00E+13 110.00 1.00E+13 35.00
HCOOH* + * <=> HCO* + OH* 1.00E+13 186.00 1.00E+13 184.00

Similar reaction and inlet conditions are evaluated as for the CO activation mechanism (see section 3.8) and
a feed stream of pure CO2 (0.33) and H2 (0.66) is used. CO2 was found to be mainly activated via direct CO2
dissociation yielding CO* and O* on the surface. For the hydrogen-assisted pathway, CO2 was observed to
be hydrogenated to formate species (HCOO*), which dissociates to form HCO* on the surface. HCO* is
then included in the CO activation pathway as explained above. Temperature analysis shows that from a
temperature of 500 K the CO2 conversion will undergo fast conversion and for a temperature around 520 K,
CO2 is completely converted. This is also visualized in the CO2 consumption turnover frequency shown in
Figure 5-19 (a). The TOFCO2 increases rapidly at 500 K and remains constant after 520 K, indicating 100 %
conversion is reached. However, CO2 conversion also includes the dissociation reaction and at high
temperatures, simulations show selectivities for CO near 100%, due to the rapid desorption of CO at
increased temperature. Figure 5-19 (b) shows the CH4 formation from the activation of CO 2 compared to CO
activation. At 520 K the H* and CO* coverages both are high, and a maximum is found for the CH 4
formation. This is in agreement with the maximum TOFCO2. Compared to the CO activation pathway an
increased CH4 production is observed, indicating the potential for CO2 activation via the hydrogen-assisted
activation pathway.
134 Microkinetic modeling

(a) (b)
0.16 0.0006 0.0006

CH4 formation (mol/s)


0.14 0.0005 0.0005
0.12 0.0004
TOFCO2 (s-1)

TOFCO (s-1)
0.1 0.0004
0.0003
0.08 0.0003
0.0002
0.06 0.0002
0.04 0.0001
0.02 0.0001 0
0 0 -0.0001
460 510 560 460 510 560
Temperature (K) Temperature (K)

Figure 5-19 (a) CO2 consumption TOF (s-1) in function of temperature and (b) CH 4 formation (mol/s) for the CO
activation and CO2 activation mechanism, with CO/H2 and CO2/H2 ratio = 2.

5.3. Conclusion
In this chapter an initial microkinetic model was composed of CO activation in Fischer-Tropsch synthesis,
and extended to chain propagation up to C 2 product fractions. Also a preliminary analysis was done for CO2
activation under FTS conditions.

The model for CO activation includes the direct dissociation, H-assisted activation, and OH-assisted
activation. For a base case (500 K, 20 bar) the dominant pathway for CO activation was found to go through
consecutive hydrogenation steps with OH* species to form COH* and CHOH*, which dissociates to yield
CH* on the surface. DFT calculated activation energies for the hydrogenation reactions with OH* are
significantly lower than for hydrogenation with H*; explaining the dominant reaction path. To evaluate the
effect of the OH* coverage, water was added to the feed. H2O dissociates on the surface and increases the
OH* surface coverage. A maximal TOFCO was observed at H2O partial pressure of 0.01, but at higher partial
pressures the TOFCO stagnates and is reduced. The dominant reaction path is shifted towards hydrogenation
of CO* with H* (as O* on the surface will block the hydrogenation with OH*), and further hydrogenation
with OH* to yield CHOH*, which dissociates to form CH* on the surface. The effect of temperature on the
activation mechanism was evaluated over a temperature range of 200 K (450 – 650 K). It is observed that
CO* is removed from the surface at higher temperature due to the higher activation energy and hence, higher
temperature dependency. At high temperature H* is the dominant surface species, and for intermediate
temperatures (540 K) a maximum is found for the OH* coverage. The TOF is also found to be maximal at
540 K, denoting the importance of the OH* coverage. However, the dominant pathway is shifted from COH*
formation via CO* hydrogenation with OH* to hydrogenation of CO* with H*, due to the increasing
temperature and accompanying higher H* coverage. Further hydrogenation of HCO* occurs via OH* and
hence, a maximal TOF is found for maximal OH* coverage.

Expansion to include for chain propagation takes in account reactions for CO insertion and the carbide
mechanism. The simulation of the base case shows a high ethylene selectivity (approximately 99%) and a
Microkinetic modeling 135

chain growth probability α of 0.5. The CO insertion mechanism was found to be the dominant pathway for
propagation and goes mainly through coupling of CO* with CH2* as this coupling reaction has the lowest
barrier (64 kJ/mol) in the CO insertion mechanism. Although the activation barrier for CH 2*-CH2* coupling
is lower, the coverages are too low for the carbide mechanism to be competitive. CH2CO* is further
hydrogenated and via scission of the C-O bond CCH3 species are obtained, which can undergo further
coupling. The CH* coverage is an important parameter as the coupling of CH* species is found to be the
dominant reaction for propagation in the carbide mechanism. Therefore, evaluation was done for the
influence of the CH* coverage on the relative importance of C-C coupling and CO insertion. The CH*
coverage was varied via modification of the barriers for CO activation and as expected, the carbide
mechanism was found to be dominant for sufficiently high CH* coverage, confirming the necessity of high
CH* for the carbide mechanism. The selectivity for CH4 is also significantly increased with CH* coverage.
The effect of temperature on the propagation mechanism was evaluated over a temperature range from 450 K
to 1000 K. Similar to the activation model, the CO* species are removed from the surface and H* is the
dominant surface species at high temperature. For intermediate temperature OH* is found to be the main
species. The maximum TOF is found at the maximal OH* coverage, which is linked to the maximal
initiation at maximal OH* coverage. The model predicts a selectivity of approximately 99% for ethylene, but
at higher temperatures the selectivities for CH4 and C2H6 are increased, due to the increased H* coverage.
Finally the H2/CO partial pressure ratio was evaluated for its effect on the propagation mechanism. For
increased ratios a higher H* coverage was observed, but other species only encountered a minor influence by
the variation of H2/CO ratio. Only the formation of CH4 was significantly increased due the higher H*
coverage, but only a limited effect on the selectivities and TOF was observed.

A preliminary analysis was performed for the activation of CO2 by incorporation of CO2 activation steps in
the activation mechanism for FTS. However, several limitations exist in the model and additional research is
required to allow for proper modeling of the CO2 activation mechanism. CO2 is found to mainly dissociate as
this has the lowest barrier (60 kJ/mol), yielding a high selectivity for CO. Hydrogen-assisted activation was
found to mainly go through formate formation. Increased CH 4 production compared to CO activation denotes
the potential for CO2 activation. However, extension of the activation mechanism and further analysis of the
propagation reactions is necessary before proper, qualitative evaluation is possible.
136 Microkinetic modeling

5.4. References

1. Neurock, M., The microkinetics of heterogeneous catalysis. By JA Dumesic, DF Rudd, LM Aparicio,


JE Rekoske, and AA Treviño, ACS Professional Reference Book, American Chemical Society,
Washington, DC, 1993, 315 pp. 1994, Wiley Online Library.

2. Kee, R.J., J.A. Miller, and T.H. Jefferson, CHEMKIN: A general-purpose, problem-independent,
transportable, FORTRAN chemical kinetics code package. 1980, Sandia Labs.

3. Getting stared with CHEMKIN, R. Design, Editor. 2015.

4. Storsæter, S., D. Chen, and A. Holmen, Microkinetic modelling of the formation of C 1 and C 2
products in the Fischer–Tropsch synthesis over cobalt catalysts. Surface Science, 2006. 600(10): p.
2051-2063.

5. Hansen, A.G., W.J. van Well, and P. Stoltze, Microkinetic modeling as a tool in catalyst discovery.
Topics in Catalysis, 2007. 45(1): p. 219-222.

6. Metaxas, K., et al., A microkinetic vision on high-throughput catalyst formulation and optimization:
development of an appropriate software tool. Topics in Catalysis, 2010. 53(1-2): p. 64-76.

7. Zhuo, M., et al., Density functional theory study of the CO insertion mechanism for Fischer−
Tropsch synthesis over Co catalysts. The Journal of Physical Chemistry C, 2009. 113(19): p. 8357-
8365.

8. Fischer, F. and H. Tropsch, The preparation of synthetic oil mixtures (synthol) from carbon
monoxide and hydrogen. Brennstoff-Chem, 1923. 4: p. 276-285.

9. Van Santen, R., et al., Mechanism and microkinetics of the Fischer–Tropsch reaction. Physical
Chemistry Chemical Physics, 2013. 15(40): p. 17038-17063.

10. Brady III, R.C. and R. Pettit, Reactions of diazomethane on transition-metal surfaces and their
relationship to the mechanism of the Fischer-Tropsch reaction. Journal of the American Chemical
Society, 1980. 102(19): p. 6181-6182.

11. Pichler, H. and H. Schulz, Recent results in synthesis of hydrocarbons from CO and H2. Chemie
Ingenieur Technik, 1970. 42(18): p. 1162-&.

12. Zhuo, M., A. Borgna, and M. Saeys, Effect of the CO coverage on the Fischer–Tropsch synthesis
mechanism on cobalt catalysts. Journal of Catalysis, 2013. 297: p. 217-226.

13. Gunasooriya, G.K.K., et al., Key role of surface hydroxyl groups in CO activation during Fischer-
Tropsch synthesis. ACS Catalysis, 2016.

14. Schulz, H. and M. Claeys, Kinetic modelling of Fischer–Tropsch product distributions. Applied
Catalysis A: General, 1999. 186(1–2): p. 91-107.

15. Cheng, J., et al., A DFT study of the chain growth probability in Fischer–Tropsch synthesis. Journal
of Catalysis, 2008. 257(1): p. 221-228.

16. Gunasooriya, G.K.K., et al., CO adsorption on cobalt: Prediction of stable surface phases. Surface
Science, 2015. 642: p. L6-L10.
Microkinetic modeling 137

17. Hibbitts, D.D., et al., Mechanistic role of water on the rate and selectivity of Fischer–Tropsch
synthesis on ruthenium catalysts. Angewandte Chemie, 2013. 125(47): p. 12499-12504.

18. Azadi, P., et al., Microkinetic Modeling of the Fischer–Tropsch Synthesis over Cobalt Catalysts.
ChemCatChem, 2015. 7(1): p. 137-143.

19. Filot, I.A.W., R.A. van Santen, and E.J.M. Hensen, The Optimally Performing Fischer–Tropsch
Catalyst. Angewandte Chemie International Edition, 2014. 53(47): p. 12746-12750.

20. Vervloet, D., et al., Fischer–Tropsch reaction–diffusion in a cobalt catalyst particle: aspects of
activity and selectivity for a variable chain growth probability. Catalysis Science & Technology,
2012. 2(6): p. 1221-1233.

21. Gani Zhi Hao, T., Kinetic modeling of the fischer-tropsch process on cobalt catalyst, in Department
of Chemical and Biomolecular Engineering. 2013, National University of Singapore. p. 70.
138

Chapter 6
Conclusions and future work
In this thesis, first a techno-economic analysis has been performed on renewable pathways for CO 2
conversion to methanol, ethylene and formic acid. Secondly, DFT calculations were done for the activation
steps in CO2 conversion on the cobalt catalyst surface, and comparison was made with copper, nickel and
gold catalyst surfaces. Finally, an initial microkinetic model was composed of CO activation and propagation
steps in FTS, with inclusion of a preliminary analysis for CO 2 activation.
Conclusion 139

6.1. Techno-economic analysis


Techno-economic analysis allows for the evaluation of the feasibility of production processes and analysis
has been performed for (1) the production of hydrogen from electrolysis of water with renewable energy; (2)
production of methanol via conversion of CO2 with hydrogen; (3) ethylene synthesis via electrochemical
reduction of CO2 and water, a modified FTS process for CO2 and H2, and the methanol-to-olefins process;
and (4) the electrochemical reduction of CO2 to formic acid.

Hydrogen production cost for alkaline and PEM electrolysis is estimated at approximately 4 €/kg H2, and is
mainly determined by the electricity feedstock cost for electrolysis and the high capital costs for electrolysis
equipment. PEM electrolysis is found to be most favorable, and is predicted to be competitive with fossil
fuels in future applications.

Methanol production is favorable for its compatibility with existing distribution infrastructure and the ability
to use it as a replacement for current liquid fuels, as well as an intermediate in chemical production
processes. Production cost for renewable methanol was estimated at 1400 €/ton, which is significantly higher
than actual methanol price of 275 €/ton. The high cost can mainly be attributed to the electricity cost of water
electrolysis for hydrogen production.

Ethylene synthesis via MTO has the advantage of a high flexibility and good control over the product
selectivity, with in particular the ability to increase the propylene to ethylene product ratio. Analysis shows a
production cost for ethylene from renewable methanol of approximately 3700 €/ton, compared to a
production cost of 935 €/ton in case of conventional methanol feedstock. Overall 1.35 ton of CO2 is recycled
per ton of product, however, carbon taxes would need to be unreasonably high for the renewable MTO
process to become feasible. Analysis of ethylene production via modified FTS yields a production cost of
approximately 2100 €/ton, which is still significantly higher than the conventional ethylene price around
1200 €/ton. The electricity cost for electrolysis remains the highest cost contributor and capital costs are also
significantly higher than in the MTO process.

Formic acid is mainly considered favorable for use in hydrogen storage and fuel cell applications. It benefits
from its low chemical consumption (possibly solely water), lower energy requirement and mild operating
conditions compared to other CO2 conversion processes Renewable formic acid production price is estimated
at 1400 €/ton and is mainly determined by the electricity and capital cost. Actual selling prices are in a range
of 600 to 700 €/ton, making it economically very difficult for renewable formic acid production to be
feasible.
140 Conclusion

6.2. Ab-initio calculations


A first evaluation has been done for the activation of CO2 on the Co(111) surface via DFT calculations. CO2
adsorption was observed to be thermodynamically unstable at 500 K and 1 bar. However, CO2 dissociation is
shown to be feasible, with a relatively low activation barrier of 60 kJ/mol and a Gibbs free energy of reaction
of -72 kJ/mol. Hydrogen is expected to be the main species involved in hydrogenation reactions and the
hydrogen-assisted pathway for CO2 activation was found to go preferentially through formate species, as this
pathway has the lowest energy barriers and endothermicity.

To obtain a more realistic catalyst model, the coverage effect on CO 2 activation was evaluated on Co(111)
with CO coverage of 1/3 ML. DFT calculations show that in situations of high coverage, CO2 has a higher
adsorption energy and the dissociation barrier increases. However, under real FT conditions the surface is
likely to be more complex and the presence of reaction intermediates might further affect the stability of the
CO2 adsorbate.

Because significantly lower activation barriers have been reported for CO dissociation reactions on step sites,
CO2 activation was also evaluated on the step sites of the Co(211) surface. Adsorption is found to be
thermodynamically stable on top sites and similar dissociation barriers were calculated as for terraces.
Hence, dissociation of CO2 on step sites is found to be thermodynamically favorable and step sites appear to
have high potential for the conversion of CO2. However, step sites are expected to be unavailable for CO2
molecules, due to their high affinity for reaction intermediates and low thermodynamic stability under FTS
conditions.

For reasons of comparison, evaluation was done for the Cu(111), Ni(111) and Au(111) surface with
inclusion of coverage effects. Adsorption energies for CO2 on Cu(111) are close to the values found for
Co(111), but DFT calculations show higher energy barriers for dissociation on the Cu(111) surface. The
nickel surface appears to have a similar capability of CO2 activation as Co(111), although the CO2 adsorbates
are thermodynamically less stable, and the dissociation barriers are slightly higher on Ni(111). The gold
catalyst surface is shown to be incapable of efficiently catalyzing the CO2 activation reactions, due to the
high DFT calculated dissociation barriers. Overall, the cobalt catalyst appears to be the preferred surface
over the Cu(111), Ni(111), and Au(111) surfaces.
Conclusion 141

6.3. Micro-kinetic modeling


An initial microkinetic model was composed of CO activation in Fischer-Tropsch synthesis, and extended to
chain propagation up to C2 product fractions. Also a preliminary analysis was done for CO2 activation under
FTS conditions.

The CO activation model predicts that the OH-assisted pathway is dominant for a base case (500 K, 20 bar),
as DFT calculations yield the lowest energy barriers. The effect of OH* coverage was evaluated by addition
of water in the feed stream, and the CO consumption TOFCO was observed to increase for low H 2O partial
pressures but is reduced at higher partial pressures. Reaction path analysis for CO activation indicates that
the dominant pathway is shifted for increasing OH* coverage towards hydrogenation of CO* with H* to
HCO* and subsequent hydrogenation with OH*, due to blocking of COH* formation by surface oxygen.
Hence, a combination of the H- and OH-assisted pathway is obtained. Next, the effect of temperature on the
activation mechanism is evaluated. A significant effect on the surface coverage is observed with a shift from
high CO* coverage to high H* coverage for increasing temperature. OH* coverage goes through a maximum
at approximately 540 K, and accompanying the TOF is also found to be maximal. Due to the increasing
temperature and hydrogen coverage the dominant pathway is shifted towards HCO* formation by
hydrogenation with H* and further hydrogenation with OH* to obtain CHOH*. The second hydrogenation
step will be maximal at high OH* coverage, explaining the maximal TOF at maximal OH* coverage.

Chain propagation is evaluated by expansion of the reaction mechanism for CO activation up to C 2.


Simulation of the base case shows a high ethylene selectivity (approximately 99%) and a chain growth
probability α of 0.5. The CO insertion mechanism was found to be the dominant pathway for propagation,
and although the activation barrier for CH2*-CH2* coupling is lower, the coverages are too low for the
carbide mechanism to be competitive. Hydrogen-assisted C-O scission then yields C2 species capable of
further coupling. The coupling of CH* species is found to be the dominant reaction for propagation in the
carbide mechanism. Modification of the energy barriers in the activation model allows for the variation of
the CH* coverage, and the carbide mechanism is observed to be dominant in case of high CH* coverage,
confirming the necessity of high CH* coverage for the carbide mechanism. The chain propagation shows a
similar temperature dependency as for the activation model, with a shift in coverage from CO* at low
temperatures to H* at high temperature and a maximal OH* coverage at intermediate temperatures.
Similarly, the maximum TOFCO is observed for the maximal OH*, linked to the maximal initiation rate. The
model predicts a high selectivity for ethylene, with increasing CH4 and C2H6 selectivity for increasing
temperature, due to the high H* coverage. Finally the H 2/CO partial pressure ratio was evaluated for its
effect on the propagation mechanism. A high H* coverage was observed for increasing ratios and the
formation of CH4 is significantly increased. Further influences of the H 2/CO ratio were observed to be rather
limited.

Incorporation of activation steps for CO2 allowed for a preliminary analysis of CO2 activation. However,
several limitations exist in the model and additional research is required to allow for proper modeling of the
142 Conclusion

CO2 activation mechanism. A high selectivity for CO is predicted, due to the low dissociation barrier for CO 2
on the Co(111) catalyst surface. Also an increased CH4 production was observed compared to CO activation,
denoting the potential for CO2 activation. However, extension of the activation mechanism and further
analysis of the propagation reactions is necessary before proper, qualitative evaluation is possible.

6.4. Future work


Future work for the techno-economic analysis of CO2 conversion consists of an extension of the different
products taken in account. Possible products such DME, higher alcohols and hydrocarbons, polycarbonates,
polyurethanes and others, could be evaluated and compared to conventional production, as well as with other
renewable conversion routes. An evaluation could be done on the future carbon tax legislation and the
growth of renewable energy generating plants, such as windmill parks or solar power stations.

Further evaluation of CO2 activation via DFT calculations could be done by examination of new catalyst
surfaces and taking in account additional reaction intermediates. Calculations of activation energies could be
performed to extent the reaction mechanism for CO2 activation and its propagation steps. For proper
evaluation of the dominant reaction mechanism for chain propagation in FTS the activation energies for the
carbide mechanism could be determined to construct a consistent kinetic model for chain growth.

The kinetic model for CO activation and chain propagation in FTS could be enhanced by implementation of
consistent barriers for the carbide mechanism, to allow for qualitative comparison with the CO insertion
mechanism. Furthermore, the model could be extended towards higher carbon species. Extension of the
model for CO2 activation could be done by incorporation of additional pathways for activation and inclusion
of consecutive propagation reactions.
143

Appendix A
Techno-economic analysis of renewable
pathways for CO2 conversion.
Table A-1: Cost breakdown for the Distributed Forecourt Production of Hydrogen from Bio-Derived Renewable
Liquids – High Temperature Ethanol Reforming. Plant design capacity is set at 1500 kg/day and a yearly
operation of 8500h [1].
2011 2015 2020
Characteristics Units
Status Target Target
Hydrogen Levelized Cost (Production Only) $/kg 6.60 5.90 2.30
Production Equipment Total Capital Investment $ 1.9M 1.4M 1.2M
Production Energy Efficiency % 68 70 75
Production Equipment Availability % 97 97 97
Ethanol Price average $/gal 2.47 2.41 0.85

Table A-2: Cost calculation for the Biomass Gasification/Pyrolysis Hydrogen Production. Production capacity is
set at 155 tH2/day. For the current case this requires 2070 t/day of biomass ($75/kg) and for the future case (2020)
this is 2000 t/day of biomass ($63/kg) [1].
2011 2015 2020
Characteristics Units
Status Target Target
Hydrogen Levelized Cost (Plant Gate) $/kg 2.20 2.10 2.00
Total Capital Investment $M 180 180 170
Energy Efficiency % 46 46 48
144 Appendix A

Table A-3: Target for Hydrogen production cost via solar-driven high-temperature thermochemical hydrogen
production. Production capacity is set at 100 000 kg H2/day [1].
2015 2020 Ultimate
Characteristics Units
Target Target Target
Solar-Driven High-Temperature Thermochemical Cycle
$/kg 14.80 3.70 2.00
Hydrogen Cost
Chemical Tower Capital Cost (installed cost) $/TPD H2 4.1MM 2.3MM 1.1MM
Annual Reaction Material Cost per TPD H2 $/ yr.-TPD H2 1.47M 89K 11K
Solar to Hydrogen (STH)
% 10 20 26
Energy Conversion Ratio
1-Sun Hydrogen Production Rate kg/s per m2 8.1E-7 1.6E-6 2.1E-6

Table A-4: Target for Hydrogen production cost via Photoelectrochemical Hydrogen Production: Photoelectrode
system with Solar Concentration. Production capacity is set at 50 000 kg H2/day [1].
2011 2015 2020 Ultimate
Characteristics Units
Status Target Target Target
Photoelectrochemical Hydrogen Cost $/kg NA 17.30 5.70 2.10
Capital cost of Concentrator & PEC Receiver (non-
installed, no electrode) $/m2 NA 200 124 63
Annual Electrode Cost per TPD H2 $/yr-TPDH2 NA 2.0M 255K 14K
Solar to Hydrogen (STH) Energy Conversion Ratio % 4 to 12% 15 20 25
1-Sun Hydrogen Production Rate kg/s per m2 3.3E-7 1.2E-6 1.6E-6 2.0E-6

Table A-5: Targets for Hydrogen cost via Photolytic Biological Hydrogen Production. Production capacity is set
at 50 000 kgH2/day [1].
2011 2015 2020 Ultimate
Characterictics Units
Status Status Target Target
Hydrogen Cost $/kg NA NA 9.20 2.00
Reactor Cost $/m² NA NA 14 11
Light utilization efficiency (% incident solar
energy that is converted into photochemical % 25 28 30 54
energy)
Duration of continuous H2 production at full
Time Units 2 min 30 min 4h 8h
sunlight intensity
Solar to H2 (STH) Energy Conversion Ratio % NA 2% 5% 17%
1-Sun Hydrogen Production Rate kg/s per m² NA 1.6E-7 4.1E-7 1.4E-6
Appendix A 145

Table A-6: Input parameters for H2A Production cases for PEM electrolysis (costs in 2007$ and in 2012$). [2]
Current Future Current Future
Parameter Forecourt Forecourt Central Central
1,500 kg/day 1,500 kg/day 50,000 kg/day 50,000 kg/day
Plant Capacity (kg/day) 1,500 1,500 50,000 50,000
Total Uninstalled Capital (2012$/kW) $940 $450 $900 $400
Stack Capital Cost (2012$/kW) $385 $171 $423 $148
Balance of Plant (BOP) Capital Cost
$555 $279 $477 $252
(2012$/kW)
Total Electrical Usage kWh/kg 54.619 50.3 54.3 50.2
Conversion Efficiency (LHV of H2) (61%) (66%) (61%) (66%)
Stack Electrical Usage (kWh/kg) 49.2 46.7 49.2 46.7
Conversion Efficiency (LHV of H2) (68%) (71%) (68%) (71%)
BOP Electrical Usage (kWh/kg) 5.4 3.6 5.1 3.5
Electrolyzer Power Consumption
3.4 3.1 113 105
at full power (MW)
Average Electricity Price over Life of
6.12 6.88 6.22 6.89
Plant20 (2007¢/kWh)
Electricity Price in Startup Year
5.74 6.59 5.74 6.59
(2007¢/kWh)
Outlet Pressure from Electrolyzer (psi) 450 1,000 450 1,000
Installation Cost (% of uninstalled capital
1222 10 12 10
cost)
Replacement Interval (years) 7 10 7 10
Replacement Cost of Major Components
15 12 15 12
(% of installed capital cost)
Plant Life (years) 20 20 40 40
Stack Current Density (mA/cm2) 1,500 1,600 1,500 1,600
Capacity Factor (%) 86 86 97 97
Total Electrical Usage kWh/kg 54.619 50.3 54.3 50.2
Conversion Efficiency (LHV of H2) (61%) (66%) (61%) (66%)
Stack Electrical Usage (kWh/kg) 49.2 46.7 49.2 46.7
Conversion Efficiency (LHV of H2) (68%) (71%) (68%) (71%)

Table A-7: Hydrogen cost, capital cost, energy efficiency and electricity price for current and future, distributed
production facilities using alkaline electrolysis systems [1].
2011 2015 2020
Characteristics Units
Status Target Target
Hydrogen Levelized Cost (Production Only) $/kg 4.20 3.90 2.30
$/kg 0.70 0.50 0.50
Electrolyzer System Capital Cost
$/kW 430 300 300
% (LHV) 67 72 75
System Energy Efficiency
kWh/kg 50 46 44
% (LHV) 74 76 77
Stack Energy Efficiency
kWh/kg 45 44 43
Electricity Price $/kWh 0.063 0.070 0.037
146 Appendix A

Table A-8: Cost contributions to the hydrogen production cost for current and future distributed production
facilities (1500 kgH2/day) using alkaline electrolysis technology. Costs are expressed on a $/kg H2 basis [1].
2011 2015 2020
Characteristics Units
Status Target Target
Cost Contribution $/kg H2 0.70 0.50 0.50
Electrolysis System
Production Equipment Availability % 98 98 98
Electricity Cost Contribution $/kg H2 3.00 3.10 1.60
Production Fixed O&M Cost Contribution $/kg H2 0.30 0.20 0.20
Production Other
Cost Contribution $/kg H2 0.10 0.10 <0.10
Variable Costs
Hydrogen Production Cost Contribution $/kg H2 4.10 3.90 2.30
Compression, Storage,
Cost Contribution $/kg H2 2.50 1.70 1.70
and Dispensing
Total Hydrogen Levelized Cost (Dispensed) $/kg H2 6.60 5.60 4.00

Table A-9: Hydrogen cost, capital cost, energy efficiency and electricity price for current and future, central
production facilities (52 300 kgH2/day) using alkaline electrolysis systems [1].
2011 2015 2020
Characteristics Units
Status Target Target
Hydrogen Levelized Cost (Plant Gate) $/kg H2 4.10 3.00 2.00
Total Capital Investment $M 68 51 40
% 67 73 75
System Energy Efficiency
kWh/kg H2 50 46 44.7
% 74 76 78
Stack Energy Efficiency
kWh/kg H2 45 44 43
Electricity Price $/kWh 0.063 0.049 0.031

Table A-10: Cost contributions to the hydrogen production cost for current and future central production
facilities (52 300 kgH2/day) using alkaline electrolysis technology. Costs are expressed on a $/kg H2 basis [1].
2011 2015 2020
Characteristics Units
Status Target Target
Capital Cost Contribution $/kg 0.60 0.50 0.40
Feedstock Cost Contribution $/kg 3.20 2.30 1.40
Fixed O&M Cost Contribution $/kg 0.20 0.10 0.10
Other Variable Cost Contribution $/kg 0.10 0.10 0.10
Total Hydrogen Levelized Cost (Plant Gate) $/kg 4.10 3.20 2.00

Table A-11: Estimated energy required to remove and recover CO 2 from coal-fired power plants using various
technologies.[3]
Process CO2 removal efficiency (%) kWh(e)/lb CO2 recovered
Improved amine absorption/ stripping
90 0.11
integrated plant
Oxygen/coal-fired power plant 100 0.15
Amine (MEA) absorption/ stripping
90 0.27
nonintegrated plant
Potassium carbonate absorption/ stripping 90 0.32
Molecular sieves adsorption/ stripping 90 0.40
Refrigeration 90 0.40
Seawater absorption 90 0.80
Membrane separation 90 0.36
Appendix A 147

Table A-12: Mass balance for the Aspen simulations performed by Mignard et al. [4]. Values are given in tons
per day. aAdditional emissions from process steam production in case no waste heat is available. bAdditional CO2
abatement in case the oxygen produced can be valorized.

In (t/day) Out (t/day)


Compound A B C D A B C D
MEA 0.98 0.47 0.53 0.47 0.98 0.47 0.53 0.47
Water 633 304 345 301 211 101 114 99.3
CO2 517 246 281 243 95.3 45.8 33 0
(+377)a (+179)a (+209)a (+188)a
(−167)b (−79)b (−102)b (−98)b
O2 555 267 302 264
Methanol 372 178 201 175

Table A-13: Energy balance for the Aspen simulations performed by Mignard et al. [4]. Values are given in tons
per day.

In (t/day) Out (t/day)

Utility A B C D A B C D
Cooling water 'In' (t/h) 3425 1750 1888 1738 85.8 41.3 46.8 40.8
Heat (≥120°C) (MW) 80.8 38.5 43.6 38
Fossil fuel electricity (MW) 15.7*8h 7.49*8h 5.67*8h 0
500*4h
Renewable electricity (MW) 100*16h 100*16h 100*16h
100*12h

Table A-14: Energy efficiency for the Aspen simulations performed by Mignard et al. [4].

Option A B C D
Fossil fuel electricity used, (kWh/kmol methanol) 10.8 10.8 7.2 0
Specific electricity use from renewable source, (kWh/kmol
274.5 288.7 255.2 292.8
methanol)
Conversion efficiency to chemical energy (%) (waste heat
61.9 59.1 67.5 58.4
scenario) (kWth/kWel)
Conversion efficiency to chemical energy (%) (process steam
54.1 52 58.3 51.4
scenario) (kWth/kWel)

Table A-15: Mass balance for the Aspen simulations performed by Van-Dal et al. [5]. Values are expressed in
tons per hour.

Compound In (t/h) Out (t/h)


CO2 88 6.6
H2O 108 41.4
Methanol 0 59.3
O2 7.1 96
MEA 0.09 0.09
148 Appendix A

Table A-16: Energy balance for the Aspen simulations performed by Van-Dal et al. [5]. Values are expressed in
tons per hour.
Unit Operation Amount
Water electrolysis power to electrolysis 645.1 MWel
CO2 capture compressing flue gas 3.9 MWel
steam to regeneration column 77.7 MWel
Methanol synthesis compressing fresh feed 15.9 MWel
compressing gas recycle 2.9 MWel
steam to methanol distillation 0 MWel
steam generated 35.7 MWel
electricity generated -1.9 MWel

Table A-17: CO2 balance for the Aspen simulations performed by Van-Dal et al. [5]. Values are expressed in tons
per hour.

In/Out Without sale of O2 (t/h) With sale of O2 (t/h)


CO2 fed to the process -88 -88
CO2 rejected by methanol synthesis 6.6 6.6
Electricity consumption 12.2 12.2
Carbon credits generated 0 -34.6
CO2 balance -69.2 -103.7

Table A-18: Energy production cost from multiple sources for 2014. Values obtained from the Annual Energy
Outlook for 2014 provided by the Energy Information Agency.[6]
Cost
Power Plant Type
$/kWh
Coal $0.10-0.14
Natural Gas $0.07-0.13
Nuclear $0.10
Wind $0.08-0.20
Solar PV $0.13
Solar Thermal $0.24
Geothermal $0.05
Biomass $0.10
Hydro $0.08
Appendix A 149

Table A-19: Price summary for energy (2013 dollars per unit) and prospects up to 2040. 1 barrel has an
approximate energy capacity of 1628 kWh, 106 Btu is equal to 293 kWh and 1 ton of coal equivalent is equal to
8141 kWh. [7]
Annual
2012 2013 2020 2025 2030 2035 2040 growth
(%)
Crude oil ($/barrel) 113 109 79 91 106 122 141 1.0%
Natural gas ($/mBtu) 2.79 3.73 4.88 5.46 5.69 6.6 7.85 2.8%
Coal ($/ton) 2.63 2.50 2.54 2.71 2.84 2.96 3.09 0.8%
Electricity (c/kWh) 10.0 10.1 10.5 11.0 11.1 11.3 11.8 0.6%

Figure A-1 Aspen simulation flowsheet of the MTO process. The mass balance is given in Table A-20 [8].
150 Appendix A

Table A-20 Stream data for the methanol to olefins (MTO) process [8].
MET METH METH ETH- PRO- ETH PROP PROD WAT PROD
Stream
H-IN -ETH -PRO RXN RXN ANE ANE -MIX ER -QUE
Molar Flow of
Components (kg/h)
17785 177859 177859 12094 74701 195645 54940. 140705
Methanol 0 0
9.872 .87 .872 4.7 .15 .86 448 .41
32014 58026 90041. 60311. 29730.
Water 0 0 0 0 0
.78 .78 562 412 15
24900 24900. 2.7446 24897.
Ethylene 0 0 0 0 0 0
.38 383 776 636
45131 45131. 8.1304 45123.
Propylene 0 0 0 0 0 0
.94 94 398 792
0.2015 1841.1
Propane 0 0 0 0 0 0 1841.4 1841.4
9436 985
1046. 0.1118 1046.2
Ethane 0 0 0 0 0 0 1046.4
4 994 881
17785 177859 177859 17785 17785 1046. 358607 11526 243344
Total Flow, kg/h 1841.4
9.872 .87 .872 9.9 9.9 4 .54 3.049 .47
Temperature, K 723.2 723.2 723.2 723.2 723.2 723.2 723.2 723.2 343.2 343.2
Pressure, atm 1.97 1.97 1.97 1.97 1.97 1.97 1.97 1.97 0.99 0.99

Table A-21 Continued


PROD PROD DEET DEET PROD- SPL- PROD- SPL-
Stream S9
-C1 -H1 H-TO H-BT ETY ETHA PRO PRPA
Molar Flow of
Components (kg/h)
14070 140705 8.34E- 140705. 14070 0.78933 140704
Methanol 0 0
5.4 .408 14 4 5.41 8 .6
29730. 29730. 5.03E- 29730.1 29730. 9.4261E 5.026E- 3.20813 29726.
Water
15 15 10 5 15 -38 10 3 95
24897. 24897. 24897.6 1.24E- 1.244 24750.1 147.521 1.24E- 3.94E-
Ethylene
64 6364 4 07 E-07 156 3 07 28
45123. 45123. 38.3660 45085.4 45085. 2.862E- 38.3660 43844.1 1241.2
Propylene
79 792 3 5 446 11 26 8 84
1841.1 1841.1 1840.80 1840.8 4.829E- 0.39236 1579.09 261.71
Propane 0.39236
98 9848 6 06 15 03 3 3
1046.2 1046.2 1046.28 1.92E- 1.917 182.018 864.269 1.92E- 6.07E-
Ethane
88 881 8 06 E-06 46 7 06 23
24334 243344 25982.6 217361. 21736 24932.1 1050.54 45427.2 171934
Total Flow, kg/h
4.5 .473 8 8 1.81 341 94 7 .6
Temperature, K 561.3 283.2 227.8 275.3 223.2 226.9 243.5 299.5 416.1
Pressure, atm 11.84 11.84 11.84 11.84 11.84 11.84 11.84 0.99 11.84
Appendix A 151

A.1 References

1. Multi-Year Research, Development, and Demonstration Plan, U.S.D.o. Energy, Editor. 2015. p. 1-
44.

2. Ainscough, C.P., D.; Miller, E., Hydrogen Production Cost From PEM Electrolysis, D.o. Energy,
Editor. 2014: United States of America. p. 11.

3. Olah, G.A., G.K.S. Prakash, and A. Goeppert, Anthropogenic Chemical Carbon Cycle for a
Sustainable Future. Journal of the American Chemical Society, 2011. 133(33): p. 12881-12898.

4. Mignard, D., et al., Methanol synthesis from flue-gas CO2 and renewable electricity: a feasibility
study. International Journal of Hydrogen Energy, 2003. 28(4): p. 455-464.

5. Van-Dal, É.S. and C. Bouallou, CO2 abatement through a methanol production process. Chemical
Engineering Transactions, 2012. 29: p. 463-468.

6. Cost, L., Levelized Avoided Cost of New Generation Resources in the Annual Energy Outlook 2014.
Energy Information Agency, 2014.

7. Annual Energy Outlook 2015 with projections to 2040, U.S.E.I. Administration, Editor. 2015, U.S.
Department of Energy: Washington.

8. Jasper, S. and M.M. El-Halwagi, A Techno-Economic Comparison between Two Methanol-to-


Propylene Processes. Processes, 2015. 3(3): p. 684-698.
152

Appendix B
Ab-initio calculations
Table B-1 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch conditions
(500 K, 20 bar, 60% conversion) for reaction intermediates on a clean Co(111) surface. Accompanying structures
are illustrated in Table B-2.
Reaction Adsorption energy Entropy of adsorbed species Gibbs free adsorption energy
intermediate (kJ/mol) (J/mol K) (kJ/mol)
C* (hcp) -625 18 53
C* (fcc) -625 18 54
O* (top) -407 58 103
O* (fcc) -542 24 -14
O* (hcp) -543 24 -16
CO* (top) -121 75 -55
CO* (bridge) -108 87 -48
CO* (hollow) -111 82 -49
CO2* (top) -15 127 50
CO2* (bridge) -16 120 53
CO2* (hollow) -15 134 46
H* (hcp) -258 7 -1
H* (fcc) -259 7 -2
OH* (top) -316 55 56
OH* (fcc) -374 41 5
OH* (hcp) -373 39 7
HCO* -170 64 68
COOH* -203 97 121
HCOO* -298 111 37
HCOOH* -43 124 115
Appendix B 153

Table B-2 Optimized configurations for reaction intermediates on a clean Co(111) surface presented in Table
B-1.
C* (hcp) C* (fcc) O* (hcp) O* (fcc)

H* (hcp) H* (fcc) OH* (top) OH* (hcp)

OH* (fcc) CO* (top) CO* (bridge) CO* (hollow)

CO2* (top) CO2* (bridge) CO2* (hollow) HCO*

COOH* HCOO* HCOOH*


154 Appendix B

Table B-3 Transition state geometries for the reactions presented in Table 4-2 and Figure 4-5 on clean Co(111).
Initial State Transition State Final State

CO* + *  C* + O*

CO2* + *  CO* + O*

CO2* + H*  COOH* + *
Appendix B 155

CO2* + H*  HCOO* + *

COOH* + H*  HCOOH* + *

COOH* + *  CO* + OH*


156 Appendix B

HCOO* + H*  HCOOH* + *

HCOO* + *  HCO* + O*

HCOOH* + *  HCO* + OH*


Appendix B 157

Table B-4 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch conditions
(500 K, 20 bar, 60% conversion) for reaction intermediates on a Co(111) surface with coverage of 1/3 ML.
Accompanying structures are illustrated in Table B-5.
Reaction Adsorption energy Entropy of adsorbed species Gibbs free adsorption energy
intermediate (kJ/mol) (J/mol K) (kJ/mol)

CO* (top) -122 71 -53

CO2* (top) -23 125 42

CO2* (bridge) -23 142 34

O* (fcc) -532 25 -5

Table B-5 Optimized configurations for reaction intermediates on Co(111) surface with coverage of 1/3 ML
presented in Table B-4.
O* (fcc) CO* (top) CO2* (top) CO2* (bridge)

Table B-6 Transition state geometry for CO2* dissociation on Co(111) with coverage of 1/3 ML, presented in
Table 4-4.
Initial State Transition State Final State

CO2* + *  CO* + O*
158 Appendix B

Table B-7 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch conditions
(500 K, 20 bar, 60% conversion) for reaction intermediates on a clean Co(211) surface. Accompanying structures
are illustrated in Table B-8.
Reaction Adsorption energy Entropy of adsorbed species Gibbs free adsorption energy
intermediate (kJ/mol) (J/mol K) (kJ/mol)

CO* (top) -204 67 -132

CO* (hollow) -169 80 -106

CO2* (top) -99 102 -27

CO2* (bridge) 272 122 339

CO2* (hollow) -49 86 29

O* (bridge) -200 20 330

O* (hollow) -587 27 -62

Table B-8 Optimized configurations for reaction intermediates on a clean Co(211) surface presented in Table
B-7.
O* (bridge) O* (hollow) CO* (top) CO* (hollow)

CO2* (top) CO2* (bridge) CO2* (hollow)


Appendix B 159

Table B-9 Transition state geometry for CO2* dissociation on a clean Co(211) surface presented in Table 4-6.
Initial State Transition State Final State

CO2* + *  CO* + O*
160 Appendix B

Table B-10 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch conditions
(500 K, 20 bar, 60% conversion) for reaction intermediates on a clean Cu(111) surface.
Reaction Adsorption energy Entropy of adsorbed species Gibbs free adsorption energy
intermediate (kJ/mol) (J/mol K) (kJ/mol)

C* (fcc) -414 21 246

O* (fcc) -455 24 73

CO* (top) -32 36 73

CO* (hollow) -27 38 76

CO2* (top) -15 64 100

CO2* (bridge) -15 55 117

CO2* (hollow) -15 53 120

H* (hcp) -234 24 6

H* (fcc) -235 23 6

OH* (fcc) -329 44 48

OH* (hcp) -328 47 48

COOH* -145 103 175

HCOO* -261 91 85
Appendix B 161

Table B-11 Optimized configurations for reaction intermediates on a clean Cu(111) surface presented in Table
B-10.
C* (fcc) O* (fcc) CO* (top) CO* (hollow)

CO2* (top) CO2* (bridge) CO2* (hollow) H* (hcp)

H* (fcc) OH* (fcc) OH* (hcp) COOH*

HCOO*
162 Appendix B

Table B-12 Transition state geometries for the reactions presented in Table 4-8 on clean Cu(111).
Initial State Transition State Final State

CO* + *  C* + O*

CO2* + *  CO* + O*

Table B-13 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch conditions
(500 K, 20 bar, 60% conversion) for reaction intermediates on a Cu(111) surface with coverage of 1/3 ML.
(Extension to Table 4-9.)
Reaction Adsorption energy Entropy of adsorbed species Gibbs free adsorption energy
intermediate (kJ/mol) (J/mol K) (kJ/mol)

CO* (top) -34 100 20

CO2* (top) -28 127 36

CO2* (bridge) -28 99 51

O* (fcc) -467 26 59
Appendix B 163

Table B-14 Optimized configurations for reaction intermediates on Cu(111) surface with CO coverage of 1/3 ML
presented in Table B-13.
O* (fcc) CO* (top) CO2* (top) CO2* (bridge)

Table B-15 Transition state geometry for CO2* dissociation on Cu(111) with CO coverage of 1/3 ML, presented
in Table 4-10.
Initial State Transition State Final State

CO2* + *  CO* + O*
164 Appendix B

Table B-16 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch conditions
(500 K, 20 bar, 60% conversion) for reaction intermediates on a clean Ni(111) surface.
Reaction Adsorption energy Entropy of adsorbed species Gibbs free adsorption energy
intermediate (kJ/mol) (J/mol K) (kJ/mol)

C* (fcc) -590 17 72

C* (hcp) -596 17 66

O* (top) -348 34 174

O* (fcc) -514 23 14

CO* (top) -113 -43 69

CO* (bridge) -117 -44 61

CO* (hollow) -122 -54 70

CO2* (bridge) -16 50 125

CO2* (hollow) -16 45 135

H* (hcp) -259 6 -3

H* (fcc) -262 7 -5

OH* (fcc) -352 39 28

OH* (hcp) -342 40 37


Appendix B 165

Table B-17 Optimized configurations for reaction intermediates on a clean Ni(111) surface presented in Table
B-16.
C* (fcc) C*(hcp) O* (top) O*(fcc)

CO* (top) CO* (bridge) CO* (hollow) CO2* (bridge)

CO2* (hollow) H* (hcp) H* (fcc) OH* (fcc)

OH* (hcp)
166 Appendix B

Table B-18 Transition state geometries for the reactions presented in Table 4-12 on clean Ni(111).
Initial State Transition State Final State

CO* + *  C* + O*

CO2* + *  CO* + O*

Table B-19 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch conditions
(500 K, 20 bar, 60% conversion) for reaction intermediates on a Ni(111) surface with coverage of 1/3 ML.
(Extension to Table 4-13.)
Reaction Adsorption energy Entropy of adsorbed species Gibbs free adsorption energy
intermediate (kJ/mol) (J/mol K) (kJ/mol)

CO* (top) -115 79 -51

CO2* (top) -24 141 33

CO2* (bridge) -24 125 41

O* (fcc) -508 23 20
Appendix B 167

Table B-20 Optimized configurations for reaction intermediates on Ni(111) surface with coverage of 1/3 ML
presented in Table B-19.
O* (fcc) CO* (top) CO2* (top) CO2* (bridge)

Table B-21 Transition state geometry for CO2* dissociation on Ni(111) with coverage of 1/3 ML, presented in
Table 4-14.
Initial State Transition State Final State

CO2* + *  CO* + O*
168 Appendix B

Table B-22 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch conditions
(500 K, 20 bar, 60% conversion) for reaction intermediates on a clean Au(111) surface.
Reaction Adsorption energy Entropy of adsorbed species Gibbs free adsorption energy
intermediate (kJ/mol) (J/mol K) (kJ/mol)

C* (fcc) -417 243 21

O* (fcc) -315 28 204

O* (hcp) -321 29 209

CO* (top) -12 119 33

CO* (bridge) -2 68 68

CO* (hollow) -3 93 55

CO2* (top) -16 124 50

CO2* (bridge) -16 137 43

CO2* (hollow) -16 138 43

H* (hcp) -203 10 50

H* (fcc) -204 10 49

OH* (fcc) -754 45 134

OH* (hcp) -732 51 152


Appendix B 169

Table B-23 Optimized configurations for reaction intermediates on a clean Au(111) surface presented in Table
B-22.
C* (fcc) O* (fcc) O*(hcp) CO* (top)

CO* (bridge) CO* (hollow) CO2* (top) CO2* (bridge)

CO2* (hollow) H* (hcp) H* (fcc) OH* (fcc)

OH* (hcp)
170 Appendix B

Table B-24 Transition state geometries for the reactions presented in Table 4-16 on clean Au(111).
Initial State Transition State Final State

CO* + *  C* + O*

CO2* + *  CO* + O*

Table B-25 Adsorption energies, entropies, and Gibbs free adsorption energies under Fischer-Tropsch conditions
(500 K, 20 bar, 60% conversion) for reaction intermediates on a Au(111) surface with coverage of 1/3 ML.
Reaction Adsorption energy Entropy of adsorbed species Gibbs free adsorption energy
intermediate (kJ/mol) (J/mol K) (kJ/mol)

CO* (top) -15 117 31

CO2* (top) -21 137 38

CO2* (bridge) -27 120 41

O* (fcc) -327 29 198


Appendix B 171

Table B-26 Optimized configurations for reaction intermediates on Au(111) surface with coverage of 1/3 ML
presented in Table B-25.
O* (fcc) CO* (top) CO2* (top) CO2* (bridge)

Table B-27 Transition state geometry for CO2* dissociation on Au(111) with coverage of 1/3 ML, presented in
Table 4-18.
Initial State Transition State Final State

CO2* + *  CO* + O*
172

Appendix C
Kinetic modeling
Table C-1 Summary of the reactions taken in account for chain propagation in FTS on the catalyst surface.
Distinction is made between the CO insertion and carbide mechanism. Values for the CO insertion mechanism
are provided by the coach of this thesis, values for the carbide mechanism were found in literature [1-4].
Forward Reverse
Forward pre- Reverse pre-
activation activation
Reactions exponential exponential
energy Eaf energy Ear
factor Af (s-1) factor Ar (s-1)
(kJ/mol) (kJ/mol)
Alfa hydride addition
C* + H* <=> CH* + * 4.72E+12 71.01 2.30E+14 108.34
CH* + H* <=> CH2* + * 9.53E+12 54.28 1.14E+13 12.93
CH2* + H* <=> CH3* + * 4.24E+12 47.25 3.60E+14 62.52
CH3* + H* <=> CH4* + * 1.07E+13 79.26 1.02E+13 2.26
C1 product desorption
CH4* <=> CH4(g) + * 9.22E+16 13.82 3.14E+08 0.00
CH2O* <=> CH2O(g) + * 9.22E+16 35.87 3.14E+08 0.00
C-C coupling (CO insertion)
CH* + CO* <=> COCH* + * 1.00E+13 96.00 1.00E+13 100.00
CH2* + CO* <=> COCH2* + * 1.00E+13 64.00 1.00E+13 84.82
CH3* + CO* <=> COCH3* + * 1.00E+13 185.00 1.00E+13 101.00
Appendix C 173

Hydrogenation reactions (CO insertion)


COCH* + H* <=> COCH2* + * 1.00E+13 105.00 1.00E+13 106.00
COCH* + H* <=> CHOCH* + * 1.00E+13 139.00 1.00E+13 97.00
CHOCH* + H* <=> CHOCH2* +* 1.00E+13 53.00 1.00E+13 56.00
COCH2* + H* <=> COCH3* + * 1.00E+13 75.00 1.00E+13 77.00
COCH2* + H* <=> CHOCH2* + * 1.00E+13 94.00 1.00E+13 95.00
CHOCH2* + H* <=> CHOCH3* + * 1.00E+13 116.00 1.00E+13 74.00
COCH3* + H* <=> CHOCH3* + * 4.46E+12 73.00 1.94E+13 14.70
COCH3* + H* <=> COHCH3* + * 4.46E+12 103.60 3.08E+12 69.90
COCH3* + OH* <=> COHCH3* + O* 1.85E+10 23.43 6.54E+10 13.53
COHCH3* + H* <=> CHOHCH3* + * 1.53E+14 95.17 1.27E+14 51.64
CHOHCH3* + H* <=> CH2OHCH3* + * 1.00E+13 70.00 1.00E+13 70.00
C-O scission reactions
COCH* + * <=> CCH* + O* 1.00E+13 180.00 1.00E+13 215.00
COCH2* + * <=> CCH2* + O* 1.00E+13 95.00 1.00E+13 167.00
COCH3* + * <=> CCH3* + O* 1.78E+13 51.49 6.50E+13 123.54
COHCH3* + * <=> CCH3* + OH* 6.72E+12 38.59 4.84E+12 118.02
C-C coupling (carbide mechanism)
2 C* <=> CC* + * 1.00E+13 138.00 1.00E+13 144.00
C* + CH* <=> CCH* + * 1.00E+13 129.00 1.00E+13 75.00
C* + CH3* <=> CCH3* + * 1.00E+13 92.00 1.00E+13 116.00
2 CH* <=> CHCH* + * 1.00E+13 86.00 1.00E+13 95.00
2 CH2* <=> CH2CH2* + * 1.00E+13 54.00 1.00E+13 70.00
Hydrogenation reactions (carbide
mechanism)
CC* + H* <=> CCH* + * 1.00E+13 104.00 1.00E+13 72.00
CCH* + H* <=> CHCH* + * 1.00E+13 66.00 1.00E+13 91.00
CCH* + H* <=> CCH2* + * 1.00E+13 79.00 1.00E+13 67.00
CHCH* + H* <=> CHCH2* + * 1.00E+13 110.00 1.00E+13 29.00
CCH2* + H* <=> CCH3* + * 1.00E+13 61.00 1.00E+13 69.00
CCH2* + H* <=> CHCH2* + * 1.00E+13 66.00 1.00E+13 22.00
CHCH2* + H* <=> CHCH3* + * 1.00E+13 53.00 1.00E+13 42.00
CCH3* + H* <=> CHCH3* + * 1.00E+13 73.00 1.00E+13 10.00
CHCH2* + H* <=> CH2CH2* + * 1.00E+13 45.00 1.00E+13 42.00
CHCH3* + H* <=> CH2CH3* + * 1.00E+13 19.00 1.00E+13 23.00
CH2CH2* + H* <=> CH2CH3* + * 1.00E+13 58.00 1.00E+13 34.00
CH2CH3* + H* <=> CH3CH3* + * 1.00E+13 112.00 1.00E+13 71.00
C2 product desorption
CH3CH3* <=> C2H6(g) + * 1.00E+13 30.00 1.00E+03 0.00
CH2CH2* <=> C2H4(g) + * 1.00E+13 40.00 1.00E+03 0.00
CHOCH3* <=> CH3HCO(g) + * 1.00E+13 41.00 1.00E+03 0.00
CH2OHCH3* <=> C2H5OH(g) + * 1.00E+13 40.15 1.00E+03 0.00
174 Appendix C

C.1 References

1. Zhuo, M., A. Borgna, and M. Saeys, Effect of the CO coverage on the Fischer–Tropsch synthesis
mechanism on cobalt catalysts. Journal of Catalysis, 2013. 297: p. 217-226.

2. Zhuo, M., et al., Density functional theory study of the CO insertion mechanism for Fischer−
Tropsch synthesis over Co catalysts. The Journal of Physical Chemistry C, 2009. 113(19): p. 8357-
8365.

3. Gunasooriya, G.K.K., et al., Key role of surface hydroxyl groups in CO activation during Fischer-
Tropsch synthesis. ACS Catalysis, 2016.

4. Filot, I.A.W., R.A. van Santen, and E.J.M. Hensen, The Optimally Performing Fischer–Tropsch
Catalyst. Angewandte Chemie International Edition, 2014. 53(47): p. 12746-12750.
175

Appendix D
List of simulations
Table D-1 List of simulations for adsorption on the cobalt catalyst surface
Component Configuration Eads [kJ/mol] Location
Co slab vasp/Cobalt/Slab
CO top -121 vasp/Cobalt/CO/
CO hollow -108 vasp/Cobalt/CO/
CO2 top -111 vasp/Cobalt/CO2/
CO2 top perp. -15 vasp/Cobalt/CO2/
CO2 bridge -8 vasp/Cobalt/CO2/
CO2 hollow -16 vasp/Cobalt/CO2/
CO2 hollow perp. -15 vasp/Cobalt/CO2/
CO2 (0.3 ML CO) 1 -9 vasp/Cobalt/Coverage/
CO2 (0.3 ML CO) 2 -23 vasp/Cobalt/Coverage/
CO2 (Co 211) 1 -23 vasp/Cobalt/211/CO2/
CO2 (Co 211) 2 -99 vasp/Cobalt/211/CO2/
CO2 (Co 211) 3 272 vasp/Cobalt/211/CO2/
H2 fcc perp. -49 vasp/Cobalt/H2/
H2 bridge -4 vasp/Cobalt/H2/
H fcc -4 vasp/Cobalt/H/
H hcp -259 vasp/Cobalt/H/
O top -258 vasp/Cobalt/O/
O hollow -407 vasp/Cobalt/O/
C top -543 vasp/Cobalt/C/
C bridge -542 vasp/Cobalt/C/
C hollow -608 vasp/Cobalt/C/
OH top -625 vasp/Cobalt/OH/
OH hollow -625 vasp/Cobalt/OH/
OH hollow -316 vasp/Cobalt/OH/
COOH bridge (1) -373 vasp/Cobalt/COOH/
COOH bridge (2) -373 vasp/Cobalt/COOH/
COOH bridge (3) -201 vasp/Cobalt/COOH/
HCOO SB Bid -184 vasp/Cobalt/HCOO/
176 Appendix D

HCOO SB Bid rot -203 vasp/Cobalt/HCOO/


HCO 1 -298 vasp/Cobalt/HCO/
HCOOH 1 -298 vasp/Cobalt/HCOOH/
CO (Co 211) top -170 vasp/Cobalt/211/CO/
CO (Co 211) bridge -43 vasp/Cobalt/211/CO/
CO (Co 211) hollow -204 vasp/Cobalt/211/CO/
O (Co 211) bridge 148 vasp/Cobalt/211/O/
O (Co 211) hollow -169 vasp/Cobalt/211/O/

Table D-2 List of transition state calculations for the cobalt catalyst surface.
Reactions Ea [kJ/mol] Location
CO* + * --> C* + O* 243 vasp/Cobalt/CO/Transition State/
CO2* + * --> CO* + O* 60 vasp/Cobalt/CO2/Transition State/
CO2* + H* --> COOH* + * 154 vasp/Cobalt/COOH/Transition State/CO2 hydrogenation/
CO2* + H* --> HCOO* + * 94 vasp/Cobalt/HCOO/Transition State/CO2 hydrogenation/
CO2* + * --> CO* + O* (0.3 ML CO) 81 vasp/Cobalt/Coverage/Transition State/
CO2* + * --> CO* + O* (Co 211) 54 vasp/Cobalt/211/CO2/Transition State/
COOH* + * --> CO* + OH* 63 vasp/Cobalt/COOH/Transition State/COOH dissociation/
COOH* + H* --> HCOOH* + * 61 vasp/Cobalt/COOH/Transition State/COOH hydrogenation/
HCOO* + * --> HCO* + O* 107 vasp/Cobalt/HCOO/Transition State/HCOO dissociation/
HCOO* + H* --> HCOOH* + * 110 vasp/Cobalt/HCOO/Transition State/HCOO hydrogenation/
HCOOH* + * --> HCO* + OH* 186 vasp/Cobalt/HCOOH/Transition State/

Table D-3 List of simulations for adsorption on the copper catalyst surface.
Component Configuration Eads [kJ/mol] Location
Cu slab vasp/Copper/Slab
CO top -32 vasp/Copper/CO/
CO hollow -27 vasp/Copper/CO/
CO2 top -15 vasp/Copper/CO2/
CO2 bridge -15 vasp/Copper/CO2/
CO2 hollow -15 vasp/Copper/CO2/
CO2 (0.3 ML CO) 1 -28 vasp/Copper/Coverage/
CO2 (0.3 ML CO) 2 -28 vasp/Copper/Coverage/
H2 hcp/fcc -4 vasp/Copper/H2/
H2 bridge -4 vasp/Copper/H2/
H fcc -234 vasp/Copper/H/
H hcp -235 vasp/Copper/H/
O bridge -458 vasp/Copper/O/
O hollow -455 vasp/Copper/O/
C top -445 vasp/Copper/C/
C bridge -430 vasp/Copper/C/
C hollow -414 vasp/Copper/C/
COOH bridge 1 -145 vasp/Copper/COOH/
COOH bridge 2 -145 vasp/Copper/COOH/
COOH bridge 3 -143 vasp/Copper/COOH/
HCOO SB bid -261 vasp/Copper/HCOO/
HCOO SB bid rot -261 vasp/Copper/HCOO/
OH top -334 vasp/Copper/OH/
OH bridge -329 vasp/Copper/OH/
OH hollow -328 vasp/Copper/OH/
Appendix D 177

Table D-4 List of transition state calculations for the copper catalyst surface.
Reactions Ea [kJ/mol] Location
CO* + * --> C* + O* 546 vasp/Copper/CO/Transition State/
CO2* + * --> CO* + O* 159 vasp/Copper/CO2/Transition State/
CO2* + * --> CO* + O* (0.3 ML CO) 156 vasp/Copper/Coverage/Transition State/
CO2* + H* --> COOH* + * 197 vasp/Copper/COOH/Transition State/
CO2* + H* --> HCOO* + * 186 vasp/Copper/HCOO/Transition State/

Table D-5 List of simulations for adsorption on the nickel catalyst surface.

Component Configuration Eads [kJ/mol] Location


Ni slab vasp/Nickel/Slab/
CO top -113 vasp/Nickel/CO/
CO bridge -117 vasp/Nickel/CO/
CO hollow -122 vasp/Nickel/CO/
CO2 top 40 vasp/Nickel/CO2/
CO2 bridge -16 vasp/Nickel/CO2/
CO2 hollow -16 vasp/Nickel/CO2/
CO2 (0.3 ML CO) 1 -24 vasp/Nickel/Coverage/
CO2 (0.3 ML CO) 2 -24 vasp/Nickel/Coverage/
H2 hcp/fcc -4 vasp/Nickel/H2/
H2 bridge -4 vasp/Nickel/H2/
H fcc -262 vasp/Nickel/H/
H hcp -259 vasp/Nickel/H/
O top -348 vasp/Nickel/O/
O bridge -514 vasp/Nickel/O/
O hollow -514 vasp/Nickel/O/
C bridge -596 vasp/Nickel/C/
C hollow -590 vasp/Nickel/C/
OH top -352 vasp/Nickel/OH/
OH bridge -342 vasp/Nickel/OH/
OH hollow -342 vasp/Nickel/OH/

Table D-6 List of transition state calculations for the nickel catalyst surface.
Ea
Reactions Location
[kJ/mol]
CO* + * --> C* + O* 289 vasp/Nickel/CO/Transition State/
CO2* + * --> CO* + O* 79 vasp/Nickel/CO2/Transition State/
CO2* + * --> CO* + O* (0.3 ML CO) 95 vasp/Nickel/Coverage/Transition State/
178 Appendix D

Table D-7 List of simulations for adsorption on the gold catalyst surface.
Component Configuration Eads [kJ/mol] Location
Au slab vasp/Gold/Slab/
CO top -12 vasp/Gold/CO/
CO bridge -2 vasp/Gold/CO/
CO hollow -3 vasp/Gold/CO/
CO2 top -16 vasp/Gold/CO2/
CO2 bridge -16 vasp/Gold/CO2/
CO2 hollow -16 vasp/Gold/CO2/
CO2 (0.3 ML CO) 1 -21 vasp/Gold/Coverage/
CO2 (0.3 ML CO) 2 -27 vasp/Gold/Coverage/
H2 hcp/fcc -5 vasp/Gold/H2/
H2 bridge -4 vasp/Gold/H2/
H fcc -204 vasp/Gold/H/
H hcp -203 vasp/Gold/H/
O top -315 vasp/Gold/O/
O bridge -321 vasp/Gold/O/
O hollow -321 vasp/Gold/O/
C bridge -417 vasp/Gold/C/
C hollow -417 vasp/Gold/C/
OH top -754 vasp/Gold/OH/
OH bridge -754 vasp/Gold/OH/
OH hollow -732 vasp/Gold/OH/

Table D-8 List of transition state calculations for the gold catalyst surface.

Reactions Ea [kJ/mol] Location


CO* + * --> C* + O* 447 vasp/Gold/CO/Transition State/
CO2* + * --> CO* + O* 310 vasp/Gold/CO2/Transition State/
CO2* + * --> CO* + O* (0.3 ML CO) 314 vasp/Gold/Coverage/Transition State/

Table D-9 CD Rom summary.

Subject Location Lab journal


Vasp results Vasp/Results/ P 88
Vasp codes Vasp/Codes/ P 88
Chemkin files Chemkin/ P 89

You might also like