Quantum Mechanics Studies of Fuel Cell Catalysts and Proton Conducting Ceramics with Validation by Experiment - CaltechTHESIS
CaltechTHESIS
A Caltech Library Service
About
Browse
Deposit an Item
Instructions for Students
Quantum Mechanics Studies of Fuel Cell Catalysts and Proton Conducting Ceramics with Validation by Experiment
Citation
Tsai, Ho-Cheng
(2015)
Quantum Mechanics Studies of Fuel Cell Catalysts and Proton Conducting Ceramics with Validation by Experiment.
Dissertation (Ph.D.), California Institute of Technology.
doi:10.7907/Z9P55KFW.
Abstract
We carried out quantum mechanics (QM) studies aimed at improving the performance of hydrogen fuel cells. This led to predictions of improved materials, some of which were subsequently validated with experiments by our collaborators.
In part I, the challenge was to find a replacement for the Pt cathode that would lead to improved performance for the Oxygen Reduction Reaction (ORR) while remaining stable under operational conditions and decreasing cost. Our design strategy was to find an alloy with composition Pt3M that would lead to surface segregation such that the top layer would be pure Pt, with the second and subsequent layers richer in M. Under operating conditions we expect the surface to have significant O and/or OH chemisorbed on the surface, and hence we searched for M that would remain segregated under these conditions. Using QM we examined surface segregation for 28 Pt
M alloys, where M is a transition metal. We found that only Pt
Os and Pt
Ir showed significant surface segregation when O and OH are chemisorbed on the catalyst surfaces. This result indicates that Pt
Os and Pt3Ir favor formation of a Pt-skin surface layer structure that would resist the acidic electrolyte corrosion during fuel cell operation environments. We chose to focus on Os because the phase diagram for Pt-Ir indicated that Pt-Ir could not form a homogeneous alloy at lower temperature. To determine the performance for ORR, we used QM to examine all intermediates, reaction pathways, and reaction barriers involved in the processes for which protons from the anode reactions react with O
to form H
O. These QM calculations used our Poisson-Boltzmann implicit solvation model include the effects of the solvent (water with dielectric constant 78 with pH 7 at 298K). We found that the rate determination step (RDS) was the O
ad
hydration reaction (O
ad
+ H
ad
-> OH
ad
+ OH
ad
) in both cases, but that the barrier for pure Pt of 0.50 eV is reduced to 0.48 eV for Pt
Os, which at 80 degrees C would increase the rate by 218%. We collaborated with the Pu-Wei Wu’s group to carry out experiments, where we found that the dealloying process-treated Pt2Os catalyst showed two-fold higher activity at 25 degrees C than pure Pt and that the alloy had 272% improved stability, validating our theoretical predictions.
We also carried out similar QM studies followed by experimental validation for the Os/Pt core-shell catalyst fabricated by the underpotential deposition (UPD) method. The QM results indicated that the RDS for ORR is a compromise between the OOH formation step (0.37 eV for Pt, 0.23 eV for Pt
2ML
/Os core-shell) and H
O formation steps (0.32 eV for Pt, 0.22 eV for Pt
2ML
/Os core-shell). We found that Pt
2ML
/Os has the highest activity (compared to pure Pt and to the Pt
Os alloy) because the 0.37 eV barrier decreases to 0.23 eV. To understand what aspects of the core shell structure lead to this improved performance, we considered the effect on ORR of compressing the alloy slab to the dimensions of pure Pt. However this had little effect, with the same RDS barrier 0.37 eV. This shows that the ligand effect (the electronic structure modification resulting from the Os substrate) plays a more important role than the strain effect, and is responsible for the improved activity of the core- shell catalyst. Experimental materials characterization proves the core-shell feature of our catalyst. The electrochemical experiment for Pt
2ML
/Os/C showed 3.5 to 5 times better ORR activity at 0.9V (vs. NHE) in 0.1M HClO
solution at 25 degrees C as compared to those of commercially available Pt/C. The excellent correlation between experimental half potential and the OH binding energies and RDS barriers validate the feasibility of predicting catalyst activity using QM calculation and a simple Langmuir–Hinshelwood model.
In part II, we used QM calculations to study methane stream reforming on a Ni-alloy catalyst surfaces for solid oxide fuel cell (SOFC) application. SOFC has wide fuel adaptability but the coking and sulfur poisoning will reduce its stability. Experimental results suggested that the Ni4Fe alloy improves both its activity and stability compared to pure Ni. To understand the atomistic origin of this, we carried out QM calculations on surface segregation and found that the most stable configuration for Ni
Fe has a Fe atom distribution of (0%, 50%, 25%, 25%, 0%) starting at the bottom layer. We calculated that the binding of C atoms on the Ni4Fe surface is 142.9 Kcal/mol, which is about 10 Kcal/mol weaker compared to the pure Ni surface. This weaker C binding energy is expected to make coke formation less favorable, explaining why Ni
Fe has better coking resistance. This result confirms the experimental observation. The reaction energy barriers for CHx decomposition and C binding on various alloy surface, Ni
X (X=Fe, Co, Mn, and Mo), showed Ni
Fe, Ni
Co, and Fe
Mn all have better coking resistance than pure Ni, but that only Ni
Fe and Fe
Mn have (slightly) improved activity compared to pure Ni.
In part III, we used QM to examine the proton transport in doped perovskite-ceramics. Here we used a 2x2x2 supercell of perovskite with composition Ba
(OH)
23
where X=Ce or Zr and M=Y, Gd, or Dy. Thus in each case a 4
X is replace by a 3
M plus a proton on one O. Here we predicted the barriers for proton diffusion allowing both includes intra-octahedron and inter-octahedra proton transfer. Without any restriction, we only observed the inter-octahedra proton transfer with similar energy barrier as previous computational work but 0.2 eV higher than experimental result for Y doped zirconate. For one restriction in our calculations is that the O
donor
-O
acceptor
atoms were kept at fixed distances, we found that the barrier difference between cerates/zirconates with various dopants are only 0.02~0.03 eV. To fully address performance one would need to examine proton transfer at grain boundaries, which will require larger scale ReaxFF reactive dynamics for systems with millions of atoms. The QM calculations used here will be used to train the ReaxFF force field.
Item Type:
Thesis (Dissertation (Ph.D.))
Subject Keywords:
Quantum Mechanics, Pt3Os Catalyst, DFT, Segregation, ORR, PtOs, Core-Shell Catalysts, UPD, Electrocatalysis, Dealloying Process, SOFC, Stream Reforming, Ni4Fe Catalyst, Coking, Proton Conducting Ceramics, Perovskite
Degree Grantor:
California Institute of Technology
Division:
Engineering and Applied Science
Major Option:
Materials Science
Minor Option:
Physics
Thesis Availability:
Public (worldwide access)
Research Advisor(s):
Goddard, William A., III
Thesis Committee:
Goddard, William A., III (chair)
Gray, Harry B.
Johnson, William Lewis
Haile, Sossina M.
Defense Date:
20 April 2015
Non-Caltech Author Email:
hctsai1103 (AT) gmail.com
Record Number:
CaltechTHESIS:06012015-203513094
Persistent URL:
DOI:
10.7907/Z9P55KFW
Default Usage Policy:
No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:
8962
Collection:
CaltechTHESIS
Deposited By:
Ho Cheng Tsai
Deposited On:
09 Jun 2015 18:23
Last Modified:
04 Oct 2019 00:08
Thesis Files
Preview
PDF
- Final Version
See Usage Policy.
6MB
Repository Staff Only:
item control page
CaltechTHESIS is powered by
EPrints 3.3
which is developed by the
School of Electronics and Computer Science
at the University of Southampton.
More information and software credits
Quantum Mechanics Studies of Fuel Cell
Catalysts and Proton Conducting Ceramics
with Validation by Experiment
Thesis by
Ho-Cheng Tsai
In Partial Fulfillment of the Requirements for the degree
of
Doctor of Philosophy
CALIFORNIA INSTITUTE OF TECHNOLOGY
Pasadena, California
2015
(Defended April 20, 2015)
ii
2015
Ho-Cheng Tsai
iii
ACKNOWLEDGEMENTS
First, I would like to express my deep appreciation for my advisor, Prof. William A. Goddard III,
for his guidance in the past 5 years at Caltech. His generosity and great patience gave me the best
support during my academic career; his intelligence and brilliance always pointed in another direction
away from the depression and maze that I encountered, and reinflamed my enthusiasm for research.
I could not even image how I could complete my PhD degree without his kind help and wise
suggestions. He made my PhD study a pleasurable period and I will never forget my life at Caltech.
I would also like to acknowledge our Materials and Process Simulation Center (MSC) fuel-cell
subgroup director, Dr. Boris. V Merinov, for his extensive help in progress discussion, manuscript
revision, and project collaboration. He also taught me how to respond to the stringent reviewers’
questions and always be confident to myself. I am so lucky to have had such a nice team leader and
to have worked with him for the past five years.
I would like to express my gratitude to my mentor, Prof. Ted H. Yu. His step-by-step teaching of
how to use computation software helped me transfer from an experimentalist to a computational
theorist. His selfless sharing of his experiences, scripts, data, and related literatures also saved me a
lot of time. We now have three joint publication but I believe that there will be more in the future. In
addition, I also want to thank the MSC colleagues, especially Dr. Hai Xiao, Dr. Qi An, Dr. Tao Chen,
Dr. Wei-Guang Liu, Dr. Mu-Jeng Cheng, Dr. Yao Sha, Mr. Fan Liu, and Mrs. Shirley Wu, for their
counsel and help.
I would like to thank my scholarship from the ministry of education, Taiwan. With this financial
help, I could complete my dreams of studying abroad at one of the top universities in the world. The
project funding from Ford, Samsung, National Science Foundation (NSF), National Science Council
(NSC at Taiwan), Defense Advanced Research Projects Agency (DARPPA) is also appreciated. My
thanks for our collaborators, Dr. Yu-Chi Hsieh, Mr. Yi-Juey Lee, Dr. Yue-Han Wu, Prof. Pu-Wei
Wu, Prof. San-Yuan Chen, Prof. Chia-Ming Yang, Dr. Chih-Kai Yang, Dr. Hays, and Dr. Pezhman
Shirvanian for discussions and experiment help. My thanks also to Prof. Morozov at South Ural State
University for the powerful supercomputers which speeded up our calculations. And thanks to Dr.
Peter A. Schultz at Sandia National Laboratories for helping solve Seqquest related issues.
I would like to thank my thesis committee, Prof. Harry B. Gray, Prof. William L. Johnson, and
Prof. Sossina M. Haile. It’s my honor to have the top scientists as my thesis committee and to give
me advice.
My thanks to my roommate and classmates, Hsieh-Chen Tsai, Wen-Hao Lee, Chun-Jen Hsueh,
iv
Tiffany Wang, Yinglu Tang, Yulia Tolstova, Min-Feng Tu, and Yu-Hung Lai. Solving problem
sets with you guys together is my most miserable but unforgettable memory at Caltech. Also, my
thanks to the Association of Caltech Taiwanese (ACT) friends. My life at Caltech became much easier
and more fun with your company.
Last but not least, I would like to thank my family and also the love of my life, Yingying. Because
of your support and devotion, I could be completely involved in my research and study for the past
five years.
ABSTRACT
We carried out quantum mechanics (QM) studies aimed at improving the performance of
hydrogen fuel cells. This led to predictions of improved materials, some of which were subsequently
validated with experiments by our collaborators.
In part I, The challenge was to find a replacement for the Pt cathode that would lead to improved
performance for the Oxygen Reduction Reaction (ORR) while remaining stable under operational
conditions and decreasing cost. Our design strategy was to find an alloy with composition Pt3M that
would lead to surface segregation such that the top layer would be pure Pt, with the second and
subsequent layers richer in M. Under operating conditions we expect the surface to have significant
O and/or OH chemisorbed on the surface, and hence we searched for M that would remain segregated
under these conditions. Using QM we examined surface segregation for 28 Pt3M alloys, where M is
a transition metal. We found that only Pt3Os and Pt3Ir showed significant surface segregation when
O and OH are chemisorbed on the catalyst surfaces. This result indicates that Pt3Os and Pt3Ir favor
formation of a Pt-skin surface layer structure that would resist the acidic electrolyte corrosion during
fuel cell operation environments. We chose to focus on Os because the phase diagram for Pt-Ir
indicated that Pt-Ir could not form a homogeneous alloy at lower temperature. To determine the
performance for ORR, we used QM to examine all intermediates, reaction pathways, and reaction
barriers involved in the processes for which protons from the anode reactions react with O2 to form
H2O. These QM calculations used our Poisson-Boltzmann implicit solvation model include the
effects of the solvent (water with dielectric constant 78 with pH 7 at 298K). We found that the rate
determination step (RDS) was the Oad hydration reaction (Oad + H2Oad OHad + OHad) in both cases,
but that the barrier for pure Pt of 0.50 eV is reduced to 0.48 eV for Pt3Os, which at 80OC would
increase the rate by 218%. We collaborated with the Pu-Wei Wu’s group to carry out experiments,
where we found that the dealloying process-treated Pt2Os catalyst showed two-fold higher activity at
vi
25OC than pure Pt and that the alloy had 272% improved stability, validating our theoretical
predictions.
We also carried out similar QM studies followed by experimental validation for the Os/Pt coreshell catalyst fabricated by the underpotential deposition (UPD) method. The QM results indicated
that the RDS for ORR is a compromise between the OOH formation step (0.37 eV for Pt, 0.23 eV for
Pt2ML/Os core-shell) and H2O formation steps (0.32 eV for Pt, 0.22 eV for Pt2ML/Os core-shell). We
found that Pt2ML/Os has the highest activity (compared to pure Pt and to the Pt3Os alloy) because the
0.37 eV barrier decreases to 0.23 eV. To understand what aspects of the core shell structure lead to
this improved performance, we considered the effect on ORR of compressing the alloy slab to the
dimensions of pure Pt. However this had little effect, with the same RDS barrier 0.37 eV. This shows
that the ligand effect (the electronic structure modification resulting from the Os substrate) plays a
more important role than the strain effect, and is responsible for the improved activity of the coreshell catalyst. Experimental materials characterization proves the core-shell feature of our catalyst.
The electrochemical experiment for Pt2ML/Os/C showed 3.5 to 5 times better ORR activity at 0.9V
(vs. NHE) in 0.1M HClO4 solution at 25OC as compared to those of commercially available Pt/C. The
excellent correlation between experimental half potential and the OH binding energies and RDS
barriers validate the feasibility of predicting catalyst activity using QM calculation and a simple
Langmuir–Hinshelwood model.
In part II, we used QM calculations to study methane stream reforming on a Ni-alloy catalyst
surfaces for solid oxide fuel cell (SOFC) application. SOFC has wide fuel adaptability but the coking
and sulfur poisoning will reduce its stability. Experimental results suggested that the Ni4Fe alloy
improves both its activity and stability compared to pure Ni. To understand the atomistic origin of
this, we carried out QM calculations on surface segregation and found that the most stable
configuration for Ni4Fe has a Fe atom distribution of (0%, 50%, 25%, 25%, 0%) starting at the bottom
vii
layer. We calculated that the binding of C atoms on the Ni4Fe surface is 142.9 Kcal/mol, which is
about 10 Kcal/mol weaker compared to the pure Ni surface. This weaker C binding energy is expected
to make coke formation less favorable, explaining why Ni4Fe has better coking resistance. This result
confirms the experimental observation. The reaction energy barriers for CHx decomposition and C
binding on various alloy surface, Ni4X (X=Fe, Co, Mn, and Mo), showed Ni4Fe, Ni4Co, and Fe4Mn
all have better coking resistance than pure Ni, but that only Ni4Fe and Fe4Mn have (slightly) improved
activity compared to pure Ni.
In part III, we used QM to examine the proton transport in doped perovskite-ceramics. Here we
used a 2x2x2 supercell of perovskite with composition Ba8X7M1(OH)1O23 where X=Ce or Zr and
M=Y, Gd, or Dy. Thus in each case a 4+ X is replace by a 3+ M plus a proton on one O. Here we
predicted the barriers for proton diffusion allowing both includes intra-octahedron and inter-octahedra
proton transfer. Without any restriction, we only observed the inter-octahedra proton transfer with
similar energy barrier as previous computational work but 0.2 eV higher than experimental result for
Y doped zirconate. For one restriction in our calculations is that the Odonor-Oacceptor atoms were kept at
fixed distances, we found that the barrier difference between cerates/zirconates with various dopants
are only 0.02~0.03 eV. To fully address performance one would need to examine proton transfer at
grain boundaries, which will require larger scale ReaxFF reactive dynamics for systems with millions
of atoms. The QM calculations used here will be used to train the ReaxFF force field.
viii
TABLE OF CONTENTS
Acknowledgements ........................................................................................................ iii
Abstract ............................................................................................................................ v
Table of Contents .......................................................................................................... vii
Section I: DFT Study of ORR on Pt-Os Catalyst Surfaces Validated by Experiment
Chapter 1: Introduction ................................................................................................ 2
Chapter 2: Methodology .............................................................................................. 7
Computational.......................................................................................................... 7
Experimental Section for Dealloyed Pt2Os Catalyst ............................................ 10
Experimental Section for Os/Pt Core-Shell Catalysts .......................................... 12
Chapter 3: Results and Discussion for Pt3Os Catalysts ............................................. 16
Surface Segregation Effect .................................................................................... 16
Binding Energy of ORR Intermediates ................................................................. 19
Reaction Barriers and Possible ORR Mechanisms............................................... 28
Reaction Barriers and ORR Mechanisms for Pt3Os in Gas Phase ....................... 30
Reaction Barriers and ORR Mechanism for Pt3Os in Solvated Phase ................. 34
Experimental Verification ..................................................................................... 38
Summary for Pt3Os Catalyst ................................................................................. 48
Chapter 4: Results and Discussion for Os/Pt Core-Shell Catalysts .......................... 49
Stability of the Os/Pt Core-Shell Structure ........................................................... 49
Binding Sites and Binding energies ...................................................................... 51
Reaction Energy Barriers .................................................................................... 58
Strain and Ligand Effects ...................................................................................... 73
Experimental Results ............................................................................................. 79
Summary for Os/Pt Core-Shell Catalysts ............................................................. 86
Chapter 5: Conclusions .............................................................................................. 88
Bibliography ............................................................................................................... 91
Section II: First-principles Modeling of Ni4M (M= Co, Fe, Mn, Mo) Alloys as SOFC Anode
Catalyst
Chapter 6: Introduction ........................................................................................... 103
Chapter 7: Theoretical Methods.............................................................................. 107
Chapter 8: Results and Discussion.......................................................................... 110
Surface Segregation in Ni4Fe Alloy .................................................................... 110
Binding Energy of CHx Species .......................................................................... 112
Reaction Energy Barriers for Methane Conversion ........................................... 121
Chapter 9: Conclusions ........................................................................................... 125
Bibliography ............................................................................................................ 126
Appendix: DFT Study of Doped Perovskite Ceramics as Proton Conducting Materials
Preface…………………………………………………………………………...130
Abstract……………………………………...…………………………....……...130
Introduction…………………………..……...…………………………………...131
Theoretical Methods…………………………..……...…………...……………...134
Results and Discussion…………………………..……...………………...……...136
Conclusions………….…………………………..……...………………...……...141
Bibliography ..……….…………………………..……...………………...……...142
Section I
DFT Study of ORR on Pt-Os Catalyst
Surfaces Validated by Experiment
Chapter 1
Introduction
A proton exchange membrane fuel cell (PEMFC) is considered to be a potential technology for
use in automotive applications, stationary/portable power supply, and as a part of hybrid energy
systems1-4 due to its low pollution emission and high efficiency advantages compared to currently
used petroleum-based energy production.1,2,5 However, the PEMFC performance is limited by the
sluggish oxygen reduction reaction (ORR) at cathode,6-8 which is much slower and more complicated
than the hydrogen oxidation reaction (HOR) at anode.8
ORR mechanisms and electronic structures of ORR catalysts are widely discussed in the
literature. For example, Hammer and Nørskov discussed the surface catalysis by considering the
interaction between local density of states of adsorbates and d-bands of surfaces.9 A similar d-band
center theory was used by Xin et al. to screen potential alloy catalysts for ORR.10,11 Holewinski and
Linic applied DFT to simulate the OH/H2O surface coverage and explained the deviation from ideal
Tafel kinetics by microkinetic modeling.12 Similar microkinetic modeling starting with surface
coverage was used to calculate the ORR mechanism and Tafel kinetics for Pt13 and surface corrosion
and kinetics for the Au-modified Pt system.14,15 Jacob et al. considered current theoretical methods
for surface reaction simulations and performed DFT calculations to elucidate the complex ORR
mechanism by different pathways.16,17
Various types of cathode catalysts, including Pt-based alloys and non-precious metal catalysts,
have been widely studied in the past few years in order to improve the ORR efficiency and reduce
the catalyst cost.2,18-21 For example, Pt based catalysts, which are alloyed with transition elements
such as Ni,22-25 Co,22-24 Ti,22-24 Zr,22 Y,22,26 Fe,23,24 V,23,24 Sc,26 and Cu,27,28 were reported in virtue of
better ORR activities than pure Pt. In our previous study,29 some Pt3M catalysts, where M is a
transition metal, show Pt segregation with formation of so-called Pt-skin on the surface, which
improves the catalytic activity of these materials compared to pure Pt.24 However, it was reported that
the components of Pt alloys would be dissolved into electrolyte at the PEMFC operation
environment,30,31 which helps form the Pt-skeleton structure,32,33 but could possibly deteriorate
because of Ostwald ripening or diffusion.30-33 This would result in descending activity of the
catalysts.31,34-36
Stabilities of Pt-alloys have been studied by different theoretical and experimental methods. For
instance, Ma et al.37 and Chen et al.38 carried out DFT studies of the Pt3M alloy surface segregation
with and without adsorbates. Wei et al.39 evaluated the PtxMo alloy stability using DFT to calculate
the Gibbs free energy change of the surface corrosion. Nørskov with co-authors26,40 computed the heat
of formation to evaluate the stability of Pt-alloys and verified the stability of Pt3Y and Pt3Sc by
experiment. The same concept was applied by Hwang et al.22 to explain the experimental results on
the stability of Pt3M (M=Y, Zr, Ti, Ni, and Co). Dai et al.41 combined surface segregation, heat of
formation, and d-band center shift calculations to show that PtW alloys may have both better activity
and superior stability. Pt-skin formed after annealing or Pt-skeleton structures built after acidtreatment or oxide-leaching after ORR cycling are responsible for the improved catalytic activity and
long-term stability of the above-mentioned materials.42-44 Our segregation study of Pt3M alloy bare
surfaces29 indicates that a few alloys show a good segregation energy favoring the Pt-skin formation.
However, the situation may change when the catalyst is under ORR environment and its surface is
exposed to ORR adsorbates, such as O or OH. In poorly segregated alloy surfaces, the non-noble
metal component may diffuse onto the surface with a risk of leaching out into the electrolyte with
subsequent degradation of PEMFC performance. Widely discussed in literature, Pt3Co, Pt3Ni, Pt3Fe
alloys are not stable under the fuel cell conditions and leach out into the electrolytes.45,46 This
experimental observation is supported by a theoretical study47 as well.
With the potential to reduce the cost, non-precious metal (NPM) catalysts became attractive
recently. For example, Bashyam and Zelenay reported a cobalt-polypyrrole-carbon (Co-PPY-C)
catalyst;19 Wu et al. described polyaniline (PANI) derived catalysts.48 However, the mass transport
loss and stability issues are unsatisfactory for NPM catalysts.21 Therefore, to achieve the desirable
high activity, low cost, and durability, new advanced Pt group metal (PGM) ORR catalyst design
approaches are still required.49
In recent years, core-shell and core-shell-like structures of Pt-based alloys have attracted
significant attention due to their much better catalytic activity than pure Pt.23-26,29,50-64 For example,
Yang et al. reported that Ru/Pt core-shell catalyst has ~4 times better activity than pure Pt.61 The
combined experimental and theoretical work of Fribel et al. compared Had and O/OHad adsorption on
2D and 3D Pt/Rh(111) surfaces and found that the H/O/OHad are destabilized in the smooth 2D
structure because of the ligand/strain effect of the Rh substrate. However, the geometry effect (both
the higher thickness and increased number of defects) balanced the strain/ligand effect and increased
the H/O/OH adsorption for the 3D structure derived from the wetting process.63 Brimaud et al.
analyzed the surface character and compared the ORR activity of Ptx-ML/Ru(0001) and PtxRu1x/Ru(0001) with Pt(111) and found that their ORR activity could exceed the pure Pt ORR activity by
modifying the Pt surface content of the PtxRu1-x/Ru(0001) alloy or adjusting the thickness of the Pt
film to the thickness of Ptx-ML/Ru(0001).62 Jackson et al. used core-shell Ru/Pt as an example to prove
the feasibility of improving the activity by tailoring the nanostructure.64
To our knowledge, Os and its alloys have rarely been studied as ORR catalysts. Most of the
literature on Os and its alloys is focused on methanol65-70 or formic acid71 oxidation in fuel cell
systems. Only Os0.2Pt0.8/Pd(111) was considered as a possible electrocatalyst for ORR,72 but the
corresponding work does not include a complete report. Therefore, we paid our attention to Pt3Os as
a catalyst because of its good bare surface segregation result,29 as well as the shorter surface Pt-Pt
bond distance, which may reduce the O/OH binding energy due to the strain effect and thus improve
its ORR catalytic activity. In addition to the Pt3Os alloy, a study of the core-shell Os/Pt catalyst was
also performed in order to investigate the ORR activity of this catalyst.
In this section, we will discuss our theoretical predictions and experimental validation for the PtOs catalysts. All methodologies, including both theoretical and experimental methods, are explained
in Chapter 2. The computational study of the Pt3Os catalyst, our efforts on synthesis of the Pt3Os
catalyst, materials characterization, and electrochemical evaluation are discussed in Chapter 3. Here,
we examined the segregation energy for 28 Pt3M alloys with adsorbed Oad and OHad on the surface
using quantum mechanics (QM) calculations. Looking ahead, we say that our calculations revealed
surface segregation to become energetically unfavorable for Pt3Co and Pt3Ni, as well as for most other
Pt binary alloys, in the presence of adsorbed Oad and OHad. However, Pt3Os and Pt3Ir remain surface
segregated and show the best energy preference among the alloys studied for both adsorbed species
on the surface. PtIr materials have been studied earlier (see, for instance, Refs 73, 74), and Ir could
not mix Pt well to form a homogeneous alloy at lower temperature. Therefore we selected the Pt3Os
system for further theoretical investigation. Binding energies of various ORR intermediates and
reaction energy barriers for various pathways of different ORR mechanisms on the Pt3Os(111) surface
were calculated, analyzed, and discussed. Furthermore, our collaborators synthesized the Pt-Os
catalyst by using the dealloying process, and performed materials characterization and
electrochemical evaluation.
In Chapter 4, our QM calculations on periodic slabs to predict catalytic activity of the Os/Pt
core-shell catalysts are analyzed. Perfect hcp (0001) and fcc (111) slabs were applied for Os, Pt, and
Pt/Os surfaces. Full kinetic modeling of complex reactions, such as ORR, on nanoparticles still
remains a challenge for atomistic first-principles simulations. Although a slab model applied in our
calculations is mostly relevant for extended surfaces and neglects natural strain, morphology, or any
defects such as edge and kink sites, multiple terraces and corners, and other peculiarities observed in
nanoparticles, several fundamental issues of the surface ORR can be analyzed using the methodology
applied, because the highly coordinated (111) facets are most conducive to the ORR on small
nanoparticles.75 Although the above-mentioned features of nanoparticles may change binding
energies of the ORR intermediates and reaction barriers compared to those on extended surfaces, the
slab approach is still sufficient enough to predict ORR trends. Moreover, reported experimental
results indicate that the higher catalytic activity of Pt3M alloy extended surfaces compared to that of
Pt is reproduced qualitatively on surfaces of Pt3M alloy nanoparticles.76-81 To validate our theoretical
predictions, we used the UPD method for the core-shell catalyst fabrication49,55,76-81 with succeeding
analysis of ORR activity of the fabricated Os/Pt core-shell catalysts.
Conclusions are in Chapter 5.
Chapter 2
Methodology
2.1 Computational
In our theoretical study, we used QM calculations to investigate surface metal segregation,
unique binding-site preferences due to the placement of sublayer alloying atoms for all
intermediates involved in the ORR on the Pt, Pt3Os(111), and Os(0001)/Pt surfaces, and the
consequent changes in the reaction barriers and ORR mechanisms.
All QM calculations were carried out using the SeqQuest software82 with optimized double zeta
plus polarization quality Gaussian-type orbitals and the Perdew-Becke-Ernzehof (PBE)83 functional
of DFT in the generalized gradient approximation (GGA).84,85 Small core Norm-conserving angular
momentum projected pseudopotentials86-90 were employed in our calculations. All calculations were
performed with spin optimization. The reciprocal space grid was 5×5×0 for the slab calculations and
12×12×12 for the bulk lattice constant calculations. We used 4 layer 2×2 cell slabs for the alloy
segregation study, while 6 layer 2×2 Pt3Os cell slabs were applied for the binding energy and most
reaction barrier calculations. Therefore, each layer contains 4 atoms and the surface coverage is ~1/4.
We consider the 1/4 coverage to be large enough to avoid irrelevant cross cell interaction, while still
keeping the unique properties of the segregated Pt3Os surface. For the O hydration step, 3×3 and 2×4
cell slabs were used for Pt and Pt3Os, respectively, to avoid artificial hydrogen bonding. The periodic
cell parameter of the slab corresponds to that of the optimized Pt3Os bulk structure with the lattice
constant of 3.94 Å, slightly smaller than 3.98 Å for Pt. All charges were obtained from the Mulliken
population analysis.
For Os/Pt core-shell catalysts, a 3×3 supercell (9 atoms per layer) with 3 to 5 layer Os slabs were
used as the core structure. 1 to 3 additional layers of Pt atoms on this core structure (the total number
of layers in all slabs was 6) were used to simulate the Os/Pt core-shell catalyst structures. In all
calculations, we fixed the coordinates of 9 Os atoms in the bottom layer and relax all other atoms in
the geometry optimization. For comparison, similar calculations were carried out on 6 layer Pt and
Os slabs. The obtained Os-Os and Pt-Pt bond lengths are 2.75 and 2.81 Å, respectively. We used the
lowest-energy surfaces in our calculations: the (0001) surface of the hexagonal closed-packed (hcp)
structure for Os and the (111) surface of the face-centered cubic (fcc) structure for Pt.
To represent the effects of solvent polarization, an implicit model91 based on the PoissonBoltzmann approximation was applied.92-94 We showed that our implicit solvation model correctly
reproduces the solvation energies of Oad, Had, OHad, and H2Oad calculated using explicit water layers.91
The Mulliken charges were used as inputs to our solvation model.
The binding energies (BE) are calculated as the energy gain for species to adsorb to the surface,
i.e.,
BEgas = Esurf + A, gas – Esurf, gas – EA, gas
where A is an adsorbate. For the solution phase, the solvent stabilization is added directly to the
binding energy in order to isolate the influence of water. This leads to
BEsolv = Esurf + A, solv – Esurf, solv – EA, gas
The solvent phase binding energy does not include the solvent effect of the adsorbate itself, because
most of the species are radicals that do not have well-defined solvation energies for comparison. The
lack of these energy data for radical species does not affect the barrier calculations since the reactants,
products, and transition states are all surface species and the differences in individual solvation
energies eventually cancel out.
All barriers are calculated as the energy difference between a transition state and surface reactant
sites, using the nudged elastic band (NEB) method.95,96 We disregard the energy that reactants need
to migrate from the globally preferred sites to the reacting sites, as these barriers would correspond
only to very low surface coverage. This condition is not typical for the conventional fuel cell
operation. However, to avoid misunderstanding and show certain details of the ORR mechanism, the
spontaneous O migration step has been added to some of our figures, representing possible ORR
mechanisms. While both Langmuir–Hinshelwood and Eley–Rideal mechanisms (see, for instance,
Refs 97, 98) may occur in ORR at electrode potentials near 1.23 eV (the reduction potential of the
ORR established by the Nernst equation),99 our Os/Pt core-shell computation results confirmed that
the conclusions derived from Langmuir-Hinshelwood type reactions are similar to the ones derived
from Eley–Rideal mechanisms.
In our calculations, zero point energy (ZPE) and entropy contributions were neglected, because
our conclusions about ORR activity are based on the rate-determining step (RDS) barriers, which
are energy differences between transition states and initial states. In this case, the above-mentioned
contributions may be cancelled.
10
2.2 Experimental Section for Dealloyed Pt2Os Catalyst
2.2.1 Pt2Os Synthesis
Carbon-supported Pt2Os nanoparticles were prepared in a chemical reduction route. First, 80
mg of carbon powders (particle size <50 nm, Sigma Aldrich) were suspended in 50 mL of deionized
water at 80 °C. Next, 36 mg of H2PtCl6·6H2O (UniRegion Bio-Tech) and 17 mg of K2OsCl6
(Sigma-Aldrich) were dissolved in 50 mL of deionized water (molar ratio of Pt/Os = 2), and the
solution was added to the carbon suspension. Subsequently, 21 mg of citric acid was added as a
chelating agent. The mixture was stirred for 30 min at 80 °C in an argon flow under reflux to
produce a homogeneous suspension. Afterward, 76 mg of NaBH4 was added to serve as a reducing
agent, and the mixture underwent further stirring for 2 hr at 80 °C in an argon flow under reflux to
ensure the complete reduction of Pt and Os ions, and the formation of Pt2Os nanoparticles
impregnated on the carbon powders. The as-synthesized sample was labeled as Pt2Os/C. Next, the
Pt2Os/C powders were filtered and washed to remove residual chloride ions. After drying at 25 °C
in air for 8 hr, a reduction treatment at 250 °C in a hydrogen flow (100% H2) was performed for 2
hr. The effective metal loading of the Pt2Os nanoparticles was 20 wt% of the Pt2Os/C sample.
2.2.2 Electrochemical Analysis
10 mg of Pt2Os/C powders underwent an ultrasonication mixing for 5 min in a solution
containing 3 mL of deionized water, 2 mL of ethanol, and 2 μL of 5 wt% Nafion ionomer solution
(Sigma-Aldrich) to render a uniform ink dispersion. Subsequently, 15 μL of the ink dispersion was
deposited on a glassy-carbon rotation disk electrode (RDE) serving as a working electrode (Pine
Research, electrode diameter is 5 mm and the electrode area is 0.1963 cm2). To initiate the selective
11
dissolution of Os atoms from the Pt2Os nanoparticles, known as a dealloying process, multiple
cyclic voltammetry (CV) scans were imposed between -0.2 and 0.8 V at 50 mV s-1 in 50 mL of
deaerated 0.1 M aqueous HClO4 solution. After the dealloying process, the sample was labeled as
DA-Pt2Os/C. The coulombic charge associated with the hydrogen underpotential deposition region
(-0.2 to 0.2 V vs. Ag/AgCl) was estimated and divided by 210 μC cmPt-2 100,101 to obtain the
electrochemical active surface area (ECSA) for the Pt2Os/C and DA-Pt2Os/C. To explore the
electrocatalytic activities for ORR, CV scans between -0.2 and 0.8 V were performed at 10 mV s-1
in 50 mL of 0.1 M aqueous HClO4 solution. Prior to the ORR experiments, the HClO4 aqueous
solution was bubbled with oxygen for 30 min to ensure it was fully saturated with the oxygen.
Durability tests were performed using CV scans at 50 mV s-1 between 0.36 and 0.76 V in 50 mL of
0.1 M aqueous HClO4 solution. The durability test lasted for 10,000 cycles and the electrolyte was
exposed to the ambient air throughout the entire cycles. The electrochemical measurements were
performed at 25 °C in a three-electrode arrangement using a Solartron 1287A electrochemical
interface. A Ag/AgCl and Pt foil (15 cm2) were used as the reference and counter electrodes,
respectively. The potential for the reversible hydrogen electrode (RHE) was -0.289 V (vs.
Ag/AgCl). In our figures, all potentials were plotted against the RHE. Lastly, identical
electrochemical tests were performed on commercially available Pt/C (20 wt% Pt on Vulcan
XC72R, BASF) for comparison purposes.
2.2.3 Materials Characterization
A high-resolution transmission electron microscope (HRTEM; JEOL JEM3000F) was
employed to observe the morphologies, sizes, and distributions of the Pt2Os/C and DA-Pt2Os/C.
The structures and composition profiles of the Pt2Os and DA-Pt2Os nanoparticles were obtained
12
using a JEOL spherical aberration corrected scanning transmission electron microscope (ARM
200F) with an Oxford energy dispersive spectrometer (EDS) in which the Lα and Mα signals from
the EDS were recorded to determine the spatial distribution of the Pt and Os atoms in the Pt2Os and
DA-Pt2Os nanoparticles. A scanning electron microscope (SEM; JEOL JSM6500F) and a total
reflection X-ray fluorescence spectrometer (TRXF; Bruker S2-PICOFOX) were employed to
estimate the atomic ratio of Pt/Os in the Pt2Os and DA-Pt2Os nanoparticles. The exact Pt amount
in our samples was determined using an inductively coupled plasma mass spectrometer (ICP-MS;
Agilent 7500ce). An X-ray diffractometer (XRD; Bruker D2 Phaser) equipped with a Cu Kα
radiation source (λ =1.54 Å) was used to identify relevant phases and crystal sizes for both Pt2Os/C
and DA-Pt2Os/C.
2.3 Experimental Section for Os/Pt Core-Shell Catalysts
2.3.1 Os Synthesis
Carbon-supported Os nanoparticles (Os/C) were prepared in a chemical reduction route. First, 50
mg of carbon powders (Ketjen Black) and 19.5 mg of OsCl3 were mixed in 100 mL of ethanol at
25°C. Next, the temperature of the mixture was raised to 110°C for 2 hr to reach a uniform dispersion.
Subsequently, 2.63 ml of 100 mM aqueous KOH solution was added. The mixture underwent further
stirring for 30 min and cooled to 25°C to filter out the Os/C. The dried Os/C was subjected to a
reduction treatment at 450°C for 1 hr in an atmosphere of 15% H2 and 85% Ar. The effective metal
loading of the Os nanoparticles was 20 wt% of the Os/C sample.
13
2.3.2 Preparation of Os/Pt Core-Shell Nanoparticles
We used the UPD method to prepare Os/Pt core-shell nanoparticles. In this method, a
monolayer of a sacrificial metal is deposited on the substrate at a potential which is more positive
than the reduction potential. Then a more noble metal substitutes the sacrificial metal using galvanic
displacement.
5 mg of Os/C powders underwent an ultrasonication mixing for 5 min in a solution containing 5
mL of ethanol, and 1 μL of 5 wt% Nafion ionomer solution (Sigma-Aldrich) to render homogeneous
ink dispersion. Next, 15 μL of the ink dispersion was deposited on a glassy-carbon rotation disk
electrode (RDE) serving as a working electrode (Pine Research, electrode diameter is 5 mm and the
electrode area is 0.1963 cm2). Subsequently, the working electrode was immersed in a deaerated
aqueous solution of 50 mM CuSO4 and 50 mM H2SO4 for 20 CV scans with a scan rate of 20 mV/s
in a potential window of 0.31 and 0.9 V (vs. RHE). The purpose for these CV scans was to
precondition the sample and identify the suitable potential for the Cu UPD process. Afterward, a Cu
UPD was imposed in which the potential was kept at 0.35 V to allow the Cu to deposit onto the
surface of the Os nanoparticles as a monolayered film. Once the depositing current was subdued and
stabilized, the sample was removed from the Cu plating solution and immersed immediately in a
deaerated aqueous solution containing 1 mM K2PtCl4 and 50 mM H2SO4. At this stage, a
displacement reaction took place in which the Cu atoms on the Os surface were oxidatively dissolved
whereas the Pt ions in the electrolyte were reduced and deposited onto the Os nanoparticles. The
sample undergoing a single Cu-UPD/displacement reaction cycle was labeled Pt1ML/Os/C and the
sample undergoing consecutive Cu-UPD/displacement reaction cycles was labeled Pt2ML/Os/C. We
should emphasize here that Pt1ML/Os/C does not necessarily mean a completely perfect Pt monolayer
14
deposited on the Os substrate; more details are reported in Chapter 4. The Os/Pt core-shell catalysts
production procedure is summarized in Figure 1.
Figure 1. The procedure for Os/Pt core-shell catalyst preparation.
To determine the ECSA, multiple CV scans were imposed between 0.05 and 1.0 V (vs. RHE) at
20 mV/s in a 50 mL of deaerated 0.1 M aqueous HClO4 solution. The Coulomb charge associated
with the hydrogen adsorption was integrated and divided by 210 μC cmPt-2 to obtain the ECSA value
for the Pt/C, Pt1ML/Os/C, and Pt2ML/Os/C, respectively. To explore the electrocatalytic activities for
the ORR, CV scans between 0.1 and 1.05 V (vs. RHE) were performed at 20 mV/s in 50 mL of 0.1
M aqueous HClO4 solution. Prior to the ORR experiments, the HClO4 aqueous solution was bubbled
with oxygen for 30 min to ensure it was fully saturated with the oxygen. The electrochemical
synthesis and ORR measurements were performed at 25°C in a three-electrode arrangement using a
Solartron 1287A electrochemical interface. Ag/AgCl and Pt foils (15 cm2) were used as the reference
and counter electrodes, respectively. In our figures, all potentials were plotted against the RHE.
15
Lastly, identical electrochemical tests were performed on commercially available carbonsupported Pt (Pt/C; 20 wt% Pt on Vulcan XC72R, BASF) for comparison purposes.
2.3.3 Materials Characterization
HRTEM (JEOL JEM3000F) and high angle angular dark field (HAADF) techniques
were employed to observe the morphologies, sizes, and distributions of the Os
nanoparticles. The structures and composition profiles of Pt1ML/Os/C and Pt2ML/Os/C were
obtained using a JEOL spherical aberration corrected scanning transmission electron
microscope (ARM 200F) with an Oxford EDS in which the Lα and Mα signals from the
EDS were recorded to determine the spatial distribution of the Pt and Os atoms. The exact
Pt amount in our samples was determined using an ICP-MS (Agilent 7500ce).
16
Chapter 3
Results and Discussion for Pt3Os Catalyst
3.1 Surface Segregation Effect
We have performed QM calculations to examine 28 Pt binary alloys and identify good
segregating alloys that maintain the favorable segregating property when the oxidative species, Oad
and OHad, are adsorbed on the surface. For calculating the segregation energy, we used the lowest
energy Oad and OHad surface binding sites from our previous Pt3Ni study.59
Figure 2. Surface segregation of Pt3M binary alloys.29
17
Figure 2 summarizes our previous bare surface segregation calculations for Pt3M alloys.29
The results showed that Re, W, Os, Tc, Mo, Ru, Ir, Co, Ni, Rh, V, Cu, Zn, Fe, and Pd could form
surface segregated Pt3M alloys. Figure 3 displays the segregation energy of the different alloy
catalysts with adsorbed Oad and OHad on the surface. In case of OH adsorbed on the surface, six
solute metals, showing favorable segregation energies for the corresponding Pt3M alloys, are Os,
Re, Ir, Ru, Tc, and Rh. For the Oad adsorbed surface, only two solute metals, Ir and Os, show
favorable segregation energies. All of these solute metals are considered difficult to oxidize and
have more positive reduction potential versus hydrogen. All but one (Tc) are considered noble
metals. As a general rule, segregation becomes unfavorable when Oad and OHad are adsorbed on a
metal which is easily oxidized.
Figure 3. Surface segregation energy with adsorbed O (left) and OH (right).
The segregation energy can vary significantly when Oad or OHad is adsorbed on the surface.
For example, without adsorbed species, the best five segregation energies were Pt alloyed with Re,
W, Os, Tc, and Mo.29 However, the segregation energy of Pt3W and Pt3Mo changes from strongly
18
favorable to strongly unfavorable with adsorbed Oad and OHad, as both W and Mo are known to
easily react to form oxides. Our segregation result is consistent with results of other theoretical
investigations, in which a similar approach was used to study some Pt-based alloy and core-shell
catalysts,73,74 as well as with the experimental results obtained by Abrams et al. for Pt deposit on
the Au foil.102 They found that the Au atoms migrated to the surface because of a lower Au surface
energy. However, exposing the sample to air makes the deactivated sample become partially
reactive again. This reactivation was ascribed to the adsorbed O from the air, which favor the Pt to
segregate back to the surface (adsorbate-induced surface segregation). Our calculation on the Pt3Au
alloy also indicates that Pt shows better segregation property with adsorbed O on the surface (see
Figure 3). We found only a few Pt binary alloys with favorable surface segregation energy in the
presence of adsorbed Oad or OHad. Out of them, only Pt3Os and Pt3Ir show surface segregation in
the presence of both adsorbed species. Therefore, these alloys are predicted to be more resistant to
solute metal leaching than Pt3Co or Pt3Ni. However, according to the Pt-Os phase diagram,103 only
20% Os can be mixed into the Pt structure, whereas Ir cannot be mixed into the Pt structure at all.104
Despite this disappointing information, synthesis of Pt-Os nanoparticles even with the molar ratio
of 1:1105 and PtIr nanoparticles with the molar ratio of 3:158,106 were reported. In this study, we
carried out computations of binding energies and energy barriers of the ORR intermediates on the
Pt3Os(111) alloy surface.
19
3.2 Binding Energy of ORR Intermediates
3.2.1 Binding Site Notation
Generally, the closest packed (111) surface of fcc structured metals has four types of sites:
1. On-top, bonded to one Pt (µ1), denoted as t,
2. Bridging, between two Pt µ2), denoted as b,
3. Bridging, between three Pt (µ3-fcc) but in the fcc position (not above atoms of the top or second
layer), denoted as f, and
4. Bridging, between three Pt (µ3-hcp) but in the hcp position (above atoms of the second layer),
denoted as h.
However, due to strong segregation, the Pt3Os(111) surface has 100% Pt in the first layer, 50%
Os and 50% Pt in the second layer, and 25% Os and 75% Pt in the four remaining layers. We find
that the binding energies of the ORR intermediates to the pure Pt layer strongly depend on the
nature of the second layer atoms. Figure 4 shows notations of these sites and details of the
differences between various sites.
For the first and second layers, there are two types of top sites: t1 with one Os neighbor in the
second layer and t2 with two Os neighbors. Considering also the third layer, we can distinguish t1a
with no Os in the third layer directly beneath the surface and t1b with one Os. All t2 sites are the
same (see Figure 4).
For the top two layers, there are four µ2 bridge sites, depending on the number of Os atoms
underneath: b0, b1, b2, and b3 with 0, 1, 2, and 3 Os atoms in the second layer. When the third layer
20
is added, there appear two subtypes for b1, b2, and b3, depending on the distance to the Os atom
in the third layer. We denote the subtypes closer to the third layer Os as b1a, b2a, and b3a, and the
others as b1b, b2b, and b3b, respectively (Figure 4).
When we consider only the top two layers, two fcc sites can be distinguished, f1 and f2, with
one and two Os atoms in the sublayer triangle, but adding the third layer splits f1 into f1a and f1b
with f1a on top of the third layer Os and f1b on top of the third layer Pt.
Similarly, there are two hcp sites related to the two top layers: h1 and h2. Here h1 is on top of
the Os sublayer, while h2 is on top of the Pt sublayer. Adding the third layer splits the h1 site into
h1a and h1b with one Os atom and without Os atoms in the projected triangle of the third layer atoms,
as shown in Figure 4.
Figure 4. Binding sites on the Pt3Os(111) surface (a). For top sites t1 and t2, the triangle indicates
the sublayer atoms (b). t1 has one Os atom beneath it, while t2 has two. For bridge sites, the bridge
itself is shown as the thick black line, while the two terminals of the black line connect the two
surface atoms, forming the bridge site. The trapezoid beneath is the sublayer atoms. b0-b3 have from
21
0 to 3 Os atoms in the sublayer. An fcc site is in the center of a surface triangle (shown as a solid
triangle). f1 and f2 differ in the sublayer triangle beneath the surface triangle: f1 has one Os atom
beneath, while f2 has two. An hcp site is also in the center of a surface triangle; it has one sublayer
atom beneath: for h1 it is Os, while for h2 it is Pt.
3.2.2 Binding Energies of ORR Intermediates
First, we studied the preference of Had, Oad, OHad, H2Oad, O2ad, OOHad, and HOOHad ORR
intermediates on various binding sites shown in Figure 4. Tables 1 and 2 list binding energies of the
above-mentioned intermediates on the Pt3Os(111) alloy surface in gas phase and solution,
respectively. It should be noted that when considering energetics (binding energies and reaction
barriers) related to ORR on the Pt3Os(111) surface and comparing them with ORR energetics on pure
Pt, more attention should be paid to the solvation phase, because this phase better corresponds to the
PEMFC operating conditions than the gas phase.
H binding. For the pure Pt surface, the binding energy of Had is 2.70−2.80 eV in gas phase and
2.81−2.87 eV in solvated phase. The strongest binding energy corresponds to the top site.
The preferred binding site for Had on the Pt3Os(111) surface is also the top site, t1b, in gas phase,
with a binding energy of 2.65 eV, followed by t2 and t1a with a binding energy of 2.57 and 2.55 eV,
respectively. Under solvation, the preferred site becomes t2 with a binding energy of 2.84 eV,
followed by t1b, b2b, b3b, t1a, f1a, b3a, f2, and b0 with binding energies of 2.63−2.79 eV. Thus, for Pt3Os,
Had can migrate relatively easily in all directions to react with other ORR species.
O binding. On pure Pt, Oad binds strongly to the fcc site with a net energy of 3.66 eV in the gas
phase and 4.36 eV in the solvated phase. The huge solvation stabilization arises from electrostatics
22
due to the appearance of the strong dipole at surface Oad atoms.
For Pt3Os in gas phase, the strongest binding of Oad is at the f1a site, 3.55 eV, followed by the f2
site with a binding energy of 3.48 eV. With solvent, the f2 and f1b sites become dominant with stronger
binding energies of 5.18 and 4.97 eV, respectively. All other binding sites are significantly less stable
than f2. This means that Oad, formed from O2ad dissociation strongly prefers to occupy the f2 site, and
most probably no further migration occurs to other sites.
Table 1. Binding energies (eV) of various ORR species at different sites on Pt3Os and Pt59 in gas
phase.
Pt3Os
OOHb
H2Oc
-1.86
-0.95 (-0.94)
-0.16
-2.50
-1.95
-1.06 (-0.94)
-0.19
-2.57
-2.57
-2.03
-1.02 (-0.84)
-0.17
b0
-2.53
-3.05
-1.84
-0.49
-0.25
b1a
-2.49
-3.03
-1.87
-0.36
-0.33
b1b
-2.51
-3.02
-1.91
-0.39
-0.26
b3a
-2.46
-3.02
-2.16
-0.53
-0.37
b3b
-2.38
-2.87
-2.06
-0.47
-0.38
b2a
-2.48
-3.11
-2.20
-0.59
-0.37
b2b
-2.44
-3.02
-1.99
-0.44
-0.32
f2
-2.52
-3.48
-1.86
-0.29
f1a
-2.49
-3.55
-2.17
-0.50
f1b
-2.42
-3.34
-2.04
-0.30
h1a
-2.45
-3.07
-2.12
-0.36
Site
OH
t1a
-2.55
-2.33
t1b
-2.65
t2
O2a
H2O2a
23
h1b
-2.40
-3.05
-2.02
-0.31
h2
-2.48
-3.21
-1.90
-0.34
Best
-2.65
-3.55
-2.20
-0.59
-1.06
-0.19
-1.06
-0.22
-0.38
Pt
-2.80
-2.50
-2.23
-2.70
-3.10
-2.25
-0.40
-2.72
-3.66
-2.22
-0.46
-2.70
-3.28
-2.28
-0.35
Best
-2.80
-3.66
-2.28
-0.46
-0.27
-1.06
-0.22
-0.27
The center of the O−O bond is used to denote the binding sites of O2 and H2O2. Thus, b means
that two O atoms are located approximately on the top of the surface atoms with the O−O bond
center at the b site. The f and h binding sites are defined as one O atom located at the top site and
the other at the b site with the O−O bond center at the f or h site, respectively.
The position of the first O atom is used to denote the binding sites of OOH with the second O
atom close to the b site. The numbers in parentheses correspond to the OOH binding at the t1a, t1b,
and t2 sites with the second O atom close to the f1b, f1a, and f2 sites.
The position of the O atom is used to denote the binding sites of H2O with the O atom at the top
site of Pt and two O−H bonds parallel to the surface.
Table 2. Binding energies (eV) of various ORR species at different sites on Pt3Os and Pt59 in
solution.
Pt3Os
Site
OH
O2a
OOHb
H2Oc
H2O2a
24
t1a
-2.67
-3.08
-2.40
-1.46 (-1.67)
-0.45
t1b
-2.79
-3.18
-2.46
-1.61 (-1.44)
-0.49
t2
-2.84
-3.34
-2.63
-1.61 (-1.32)
-0.52
b0
-2.63
-3.75
-2.32
-0.77
-0.53
b1a
-2.57
-4.36
-2.36
-0.68
-0.58
b1b
-2.59
-4.78
-2.37
-0.84
-0.52
b3a
-2.64
-3.87
-2.50
-0.85
-0.62
b3b
-2.68
-4.24
-2.34
-0.74
-0.60
b2a
-2.62
-3.86
-2.60
-0.94
-0.62
b2b
-2.69
-4.46
-2.34
-0.77
-0.54
f2
-2.63
-5.18
-2.41
-0.77
f1a
-2.66
-4.25
-2.56
-1.06
f1b
-2.54
-4.97
-2.59
-0.79
h1a
-2.58
-4.21
-2.56
-0.75
h1b
-2.42
-4.63
-2.39
-0.59
h2
-2.63
-3.88
-2.32
-0.74
Best
-2.84
-5.18
-2.63
-1.06
-1.67
-0.52
-1.52
-0.58
-0.62
Pt
-2.87
-3.09
-2.77
-2.82
-3.73
-2.63
-0.73
-2.85
-4.36
-2.57
-0.87
-2.81
-3.92
-2.64
-0.70
Best
-2.87
-4.36
-2.77
-0.87
-0.61
-1.52
-0.58
-0.61
The center of the O-O bond is used to denote the binding sites of O2 and H2O2. Thus, b means
that two O atoms are located approximately on the top of the surface atoms with the O−O bond
25
center at the b site. The f and h binding sites are defined as one O atom located at the top site and
the other at the b site with the O−O bond center at the f or h site, respectively.
The position of the first O atom is used to denote the binding sites of OOH with the second O
atom close to the b site. The numbers in parentheses correspond to the OOH binding at the t1a, t1b,
and t2 sites with the second O atom close to the f1b, f1a, and f2 sites.
The position of the O atom is used to denote the binding sites of H2O with the O atom at the top
site of Pt and two O−H bonds parallel to the surface.
OH binding. On pure Pt, OHad has almost the same binding energy at all sites, with 2.22−2.28
eV in the gas phase and 2.57−2.77 eV in solution.
For Pt3Os, the most preferred site in gas phase is b2a followed by f1a, b3a, h1a, and b3b, with binding
energies of 2.06−2.20 eV. In solution t2 is the most preferred site, with a binding energy of 2.63 eV,
followed by slightly weaker binding at the b2a (2.60 eV ), f1b (2.59 eV), f1a and h1a (both 2.56 eV), and
b3a (2.50 eV). The binding energy of OH at all possible sites ranges within ~0.2 and 0.4 eV for Pt and
Pt3Os, respectively. This probably indicates the ability for easier OHad migration on pure Pt and harder
OHad migration on the Pt3Os surface.
In gas phase, both O and OH binding on Pt3Os are weaker than on Pt, 3.55 vs. 3.66 eV and 2.20
vs. 2.28 eV, respectively. In solution, OH still binds more weakly on the Pt3Os (111) surface (2.63
vs. 2.77 eV), whereas the O binding becomes stronger (5.18 vs. 4.36 eV). A possible reason for this
change in solution might be a strong dependence of the solvent effect on a charge dipole. The dipole
related to the O binding on Pt3Os is obviously stronger than that related to the OH binding (Figure 5).
Furthermore, the charge dipole between Os and Pt makes the solvent effect for Pt3Os more significant
than for an almost uniformly distributed slab of pure Pt.
26
Figure 5. Charge dipoles related to the O and OH binding on Pt3Os(111) surface.
O2 binding. For pure Pt, we find that the binding energy of O2ad is 0.46 eV in gas phase and 0.87
eV in solution, with the difference for various sites ranging within 0.11 and 0.17 eV for gas phase
and solution, respectively.
For Pt3Os, O2ad prefers to bind to the surface at the b2a site with a binding energy of 0.59 eV in
gas phase followed by b3a, f1a, b0, and b3b with similar binding energies of 0.53, 0.50, 0.49, and 0.47
eV, respectively. Overall, the corresponding binding energies vary from 0.29 to 0.59 eV. In the
solvated Pt3Os phase, the strongest binding energy, 1.06 eV, is for the f1a site of O2. The following
energetically favored sites are b2a (0.94 eV), b3a and b1b (0.85 and 0.84 eV, respectively), and f1b (0.79
eV). Such a strong binding of O2ad at the f1a site should impede O2ad migration to other sites.
OOH binding. For pure Pt, OOHad binds to the top sites with the terminal O bonded to the Pt and
the OOHad plane parallel to the surface. OOHad prefers to have the O−O bond heading to an adjacent
Pt atom. This leads to a binding energy of 1.06 eV in the gas phase and 1.52 eV in solution.
27
On the Pt3Os(111) surface, OOHad binds to the surface at the t1b site with a similar strength,
1.06 eV, in gas phase, as in pure Pt. The corresponding binding energy of OOHad at other stable
binding sites, t1a and t2, are 0.95 and 1.02 eV, respectively. In solution, OOHad binds to the Pt3Os(111)
surface significantly stronger than in gas phase at all stable binding sites, by 0.5~0.6 eV. The preferred
site is t2 with a binding energy of 1.67 eV which is by ~0.15 eV stronger than the corresponding
binding energy for pure Pt. It should also be noted that the OOHad binding at the t2 site with the O-O
bond toward f2 is more stable than heading to another Pt atom, but the first O atom still binds to the
t2 top site. It is unstable for the first O atom to bind to other sites. Thus, once formed, OOHad will
probably not migrate on the surface, but the O−O bond could be differently orientated.
H2O binding. Similar to the case of pure Pt, H2Oad binds only to the top sites (t1a, t1b, and t2) on
the Pt3Os(111) surface with very similar binding energies, 0.16−0.19 eV in gas phase, and 0.45−0.52
eV in solution. These values are very close to the corresponding values for pure Pt, 0.22 eV in gas
phase and 0.58 eV in solution. The difference between the binding energies in gas phase and solution
is close to the value of the solvent stabilization of bulk H2Oad, 0.40 eV. Since the surface H2Oad does
not bind to bridge, fcc, or hcp sites, migration of H2Oad from one top site to the others is through
adsorption and dissociation. The migration barrier can be estimated to be 0.10~0.20 eV (the 0.50~0.60
eV binding energy minus the 0.40 eV solvation of H2Oad, H2O is considered always solvated).
HOOH binding. Only the bridge sites are available for the HOOHad binding on the Pt(111) and
Pt3Os(111) surfaces. Two oxygen atoms bind to two neighboring top sites with the O−O bond parallel
to the Pt-Pt bond. For Pt3Os, the binding energies vary from 0.25 to 0.38 eV and 0.52 to 0.62 eV in
gas phase and solution, respectively. Both are close to the HOOHad binding energy of pure Pt, 0.27
eV in gas phase and 0.61 eV in solution.
28
Summarizing this section, we can say that the Pt3Os binding energies show strong site
dependence. However, we do not find any obvious trend for the binding energies of the Ocontaining species, in particular for the O and OH binding energy in solution. Thus, in the case of
the Pt3Os(111) alloy surface, it is hard to decide whether Pt3Os is a good ORR catalyst based only
on the binding energies of the ORR intermediates. Reaction barrier calculations might help to make
a more thorough conclusion.
3.3 Reaction Barriers and Possible ORR Mechanisms
Generally, eight fundamental steps (Figure 6) for the Langmuir–Hinshelwood mechanism of the
ORR can be considered:
(1) H2 dissociation: H2ad → 2Had
(2) O2 dissociation: O2ad → 2Oad
(3) OH formation: Oad + Had → OHad
(4) O hydration: Oad + H2Oad → 2OHad
(5) OOH formation: O2ad + Had → OOHad
(6) OOH dissociation: OOHad → OHad + Oad
(7) H - OOH dissociation: OOHad + Had→ 2OHad
(8) H2O formation: OHad + Had → H2Oad
By including these fundamental steps into an overall ORR mechanism, we distinguish three
chemical processes:107
O−O bond activation, which can occur via two mechanisms: O2 dissociation (2) and OOH
formation (5) followed by OOH dissociation (6).
29
Figure 6. Eight ORR fundamental steps.
OH formation proceeds via three mechanisms: OH formation (3), O hydration (4), and H−OOH
dissociation (7).
OH consumption: There is only one mechanism (H2O formation (8)) for this process. A good
catalyst must provide low barriers for all three of these processes and for pathways connecting them.
Starting from the preferred sites, we calculated the barriers for all eight steps on the Pt3Os(111)
surface in gas phase and solution. These barriers and the corresponding barriers for pure Pt59 are
shown in Tables 3 and 4. The pathway with the lowest reaction barriers is used for the final
determination of the ORR mechanism. The potential energy surface, including barriers and
geometry insets for the OOH-form-hydr mechanism and O2-diss-hydr mechanism108 for Pt3Os and
Pt, is shown in Figure 7.
30
3.4 Reaction Barriers and ORR Mechanisms for Pt3Os in Gas Phase
O-O bond activation. For pure Pt, OOHad formation with a barrier of 0.28 eV is followed by
OOHad dissociation with a lower barrier of 0.14 eV, whereas the barrier for the direct dissociation is
0.58 eV.
In the Pt3Os case, the OOHad formation barrier is 0.34 eV and the barrier for the OOHad
dissociation is only 0.09 eV, whereas the barrier for the direct O2ad dissociation is 1.36 eV, much
higher than the barrier for the OOHad formation. Thus, it is preferable for the ORR pathway to proceed
via the OOH formation step.
OH formation. The barrier for the direct OHad formation is 0.72 and 0.57 eV for Pt and Pt3Os,
respectively. This is significantly higher compared to the OHad formation via the O hydration step,
0.29 and 0.23 eV for Pt and Pt3Os, respectively.
OH consumption. The H2Oad formation step proceeds with a small barrier of 0.11 eV for Pt and
0.09 eV for Pt3Os.
Summarizing these three processes, we propose the following ORR mechanism for Pt and Pt3Os
in gas phase:
O2 + H → OOH (Ea = 0.28 eV for Pt and 0.34 for Pt3Os)
OOH → O + OH (Ea = 0.14 eV for Pt and 0.09 eV for Pt3Os)
O + H2O → 2OH (Ea = 0.29 eV for Pt and 0.23 eV for Pt3Os)
OH + H → H2O (Ea = 0.11 eV for Pt and 0.09 eV for Pt3Os)
31
Table 3. Reaction barriers (eV) for Pt59 and Pt3Os in gas phase.
Reaction Barriers
Pt
Pt3Os
H2 Dissociation
0.00
0.03
O2 Dissociation
0.58
1.36
OH Formation
0.72
0.57
O Hydration
0.29
0.23
OOH Formation
0.28
0.34
OOH Dissociation
0.14
0.09
H-OOH Dissociation
0.18
0.24
H2O Formation
0.11
0.09
Table 4. Reaction barriers (eV) for Pt59 and Pt3Os in solvated phase.
Reaction Barriers
Pt
Pt3Os
H2 Dissociation
0.00
0.05
O2 Dissociation
0.00
0.16
OH Formation
1.09
0.90
O Hydration
0.50
0.48
OOH Formation
0.19
0.00
OOH Dissociation
0.00
0.00
H-OOH Dissociation
0.04
0.41
H2O Formation
0.17
0.35
32
33
Figure 7. Potential energy surfaces, including reaction barriers for the OOH-form-hydr mechanism
in gas phase (a), OOH-form-hydr mechanism in solution (b), and O2-diss-hydr mechanism in
solution (c) for Pt and Pt3Os.
Figure 7a shows the potential energy surface of this mechanism for Pt and Pt3Os. It starts from
O2 gas labeled as O2(g), and represents the sequential steps of the ORR that follows the fourelectron mechanism: O2 + 4H+ + 4e- → 2H2O. For the purpose of conservation of atoms, we include
4H+ in the first step, although the proton successively comes to the cathode from the electrolyte.
The reactants and products involved in each reaction step are marked by bold font. The ratedetermining step (RDS) is O-hydration with a barrier of 0.29 eV for Pt and OOH formation with a
barrier of 0.34 eV for Pt3Os. Although the RDS barrier for Pt3Os is higher than for Pt, the energy
barrier for the O hydration reaction on the Pt3Os(111) surface is lower (0.23 eV) than for Pt.
34
3.5 Reaction Barriers and ORR Mechanism for Pt3Os in Solvated Phase
O-O bond activation. Since solvent strongly favors O2 dissociation, this mechanism becomes
preferable for the materials considered here. O2 can easily dissociate to form Oad at the fcc site with
no barrier for Pt and with a barrier of 0.16 eV for Pt3Os, where Oad is at the top site (Figure 8). The
solvent effect on the O2 dissociation is stronger on Pt3Os than on Pt (see Tables 3 and 4), which is
related to the stronger solvent effect on the Oad binding on Pt3Os than on Pt. On Pt3Os, O2 first
migrates from the f1a site to the b2a site, and then dissociates to two adsorbed oxygen at the t1b and t2
sites. The O atom adsorbed at the top site generates a stronger dipole than the O atom adsorbed at the
fcc site, which results in the stronger solvent effect on Pt3Os compared to pure Pt.
Figure 8. O2 dissociation on Pt(111) and Pt3Os(111) surfaces.
For comparison, OOHad formation and OOHad dissociation on the Pt3Os surface is barrierless,
which favors the ORR pathway to go via this step.
35
OH formation. Proceeding from the above conclusion, the second step should be O hydration,
because the barrier for the direct OHad formation is 1.09 and 0.90 eV for Pt and Pt3Os, respectively.
The O hydration step with a barrier of 0.50 eV for Pt and 0.48 eV for Pt3Os is significantly favorable.
OH consumption. The H2O formation reaction on the Pt3Os (111) surface occurs with a barrier
of 0.35 eV, while pure Pt has a lower barrier of 0.17 eV.
Summarizing the above discussion, we come to the conclusion that the O2-diss-hydr
mechanism108 is favorable for Pt, but for Pt3Os, both the OOH-form-hydr mechanism and O2-disshydr mechanism in solvated phase are feasible, although the OOH-form-hydr mechanism with the
barrierless OOHad formation and OOHad dissociation steps looks more preferable than the O2-disshydr mechanism:
O-O bond activation: O2 → 2O (Ea = 0.00 eV for Pt, and 0.16 eV for Pt3Os)
O2 + H → OOH (Ea = 0.19 eV for Pt and 0.00 eV for Pt3Os)
OOH → O + OH (Ea = 0.00 eV for Pt and 0.00 eV for Pt3Os)
OH formation: O + H2O → 2OH (Ea = 0.50 eV for P and 0.48 eV for Pt3Os)
OH consumption: OH + H → H2O (Ea = 0.17 eV for Pt and 0.35 eV for Pt3Os)
Figures 7b and c show the potential energy surfaces of the OOH-form-hydr and O2-diss-hydr
mechanisms for Pt and Pt3Os in solution. The RDS for both mechanisms is the O hydration step with
a barrier of 0.50 and 0.48 eV for Pt and Pt3Os, respectively. Spontaneous Oad migration to the most
stable fcc site after the OOH dissociation step or the O2 dissociation step is included in the figure as
well.
36
Based on this result, the ORR catalytic activity of the Pt3Os alloy can be estimated and
compared to that of pure Pt. We find that Pt3Os should show about 2 times better catalytic activity
than pure Pt (e-0.48ev/kT/e-0.50ev/kT = 2.18) at 80oC. Nilekar et al.50 calculated the OH−OH repulsive
energy and correlated it with the measured kinetic current density for Pt0.8M0.2/Pd(111), where M
= Au, Pd, Rh, Re, and Os. Under 0.8 V, the current density for Pt0.8Os0.2/Pd (111) is about 3 times
better than for Pt/Pd(111) (47 vs. 17 mA/cm2). Pt alloys with Os and Re were found to be more
reactive than the others due to their high OH−OH and OH−O repulsive energy. Another possible
way to explain the better performance of Pt3Os is either by the smaller size of the Os atom, which
results in a compressed strain (strain effect) for the surface Pt atoms, or the electronic interaction
(ligand effect) of the subsurface Os atoms with the surface Pt atoms. In general, the strain and
ligand effects occur simultaneously and it is hard to separate them. Both effects are manifested in
the interatomic matrix element describing bonding interaction between an atom and its nearest
neighbors.109 For the Pt3Os catalyst, the electronic structure of the surface Pt is modified by the
underlying Os atoms, and the Pt−Pt bond distance is compressed to 2.76 Å compared to 2.81 Å for
the Pt bulk structure. Nørskov et al. developed the d-band model,9,109,110 which was applied to
connect the surface chemical properties of bimetallic alloy catalysts with their electronic structures.
For compressive strain, the interatomic distances become smaller and the overlapping of metal dstates increases. The increased d-state overlapping results in the increased d-band width, which is
highly correlated with the position of the d-band center due to the fact that the d-band filling
changes negligibly upon the formation of the bimetallic surfaces. To maintain the filling, the
broader d-band downshifts the d-band center, which results in weaker adsorbate bonding. On the
Pt3Os(111) surface, the downshift of the surface d-band center relative to pure Pt(111) is ~0.35 eV
(Figure 9), whereas according to Stamenkovic et al.,23 this downshift should be ~0.2 eV and the
O/OH bindings should be by 0.1−0.2 eV weaker23,26,28 to reach the maximum oxygen reduction
37
activity. Both above-mentioned effects reduce the O and OH binding energies (the O binding
energy on the Pt3Os(111) surface is by 0.11 eV weaker than for Pt(111), 3.55 vs. 3.66 eV, and the
OH binding energy for Pt3Os(111) is by 0.08 eV weaker than for Pt(111), 2.20 vs. 2.28 eV; see
Table 1) in gas phase. However, in solution, which is more relevant to the PEMFC operating
conditions, O binds stronger on Pt3Os than on pure Pt (see Table 2), but the computed RDS barrier
is lower for Pt3Os, 0.48 vs. 0.50 eV for Pt. Therefore, knowing the binding energies of the
intermediates is not always sufficient to reliably predict the ORR activity. Reaction barriers have
to be considered as well to better understand the ORR kinetics. The result obtained in our study
indicates that Pt-Os materials might be considered as potential ORR catalysts.
Figure 9. Densities of states and d-band centers for Pt(111) and Pt3Os(111) surface layers.
38
3.6 Experimental Verification
To verify our simulation results, we performed the dealloying method to synthesize Pt3Os
nanoparticles. Table 5 lists the structural data and composition for our synthesized Pt2Os, DA-Pt2Os,
and commercial Pt nanoparticles. Based on SEM-EDS analysis, the atomic ratio for our DA-Pt2Os
electrocatalyst is ~ Pt : Os = 4 :1, slightly deviating but not far from our target Pt : Os =3 :1.
Table 5. Structural parameters and composition of Pt/C, Pt2Os/C, and DA-Pt2Os/C electrocatalysts.
Sample
Peak
angle
(2θ)
Lattice
constant
(Å)
Particle
sizea (nm)
Particle sizeb
(nm)
Atomic
ratioc
Atomic
ratiod
Pt/Os
Pt/Os
Pt/C
39.62
3.94
2.09
2.3±0.4
Pt2Os/C
41.01
3.81
3.65
3.1±0.7
68/32
67/33
DAPt2Os/C
40.43
3.85
3.7
3.1±1
93/7
81/19
a. Mean particle size calculated from Debye-Scherrer equation on Pt (111) and Pt2Os (220) planes.
b. Mean particle size observed from TEM images.
c. Atomic ratio determined by TXRF.
d. Atomic ratio determined by SEM-EDS.
The elemental profiles of Pt and Os in the Pt2Os and DA-Pt2Os nanoparticles were obtained by
the STEM-EDS measurements with a probe size of about 1.5 Å. Figure 10 (a) exhibits a STEM image
and the corresponding line-scan profile across a Pt2Os nanoparticle in 4.5 nm size. The result
suggested an alloy state for the as-synthesized Pt2Os nanoparticles. Figure 10 (b) shows a STEM
image and the line-scan profile of a DA-Pt2Os nanoparticle of 7 nm in diameter. Apparently, the Os
atoms residing on the surface regime were mostly removed and the DA-Pt2Os (Pt4Os) revealed a
quasi-core-shell structure in which the core retained some Os atoms, whereas the shell was occupied
by the Pt atoms exclusively.
39
Figure 10. The STEM images and EDS line-scan of (a) Pt2Os (b) DA-Pt2Os nanoparticles.
Figure 11 demonstrates the ORR CV curves for Pt2Os/C, DA-Pt2Os/C, and Pt/C. In the
literature, at potential below 0.6 V, the ORR response is under mass transport control limited by the
diffusion of the dissovled oxygen in the electrolyte, whereas at potential between 0.8 and 1 V the
ORR response is dominated by kinetics (the electrocatalytic activity of the electrocatalyst involved
in the ORR process).111 Hence, a simple method to quickly evaluate the ORR behavior of a potential
40
electrocatalyst is the reading of half-wave potential, which is defined as the potential at which the
magnitude of the current is half of the limiting current. In general, the larger the half-wave potential,
the greater the ORR activity. As shown, the half-wave potentials for the Pt/C, Pt2Os/C, and DAPt2Os/C were 891, 837, and 908 mV, respectively. Moreover, the DA-Pt2Os/C exhibited the highest
on-set potential of 1 V, and this provided more evidence of better electrocatalytic ability among these
samples.
Figure 11. The ORR curves of Pt/C, Pt2Os/C, and DA-Pt2Os/C in apparent current density.
Figure 12 (a) demonstrates the ORR curves from the DA-Pt2Os/C at various rotation speeds of
RDE. Among these curves, the ORR responses at voltage below 0.6 V were stabilzed at the limiting
41
currents, whose values were proportional to the rotation speed as expected. To extract the kinetic
information, we employed the Koutecky-Levich equation listed below,112
1/i = 1/ikinetic + 1/idiffusion limit = 1/ikinetic + 1/0.62nFADO22/3ω1/2v-1/6CO2
(1)
where i is the experimentally-measured current, idiffusion limit is the diffusion limiting current due to the
limitation of mass transport of dissolved oxygen in the 0.1 M aqueous HClO4 solution, ikinetic is the
kinetic current associated with the ORR activity of DA-Pt2Os/C, n is the number of electron
transferred in the ORR process, F is the Faraday constant, A is the reaction area of the RDE (0.196
cm2), DO2 is the diffusivity of dissolved oxygen in the 0.1 M aqueous HClO4 solution (1.93 × 10-5 cm2
s-1),113 ω is the rotation speed of the RDE, v is the kinematic viscosity of the 0.1 M aqueous HClO4
solution (1.009 × 10-2 cm2 s-1),112 and CO2 is the concentration of dissolved oxygen in the 0.1 M
aqueous HClO4 solution (1.26 × 10-3 mol L-1).113 Figure 12 (b) provides the Koutecky-Levich plots at
different potentials. Obviously, these curves showed a consistent pattern and the average slope was
11.11 mA-1 s-1/2. Thus, the resulting n value became 3.94. Therefore, we concluded that the DAPt2Os/C nanoparticles adopted a four-electron route to catalyze the ORR process, similar to that of
Pt.
The values of ikinetic obtained from the Koutecky-Levich equation were used to calculate the
mass activity and specific activity shown in Figure 13. These Tafel plots were obtained using ikinetic
values from 0.95 to 0.8 V divided by the effective mass of the Pt in the RDE (from ICP-MS) and
42
Figure 12. (a) The ORR curves of DA-Pt2Os/C at various rotation speeds. The electrolyte was
oxygen-saturated 0.1 M aqueous HClO4 solution and the scan rate was 10 mV s-1. (b) The KouteckyLevich plot of DA－Pt2Os/C at different voltages.
43
ECSA (cmPt2) values. For the Pt/C, its mass activity and specific activity at 0.9 V were 0.14 mA
μgPt-1 and 0.28 mA cmPt-2, respectively. However, the Pt2Os/C exhibited a lower mass activity (0.05
mA μgPt-1) and a greater specific activity (0.52 mA cmPt-2), as compared to those of Pt/C. The reduced
mass activity is attributed to its relatively larger size as compared to that of Pt/C because the
predominant share of the Pt atoms was residing at the core. However, the 250% increment in the
specific activity suggested the complementary role of Os atoms to the Pt atoms for ORR activity. This
is because the surface of the Pt2Os nanoparticle was partially occupied by the Os atoms, and those Os
atoms were oxyphilic and thus promote the ORR process via bifunctional model and electronic
mechanism. In the bifunctional model, the Os atoms exhibit a strong affinity toward the adsorption
of OH and as a result lead to a considerable reduction in the Pt-OH coverage.72 In general, a larger
percentage of OH coverage on the Pt sites is considered disadvantageous for the ORR action.
According to Zhang et al., two OH species can adsorb onto a single Os atom and subsequently initiate
a spontaneous breakup of the O-H bonding forming an oxygen atom and water.72
Once the dealloying process was completed, the DA-Pt2Os demonstrated a significantlyenhanced mass activity (0.29 mA μgPt-1) and specific activity (1.03 mA cmPt-2). We realized that the
removal of Os atoms engendered a Pt-enriched surface with a substantially-enlarged ECSA. At this
stage, the contributory role played by the Os atoms can be explained mostly by the electronic effect
because the electronic structure of the surface Pt atoms was expected to be altered by the Os atoms
underneath, as our previous discussion in section 3.5 describes.
For durability tests, the DA-Pt2Os and Pt/C samples were subjected to multiple CV scans in a
potential window of 0.65 and 1.05 V at 50 mV s-1 in a 0.1 M aqueous HClO4 solution following
procedures used by Wang et al.115 Figure 14 shows the ORR responses before and after the
durability test for DA-Pt2Os/C and Pt/C, (a) and (b), respectively. Apparenttly, for both samples
44
Figure 13. The Tafel plots of the ORR curves (Figure 11) in (a) mass activity and (b) specific activity.
45
Figure 14. The ORR curves before and after the durability test for (a) DA-Pt2Os/C and (b) Pt/C. The
electrolyte was oxygen-saturated 0.1 M aqueous HClO4 solution and the scan rate was 10 mV s-1.
46
there appeared a very subdued degradation during the durability tests, and both the onset potentials
and diffusion-limiting currents remained almost unchanged. The minor degradation is manifested in
the slight negative shift of the half-wave potential at 5 and 9 mV for DA-Pt2Os/C and Pt/C,
respectively. These CV curves suggested that the stability of DA-Pt2Os/C was comparable or even
better than that of commercially available Pt nanoparticles. Since the Os atoms are easily oxidized in
the electrolyte, a stable Pt-shell structure formed for DA-Pt2Os/C catalyst.
The DFT study of the Pt segregation on the Pt4Os slab is performed to support the experimental
observation and the results were summarized in Figure 15. The five layer 2×2 cell slab contains 16 Pt
atoms and 4 Os atoms, providing the 4:1 ratio. 4 Os atoms cannot be equally distributed in five layers,
so one of the layers should be pure Pt. The a slab shows a uniform Os distribution in four lower layers
with the pure Pt top layer. This structure was chosen because from our previous calculations29 we
know that the Pt3Os alloy demonstrates strong surface segregation and the most stable Pt3Os slab
structure has 100% Pt at the top surface layer, 50% Pt in the second layer, and 75% Pt in the following
layer.
The b structure has pure Pt at the top and in the second layer, but with 50% Os in the third layer,
and 25% Os in the following layers.
The c structure also has two top layers of Pt but is Os enriched in the bottom layer rather than in
the third layer.
Our calculation shows that the most stable structure is b, followed by a and c (see Figure 15).
This is consistent with our previous segregation study of Pt-based binary alloys,29 which found that
the most stable structure of the segregated alloys has Pt-skin on the surface. The 6 layer 2×2 and 4
layer 3×3 cell models provide a similar result. Therefore, both experiment (STEM-EDS analysis) and
47
theory (computational modeling) demonstrate that the DA-Pt2Os (Pt4Os) prefers the Pt-skin surface
structure. The Pt segregation is expected to influence the electronic structure of the alloy and affects
its electrochemical activity as we mentioned before.
Top view
Side view
0.85 eV
0 eV
0.88 eV
Figure 15. The surface segregation models for 5 layers of DA-Pt2Os electrocatalysts. (a) the
uniformly distributed structure, (b) the segregated structure with Os enriched in the third layer, and
(c) the segregated structure with Os enriched in the bottom layer. The energies are relative to the
energy of the most stable structure b, at -48849.55 eV.
48
3.7 Summary for Pt3Os Catalyst
Surface segregation of a number of Pt3M alloys was investigated. Only a few Pt binary alloys
demonstrate favorable surface segregation energy in the presence of adsorbed O or OH. Out of them,
only Pt3Os and Pt3Ir show surface segregation in the presence of both adsorbed species. This assumes
that unlike other Pt-based binary alloys, Pt3Os and Pt3Ir might be stable under fuel cell operating
conditions. We systematically studied the binding site preference of all reaction intermediates
involved in ORR on Pt3Os in gas phase and solution. The binding energies of adsorbates on the alloy
surface show the strong sublayer dependence. Reaction barriers for the eight ORR fundamental steps
were also calculated for Pt3Os both in gas phase and solution, and compared to those for pure Pt.
According to our result, Pt3Os has a slightly lower energy barrier for the ORR RDS than pure Pt,
which should result in better catalytic activity compared to pure Pt. This conclusion is in agreement
with our earlier performed experimental study of Pt-Os catalysts, in which we found that the ORR
mass and specific activities of dealloyed Pt2Os materials are 2 and 3.5 times better than the
corresponding activities of pure Pt. In addition, these materials show good electrochemical stability.
Thus, Pt-Os alloys might be considered as promising ORR catalysts.
49
Chapter 4
Results and Discussion for Os/Pt Core-Shell
Catalysts
4.1 Stability of the Os/Pt Core-Shell Structure
The crystal structure of Os is hcp, while that of Pt is fcc. The stacking sequences for the closest
packed planes of hcp (0001) and fcc (111) are ABABAB and ABCABC, respectively. To determine
structures with different numbers of Pt layers deposited on the Os core, we calculated all the
stacking sequences and compared their energies (see Table 6). Each letter in the “Pt/Os surface
structure” row represents a specific layer in the structure. The capital letters represent Os layers,
while the lower cases denote Pt layers. For example, the label b/ABA means that the structure has
a Pt monolayer with the b stacking on the three Os layer substrate with the ABA stacking sequence.
The lowest energy stacking sequence result conforms to the similar DFT result for the
Ru(hcp)/Pt(fcc) core-shell catalyst.61,116 In addition, our recent study using HAADF STEM
technique proves the energetically favorable cb/ABAB stacking sequence for Pt2ML/Ru obtained
from the DFT calculation.116 We used the lowest energy stacking sequences as the stable structures
for calculations of binding energies and reaction energy barriers on Pt/Os slabs.
50
Table 6. Relative energies (eV) for Pt/Os surface structures with different stacking sequences.
Pt/Os surface structure
Relative Energy
Pt1ML/Os
b/ABA
0.07
c/ABA
0.00*
Pt2ML/Os
ab/ABA
0.53
cb/ABA
0.00*
bc/ABA
0.40
ac/ABA
0.81
Pt3ML/Os
bab/ABA
0.82
cab/ABA
0.32
acb/ABA
0.00*
bcb/ABA
0.52
abc/ABA
0.29
cbc/ABA
0.66
bac/ABA
0.64
cac/ABA
1.03
*Reference energy: c/ABA: -76283.72 eV; cb/ABA: -99327.96 eV; acb/ABA: -122371.54 eV. All
energies are relative to the lowest one (bold numbers).
51
4.2 Binding Sites and Binding Energies
4.2.1 Binding Site Notations
Figure 16 shows the binging sites on Os and Pt/Os surfaces. There are four types of sites on
the closed-packing plane:
• On top, bonded to one Os or Pt atom (μ1), denoted as t
• Bridging between two Os or Pt atom (μ2), denoted as b
• An fcc position between three Os or Pt atom (μ3), denoted as f
• An hcp position between three Os or Pt atom (μ3), denoted as h
Figure 16. Binding sites on Os (left) and Pt/Os (right) slab surfaces.
In our notations, capital letters represent the binding species, such as H, O, O2, etc. The lower
case letters denote the binding sites. For OH and OOH, two and three lower case labels are applied
to distinguish the different orientations of the species. Figure 17 illustrates some examples for the
labels of OH and OOH species on the Pt/Os catalyst surface. OH/b-f means OH with O adsorbed
52
at the b bridge site and H at the f site. OOH/t-b-f denotes OOH with the first O at the top site,
the second O at the b bridge site, and the H atom at the f site. OOH/t-f labels only the O positions:
the first O at the top site and the second O at the f site. Similar notation rules are applied to other
ORR intermediates. Figure 18 shows binding energies of the ORR intermediates at the most stable
sites on the Os, Pt, and Pt/Os surfaces in vacuum and solvent. All binding energies for the ORR
species at various surface sites are summarized in Tables 7 and 8. It should be noted that after
geometry optimization, the OH/f and OH/h spontaneously move to the sites which can be
considered as OH/b-f and OH/b-h with similar corresponding energy values.
Figure 17. OH and OOH species on the Pt/Os catalyst surface: (a) OH/b-f (b) OOH/t-b-f and (c)
OOH/t-f.
H binding. On the pure Os surface, both in gas phase and solvent, H prefers the f site, but the
binding energy difference between various sites is not significant.
For Pt and Pt/Os surfaces, the t site is most preferable for the H binding. The binding energy
values lie between -2.45 and -2.81 eV in gas phase and -2.61 and -2.90 eV in solvation phase. It is
worthy to notice that the binding energy rises with an increasing number of deposited Pt layers and
approaches the binding energy for the pure Pt surface. A similar trend is observed for most of the
ORR species.
53
Figure 18. Binding energies of the ORR intermediates at the most stable sites on Os, Pt, and Pt/Os
surfaces in gas phase (a) and solution (b).
54
Table 7. Binding energiesa (eV) of ORR species at various binding sites on Os, Pt, and Pt/Os
surfaces in gas phase.
Species
pure Os
pure Pt
Pt1ML/Os
Pt2ML/Os
Pt3ML/Os
H/b
-2.72
-2.65
-2.34
-2.53
-2.69
H/f
-2.80
-2.66
-2.35
-2.53
-2.68
H/h
-2.71
-2.62
-2.34
-2.53
-2.66
H/t
-2.69
-2.73
-2.45
-2.64
-2.81
O/b
-4.57
-3.20
-2.90
-2.98
-3.16
O/f
-4.87
-3.68
-3.16
-3.29
-3.55
O/h
-5.21
-3.33
-2.97
-3.08
-3.27
O/t
-4.26
-2.62
-2.36
-2.29
-2.33
OH/b-f
-3.24
-2.31
-2.26
-2.17
-2.28
OH/b-h
-3.20
-2.30
-2.26
-2.17
-2.27
OH/f
Unstableb
OH/h
Unstableb
OH/t
-3.09
-2.28
-2.27
-2.21
-2.29
O2/b
-1.80
-0.56
-0.33
-0.28
-0.45
O2/f
-1.78
-0.49
-0.28
-0.19
-0.43
O2/h
-1.89
-0.43
-0.30
-0.24
-0.44
OOH/t-b-f
unstable
-1.09
-1.02
-0.99
-1.11
OOH/t-b-h
unstable
-1.09
-1.02
-0.99
-1.11
OOH/t-f
-1.73
-1.01
-0.92
-0.91
-1.05
OOH/t-h
-1.68
-1.03
-0.91
-0.91
-1.05
HOOH/b
-0.55
-0.27
-0.26
-0.29
-0.32
HOH/t
-0.50
-0.26
-0.23
-0.26
-0.30
55
HOH-down/t
-0.05
-0.02
-0.05
-0.06
-0.09
The estimated value for the basis set superposition error (BSSE) is ∼0.05 eV.
After geometry optimization, OH/f and OH/h moved to the bridge sites OH/b-f and OH/b-h.
Table 8. Binding energiesa (eV) of ORR species at various binding sites on Os, Pt, and Pt/Os surfaces
in solution.
Species
pure Os
pure Pt
Pt1ML/Os
Pt2ML/Os
Pt3ML/Os
H/b
-2.90
-2.76
-2.46
-2.63
-2.81
H/f
-2.94
-2.79
-2.46
-2.62
-2.78
H/h
-2.84
-2.75
-2.44
-2.61
-2.75
H/t
-2.83
-2.82
-2.61
-2.72
-2.90
O/b
-5.08
-3.95
-3.59
-3.65
-3.87
O/f
-5.26
-4.49
-3.90
-4.11
-4.27
O/h
-5.67
-4.05
-3.67
-3.82
-4.01
O/t
-5.17
-3.47
-3.59
-3.22
-3.19
OH/b-f
-3.50
-2.72
-2.79
-2.51
-2.62
OH/b-h
-3.38
-2.71
-2.70
-2.61
-2.65
OH/f
Unstableb
OH/h
Unstableb
OH/t
-3.72
-2.83
-3.10
-2.80
-2.83
O2/b
-2.56
-1.05
-0.86
-0.61
-0.87
O2/f
-2.21
-0.94
-0.78
-0.59
-0.85
O2/h
-2.40
-0.83
-0.79
-0.57
-0.72
OOH/t-b-f
unstable
-1.55
-1.73
-1.45
-1.56
OOH/t-b-h
unstable
-1.58
-1.67
-1.52
-1.58
OOH/t-f
-2.24
-1.53
-1.52
-1.46
-1.49
OOH/t-h
-2.35
-1.50
-1.66
-1.39
-1.52
56
HOOH/b
-0.84
-0.68
-0.58
-0.66
-0.71
HOH/t
-0.90
-0.67
-0.63
-0.65
-0.71
HOH-down/t
-0.51
-0.46
-0.60
-0.50
-0.49
The estimated value for the BSSE is ∼0.05 eV.
After geometry optimization, OH/f and OH/h moved to the bridge sites OH/b-f and OH/b-h.
O binding. On the pure Os surface, O binds stronger than on the Pt or Pt/Os surfaces. The
binding energy on the Os surface is -5.21 eV at the h site and ranges from -3.16 to -3.68 eV at the
f site for Pt and Pt/Os.
In solvation phase, the most stable binding sites are the same as in gas phase. The differences
between the O binding energy values on the Os and Pt, Pt/Os surfaces are smaller.
OH binding. In gas phase, the most stable OH binding site on the Os and Pt surfaces is the b site,
whereas the t site is most stable for the Pt/Os surfaces. The binding energy on the Os surface in gas
phase is -3.24 eV, which is lower than -2.31 and -2.21 to -2.29 eV for the Pt and Pt/Os surfaces,
respectively. Interestingly, for all catalysts OH shows no significant binding site preference in gas
phase, unlike the solvation phase where the t site is clearly preferable.
In solvation phase, the t site is the most stable for the all catalysts. The OH binding energy for
the Pt1ML/Os is -3.10 eV, higher than on the Os surface (-3.72 eV) but lower than on the Pt surface (2.83 eV). The OH binding energy does not exhibit an obvious trend versus different numbers of
deposited Pt layers, as the values vary from -2.80 to -3.10 eV.
O2 binding. We use the center of the O-O bond to denote the binding sites of O2. Thus, the O2/b
means that two O atoms are located approximately on the top of the surface Os or Pt atoms with the
O-O bond center at the b site. The O2/f and O2/h binding sites are defined as one O atom located on
57
top of the Os or Pt atom, and the other at the b site with the O-O bond center at the f or h site,
respectively. The O2 binding on the Os surface is stronger than on the Pt and Pt/Os surfaces both in
gas phase and solution. The bridge site is the most stable for the O2 adsorption on all surfaces, except
for the Os surface in gas phase, where O2 binds most strongly at the h site.
We find that the O2 binding energy differences for the b and f sites on Pt2ML/Os and Pt3ML/Os
surfaces are as small as 0.02 eV in solvent, which means that during the ORR process, O2 can
probably be adsorbed at both sites. It was also found that the O2 binding energy for the Pt2ML/Os
catalyst is higher than the O2 binding energy for the Pt1ML/Os and Pt3ML/Os catalysts. The reason for
this violation of the expected linear trend is not clear yet.
OOH binding. For notations of the OOH binding sites we used two or three letters. The first letter
denotes a position of the first oxygen atom, the second letter denotes a position of the second oxygen
atom, and the third letter denotes a position of the hydrogen atom. For example, OOH/t-b-f means
that the first oxygen is located on the top of Pt or Os atom, the second oxygen atom is at the b site
(actually, it is between the top and bridge sites), and the hydrogen is at the f site (see Figure 17b). For
OOH/t-f or OOH/t-h, the symmetry makes the O-H bond toward right or left be the same, and
therefore only two letters can be used to denote positions of the two oxygen atoms (see Figure 17c).
For the OOH binding on the Os surface in gas and solvation phases, the OOH/t-b-f and OOH/t-b-h
sites are both unstable and OOH spontaneously decomposes into O and OH at the top sites. We find
that the OOH binding for the Pt2ML/Os catalyst is weaker than that for the Pt1ML/Os and Pt3ML/Os
catalysts, similar to the O2 binding case.
HOOH and HOH binding. There is only one site for the HOOH binding: the two O atoms bind
to two neighboring Pt atoms similar to the O2/b case with the O-O bond parallel to the surface Pt-Pt
bond. For the HOH binding, the O atom binds at the top site with the two O-H bonds parallel to the
58
surfaces. For the HOH-down binding, one H atom binds at the top site with the remaining O-H
bond almost parallel to the surface. The HOOH and HOH binding energies both in gas and solvation
phases fit the general trend described before, i.e. the binding energies rise with the increasing number
of deposited Pt layers and approach the binding energies for the pure Pt surface. Furthermore, the
corresponding binding energies on the Os and Pt3ML/Os surfaces are lower than on the pure Pt surface.
Summarizing, we can say that the binding on the Os surface is stronger than on the Pt or Pt/Os
surfaces in gas and solvated phases. The general trend both in gas phase and solution is that the
binding becomes stronger when the number of deposited Pt layers increases.
4.3 Reaction Energy Barriers
Based on our previously obtained results,59,91,108 the ORR could be divided into three
fundamental stages:
I. O2O: dissociation of O2 to become O, which could be either via direct O2 dissociation:
O2O or OOH dissociation: OOHO+OH following OOH formation: O2+HOOH
II. OOH: OH formation. This step could be OH formation, O hydration, or H-OOH
dissociation: O+HOH, O+H2O2OH, H+OOH2OH
III. OHH2O: H2O formation from OH generated in the second stage: OH+HH2O
Here we consider the reactions which proceed via the Langmuir–Hinshelwood mechanism.
According to our previous result, the estimated energy barrier for a hydronium ion adsorbed on the
Pt surface is 0.25 eV,108 which is consistent with other published results117,118 and lower than the RDS
barrier for pure Pt, 0.50 eV (0.37 eV, if the H-OOH-diss mechanism is realized for Pt). A co-adsorbed
hydronium ion with anions has also been observed experimentally at potential higher than 0.6V
(RHE).119 Figure 19 shows the ORR pathway and potential energy surface for Os, Pt, and Os/Pt core-
59
shell catalysts in gas phase. The reaction energy barriers for each ORR step in gas phase are listed
in Table 9.
60
Figure 19. Potential energy surface including reaction barriers for the O2-diss-hydr mechanism for
Os (a), OOH-form-hydr mechanism (b), and H-OOH-diss mechanism (c) for Pt and Pt/Os catalysts
in gas phase.
In stage I, the O2 dissociation energy barrier is higher than the OOH formation/dissociation
barriers for the Pt and Os/Pt core-shell catalysts. Thus, it is more difficult for the ORR to start from
the direct O2 dissociation, but for Os this step is favorable. This is due to the stronger O binding on
the Os surface, which prompts direct O2 dissociation.
In stage II, the direct OH formation step has a much higher barrier, which makes the oxygen
hydration reaction and H-OOH dissociation more feasible, because the reaction proceeds along the
lowest energy path.
61
Table 9. Reaction energy barriers (eV) for ORR steps on Os, Pt, and Pt/Os surfaces in gas phase.
Step
Os
Pt
Pt1ML/Os
Pt2ML/Os
Pt3ML/Os
HH dissociation
0.16
0.00
0.24
0.20
0.13
O2 dissociation
0.00
0.56
0.90
0.77
0.63
OH formation
1.38
0.74
0.50
0.55
0.68
H2O formation
0.69
0.26
0.36
0.23
0.19
OOH formation
0.81
0.31
0.18
0.23
0.29
OOH dissociation
0.00
0.09
0.29
0.27
0.17
H-OOH dissociation 0.00
0.06
0.29
0.25
0.23
O hydration
0.27
0.04
0.20
0.19
0.35
In stage III, we find that the H2O formation reaction energy barrier decreases, while the OOH
formation energy barrier increases with the increasing number of deposited Pt layers. Thus, the RDS
is a compromise between the OOH formation reaction (stage I) with the energy barrier from 0.18 to
0.31 eV and the H2O formation reaction (stage III) with the energy barrier from 0.19 to 0.36 eV for
the Pt and Os/Pt catalysts in gas phase. For pure Os, the O2 dissociation is barrierless, but the H2O
formation has a barrier of 0.69 eV, much higher than the corresponding values for the Pt or Os/Pt
catalysts. This is because of the much stronger OH binding, which makes difficult the H2O formation
reaction. Therefore, the RDS for pure Os is the H2O formation reaction.
The ORR potential energy surface and reaction barriers in solution are shown in Figure 20 and
listed in Table 10. Although the O2 dissociation barriers are greatly reduced due to the solvent effect,
the O2 dissociation reaction and oxygen hydration reaction barriers show opposite trends versus the
increasing number of deposited Pt layers. This occurs because the O binding energy in solvent rises
62
as the number of deposited Pt layer increases. The lower O binding energy benefits the O2
dissociation reaction but hinders the O hydration reaction. Similar phenomena have been reported for
pure metals in stage I and III.107 The reaction energy barriers for the direct O2 dissociation followed
by the oxygen hydration reaction are higher than the barriers for the path starting from the OOH
formation reaction and followed by the H-OOH dissociation reaction. Thus, the reaction path via the
OOH formation is favorable for pure Pt and Os/Pt catalysts, which is in agreement with results
obtained by other theoretical groups13,117 that proposed a similar ORR mechanism for Pt. The RDS
barrier is again a compromise between stage I and stage III.
Figure 21 shows the reaction energy barriers for the OOH formation reaction and H2O formation
reaction in solvated phase. The values are from Table 10. The OOH formation barrier increases with
an increasing number of Pt deposited layers, while the H2O formation barrier decreases. The two lines
intersect at about 0.22 eV, which approximately corresponds to 2 ML deposited Pt. The RDS is a
compromise between stage I and stage III. The best activity is observed for ~2 layers of Pt deposition.
For pure Os, the OOH formation barrier is much higher because of the stronger O2 binding.
Thus, the ORR proceeds through the direct O2 dissociation and then the O hydration reaction. The
RDS is 0.64 eV for Os, while for pure Pt, Pt1ML/Os, Pt2ML/Os, and Pt3ML/Os the RDS is 0.37, 0.26,
0.23, and 0.35 eV, respectively. Therefore, we may expect that the ORR catalytic activity obeys
the following trend, Pt2ML/Os>Pt1ML/Os>Pt3ML/Os> Pt>Os, which allows us to consider the Os/Pt
core-shell materials as potential candidates for the ORR catalysts.
63
Figure 20. Potential energy surface including reaction barriers for the O2-diss-hydr mechanism for
Os (a) and H-OOH-diss mechanism (b) for Pt and Pt/Os catalysts in solution.
64
Table 10. Reaction energy barriers (eV) for ORR steps on Os, Pt, and Pt/Os surfaces in solution.
Step
Os
Pt
Pt1ML/Os
Pt2ML/Os
Pt3ML/Os
HH dissociation
0.14
0.00
0.15
0.15
0.09
O2 dissociation
0.00
0.00
0.37
0.21
0.00
OH formation
1.92
1.09
0.61
0.83
0.92
H2O formation
0.64
0.32
0.26
0.22
0.16
OOH formation
0.87
0.37
0.15
0.23
0.35
OOH dissociation
0.00
0.00
0.00
0.00
0.00
H-OOH
dissociation
0.00
0.00
0.00
0.04
0.00
O hydration
0.42
0.45
0.06
0.37
0.42
Figure 21. Reaction energy barriers for the OOH formation and H2O formation reactions vs. a
number of deposited Pt layers in solution.
65
Although we do not consider here the influence of the electrode potential on the ORR in
detail, we have briefly analyzed the thermodynamic effect of the electrode potential by examining the
Eley-Rideal mechanism. For this, we applied the approach developed by Norskøv et al.120-122 In this
approach, the reaction energy barriers are not considered properly, but assumed to be equal at least to
the energy differences of the corresponding endothermic reactions, whereas the exothermic reactions
are regarded as spontaneous and barrierless.
Since O2 is not well described by DFT PBE,121 we set the energy of the reaction
H2(g)+1/2O2(g)H2O(g) to the experimental value of the Gibbs free energy, −2.46 eV. Therefore, the
O2 energy could be determined by calculating the H2 and H2O energies as reference energies. The
binding energies of the ORR species in solution (Table 8) were applied for each ORR step of the O2diss-hydr and H-OOH-diss mechanisms. Three potentials, 0.00, 0.80, and 1.23 V, were applied in our
calculations. The ORR potential energy surfaces for the Eley-Rideal reactions are shown for the O2diss-hydr and H-OOH-diss mechanisms in Figures 22 and 23, respectively.
The ORR pathways obtained for the Os, Pt, and Pt/Os surfaces using the Eley-Rideal mechanism
are consistent with those resulted from the Langmuir–Hinshelwood mechanism. On the Os surface,
the O2-diss-hydr mechanism is preferable, whereas on the Pt and Pt/Os surfaces, the H-OOH-diss
mechanism dominates. At potential lower than 0.80 V, the ORR is almost barrierless for the Pt/Os
catalysts, with a slightly higher barrier for pure Pt (0.17 eV at 0.80 V). At 1.23 V, the RDS for
Pt1ML/Os is the H2O formation reaction with a barrier of 0.45 eV, while for Pt2ML/Os, Pt3ML/Os, and
Pt, the RDS is the OOH formation reaction with a barrier of 0.23 eV, 0.42, and 0.60 eV, respectively.
The RDS is again a compromise between the H2O formation and OOH-formation reactions and
Pt2ML/Os is the best among the catalysts considered here.
Figure 22 (a) shows the O2 reduction via the O2 dissociation (O22O) and O-hydration (O+H2O
66
2OH) on Os slab at the three different potentials. The energy difference between the first and
last steps at 1.23 V is the H2O binding energy, since we have set the energy of the reaction
H2(g)+1/2O2(g) H2O(g) to −2.46 eV. We find that the potential-independent O-hydration reaction
(0.27 eV) determines the ORR rate at 0 V, while the H2O formation reaction is the RDS at higher
potentials (0.37 eV at 0.80 V and 0.80 eV at 1.23 V).
If we consider the O2-diss-hydr mechanism on the Pt and Pt/Os surfaces, the O-hydration reaction
is the RDS for almost all catalysts (0.65 eV for Pt, 0.31 eV for Pt2ML/Os, and 0.48 eV for Pt3ML/Os),
except for Pt1ML/Os, where the O-hydration reaction is practically barrierless and the more sluggish
H2O formation reaction retards the ORR (Figure 22 b-e and Table 10). The high barrier for the Ohydration reaction makes the ORR run via the OOH formation, H-OOH dissociation, and H2O
formation steps (Figure 23). For Os, the potential-dependent OOH formation reaction has a higher
barrier than the H2O formation reaction. Thus, the O2-diss-hydr mechanism is more relevant for the
ORR on the Os surface.
On the pure Pt surface, the OOH formation reaction, which is the RDS, has a lower barrier than
the O-hydration reaction and, therefore, the H-OOH-diss mechanism is preferable on the Pt surface.
On the Pt/Os surfaces, the ORR are almost barrierless at 0 V and 0.80 V. More important is the
barrier at a higher potential, because the activation region of the ORR polarization curves begins from
~ 1.00 V vs. RHE (see the ORR curves in section 4.5 Experimental Results). At 1.23 V, a compromise
between the H2O and OOH formation reactions determines the ORR rate (similar to Figure 21). The
barrier for the H2O formation reaction decreases with the increasing number of the Pt layers deposited,
at 0.45, 0.12 and 0.09 eV for Pt1ML/Os, Pt2ML/Os, and Pt3ML/Os, respectively. On the other hand, the
barrier for the OOH formation reaction generally increases with the increasing number of the
deposited Pt layers, at 0.26, 0.23 and 0.42 eV for Pt1ML/Os, Pt2ML/Os, and Pt3ML/Os, respectively. The
67
best catalyst is, therefore, Pt2ML/Os. This conclusion is consistent with our result for the Langmuir–
Hinshelwood mechanism.
Summarizing, we can say that the result obtained for the Eley-Rideal mechanism generally
agrees with that obtained for the Langmuir–Hinshelwood mechanism and does not affect our
conclusions.
68
69
Figure 22. Potential energy surfaces for the O2-diss-hydr mechanism for Os (a), Pt (b), Pt1ML/Os
(c), Pt2ML/Os (d), and Pt3ML/Os (e) in solution.
70
71
72
Figure 23. Potential energy surfaces for the H-OOH-diss mechanism for Os (a), Pt (b), Pt1ML/Os
(c), Pt2ML/Os (d), and Pt3ML/Os (e) in solution.
73
4.4 Strain and Ligand Effects
There are many publications both experimental and theoretical (see, for instance, refs
109,123,124) that discuss the strain and ligand effects for the heteroepitaxial metal layers (bimetallic
overlayers). The strain effect arises due to the bond length difference between the deposited layers
and the substrate, whereas the ligand effect describes the effect owing to the heterometallic bonding
between the surface atoms and the substrate atoms which changes the electronic structure.
To study the strain effect, we compress the Pt bond lengths to the Os bond length value and
compare the binding energies of different ORR species and reaction barriers for Pt and Pt3ML/Os.
Figures 24 compares the binding energies of the ORR species on the pure Pt, compressive Pt,
Pt1ML/Os, and Pt3ML/Os surfaces in gas phase and solution (the corresponding values are listed in
Tables 11 and 12, respectively). In both phases, all binding energies for Pt3ML/Os are similar to those
for compressive Pt, which includes the strain effect but excludes the ligand effect. On the other hand,
the binding energy differences between compressive Pt and Pt1ML/Os are larger. This phenomenon is
more obvious in solvation phase than in gas phase.
Since the ligand effect descends versus the increasing number of deposited Pt layers, the ligand
effect is more significant in Pt1ML/Os than in Pt3ML/Os. This result supports the viewpoint that the
ligand effect of the Os substrate plays an important role. A similar conclusion could be made from
the reaction energy barriers (Figure 25, Tables 13 and 14). The barrier differences between pure Pt,
Pt3ML/Os, and compressive Pt are lower than the differences between Pt1ML/Os and compressive Pt.
This implies that the ligand effect of the Os substrate rather than the strain effect is responsible for
the improved ORR catalytic activity. The RDS barrier becomes higher and approaches the value
74
Figure 24. Binding energies of ORR species for Pt, compressive Pt, and Pt/Os surfaces in gas
phase (a) and solution (b).
75
Table 11. Binding energiesa (eV) of ORR species on Pt, compressive Pt, and Pt/Os surfaces in
gas phase.
Species
Pt
compressive Pt Pt1ML/Os
Pt3ML/Os
-2.73
-2.76
-2.45
-2.81
-3.68
-3.45
-3.16
-3.55
OH
-2.31
-2.23
-2.27
-2.29
O2
-0.56
-0.33
-0.33
-0.45
OOH
-1.09
-1.02
-1.02
-1.11
HOOH
-0.27
-0.27
-0.26
-0.32
HOH
-0.26
-0.25
-0.23
-0.30
The estimated value for the BSSE is ∼0.05 eV.
Table 12. Binding energiesa (eV) of ORR species on Pt, compressive Pt, and Pt/Os surfaces in
solution.
Species
Pt
compressive Pt Pt1ML/Os
Pt3ML/Os
-2.82
-2.84
-2.61
-2.90
-4.49
-4.23
-3.90
-4.27
OH
-2.83
-2.79
-3.10
-2.83
O2
-1.05
-0.77
-0.86
-0.87
OOH
-1.58
-1.53
-1.73
-1.58
HOOH
-0.68
-0.70
-0.58
-0.71
HOH
-0.67
-0.69
-0.63
-0.71
The estimated value for the BSSE is ∼0.05 eV.
76
for pure Pt with the increasing number of deposited Pt layers (see Table 10). Earlier, it was
explained by weakening of the ligand effect, when the number of deposited Pt layers increases.124
The ligand effect can be seen in Figure 24, Tables 11 and 12. We find that Pt3ML/Os, which
experiences a weaker ligand effect than Pt1ML/Os, has similar binding energy values with
compressive Pt. In general, the ligand effect, as well as the strain effect, weakens the ORR species
binding on the Os/Pt core-shell structure compared to pure Pt. However, too weak binding of certain
ORR intermediates may result in a higher barrier for the critical reaction. That is why Pt2ML/Os,
which has the binding energy closer to the optimal value than Pt1ML/Os, shows better ORR activity
than Pt1ML/Os (see Figure 21, which shows a compromise between the OOH formation and H2O
formation steps).
77
Figure 25. Potential energy surface including reaction barriers of H-OOH-diss-hydr mechanism
for Pt, compressive Pt, and Pt/Os catalysts in gas phase (a) and solution (b).
78
Table 13. Reaction energy barriers (eV) for ORR steps on Pt, compressive Pt, and Pt/Os surfaces
in gas phase.
Step
Pt
Compressive Pt
Pt1ML/Os
Pt3ML/Os
HH dissociation
0.00
0.00
0.24
0.13
O2 dissociation
0.56
0.63
0.90
0.63
OH formation
0.74
0.69
0.56
0.68
H2O formation
0.26
0.22
0.36
0.19
OOH formation
0.31
0.31
0.18
0.29
OOH dissociation
0.09
0.14
0.29
0.17
H-OOH dissociation
0.06
0.24
0.29
0.23
O hydration
0.27
0.27
0.04
0.19
Table 14. Reaction energy barriers (eV) for ORR steps on Pt, compressive Pt, and Pt/Os surfaces
in solution.
Step
Pt
Compressive Pt
Pt1ML/Os
Pt3ML/Os
HH dissociation
0.00
0.00
0.15
0.09
O2 dissociation
0.00
0.00
0.37
0.00
OH formation
1.09
0.92
0.61
0.92
H2O formation
0.32
0.21
0.26
0.16
OOH formation
0.37
0.37
0.15
0.35
OOH dissociation
0.00
0.00
0.00
0.00
H-OOH dissociation
0.00
0.03
0.00
0.00
O hydration
0.45
0.47
0.06
0.42
79
4.5 Experimental Results
A HAADF STEM image for the Pt2ML/Os/C particle is shown in Figure 26 (a). Although the
HAADF technique could supply clear contrast for interface/locations because of various element
distributions,55 some HAADF STEM images do not show clear core-shell structure contrast, like
in our case where the atomic number variation (Z contrast) is not significant, as was discussed
earlier.55,125,126 Figure 26 (b) shows the element profile analysis for a Pt2ML/Os/C nanoparticle from
the STEM-EDS measurement with a probe size of about 1.5 Å. As shown, Pt2ML/Os/C revealed a
core-shell structure in which the core consisted of Os atoms whereas the shell was predominately
occupied by the Pt atoms.
Figure 26. The HAADF STEM image (a) and EDS line-scan of a Pt2ML/Os/C nanoparticle (b).
Figure 27 shows the CV curves of Os/C, Pt1ML/Os/C, Pt2ML/Os/C, and Pt/C for ECSA
determination by hydrogen adsorption. The ECSA values for Pt/C, Pt1ML/Os/C, and Pt2ML/Os/C
80
were 2.39, 1.18, and 1.88 cmPt2, respectively. The Pt/Os/C catalysts and pure Pt have different
adsorption behavior, because the modified structure weakens the interaction between adsorbates
and catalyst surfaces,127 which can also be seen in Tables 7 and 8. By increasing the thickness of
the Pt overlayer, the current densities at the double layer region (0.3 ~ 0.6 V vs. RHE) are decreased
(see Figure 28). The increased current densities at the double layer region of Pt1ML/Os/C is possibly
due to the oxidation of incompletely covered Os, similar to the Ru/Pt catalyst.61 Therefore we
reasonably concluded that the Pt in Pt1ML/Os/C did not cover the entire Os surface, but in the case
of Pt2ML/Os/C, the surface was more completely covered by the Pt atoms. This is anticipated
because even in a straightforward displacement reaction process, the stoichiometric ratio of Cu:Pt
was 1:1, and previous literature reported that the galvanic displacement did not form a continuous
layer but a fine structure with some nano-voids or 2ML high deposit at some spots76 or the deposit
formed by interconnected Pt islands with some holes.78,128 A similar 3D island structure was also
observed by using the extended X-ray adsorption fine structure (EXAFS) analysis applied for
Pt/Rh(0001)63 and Au/Pd129 samples, which were prepared via the Cu UPD displacement process.
The island structure formation might be due to the incomplete Cu UPD shell structure before
galvanic displacement.130 Furthermore, some references reported the Cl- as a strong complexing
ligand, promoting the stability of Cu(I) over Cu(II),131,132 which makes the stoichiometric ratio
greater than one. Therefore, the Os bare surface of the Pt1ML/Os/C catalyst is not fully covered by
Pt deposit, but the Pt2ML/Os/C has much better Pt coverage. The incomplete coverage decreases the
stability of the Pt/Os/C catalysts. However, in Chapter 3, we reported that the stability of dealloyed
Pt-Os nanoparticles after 10000 CV cycles was better than that of pure Pt. This implies that with
complete Pt coverage, which could be reached by using another technique, Pt/Os core-shell
catalysts may have a good enough stability.
81
Figure 27. The CV curves for ECSA determination for Os/C, Pt/C, Pt1ML/Os/C, and Pt2ML/Os/C,
respectively. The electrolyte was deaerated 0.1 M aqueous HClO4.
Figure 28. The CV curves for ECSA determination for Os/C, Pt1ML/Os/C, and Pt2ML/Os/C,
respectively. The electrolyte was deaerated 0.1 M aqueous HClO4.
82
Figure 29 demonstrates the ORR CV curves in apparent current density for Pt1ML/Os/C,
Pt2ML/Os/C, and Pt/C. At potential below 0.6 V, the ORR response is under mass transport control
limited by the diffusion of the dissolved oxygen in the electrolyte, whereas at potential between 0.8
and 1 V, the ORR response is dominated by kinetics (the electrocatalytic activity of the
electrocatalyst involved in the ORR process).111 Therefore, a simple method to quickly evaluate the
ORR behavior of a potential electrocatalyst is the reading of half-wave potential, which is defined
as the potential at which the magnitude of the current is half of the limiting current. In general, the
larger the half-wave potential, the greater the ORR activity. As shown, the half-wave potentials for
the Pt/C, Pt1ML/Os/C, and Pt2ML/Os/C were 876, 795, and 918 mV, respectively. It should be noted
that the diffusion-limiting current for these samples at a rotation speed of 1600 rpm was close to 6
mA cm-2, which is consistent with the value earlier reported.133 This consistence indicated that our
ORR experiments were carried out properly.
To extract the kinetic information, we employed the Kouteck-Levich equation (see section 3.6)
again. The respective kinetic parameters are listed in Table 15. As it was mentioned above, the Os
core of Pt1ML/Os/C is not fully covered by Pt atoms. According to our solvation calculation (Table
10), the ORR at the Os surface is very sluggish because of the high H2O formation reaction energy
barrier due to the strong OH binding. This causes the catalytic performance of Pt1ML/Os/C, which is
not completely covered by Pt, to be even worse than that of the commercial Pt catalyst. The
Pt2ML/Os/C sample, which is more completely covered by Pt atoms and shows a core-shell structure,
exhibits excellent catalytic activity. It is in agreement with the low reaction energy barriers for the
Pt1ML/Os and Pt2ML/Os catalysts (Table 10) from our computational evaluations.
83
Figure 29. The ORR curves of Pt/C, Pt1ML/Os/C, and Pt2ML/Os/C in apparent current density. The
electrolyte is oxygen-saturated 0.1 M aqueous HClO4 solution and the scan rate is 20 mV/s.
Table 15. The comparison of kinetic data for ORR Os/Pt core-shell catalysts.
Catalyst
Pt
loading
(µg)
Imass
Imass
ispecific
(mA µgPt-1)
(mA µgPt+Os-1)
(mA cmPt-2)
-0.20
-0.20
-0.26
0.876
Pt1ML/Os/C
1.42
0.795
Pt2ML/Os/C
2.82
-0.70
-0.45
-1.33
0.918
Pt/C
Half wave potential
(V vs RHE)
84
Figure 30 shows correlation between the experimental half wave potential and theoretically
calculated OH binding energy. Previous experimental studies28 indicate good agreement with a
theoretical study in which the maximum activity corresponds to the surface that could bind OH by
~0.1 eV weaker than Pt.134,135 Our results indicate that the highest activity occurs at Pt2ML/Os/C, where
the corresponding OH binding is slightly (by ~0.03 eV) weaker than for Pt. This is a relatively rough
approximation, but still a good way to find correlation between the theoretical prediction and
experimental result. In Figure 31, we use the calculated RDS for comparison with the experimental
half wave potential. To fix the UPD coverage issue (Pt atoms partially cover the Os core), we used
the average value of the RDS barriers for pure Os and Pt1ML/Os (Table 10) as the RDS barrier for
Pt1ML/Os/C and the average value of the RDS barriers for Pt1ML/Os and Pt2ML/Os as the RDS barrier
for Pt2ML/Os/C. The higher barrier corresponds to the smaller half wave potential, and the
correspondence looks good enough, although it is based on a rough approximation and assumption.
Indeed, Pt2ML/Os/C shows 3.5 times better mass activity, when considering only Pt loading, and 5
times better specific activity for ORR compared to those of Pt/C (Table 15 and Figure 32).
Considering a total precious metal loading (Pt+Os), Pt2ML/Os/C shows over 2 times better activity
(Figure 32).
85
Figure 30. Experimental half wave potential vs. the theoretically predicted OH binding energy in
solution.
Figure 31. Experimental half wave potential versus the theoretical RDS reaction barrier in solution.
86
Figure S8. The total precious mass (left) and specific (right) activity of Pt/C and Pt2ML/Os/C at 0.9
V (vs. NHE).
4.6 Summary for Os/Pt Core-Shell Catalysts
We applied both QM calculations and electrochemical measurements to evaluate the catalytic
activity of the Pt/Os/C catalysts. Using QM calculations, we identified the most stable structures
for various layer Pt deposits on the Os substrate and these structures were applied to calculate
binding energies and reaction energy barriers for the ORR species. In general, the binding energy
increases with the number of Pt deposit layers and approaches the values on the pure Pt surface
both in gas phase and solution. This result confirms that the ligand effect gradually decreases as
the number of Pt deposit layers increases. The calculated RDS barriers predict the ORR catalytic
activity as following: Pt2ML/Os>Pt1ML/Os>Pt3ML/Os>Pt>Os. To further understand the origin of the
higher catalytic activity of the Pt/Os surfaces, a non-physical compressive Pt system was built to
compare its reaction energy barriers and binding energies to Pt, Pt1ML/Os and Pt3ML/Os. The results
implied that due to changes in the electronic structure of the Pt/Os catalysts, the ligand effect might
be more important for the improved ORR activity than the purely compressive strain effect.
87
Pt/Os/C catalysts were fabricated by the chemical synthesis and UPD method. TEM
technologies were applied for materials characterization. The CV curve for Os, Pt1ML/Os/C, and Pt
showed that in Pt1ML/Os /C, the Os substrate is only partially covered by Pt atoms. The more
complete Pt coverage is formed after the second UPD process. The SEM-EDS analysis proved the
existence of the core-shell structure for our Pt2ML/Os/C catalyst. The ORR CV curves demonstrate
that the Pt2ML/Os/C structure shows 3.5 to 5 times better catalytic activity than Pt/C, which allows
the Os/Pt core-shell structured materials to be considered as a potential ORR catalyst. The
experimentally observed ORR catalytic activity follows the sequence: Pt2ML/Os/C > Pt/C >
Pt1ML/Os/C, which agrees with our theoretical prediction, based on the ORR energy barrier
calculations.
88
Chapter 5
Conclusions
The fuel cell technology has a potential to solve many energy related problems due to its high
efficiency and low pollution merits. However, the sluggish ORR at cathode limits its performance.
Therefore, various catalysts, including Pt based metal catalysts and NPM catalysts, were developed
to improve the ORR activity. Even though the high cost of the Pt based metal catalysts is a big
problem for wide commercialization of these catalysts, the unsatisfactory stability issues of NPM
catalysts limits the application of NPM catalysts to replace Pt based catalysts. To date, developing a
low cost, high efficiency, and great stability catalyst is still the most important target in the fuel cell
area.
Among the Pt based catalysts, Pt3M alloys demonstrates superior activity when compared to
pure Pt because of the formation of Pt-skin structure. Unfortunately, these reported Pt3M alloys
showed fast performance degradation due to the dissolution of their NPM elements during fuel cell
operation. The dissolution is a consequence of the NPM element migration to the catalyst surface
and subsequent reactions with acidic electrolyte during the ORR operation. Therefore, Pt3M alloys
that show a good surface segregation with ORR species adsorbed on their surfaces are potential
candidates for the ORR catalyst. To find these potential candidates, we performed the surface
segregation study of 28 Pt3M alloys and found that only Pt3Os and Pt3Ir may show a good surface
segregation under the fuel cell operation environment. However, Pt-Ir could not formed a
homogeneous alloy at lower temperature. Thus, we focused on the Pt3Os alloy considering it as a
possible candidate for the ORR catalyst.
89
Our studies of the ORR species binding energies and reaction energy barriers predicted that
Pt3Os has slightly higher activity than pure Pt. The improved activity could be ascribed to the
synergetic effects of compressive strain and the electronic structure modification by the underlying
Os atoms. The d-band center shift explained the reduced O/OH binding energy and the better
activity of Pt3Os catalyst. Our barrier calculation showed the RDS for Pt3Os and Pt is O hydration
step ((Oad + H2Oad OHad + OHad) for both cases. The barrier for Pt of 0.50 eV is decreased to
0.48 eV for Pt3Os, which would improve the activity by 218%. To validate our theoretical
prediction, we prepared the DA-Pt2Os catalysts by the dealloying method. The electrochemical
analysis showed that the DA-Pt2Os has ~2 times better activity than pure Pt and possesses 272%
better stability, which confirmed our theoretical predictions.
To further improve the Pt-Os catalyst activity, we modified the surface structure of this type of
catalysts. We used UPD method to fabricate Os/Pt core-shell catalysts. The surface structure energy
calculation determined the stacking series of our Os/Pt slab. The ORR species binding energy
calculation, as well as ORR barrier analysis, were performed to evaluate the catalyst activity. Our
theoretical study showed that the ORR activity is a compromise between the OOH formation step
(0.23 eV for Pt2ML/Os and 0.37 eV for Pt) and H2O formation step (0.22 eV for Pt2ML/Os and 0.32
eV for Pt). With too weak O/OH binding energy, the OOH formation step dominates ORR, while
for too strong O/OH binding energy the high H2O formation barrier limits catalyst activity. The
results predicted that the Pt2ML/Os catalyst, which has the optimum O/OH binding energy because
of suitable ligand effect resulting from the Os core, have better activity than Pt, Pt1ML/Os, Pt3ML/Os,
or Pt3Os. We concluded that our results based on the Langmuir–Hinshelwood mechanism are
consistent with the results obtained by other groups for the Eley-Rideal mechanism. Experimental
material characterizations for our catalyst proved its core-shell structure feature. The electrochemical
analysis showed that the activity of Pt2ML/Os/C in 0.1M HClO4 solution at 25OC has 3.5 to 5 times
90
better activity than commercial Pt catalyst, which validated our theoretical prediction, while
Pt1ML/Os/C has worse activity than Pt because its Os core was incompletely covered by Pt. The
correlation between experimental half potential and theoretical OH binding energy/RDS energy
barrier showed good correspondence, although it was based on rough approximation and assumption.
In this thesis, we successfully predict the Pt-Os catalyst activity and validated by experimental
results. Our Pt-Os catalyst case demonstrates the possibility of using the basic QM method and simple
Langmuir–Hinshelwood mechanism to evaluate the catalyst performance without losing the
accuracy.
91
BIBLIOGRAPHY
[1] Borup, R.; Meyers, J.; Pivovar, B.; Kim, Y. S.; Mukundan, R.; Garland, N.; Myers, D.; Wilson,
M.; Garzon, F.; Wood, D.; Zelenay, P.; More, K.; Stroh, K.; Zawodzinski, T.; Boncella, J.; McGrath,
J. E.; Inaba, M.; Miyatake, K.; Hori, M.; Ota, K.; Ogumi, Z.; Miyata, S.; Nishikata, A.; Siroma, Z.;
Uchimoto, Y.; Yasuda, K.; Kimijima, K. I.; Iwashita, N. Chem. Rev. 2007, 107, 3904-3951.
[2] Wang, Y.; Chen, K. S.; Mishler, J.; Cho, S. C.; Adroher, X. C. Appl. Energy 2011, 88, 981-1007.
[3] Yasuda, K.; Taniguchi, A.; Akita, T.; Ioroi, T.; Siroma, Z. Phys. Chem. Chem. Phys. 2006, 8, 746752.
[4] Eroglu, M.; Dursun, E.; Sevencan, S.; Song, J.; Yazici, S.; Kilic, O. Int. J. Hydrogen Energy 2011,
36, 7985-7992.
[5] Steele, B. C. H.; Heinzel, A. Nature 2001, 414, 345-352.
[6]
Kordesch, K.; Simader, G. Fuel Cells and Their Applications; Wiley-VCH: New York, 1996.
[7] Appleby, A. J.; Foulkes, F. R. Fuel Cell Handbook; Van Nostrand Reinhold: New York, 1989.
[8] O’Hayre, R. P.; Cha, S. W.; Colella, W.; Prinz, F. B. Fuel Cell Fundamentals; 2nd ed.; John Wiley
& Sons: New York, 2006.
[9] Hammer, B.; Nørskov, J. K. Adv. Catal. 2000, 45, 71-129.
[10] Xin, H.; Holewinski, A.; Schweitzer, N.; Nikolla, E.; Linic, S. Top. Catal. 2012, 55, 376390.
[11] Xin, H.; Holewinski, A.; Linic, S. ACS Catal. 2011, 2, 12-16.
[12] Holewinski, A.; Linic, S. J. Electrochem. Soc. 2012, 159, H864-H870.
[13] Wei, G. F.; Fang, Y. H.; Liu, Z. P. J. Phys. Chem. C 2012, 116, 12696-12705.
[14] Fang, Y. H.; Liu, Z. P. J. Phys. Chem. C 2011, 115, 17508-17515.
92
[15] Wei, G. F.; Liu, Z. P. Phys. Chem. Chem. Phys. 2013, 15, 18555-18561.
[16] Jacob, T.; Goddard, W. A. ChemPhysChem 2006, 7, 992-1005.
[17] Keith, J. A.; Jerkiewicz, G.; Jacob, T. ChemPhysChem 2010, 11, 2779-2794.
[18] Wang, B. J. Power Sources 2005, 152, 1-15.
[19] Bashyam, R.; Zelenay, P. Nature 2006, 443, 63-66.
[20] Othman, R.; Dicks, A. L.; Zhu, Z. Int. J. Hydrogen Energy 2012, 37, 357-372.
[21] Debe, M. K. Nature 2012, 486, 43-51.
[22] Hwang, S. J.; Kim, S. K.; Lee, J. G.; Lee, S. C.; Jang, J. H.; Kim, P.; Lim, T. H.; Sung, Y. E.;
Yoo, S. J. J. Am. Chem. Soc. 2012, 134, 19508-19511.
[23] Stamenkovic, V.; Mun, B. S.; Mayrhofer, K. J. J.; Ross, P. N.; Markovic, N. M.; Rossmeisl,
J.; Greeley, J.; Nørskov, J. K. Angew. Chem. 2006, 118, 2963-2967.
[24] Stamenkovic, V. R.; Mun, B. S.; Arenz, M.; Mayrhofer, K. J.; Lucas, C. A.; Wang, G.; Ross,
P. N.; Markovic, N. M. Nat. Mater. 2007, 6, 241-247.
[25] Stamenkovic, V. R.; Fowler, B.; Mun, B. S.; Wang, G.; Ross, P. N.; Lucas, C. A.; Markovic,
N. M. Science 2007, 315, 493-497.
[26] Greeley, J.; Stephens, I. E. L.; Bondarenko, A. S.; Johansson, T. P.; Hansen, H. A.; Jaramillo,
T. F.; Rossmeisl, J.; Chorkendorff, I.; Nørskov, J. K. Nat. Chem. 2009, 1, 552-556.
[27] Strasser, P.; Koh, S.; Anniyev, T.; Greeley, J.; More, K.; Yu, C. F.; Liu, Z. C.; Kaya, S.;
Nordlund, D.; Ogasawara, H.; Toney, M. F.; Nilsson, A. Nat. Chem. 2010, 2, 454-460.
93
[28] Stephens, I. E. L.; Bondarenko, A. S.; Perez-Alonso, F. J.; Calle-Vallejo, F.; Bech, L.;
Johansson, T. P.; Jepsen, A. K.; Frydendal, R.; Knudsen, B. P.; Rossmeisl, J.; Chorkendorff, I. J.
Am. Chem. Soc. 2011, 133, 5485-5491.
[29] Yu, T. H.; Sha, Y.; Merinov, B. V.; Goddard, W. A. J. Phys. Chem. C 2010, 114, 1152711533.
[30] Colón-Mercado, H. R.; Popov, B. N. J. Power Sources 2006, 155, 253-263.
[31] Colón-Mercado, H. R.; Kim, H.; Popov, B. N. Electrochem. Commun. 2004, 6, 795-799.
[32] Maillard, F.; Dubau, L.; Durst, J.; Chatenet, M.; André, J.; Rossinot, E. Electrochem.
Commun. 2010, 12, 1161-1164.
[33] Chen, S.; Ferreira, P. J.; Sheng, W. C.; Yabuuchi, N.; Allard, L. F.; Shao-Horn, Y. J. Am.
Chem. Soc. 2008, 130, 13818-13819.
[34] Watanabe, M.; Tsurumi, K.; Mizukami, T.; Nakamura, T.; Stonehart, P. J. Electrochem. Soc.
1994, 141, 2659-2668.
[35] Antolini, E.; Salgado, J. R. C.; Gonzalez, E. R. J. Power Sources 2006, 160, 957-968.
[36] Wakabayashi, N.; Takeichi, M.; Uchida, H.; Watanabe, M. J. Phys. Chem. B 2005, 109, 58365841.
[37] Ma, Y. G.; Balbuena, P. B. Surf. Sci. 2008, 602, 107-113.
[38] Chen, W.; Schneider, W. F.; Wolverton, C. J. Phys. Chem. C 2014, 118, 8342-8349.
[39] Wei, G. F.; Liu, Z. P. Energy Environ. Sci. 2011, 4, 1268-1272.
[40] Mayrhofer, K. J.; Arenz, M. Nat. Chem. 2009, 1, 518-519.
94
[41] Dai, Y.; Ou, L.; Liang, W.; Yang, F.; Liu, Y.; Chen, S. J Phys. Chem. C 2011, 115, 21622168.
[42] Brushett, F. R.; Duong, H. T.; Ng, J. W.; Behrens, R. L.; Wieckowski, A.; Kenis, P. J. A. J.
Electrochem. Soc. 2010, 157, B837-B845.
[43] Nishimura, T.; Morikawa, T.; Yokoi, M.; Iwakura, C.; Inoue, H. Electrochim. Acta 2008, 54,
499-505.
[44] Johansson, T. P.; Ulrikkeholm, E. T.; Hernandez-Fernandez, P.; Malacrida, P.; Hansen, H. A.;
Bandarenka, A. S.; Nørskov, J. K.; Rossmeisl, J.; Stephens, I. E. L.; Chorkendorff, I. Top. Catal.
2014, 57, 245-254.
[45] Fernandes, A. C.; Paganin, V. A.; Ticianelli, E. A. J. Electroanal. Chem. 2010, 648, 156-162.
[46] Chen, S.; Gasteiger, H. A.; Hayakawa, K.; Tada, T.; Shao-Horn, Y. J. Electrochem. Soc. 2010,
157, A82-A97.
[47] Ma, Y.; Balbuena, P. B. Surf. Sci. 2009, 603, 349-353.
[48] Wu, G.; More, K. L.; Johnston, C. M.; Zelenay, P. Science 2011, 332, 443-447.
[49] Yang, H. Angew. Chem., Int. Ed. 2011, 50, 2674-2676.
[50] Nilekar, A. U.; Xu, Y.; Zhang, J.; Vukmirovic, M. B.; Sasaki, K.; Adzic, R. R.; Mavrikakis,
M. Top. Catal. 2007, 46, 276-284.
[51] Adzic, R. R.; Zhang, J.; Sasaki, K.; Vukmirovic, M. B.; Shao, M.; Wang, J. X.; Nilekar, A.
U.; Mavrikakis, M.; Valerio, J. A.; Uribe, F. Top. Catal. 2007, 46, 249-262.
95
[52] Zhou, W. P.; Yang, X. F.; Vukmirovic, M. B.; Koel, B. E.; Jiao, J.; Peng, G. W.;
Mavrikakis, M.; Adzic, R. R. J. Am. Chem. Soc. 2009, 131, 12755-12762.
[53] Wang, J. X.; Inada, H.; Wu, L. J.; Zhu, Y. M.; Choi, Y. M.; Liu, P.; Zhou, W. P.; Adzic, R.
R. J. Am. Chem. Soc. 2009, 131, 17298-17302.
[54] Gong, K. P.; Su, D.; Adzic, R. R. J. Am. Chem. Soc. 2010, 132, 14364-14366.
[55] Sasaki, K.; Wang, J. X.; Naohara, H.; Marinkovic, N.; More, K.; Inada, H.; Adzic, R. R.
Electrochim. Acta 2010, 55, 2645-2652.
[56] Knupp, S. L.; Vukmirovic, M. B.; Haldar, P.; Herron, J. A.; Mavrikakis, M.; Adzic, R. R.
Electrocatal. 2010, 1, 213-223.
[57] Sasaki, K.; Naohara, H.; Cai, Y.; Choi, Y. M.; Liu, P.; Vukmirovic, M. B.; Wang, J. X.; Adzic,
R. R. Angew. Chem., Int. Ed. Engl. 2010, 49, 8602-8607.
[58] Bing, Y.; Liu, H.; Zhang, L.; Ghosh, D.; Zhang, J. Chem. Soc. Rev. 2010, 39, 2184-2202.
[59] Sha, Y.; Yu, T. H.; Merinov, B. V.; Shirvanian, P.; Goddard, W. A., III J. Phys. Chem. C
2012, 116, 21334-21342.
[60] Adzic, R. R. Electrocatal. 2012, 3, 163-169.
[61] Yang, L.; Vukmirovic, M. B.; Su, D.; Sasaki, K.; Herron, J. A.; Mavrikakis, M.; Liao, S.;
Adzic, R. R. J. Phys. Chem. C 2013, 117, 1748-1753.
[62] Brimaud, S.; Engstfeld, A.; Alves, O.; Hoster, H.; Behm, R. Top. Catal. 2014, 57, 222-235.
[63] Friebel, D.; Viswanathan, V.; Miller, D. J.; Anniyev, T.; Ogasawara, H.; Larsen, A. H.;
O’Grady, C. P.; Nørskov, J. K.; Nilsson, A. J. Am. Chem. Soc. 2012, 134, 9664-9671.
96
[64] Jackson, A.; Viswanathan, V.; Forman, A. J.; Larsen, A. H.; Nørskov, J. K.; Jaramillo, T.
F. ChemElectroChem. 2014, 1, 67-71.
[65] Ley, K. L.; Liu, R. X.; Pu, C.; Fan, Q. B.; Leyarovska, N.; Segre, C.; Smotkin, E. S. J.
Electrochem. Soc. 1997, 144, 1543-1548.
[66] Kua, J.; Goddard, W. A., III J. Am. Chem. Soc. 1999, 121, 10928-10941.
[67] Gurau, B.; Viswanathan, R.; Liu, R. X.; Lafrenz, T. J.; Ley, K. L.; Smotkin, E. S.; Reddington,
E.; Sapienza, A.; Chan, B. C.; Mallouk, T. E.; Sarangapani, S. J. Phys. Chem. B 1998, 102, 999710003.
[68] Crown, A.; Moraes, I. R.; Wieckowski, A. J. Electroanal. Chem. 2001, 500, 333-343.
[69] Huang, J. J.; Yang, H.; Huang, Q. H.; Tang, Y. W.; Lu, T. H.; Akins, D. L. J. Electrochem.
Soc. 2004, 151, A1810-A1815.
[70] Santos, V. P.; Del Colle, V.; Bezerra, R. M.; Tremiliosi-Filho, G. Electrochim. Acta 2004, 49,
1221-1231.
[71] Liu, W.; Huang, J. J. Power Sources 2009, 189, 1012-1015.
[72] Zhang, J. L.; Vukmirovic, M. B.; Sasaki, K.; Nilekar, A. U.; Mavrikakis, M.; Adzic, R. R. J.
Am. Chem. Soc. 2005, 127, 12480-12481.
[73] Ma, Y.; Balbuena, P. B. J. Electrochem. Soc. 2010, 157, B959-B963.
[74] Ramírez-Caballero, G. E.; Ma, Y.; Callejas-Tovar, R.; Balbuena, P. B. Phys. Chem. Chem.
Phys. 2010, 12, 2209-2218.
97
[75] Wang, J. X.; Inada, H.; Wu, L. J.; Zhu, Y. M.; Choi, Y. M.; Liu, P.; Zhou, W. P.; Adzic, R.
R. J. Am. Chem. Soc. 2009, 131, 17298-17302.
[76] Brankovic, S. R.; Wang, J. X.; Adzic, R. R. Surf. Sci. 2001, 474, L173-L179.
[77] Herrero, E.; Buller, L. J.; Abruna, H. D. Chem. Rev. 2001, 101, 1897-1930.
[78] Zhang, J.; Mo, Y.; Vukmirovic, M. B.; Klie, R.; Sasaki, K.; Adzic, R. R. J. Phys. Chem. B
2004, 108, 10955-10964.
[79] Zhang, J.; Lima, F. H. B.; Shao, M. H.; Sasaki, K.; Wang, J. X.; Hanson, J.; Adzic, R. R. J.
Phys. Chem. B 2005, 109, 22701-22704.
[80] Zhang, J.; Vukmirovic, M. B.; Xu, Y.; Mavrikakis, M.; Adzic, R. R. Angew. Chem., Int. Ed.
Engl. 2005, 44, 2132-2135.
[81] Zhang, J.; Sasaki, K.; Sutter, E.; Adzic, R. R. Science 2007, 315, 220-222.
[82] Schultz, P. SeqQuest, Electronic Structure Code; Sandia National Laboratory, Albuquerque,
NM, http://dft.sandia.gov/Quest/.
[83] Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865-3868.
[84] Perdew, J. P.; Zunger, A. Phys. Rev. B 1981, 23, 5048-5079.
[85] Ceperley, D. M.; Alder, B. J. Phys. Rev. Lett. 1980, 45, 566-569.
[86] Goddard, W. A., III Phys. Rev. 1968, 174, 659-662.
[87] Hamann, D. R. Phys. Rev. B 1989, 40, 2980-2987.
[88] Melius, C. F.; Goddard, W. A., III Phys. Rev. A 1974, 10, 1528-1540.
98
[89] Melius, C. F.; Olafson, B. D.; Goddard, W. A., III Chem. Phys. Lett. 1974, 28, 457-462.
[90] Redondo, A.; Goddard, W. A., III; McGill, T. C. Phys. Rev. B 1977, 15, 5038-5048.
[91] Sha, Y.; Yu, T. H.; Liu, Y.; Merinov, B. V.; Goddard, W. A., III J. Phys. Chem. Lett. 2010,
1, 856-861.
[92] Baker, N. A.; Sept, D.; Joseph, S.; Holst, M. J.; McCammon, J. A. Proc. Natl. Acad. Sci.
U.S.A. 2001, 98, 10037-10041.
[93] Holst, M.; Saied, F. J. Comput. Chem. 1993, 14, 105-113.
[94] Holst, M. J.; Saied, F. J. Comput. Chem. 1995, 16, 337-364.
[95] Mills, G.; Jonsson, H.; Schenter, G. K. Surf. Sci. 1995, 324, 305-337.
[96] Mills, G.; Jonsson, H. Phys. Rev. Lett. 1994, 72, 1124-1127.
[97] Bonnet, N.; Otani, M.; Sugino, O. J. Phys. Chem. C 2014, 118, 13638-13643.
[98] Jinnouchi, R.; Kodama, K.; Hatanaka, T.; Morimoto, Y. Phys. Chem. Chem. Phys. 2011, 13,
21070-21083.
[99] Keith, J. A.; Anton, J.; Kaghazchi, P.; Jacob T. Modeling Catalytic Reactions on Surfaces
with Density Functional Theory In Deutschmann, O. Modeling and Simulation of Heterogeneous
Catalytic Reactions: From the Molecular Process to the Technical System; John Wiley & Sons,
2013.
[100] K. Kinoshita, P. Stonehart, in: J.O.M. Bockris, B.E. Conway (Eds.), Modern Aspects of
Electrochemistry, Springer US, 1977.
[101] F. Gloaguen, J.M. Le´Ger, C. Lamy, J. Appl. Electrochem. 1997, 27, 1052-1060.
99
[102] Abrams, B. L.; Vesborg, P. C. K.; Bonde, J. L.; Jaramillo, T. F.; Chorkendorff, I. J.
Electrochem. Soc. 2009, 156, B273-B282.
[103] Voronova, L. I.; Polyakova, V. P.; Savitskii, E. M. Russ. Metall. 1984, 5, 201-203.
[104] Raub, E.; Plate, W. Zeit Metallkunde 1956, 47, 688-693.
[105] Moore, J. T.; Chu, D.; Jiang, R. Z.; Deluga, G. A.; Lukehart, C. M. Chem. Mater. 2003, 15,
1119-1124.
[106] Hu, S.; Xiong, L. P.; Ren, X. B.; Wang, C. B.; Luo, Y. M. Int. J. Hydrogen Energy 2009, 34,
8723-8732.
[107] Yu, T. H.; Hofmann, T.; Sha, Y.; Merinov, B. V.; Myers, D. J.; Heske, C.; Goddard, W. A.,
III J. Phys. Chem. C 2013, 117, 26598-26607.
[108] Sha, Y.; Yu, T. H.; Merinov, B. V.; Shirvanian, P.; Goddard, W. A., III J. Phys. Chem. Lett.
2011, 2, 572-576.
[109] Kitchin, J. R.; Nørskov, J. K.; Barteau, M. A.; Chen, J. G. Phys. Rev. Lett. 2004, 93, 156801.
[110] Kitchin, J. R.; Nørskov, J. K.; Barteau, M. A.; Chen, J. G. J. Chem. Phys. 2004, 120, 1024010246.
[111] Paulus, U. A.; Schmidt, T. J.; Gasteiger, H. A.; Behm, R. J. J. Electroanal. Chem. 2001, 495,
134-145.
[112] Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Fundamentals and Applications, John
Wiley & Sons, 2001.
100
[113] Chen S.; Sheng, W.; Yabuuchi, N.; Ferreira, P. J.; Allard, L. F.; Shao-Horn, Y. J. Phys.
Chem. C 2009, 113, 1109–1125.
[114] Markovic, N. M.; Gasteiger, H. A.; Ross, P. N. J. Phys. Chem. C 1995, 99, 3411–3415.
[115] Wang, J. X.; Ma, C.; Choi, Y. M.; Su, D.; Zhu, Y. ; Liu, P.; Si, R.; Vukmirovic, M. B.; Zhang,
Y.; Adzic, R. R. J. Am. Chem. Soc. 2011, 133, 13551-13557.
[116] Hsieh, Y. C.; Zhang, Y.; Su, D.; Volkov, V.; Si, R.; Wu, L. J.; Zhu, Y. M.; An, W.; Liu, P.;
He, P.; Ye, S. Y.; Adzic, R. R.; Wang, J. X. Nat. Commun. 2013, 4, 2466.
[117] Tripković, V.; Skúlason, E.; Siahrostami, S.; Nørskov, J. K.; Rossmeisl, J. Electrochim. Acta
2010, 55, 7975-7981.
[118] Anderson, A. B.; Sidik, R. A.; Narayanasamy, J.; Shiller, P. J. Phys. Chem. B 2003, 107, 46184623.
[119] Hirota, K.; Song, M. B.; Ito, M. Chem. Phys. Lett. 1996, 250, 335-341.
[120] Nørskov, J. K.; Rossmeisl, J.; Logadottir, A.; Lindqvist, L.; Kitchin, J. R.; Bligaard, T.;
Jonsson, H. J. Phys. Chem. B 2004, 108, 17886-17892
[121] Karlberg, G.; Rossmeisl, J.; Nørskov, J. K. Phys. Chem. Chem. Phys. 2007, 9, 5158-5161.
[122] Viswanathan, V.; Hansen, H. A.; Rossmeisl, J.; Nørskov, J. K. ACS Catal. 2012, 2, 1654-1660.
[123] Mavrikakis, M.; Hammer, B.; Nørskov, J. K. Phys. Rev. Lett. 1998, 81, 2819-2822.
[124] Schlapka, A.; Lischka, M.; Groß, A.; Käsberger, U.; Jakob, P. Phys. Rev. Lett. 2003, 91,
016101.
101
[125] Garcia-Gutierrez, D.; Gutierrez-Wing, C.; Miki-Yoshida, M.; Jose-Yacaman, M. Appl.
Phys. A-Mater. 2004, 79, 481-487.
[126] Liu, C. P.; Twesten, R. D.; Gibson, J. M. Ultramicroscopy 2001, 87, 79-88.
[127] van der Vliet, D. F.; Wang, C.; Li, D.; Paulikas, A. P.; Greeley, J.; Rankin, R. B.; Strmcnik,
D.; Tripkovic, D.; Markovic, N. M.; Stamenkovic, V. R. Angew. Chem., Int. Ed. 2012, 51, 31393142.
[128] Vukmirovic, M. B.; Bliznakov, S. T.; Sasaki, K.; Wang, J. X.; Adzic, R. R. ECS Interface
2011, 20, 33-40.
[129] Price, S. W.; Rhodes, J. M.; Calvillo, L.; Russell, A. E. J. Phys. Chem. C 2013, 117, 2485824865.
[130] Price, S. W.; Speed, J. D.; Kannan, P.; Russell, A. E. J. Am. Chem. Soc. 2011, 133, 1944819458.
[131] Gokcen, D.; Bae, S. E.; Brankovic, S. R. J. Electrochem. Soc. 2010, 157, D582-587.
[132] Lee, H. P.; Nobe, K. J. Electrochem. Soc. 1986, 133, 2035-2043.
[133] Paulus, U. A.; Wokaun, A.; Scherer, G. G.; Schmidt, T. J.; Stamenkovic, V.; Radmilovic, V.;
Markovic, N. M.; Ross, P. N. J. Phys. Chem. B 2002, 106, 4181-4191.
[134] Rossmeisl, J.; Nørskov, J. K.; Taylor, C. D.; Janik, M. J.; Neurock, M. J. Phys. Chem. B
2006, 110, 21833-21839.
[135] Rossmeisl, J.; Karlberg, G. S.; Jaramillo, T.; Nørskov, J. K. Faraday Discuss. 2009,
140, 337-346.
102
Section Ⅱ
First-principles Modeling of Ni4M (M=
Co, Fe, Mn, Mo) Alloys as Solid Oxide
Fuel Cell Anode Catalyst
103
Chapter 6
Introduction
The solid oxide fuel cells (SOFCs) are considered as a potential technology for power generation
systems due to a number of advantages which this technology can offer.1-3 For example, it generates
electricity from oxidation of fuels via electrochemical reactions instead of mechanical conversion,
provides lower noise1 and higher efficiency which is not limited by the Carnot cycle.1,3-5 Furthermore,
since the SOFCs operate at high temperatures (typically above 800oC), close to the temperature of the
steam reforming process, the SOFC systems can use the waste heat for internal steam reforming.3,6,7
In addition, SOFCs have wide fuel adaptability,1,7 i.e. they may operate with various fuels, such as
hydrogen,2,4,8-10 carbon monoxide,11 ammonia,12 hydrocarbons,3,5,8,13,14 and their combinations.11,13,15
In this type of fuel cells, oxygen-ion-conducting yttria-stabilized zirconia (YSZ) ceramics are
widely used as electrolyte with perovskite-type oxides cathode and Ni/YSZ cermets anode.3,16,17
SOFCs operate at ∼800oC because the YSZ electrolyte reaches the desirable conductivity only at this
high temperature. Such an operating condition creates certain advantages and disadvantages. One of
the most important advantages is that SOFCs use non-precious metal catalysts which make them more
economically competitive than, for example, proton exchange membrane fuel cells (PEMFCs).
Another significant advantage of SOFCs is their potential fuel flexibility, as we mentioned before.
However, the high operating temperature causes certain problems, such as the long start-up time,4
crack formation resulting from the stress because of different thermal expansion coefficients of cell
components,10,18-20 inter-diffusion at the electrolyte/electrode interfaces,10,19-21 and the anode particle
coarsening.20 In the past years, many efforts were focused on reducing the operating temperature to
develop intermediate or low-temperature SOFCs by using other high conductivity electrolytes, such
104
as samarium or gadolinium-doped ceria (SDC, GDC),7,10,18,19,21 lanthanum gallate (LaGaO3) based
materials,4,17,18,20 and scandium stabilized zirconia (ScSZ).22,23
Although SOFCs could perform internal steam reforming, utilizing hydrocarbon as fuels causes
carbon deposition,3,6,8,13,14,24 sulfur poisoning,24-27 etc. on the conventional Ni-anode catalyst surfaces.
The carbon growth mechanism includes the decomposition of hydrocarbons on Ni surface, dissolution
of carbon into the bulk metals, and precipitation as carbon filament.6,13,16 While moderate carbon may
help the electronic connection between separate Ni particles,28 it is generally believed that further
growth of carbon filaments lifts the metal particles3,29 and causes the fracture of materials resulting
from the stress induced by carbon filaments.3,6,8,16 For sulfur poisoning, either the elementary sulfur
deposition or nickel sulfide formation resulting from the interaction between hydrogen sulfide (H2S)
in the hydrocarbon fuels and Ni induced the degradation of the anode performance.24-27
To mitigate the carbon deposition without sacrificing the anode activity, certain Ni alloys were
used to replace pure Ni. For instances, Kim et al. reported that the carbon deposition could be greatly
suppressed by using Ni-Cu cermet anode.28 Fu et al. examined the microstructure, performance, and
stability of Ni0.75Fe0.25-GDC anode for intermediate-temperature SOFC.21 da Paz Fiuza et al. evaluated
the conversion rate of ethanol steam reforming for various Ni-Fe alloy anode compositions and
concluded that the catalyst performance was not significantly changed with the Fe substitution for Ni,
but the carbon deposition could greatly be suppressed if most Ni were replaced by Fe.7 The coking
suppression was also observed for the Fe-Ni/ScSZ system by Huang et al.22 However, the results
published by Zhu et al. indicate that catalyst performance of a methane-fueled SOFC strongly depends
on the Fe composition below 750oC and Ni4Fe-ZrO2 show the catalytic ability similar to Ni with
somewhat lower carbon deposition.30 The observed activity dependence does not look surprising
because the earlier study performed by Horita et al.31 for the methane steam oxidation and reforming
105
on Ni/YSZ and Fe/YSZ anodes revealed that the steam reforming activity on Ni/YSZ was much
higher than that on Fe/YSZ. The work completed by Kan et al.19 reported as well that the addition of
Fe improved the stability of the Ni-Fe/GDC anode stability. In addition, the investigation of Ni-Fe +
SDC composite anode for a hydrogen-fueled SOFC by Lu et al.20 also pointed out the improved
electrochemical properties of the Ni4Fe composite anode.
Along with the experimental studies, computational modeling [mostly density-functional theory
(DFT) studies] of chemical reactions in SOFCs was performed as well. For example, Rossmeisl and
Bessler calculated the hydrogen atom, oxygen atom, and hydroxyl radical binding energies on
surfaces of various metals and computed the reaction energy changes for three fundamental steps of
the SOFC operation.32 They found that that the activity of the metal catalysts versus O/OH binding
energy obeys a volcano shape dependence with the apex at the optimum O binding energy, which is
a compromise between the O adsorption and OH formation reactions. An et al.33 carried out a similar
DFT study of the binding energies of O, S, C, and H on various bimetallic Ni alloys surfaces and
concluded that Cu-Ni, Fe-Ni, and Co-Ni have a better activity for the anode oxidation reaction. In
addition, Mo-Ni has better C and S deposition resistance and still keep a good catalytic activity at the
same time.33 Nikolla et al.34 employed DFT calculations to examine the activation energy barriers for
C and O atom diffusion pathways, subsequent C atom attachment to a graphite sheet, and the C-O
bond formation on both Ni(111) and Sn/Ni(111) surfaces. They predicted that the growth of carbon
deposition is more difficult on the Sn/Ni(111) surface than on the Ni(111) surface and validated this
prediction by experimental evidences. Ammal et al.35 combined DFT and microkinetics modeling to
study the hydrogen oxidation reaction at the Ni/YSZ interface and concluded that the rate-determining
step is the bulk oxygen diffusion in YSZ at low temperatures and H transfer from Ni to YSZ to form
water at high temperatures.
106
In summary we can say that according to the results published,7,19-21,30 certain binary metal
alloys, such as Ni-Fe catalysts with 10~25% Fe atomic compositions, show improved catalytic
activity and/or better coking resistance and, therefore, it would be useful to better understand the
mechanisms underlying these chemical processes.
In this work, we report results of our computational modeling of Ni4M (M = Fe, Co, Mn, and
Mo) binary alloys using DFT calculations to evaluate the activity and coking resistance of these
catalysts. We calculated the segregation energy for different atomic configurations of the Ni4Fe(111)
structure, and used the lowest energy configuration as the slab for binding energy and methane
conversion reaction barriers calculations for the above-mentioned alloys.
107
Chapter 7
Theoretical Methods
For our QM two-dimensional slab periodic calculations, we use the SeqQuest code36 with the
Perdew, Burke, and Ernzerhof (PBE)37 flavor of DFT with generalized gradient approximation
(GGA)38,39 and an optimized double zeta plus polarization Gaussian-type basis set for periodic
calculations. Angular-momentum-projected norm-conserving pseudopotentials40-44 are used to
substitute for the core electrons in Seqquest code package. All calculations were performed with spin
optimization. Since both Ni and Fe are ferromagnetic materials, the spin of Ni4Fe is an important
factor that affects its energetic. The SeqQuest program package allows the spin to be optimized
simultaneously with the geometric parameters. To be confident that the program properly describes
the spins, we performed manual calculations of the spins, optimizing only the geometry for each fixed
spin. The energy difference between the optimized structures with the automatically (the obtained
spin is 22.3) and manually calculated (the optimal spin is 23) spins is only 0.06 eV, the comparing
results are summarized in Figure 1. This result confirmed the reliability of the automatic spin
optimization procedure in the SeqQuest package that was widely used in our further calculations.
To describe the surface segregation and binding energies of intermediates involved in
hydrocarbon fuel conversion on the Ni4Fe(111) surface, we assumed the closest fcc packed structure
and used the 2 × 2 hexagonal periodic unit cell in the a and b directions based on the bulk lattice
constant of Ni3Fe, where Ni atoms are at the face centered sites and Fe at the corner sites. The
calculated Ni3Fe lattice constant is 3.56 Å, which is slightly greater than the Ni fcc structure lattice
constant 3.53 Å. For calculating energetics, we considered a five-layer slab with 4 atoms per layer, in
108
which the top four layers are allowed to relax, but the bottom layer was fixed with the atoms in
their bulk structure positions (Figure 2). There are 16 Ni and 4 Fe, totally 20 atoms per unit cell. The
stacking of atoms follows the fcc(111) stacking rule, that is the abcabc… stacking series, shown as
Figure 2.
The reaction energy barriers for each ORR step were calculated using the Nudged Elastic Band
(NEB) method45,46 implemented into the SeqQuest code, and the energy difference between the
initial step and the highest NEB image (transition state) was considered as an energy barrier. The
initial and final states were determined from the binding energies of the ORR species.
Figure 1. The energy versus spin polarization relationship of Ni4Fe.
109
Figure 2. The atomic positions for the 5 layer slab with abcab stacking sequence.
110
Chapter 8
Results and Discussion
8.1 Surface Segregation in Ni4Fe Alloy
To study the segregation effect for the Ni4Fe alloy, we have examined different slab structures,
which can be built by placing the Fe atoms in different positions. For example, the lowest energy
structure is the structure which distributes the Fe atoms in 5, 6, 12, and 13 atomic positions, as shown
in Figure 2. The results of our DFT calculations on the segregation effect for the Ni4Fe alloys are
listed in Table 1. Since mirror symmetry relationship exists between the structures, some structures
are ignored in our calculation. For example, 5-6-12-13 and 8-10-15-16 are mirror symmetrical to each
other via XY plane, so does 5-7-12-13 and 8-10-14-16. Though with symmetrical relationship, the
fixed atom number 1 to 4 make the structures with two Fe atoms in the 2nd layer (atom number 5~8)
have ~0.2 eV lower energy than the structure with two Fe atoms in the 4th layer (atom number 1316). Therefore, our slab calculations ignored some mirror symmetrical configurations without losing
the generality.
The calculated spin obtained for Ni4Fe, using automatic spin optimization procedure, is ~0.6 per
a Ni atom and ~3.0 per a Fe atom. To summarize, we find that Ni atoms show segregation preference
for the surface layer and the most favorable Ni4Fe structure has the (02110) Fe atom distribution (56-12-13). The 5-6-12-13 is also the structure that separate the Fe atoms farthest with surface
segregation at the top and bottom layer. This 5-6-12-13 structure was used for our further QM binding
energy and reaction barriers calculations.
111
Table 1. Ni4Fe surface segregation energies (eV).
Number of Fe atoms in each layer*
40000
00400
20002
02020
01111
11101
11011
01210
10201
01120
02110
Position of Fe atoms Relative energy (eV)**
1-2-3-4
2.66
9-10-11-12
1.36
1-2-19-20
1.64
2-3-17-20
1.74
6-7-14-15
0.62
2-7-10-15
0.57
2-7-9-13
0.76
2-8-10-16
0.78
2-7-9-18
0.90
1-8-13-20
1.24
6-10-11-14
0.58
7-9-12-14
0.57
7-9-11-14
0.29
2-9-11-19
1.17
6-12-13-14
0.19
8-10-14-16
0.52
8-10-15-16
0.20
5-7-12-13
0.28
5-6-12-13
0.00
The number from left to right indicates the number of Fe atoms from the bottom layer
(1st layer) to the top layer (5th layer)
**
All energies are relative to 5-6-12-13 configuration (-65780.53 eV)
112
8.2 Binding Energy of CHx Species
8.2.1 Binding Site Notation
Generally, a closest packed (111) surface of fcc structured metals has four types of sites:
1. On-top, bonded to one Ni (µ1), denoted as t,
2. Bridging, between two Ni (µ2), denoted as b,
3. Bridging, between three Ni (µ3-fcc) but in the fcc position (not above atoms of the top or 4th
layer), denoted as f.
4. Bridging, between three Ni (µ3-hcp) but in the hcp position (above atoms of the 4th layer),
denoted as h.
Figure 3. Notations of binding sites (left) and schematic representation possible binding sites
(right) for the Ni4Fe alloy.
113
However for the Ni4Fe surface, we need to take into account that the 3rd and 4th layer is 25%
Fe and 75% Ni, while the 2nd layer 50% Fe and 50 % Ni. We find that the binding energies to the top
pure Ni layer depend strongly on the nature of the 2nd, 3rd, and 4th layer Fe atoms (see Figure 3 for
details).
Considering only the 4th and 5th layers, we have three types of top sites: t1 and t3 with two Ni
and one Fe neighbor in the 4th layer, while t2 with three Ni neighbor atoms. As it can be seen from
Figure 2, t1 and t3 sites are different because the Fe atoms in the 2nd layer are located vertically below
t1 and t2 sites, on the other hand, the atoms directly below t3 site at 2nd layer are Ni atoms.
Considering only top two layers, there are seven bridge sites, depending on the number of Fe
atoms underneath: b1, b2, b3, b4, b6 and b5, b7 with 0 and 1 Fe atoms in the 4th layer, respectively. If
we take into account the 3rd layer as well, then two subtypes for b2, b4 and b1, b3, b6 depending on the
number of Fe atoms underneath in the 3rd layer, can be distinguished. If we also consider the 2nd
layer, the underneath of the b2 site connects one Ni and one Fe atoms, while the underneath of b4 site
connects two Ni atoms in the 2nd layer. Similarly, the underneath of b1 connects two Fe atoms, while
the underneath of b3 and b6 has one Fe and one Ni atom but in different positions. The underneath of
b5 has two Ni atoms, whereas the underneath of b7 has one Fe and one Ni atom in the 2nd layer.
Considering the top two layers, three fcc sites can be distinguished: f1 and f2, f3 with zero and one
Fe atom in the sublayer (4th layer) triangle. If we take into account the 3rd layer as well, then f1 is on
top of the 3rd layer Fe, while f2 and f3 are on top of the 3rd layer Ni. With the 2nd layer, f2 has one Fe
atom in the 4th layer triangle, but f3 has two Fe atoms.
Similarly considering the top two layers, three hcp sites, h1, h2 and h3, can be distinguished. Here
h1, h2 are on top of the sublayer Ni, while h3 is on top of the sublayer Fe. Adding the 2nd layer allows
114
the h1 and h2 sites to be distinguished. h1 has two Fe atoms in the 2nd layer, while h2 has only
one Fe atom in the projected triangle of the 2nd layer atoms (Figure 3).
8.2.2 Binding Energies of Methane Reforming Intermediates
First, we studied the methane reforming intermediates, CH3, CH2, CH, C, and H binding on
various sites shown in Figure 3. Tables 2-6 list binding energies for CH3, CH2, CH, C and H species
and their comparison with similar binding energies for pure Ni(111).47 For CH3 binding (see Table 3),
the most and 2nd stable site is f2 and hi site with 41.4 and 40.5 Kcal/mol binding energy, respectively.
CH3 binding energies on other fcc and hcp sites are between 37 to 40 Kcal/mol, which are greater
than the top site or bridge site binding energies (~35 to 37 Kcal.mol). Therefore, the fcc and hcp sites
are energetically more preferable than top and bride sites, very similar to the pure Ni case. However,
the addition of Fe atoms make the binding energy of Ni4Fe slightly lower than binding energy of Ni,
which possibly due to the ligand effect of Fe atoms or the strain effect resulting from lattice constant
change. Since similar discussion for ligand effect and strain effect for Pt alloys has been discussed in
our previous studies,48,49 and we concluded that the ligand effect is more significant than strain effect
in modifying the binding energy/activity.
For CH2, the most stable binding sites are f3 and h1 site, while the fcc/hcp site with Fe atom
directly underneath (h3/f1) is the least preferable fcc/hcp site. The top and bridge sites are relatively
difficult for CH2 binding, CH2 is even unstable to bind at some of these sites and migrate to the
neighboring fcc/hcp sites (see Table 3).
The most stable CH binding site is pretty similar to CH2 binding. But CH binding at the most
stable site, h1, is 0.3 Kcal/mole greater than f3, consistent with pure Ni case, in which hcp site is also
slightly more stable than the fcc site. The CH binding energies are ~10 Kcal/mol smaller than binding
115
on pure Ni surface, while the differences for CH3 and CH2 binding between Ni4Fe and Ni surfaces
are only 2~4 Kcal/mol.
For C binding on Ni4Fe surface, binding at top sites and bridge sites are unfavorable and C will
migrate to surrounding fcc/hcp sites after geometry optimization. h1 and h2 sites are the most two
preferred binding sites, consistent with our previous Ni result in which hcp site is the most favorable
C binding site. But again, C binding on Ni4Fe surface is ~10 Kcal/mol lower than on the pure Ni
surface. The lower C binding energy indicates that C removal is easier and the coke formation is less,
which conform to previous experimental literature.7,19,30 For H binding, the most preferable binding
sites are f2 and h1, with 60.4 and 60.2 Kcal/mol binding energy, respectively. The only bridge site that
H adsorbed is b3 site, which located between f2 and h1 site. Top sites are more disliked for H
adsorption, and H binding energy at t2 site is greater than t3 site. If we combine H binding results of
bridge and top sites, we could find that the sites which neighbors to Fe atoms is not ideal for H
adsorption. For H binding on pure Ni, the fcc site is also slightly favorable than hcp site, and both fcc
and hcp sites are more stable than bridge and top sites. This result is consistent with our Ni4Fe results.
Based on the results obtained, we can assume that the most stable sites for all species are either
fcc or hcp for both Ni4Fe and pure Ni. The species at the top and bridge sites are bound more weakly
to the Ni4Fe surface than at the fcc and hcp sites.
Summarizing the data for the binding energy of the species involved in the methane conversion
reaction, we find that the Ni4Fe surface bind CH3, CH2, and H by 1~5 kcal/mol more weakly than
pure Ni, whereas the binding energy of CH and C is ~10 kcal/mol weaker for Ni4Fe compared to pure
Ni. The weaker binding energy of C means easier removal of carbon deposits from the surface,
providing better coking resistance. However, to make a reliable conclusion about the catalytic activity
116
of the Ni-based binary alloys compared to that of pure Ni, we need information about the reaction
energy barriers for the methane conversion on the metal alloy surfaces.
Table 2. CH3 binding energies.
Ni4Fe
Opt.
spin
Ni ave.
spin
Fe ave.
spin
t1
Ebond
(Kcal/mol
35.2
22.25
0.63
3.07
t2
36.2
22.35
0.63
3.09
t3
35.6
22.25
0.63
3.07
b1
37.7
22.31
0.63
3.09
Sit
es
sit
es
Ebond
(Kcal/mol
Calc.
spin
Opt.
spin
Ni ave.
spin
37.2
12
11.80
0.78
39.3
12
11.69
0.78
42.7
12
11.54
0.79
42.3
12
11.60
0.80
unstablea
b2
H3
Ni
b3
36.0
22.36
0.63
3.08
b4
34.5
22.34
0.63
3.06
b5
35.9
22.26
0.63
3.07
b6
37.3
22.31
0.63
3.08
b7
34.5
22.26
0.63
3.08
f1
37.3
22.29
0.63
3.06
f2
41.4
22.25
0.62
3.08
f3
40.4
22.24
0.62
3.09
h1
40.5
22.32
0.63
3.08
h2
40.0
22.34
0.63
3.08
h3
37.9
22.17
0.62
3.07
CH3 moves to the h1 site after geometry optimization.
117
Table 3. CH2 binding energies.
Ni4Fe
Opt.
spin
Ni ave.
spin
Fe ave.
spin
t1
Ebond
(Kcal/mol
62.6
22.62
0.64
3.07
t2
64.5
22.72
0.64
3.08
Sit
es
t3
unstablea
b1
unstableb
b2
unstablec
b3
H2
Ni
82.8
21.90
0.60
b4
unstabled
b5
unstablea
b6
82.5
21.91
0.60
3.08
b7
79.5
21.90
0.60
3.07
f1
78.3
22.09
0.61
3.09
f2
84.1
22.00
0.61
3.08
f3
85.7
21.99
0.60
3.09
h1
84.3
21.91
0.60
3.08
h2
81.2
21.94
0.60
3.08
h3
79.1
21.88
0.60
3.09
CH2 moves to f2 site after geometry optimization.
CH2 moves to f3 site after geometry optimization.
CH2 moves to h1 site after geometry optimization.
Ebond
(Kcal/mol
Calc.
spin
Opt.
spin
Ni ave.
spin
66.0
12
11.62
0.77
83.9
12
11.04
0.76
89.3
11
10.88
0.71
88.6
11
10.77
0.71
3.08
sit
es
CH2 moves to h2 site after geometry optimization.
118
Table 4. CH binding energies.
Ni4Fe
Sit
es
Ebond
(Kcal/mol
Opt.
spin
Ni
Ni ave.
spin
t1
unstablea
t2
unstableb
t3
unstablec
b1
unstablec
b2
unstablec
b3
unstablec
b4
unstabled
b5
unstablee
b6
unstabled
b7
unstablea
Fe ave.
spin
f1
129.9
21.79
0.58
3.11
f2
138.9
21.34
0.57
3.09
f3
139.0
21.43
0.57
3.09
h1
139.3
21.22
0.56
3.09
h2
138.8
21.25
0.56
3.08
h3
133.6
21.16
0.55
3.10
sit
es
Ebond
(Kcal/mol
Calc.
spin
Opt.
spin
Ni ave.
spin
99.5
11
10.92
0.72
139.4
10
10.42
0.65
148.0
10
10.24
0.63
148.9
10
10.06
0.65
CH moves to the h3 site after geometry optimization.
CH moves to the f3 site after geometry optimization.
CH moves to the h1 site after geometry optimization.
CH moves to the h2 site after geometry optimization.
CH moves to the f2 site after geometry optimization.
119
Table 5. C binding energies.
Ni4Fe
Sit
es
Ebond
(Kcal/mol
Opt.
spin
Ni
Ni ave.
spin
t1
unstablea
t2
unstableb
t3
unstablea
b1
unstablec
b2
unstablec
b3
unstablec
C b4
unstabled
b5
unstablee
b6
unstabled
b7
unstableb
Fe ave.
spin
f1
137.0
20.59
0.52
3.08
f2
141.7
20.50
0.51
3.08
f3
141.5
20.51
0.51
3.08
h1
142.9
20.35
0.51
3.06
h2
141.8
20.32
0.51
3.06
h3
140.0
20.34
0.50
3.10
sit
es
Ebond
(Kcal/mol
Calc.
spin
Opt.
spin
Ni ave.
spin
103.6
11
10.96
0.71
143.1
10
10.04
0.64
153.2
10
9.87
0.64
154.8
10
9.82
0.63
C moves to the h3 site after geometry optimization.
C moves to the f3 site after geometry optimization.
C moves to the h1 site after geometry optimization.
C moves to the h2 site after geometry optimization.
C moves to the f2 site after geometry optimization.
120
Table 6. H binding energies.
Ni4Fe
Sit
es
Ebond
(Kcal/mol
Opt.
spin
Ni
Ni ave.
spin
Fe ave.
spin
sit
es
Ebond
(Kcal/mol
Calc.
spin
Opt.
spin
Ni ave.
spin
52.7
12
11.93
0.79
62.6
12
11.79
0.79
65.7
12
11.77
0.79
65.4
12
11.78
0.79
unstablea
t1
t2
53.9
22.52
0.64
3.09
t3-2
51.9
22.42
0.64
3.06
b1
unstableb
b2
unstablec
b3
58.0
22.30
0.63
H b4
unstabled
b5
unstablee
b6
unstableb
b7
unstableb
3.08
f1
58.1
22.40
0.64
3.07
f2
60.4
22.33
0.63
3.08
f3
59.2
22.24
0.63
3.07
h1
60.2
22.35
0.63
3.08
h2
59.6
22.35
0.63
3.08
h3
58.1
22.27
0.63
3.08
C moves to the h3 site after geometry optimization.
C moves to the f3 site after geometry optimization.
C moves to the h1 site after geometry optimization.
C moves to the h2 site after geometry optimization.
C moves to the f2 site after geometry optimization.
121
8.3 Reaction Energy Barriers for Methane Conversion
8.3.1 Ni4Fe Catalyst
Generally, efficiency of a catalyst much depends on reaction energy barriers. Having the binding
energies calculated for the intermediates at various sites, we can now estimate barriers for the steps
of the methane reforming reaction to describe the overall reaction mechanism.
Since the experimental barrier for CH4 dehydrogenation is 17.7 Kcal/mol,47 which is smaller than
the subsequent reaction, we will ignore CH4 dehydration but focusing on the subsequent reactions.
We used CH3 bound at the most preferable site, i.e. f2 as the initial state and the decomposed CHx
which stayed at f2 site as the initial state of the subsequent reactions because f2 is still energy preferable
site even though not the most one. The reason is because migration from f2 site to the neighboring h1
site will pass through the relatively high energy b3 site, and the energy cost to overcome this barrier
make this route is not preferable.
Table 7 summarize the reaction energy barriers for methane conversion on Ni4Fe surface and
make a comparison with pure Ni. The reaction barrier for CH3 (ad) CH2 (ad) + H (ad) on Ni4Fe surface
is 21.4 Kcal/mol, slightly higher than on pure Ni surface. However, for the rate- determination step
(RDS), CH (ad) C (ad) + H (ad), the reaction energy barrier for Ni4Fe is smaller than on pure Ni surface.
Therefore, the catalytic activity of Ni4Fe for CH4 conversion may be greater than Ni activity, which
was also reported in previous experimental result.20 Furthermore, the lower C binding energy implies
that the coking resistance on Ni4Fe surface is better than on pure Ni, which conforms to previous
experimental literature.19,30
122
Table 7. Reaction energy barriers (kcal/mol) for methane decomposition and carbon binding
energy on Ni, Ni4Fe, Ni4Co, Ni4Mo, and Ni4Mn catalyst surfaces.
Reaction
Ni47
Ni4Fe
Ni4Co
Ni4Mo
Ni4Mn
CH3→CH2+H
18.4
21.2
21.4
17.9
23.8
CH2→CH+H
8.3
9.4
10.2
8.1
10.9
CH→C+H
32.8
31.4
33.7
34.9
29.9
Carbon binding
energy
155
143
143
154.1
144.1
8.3.2 Ni4Co, Ni4Mo, and Ni4Mn Catalysts
Because better activity for fuel oxidation at SOFC anode by some Ni alloys was predicted by
previous theoretical literature,33 we performed NEB calculations to evaluate the activity of Ni4Co,
Ni4Mo, and Ni4Mn alloys for CH4 reforming reaction. Our segregation energy comparison of these
three alloys between 5-6-12-13 (the segregated one) and 2-7-10-15 (non-segregated one)
configuration showed that the segregated configuration still has lower energy than non-segregated
one (see Table 8), therefore, we used the same 5-6-12-13 configuration as the slab structure for CH4
reforming reaction. We did not perform the binding energy calculation for these three alloys because
of the computational cost but assume that the CHx species have the same most preferable binding
sites as Ni4Fe. Our simple tests for some fcc/hcp/bridge sites binding showed that the binding at fcc
and hcp sites is more stable than on bridge sites and the difference between fcc and hcp site is
insignificant as we observed at Ni4Fe catalyst.
The reaction energy barriers for these three catalysts are listed in Table 7. The values of each step
are similar to the values of Ni4Fe and pure Ni catalysts. The RDS for CH4 conversion on these
123
catalysts is CHad Cad + Had, the same as Ni4Fe and Ni. Only Ni4Mn possesses both lower RDS
barrier and C binding energy as Ni4Fe. Ni4Co has lower C binding energy but slightly higher RDS
barrier than pure Ni. The results are summarized by Figure 4. Differs to the previous study,33 our
result on Ni4Mo shows that this alloy should have a similar activity and C binding energy as pure Ni,
and could not be proposed as a potential catalyst with better activity and coking resistance. This
contradiction may be due to the surface segregation effect that did not taken into consideration or
using only binding energy as the indicator for activity in previous literature.35 To summarize, Ni4Fe
and Ni4Mn have higher activity for CH4 conversion and better coking resistance, thus could be
potential catalysts for CH4 reforming at SOFC anode.
Table 8. Ni4X alloy surface segregation energies (eV).
Ni4X alloys
Position of the X atoms
Relative energy (eV)*
Ni4Fe
5-6-12-13
0.00
2-7-10-15
0.57
5-6-12-13
0.00
2-7-10-15
0.35
5-6-12-13
0.00
2-7-10-15
0.55
5-6-12-13
0.00
2-7-10-15
0.46
Ni4Co
Ni4Mn
Ni4Mo
The energy is relative to 5-6-12-13 configuration (the surface segregated structure) of each alloy.
124
Figure 4. Energy pathway for CH4 decomposition on Ni4X surfaces (the insets show the
decomposition occur on Ni4Fe surface).
125
Chapter 9
Conclusions
SOFC is a clean energy technology with wide fuel adaptability. However, using hydrocarbon as
the fuel has the coking problem and induce the poor stability. We studied the surface segregation for
Ni4Fe, and found that a Fe atom distribution of (0%, 50%, 25%, 25%, 0%) starting at the bottom layer
with Ni segregation has the lowest energy. The binding energy calculations of CHx species at various
binding sites were performed and used as the initial state for CH4 conversion reaction energy barrier
calculation. The fcc and hcp sites are more preferable for binding than top and bridge sites. The
binding energy results for CHx species are pretty similar to pure Ni results but the C binding energy
on Ni4Fe is ~10 Kcal/mol less than on pure Ni surface. Our NEB calculation showed that Ni4Fe also
has lower RDS energy barrier. Further calculation for other Ni alloys and comparison with pure Ni
results indicated that Ni4Fe, Ni4Co, and Fe4Mn all have better coking resistance than pure Ni, but only
Ni4Fe and Ni4Mn have slightly better activity than Ni, maybe good catalysts for CH4 reforming at
SOFC anode.
126
BIBLIOGRAPHY
[1] Singhal, S. Solid State Ionics 2000, 135, 305-313.
[2] Lu, X. C.; Zhu, J. H.; Bi, Z. H. Solid State Ionics 2009, 180, 265-270.
[3] McIntosh, S.; Gorte, R. J. Chem. Rev. 2004, 104, 4845-4866.
[4] Ishihara, T.; Yan, J.; Shinagawa, M.; Matsumoto, H. Electrochim. Acta 2006, 52, 1645-1650.
[5] Sheng, C. Y.; Dean, A. M. J. Phys. Chem. A 2004, 108, 3772-3783.
[6] Gorte, R. J.; Vohs, J. M. J. Catal. 2003, 216, 477-486. 305-313.
[7] da Paz Fiuza, R.; Aurélio da Silva, M.; Boaventura, J. S. Int. J. Hydrogen Energy 2010, 35,
11216-11228.
[8] Lin, Y.; Zhan, Z.; Liu, J.; Barnett, S. A. Solid State Ionics 2005, 176, 1827-1835.
[9] Ringuedé, A.; Fagg, D.; Frade, J. J. Eur. Ceram. Soc. 2004, 24, 1355-1358.
[10] Xie, Z.; Zhu, W.; Zhu, B.; Xia, C. Electrochim. Acta 2006, 51, 3052-3057. 1358.
[11] Zhu, H.; Kee, R. J.; Janardhanan, V. M.; Deutschmann, O.; Goodwin, D. G. J. Electrochem.
Soc. 2005, 152, A2427-A2440.
[12] Ni, M.; Leung, M. K.; Leung, D. Y. International Journal of Energy Research 2009, 33, 943959.
[13] Kim, T.; Liu, G.; Boaro, M.; Lee, S.-I.; Vohs, J. M.; Gorte, R. J.; Al-Madhi, O.; Dabbousi, B.
J. Power Sources 2006, 155, 231-238.
[14] Huang, T.-J.; Wang, C.-H. J. Power Sources 2006, 163, 309-315.
[15] Hecht, E. S.; Gupta, G. K.; Zhu, H.; Dean, A. M.; Kee, R. J.; Maier, L.; Deutschmann, O. Appl.
Catal., A 2005, 295, 40-51.
[16] Gross, M. D.; Vohs, J. M.; Gorte, R. J. J. Mater. Chem. 2007, 17, 3071-3077.
[17] Ju, Y.-W.; Eto, H.; Inagaki, T.; Ida, S.; Ishihara, T. J. Power Sources 2010, 195, 6294-6300.
127
[18] Tucker, M. C. J. Power Sources 2010, 195, 4570-4582.
[19] Kan, H.; Lee, H. Catal. Commun. 2010, 12, 36-39.
[20] Lu, X.; Zhu, J. J. Power Sources 2007, 165, 678-684.
[21] Fu, C.; Chan, S.; Ge, X.; Liu, Q.; Pasciak, G. Int. J. Hydrogen Energy 2011, 36, 13727-13734.
[22] Huang, B.; Wang, S.; Liu, R.; Wen, T. J. Power Sources 2007, 167, 288-294.
[23] Cowin, P. I.; Petit, C. T.; Lan, R.; Irvine, J. T.; Tao, S. Adv. Energy Mater. 2011, 1, 314-332.
[24] Gorte, R.; Vohs, J. Curr. Opin. Colloid Interface Sci. 2009, 14, 236-244.
[25] Cheng, Z.; Wang, J.-H.; Choi, Y.; Yang, L.; Lin, M.; Liu, M. Energy Environ. Sci. 2011, 4,
4380-4409.
[26] Gong, M.; Liu, X.; Trembly, J.; Johnson, C. J. Power Sources 2007, 168, 289-298.
[27] Yang, C.; Yang, Z.; Jin, C.; Xiao, G.; Chen, F.; Han, M. Adv. Mater. 2012, 24, 1439-1443.
[28] Kim, H.; Lu, C.; Worrell, W.; Vohs, J.; Gorte, R. J. Electrochem. Soc. 2002, 149, A247-A250.
[29] Baker, R. Carbon 1989, 27, 315-323.
[30] Zhu, H.; Wang, W.; Ran, R.; Su, C.; Shi, H.; Shao, Z. Int. J. Hydrogen Energy 2012, 37, 98019808.
[31] Horita, T.; Sakai, N.; Kawada, T.; Yokokawa, H.; Dokiya, M. J. Electrochem. Soc. 1996, 143,
1161-1168.
[32] Rossmeisl, J.; Bessler, W. G. Solid State Ionics 2008, 178, 1694-1700.
[33] An, W.; Gatewood, D.; Dunlap, B.; Turner, C. H. J. Power Sources 2011, 196, 4724-4728.
[34] Nikolla, E.; Holewinski, A.; Schwank, J.; Linic, S. J. Am. Chem. Soc. 2006, 128, 11354-11355.
[35] Ammal, S. C.; Heyden, A. J. Phys. Chem. Lett. 2012, 3, 2767-2772.
[36] Schultz, P. SeqQuest, Electronic Structure Code; Sandia National Laboratory, Albuquerque,
NM, http://dft.sandia.gov/Quest/.
[37] Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865-3868.
[38] Perdew, J. P.; Zunger, A. Phys. Rev. B 1981, 23, 5048-5079.
128
[39] Ceperley, D. M.; Alder, B. J. Phys. Rev. Lett. 1980, 45, 566-569.
[40] Goddard, W. A., III Phys. Rev. 1968, 174, 659-662.
[41 Hamann, D. R. Phys. Rev. B 1989, 40, 2980-2987.
[42] Melius, C. F.; Goddard, W. A., III Phys. Rev. A 1974, 10, 1528-1540.
[43] Melius, C. F.; Olafson, B. D.; Goddard, W. A., III Chem. Phys. Lett. 1974, 28, 457-462.
[44] Redondo, A.; Goddard, W. A., III; McGill, T. C. Phys. Rev. B 1977, 15, 5038-5048.
[45] Mills, G.; Jonsson, H. Phys. Rev. Lett. 1994, 72, 1124-1127.
[46] Mills, G.; Jonsson, H.; Schenter, G. K. Surf. Sci. 1995, 324, 305-337.
[47] Mueller, J. E.; van Duin, A. C.; Goddard III, W. A. J. Phys. Chem. C. 2009, 113, 20290-20306.
[48] Tsai, H.-C.; Yu, T.; Sha, Y.; Merinov, B. V.; Wu, P.-W.; Chen, S.-Y.; Goddard III, W. A. J.
Phys. Chem. C. 2014.
[49] Tsai, H.-C.; Hsieh, Y.-C.; Yu, T.; Lee, Y.-J.; Wu, Y.-H.; Merinov, B. V.; Wu, P.-W.; Chen,
S.-Y.; Adzic, R. R.; Goddard III, W. A. ACS Catal. 2015.
129
Appendix
DFT Study of Doped Perovskite
Ceramics as Proton Conducting Materials
130
Preface
The last part of my thesis includes preliminary results of work that is still in progress and
certain issues still need to be clarified. Nevertheless, a summary of this study may be of interest for
researchers working in the field of proton-conducting perovskite-ceramics. Since the results are
preliminary, I decided to place them in the appendix rather than in the main body of the thesis.
Abstract
We used quantum mechanics calculations to study the doped perovskite oxides as protonconducting ceramics for hydrogen permeation. A 2x2x2 Ba8X7(OH)1M1O23 (X=Ce, Zr; M=Y, Gd,
Dy) cell was used to study proton diffusion that both includes intra-octahedron and inter-octahedra
proton transfers. Without any calculation restriction, we only observed intra-octahedron proton
transfer with 0.26 eV barrier, which is smaller than experimental value 0.44 eV. For fixed OdonorOacceptor distance, our calculations show that the Gd-doped BaCeO3 and Y-doped BaZrO3 oxides
have slightly lower (0.03 eV) proton transfer barriers than the others, but the difference is not
significant. It is well-known that the proton transfer barrier is proportional to the Odonor-Oacceptor
distance. During the thermal librations/vibrations, the Odonor-Oacceptor distance becomes short enough
and the proton may jump from one oxygen to another, i.e. the proton transfer will occur. As a matter
of fact, the bulk proton conductivity of the oxides considered here are similar, therefore, a more
131
comprehensive study by ReaxFF reactive dynamics which includes the grain boundary
conduction will be performed in the future based on our QM results.
Introduction
Integrated gasification combined cycle (IGCC) is an electricity generation technology that
integrates the coal gasification and combined cycle. The gasification process transfers the coal or
carbon based fuel into the syngas. Then the purification process removes the impurities and sulfur
from the syngas before combustion. Therefore, IGCC system has lower emission with equal or higher
efficiency than general coal combustion power plants.1 Our strategic target is to develop a system
that could capture the CO2 for photocatalytic conversion and separate the useful H2 gas for fuel cell
operation, as illustrated in Figure 1. Figure 2 shows the system that we designed for H2 separation.
The cermet with porous ceramic substrate tube is used for H2 permeation. The H2 gas will be
decomposed and oxidized to protons by the metal catalysts within the cermet. The protons then diffuse
through the proton conducting ceramics and are reduced to hydrogen gas by combing with the
electrons at the interface between the cermet and porous ceramic tube. The driving force is the
concentration difference between two sides of the cermet membrane.
The perovskite-type oxides have attracted much attention because of various possible
applications in various devices, such as sensors, hydrogen pumping/separation/extraction, steam
electrolysis, fuel cells etc. due to their high protonic conductivity at elevated temperature.2-9 Among
these ABO3-type perovskite oxides, BaCeO3, SrCeO3, BaZrO3, and CaZrO3 exhibit good
conduction properties under hydrogen containing atmosphere at temperature which is lower than
the operating temperature of solid oxide fuel cells based on O2- ion conducting ceramics.3,10,1110,12
132
However, some of these oxides, for instance BaCeO3, are not stable in the CO2 containing
environment10,12,13 because of the following reaction:
BaCeO3 + CO2 BaCO3 + CeO2
In contrast to BaCeO3, BaZrO3 is much more stable in the CO2 containing environment.10,12,13 But
the latter oxide has a lower total proton conductivity because of high resistivity resulting from the
large amount of grain boundaries observed in this ceramics.14 Therefore, to reach both good stability
and high proton conductivity, combinations of these two oxides were used in some previous
studies.10,12,13 In addition to Y-doped BaCeO35,10,13,15 and BaZrO3,4,16-18 proton-conducting oxides
doped with other transition elements, such as Gd,11,12,18-21 Nd,12,18,22 or Dy,23 were also reported in
literature. Our study focuses on the Gd- and Dy-doped ceramics because high conductivity was
reported for these doped perovskite oxides.12,23 The composition of our perovskite ceramics for proton
conduction is BaCe0.4Zr0.4Gd0.1Dy0.1O3-x. The cermet applied for hydrogen permeation is a
combination of BaCe0.4Zr0.4Gd0.1Dy0.1O3-x and Pd. However, the simulation that includes both
ceramics and metals is extremely difficult, so we only performed the calculation for the perovskite
oxides. Furthermore, it is hard to simulate grain boundary proton conduction using the DFT method,
so here we only discuss the influences of dopants on the bulk conduction property. In this work, we
studied the proton diffusion in Gd- and Dy-doped barium ceria/zirconia and compared it to proton
diffusion in Y-doped barium ceria/zirconia.
133
Figure 1. Illustration for the CO2 capture/separation and fuel conversion idea.
Figure 2. The design for H2 permeation membrane.
134
Theoretical Methods
For our QM three dimensional periodic bulk calculation, we used Perdew−Becke−Ernzehof
(PBE)24 functional including low gradient London dispersion correction (PBE-ulg) in the generalized
gradient approximation (GGA)25,26 implemented in Vienna Ab-Initio Simulation Package (VASP)2729
to calculate the lattice constants and proton conduction barriers. The projector-augmented wave
method was used to calculate the interaction between the core and valence electrons.30 The reciprocal
lattice is 4x4x4 Γ-centered Monkhorst−Pack grid with zero shift in our calculations. The cutoff energy
for plane wave basis is 500 eV. The convergence criteria for electronic wave function and geometry
optimization are 1x10-4 eV and 1x10-2 eV/Å force, respectively. For our calculation, 1x1x1 ABO3
unit cells were used for non-doped perovskite lattice constant calculation, while 2x2x2 ABO3 unit
cells were used for doped perovskite lattice constants and proton conduction barrier calculations. Our
BaCeO3 and BaZrO3 lattice constants are 4.47 Å and 4.26 Å, compared to experimental result 4.44
Å31 and 4.19 Å18, respectively ( see Table 1).
BaCe0.4Zr0.4Gd0.1Dy0.1O3-x is not a simple stoichiometric ratio that could be constructed by a
2x2x2 cell; furthermore, there are too many possible configurations for BaCe0.4Zr0.4Gd0.1Dy0.1O3-x
in the 2x2x2 cell. Therefore, we simplified our system to only one doped element in either cermet
or zirconate (e.g. Ba8Ce7GdHO24) and discussed the influence of different doping elements on
proton diffusion. The hydrogen atom was placed into a position neighboring to an oxygen atom to
form an OH bond, then the configuration built was computationally optimized. The optimized
structures were then used as the initial and final states for a proton diffusion path in our nudged
elastic band32,33(NEB) calculation. The energy difference between the transition state found by
NEB and the initial state was used as the barrier for proton diffusion along this path.
135
Table 1. Lattice constants comparison of barium cerate and barium zirconate.
Calculated (Å)
Experimental (Å)
BaCeO3
4.47
4.4431
BaZrO3
4.26
4.1918
Table 2. Lattice constants of doped perovskites.
Composition
Lattice Constants (Å)
BaCeO3 (1x1x1)
4.47
Ba8Ce7YO24 (2x2x2)
8.92
Ba8Ce7GdO24 (2x2x2)
8.93
Ba8Ce7DyO24 (2x2x2)
8.92
BaZrO3 (1x1x1)
4.26
Ba8Zr7YO24 (2x2x2)
8.55
Ba8Zr7GdO24 (2x2x2)
8.56
Ba8Zr7DyO24 (2x2x2)
8.55
136
Results and Discussion
The lattice constants for Ba8X7MHO24 (X=Ce or Zr, M=Y, Gd, Dy), are listed in Table 2. Doping
does not significantly change the lattice constants compared to those for the undoped oxides.
In Ba8X7MHO24 (X=Ce, Zr; M=Y, Gd, Dy), there are five various conduction paths for the
proton transfer from one oxygen atom to another. These paths are shown in Figure 3. In path 1, the
proton migrates along an edge of a ZrO6-octahedron in the Zr-OH-Y configuration. The path 2
describes a similar proton transfer but in the Zr-OH-Zr configuration. Path 3 describes the intraYO6-octahedron proton transfer. Path 4 and 5 are inter-octahedra proton transfers between a ZrO6
octahedron and YO6 octahedron and two ZrO6 octahedra, respectively. Complete proton diffusion
paths may be composed from combination of paths 1-5 in which the proton returns to the
geometrically equivalent position in our three dimension periodic cell. For example, a proton
conduction route composed by paths 1331 is a complete route because the proton attached
to the crystallographically equivalent oxygen atom as shown in Figure 4. If we used the Boltzmann
population (Fstate ∝ e-E/kT) of the initial state energy of each path to determine the probability for
proton to occupy a certain site at 700OC, the probability for the proton at the initial site of path 1,
2, 3, 4 is 0.246; whereas for path 5 it is only 0.014. Therefore, it is practically impossible for the
proton to use path 5. The energy barriers for different proton transfer paths in Y-doped BaZrO3 are
tabulated in Table 3. Our NEB calculation also shows that the OH bond reorientation has 0.07 eV
energy difference in the last two steps, which is comparable with the earlier reported value of 0.02
eV.2
137
Figure 3. The proton conduction path in Ba8Zr7YHO24 cell (copied and revised from Ref. 17).
138
Figure 4. The proton conduction route example in Ba8Zr7YHO24 cell.
Table 3. The relative energy corresponding to each step of various proton conduction paths in
Ba8Zr7YHO24.
Step\Path*
barrier
Path_1
0.00
0.22
0.04
0.24
0.19
0.07
0.00
0.24
Path_2
0.00
0.21
0.29
0.13
0.44
0.11
0.00
0.44
Path_3
0.00
0.07
0.19
0.32
0.19
0.07
0.00
0.32
Path_4
0.00
0.09
0.05
0.11
0.26
0.05
0.00
0.26
Path_5
0.00
0.06
0.30
0.08
0.08
0.08
0.03
0.30
*All energies are relative to “step 0” (initial state) energy of each path.
It was also found in our calculation that instead of the inter-octahedra proton transfer (path 4 and
5), we got two successive intra-octahedron proton transfers, as illustrated in Figure 5. Figure 6 shows
139
the potential energy surface of proton transfer in doped cerates with fixed donor oxygen and
acceptor oxygen distances. We plot the energy versus the distance between the proton and donor
oxygen atoms of barium ceria doped with Dy, Gd, and Y. As expected, the proton transfer barrier is
proportional to the donor oxygen−acceptor oxygen distance. For 2.46 Å, the barrier is about 0.1 eV,
while for 2.69 Å, the barrier is about 0.45 eV. The Gd-doped barium ceria has the slightly lower
barrier than the Y- and Dy-doped analogues, which is in agreement with the reported higher proton
conductivity of the Gd-doped BaCeO3 oxide.12
Figure 5. The calculated successive intra-octahedra proton transfer path in our Ba8Zr7YHO24 cell.
Figure 6. The energy versus Odonor-H distance of doped cerates with different fixed Odonor-Oacceptor
distance.
140
As for doped barium zirconates, the Y-doped sample seems to have the slightly lower proton
diffusion barrier than the Gd- and Dy-doped zirconates. As the distance between the donor and
acceptor oxygen atoms increases to ~2.69 Å, the proton diffusion barrier value becomes similar to the
experimentally reported activation energy of 0.44 eV.16 Furthermore, the doped zirconates have
similar proton conductivity values. This is probably because our calculations consider only the bulk
diffusion and do not include the proton diffusion through grain boundaries. Generally, the proton
diffusion barriers for all materials studied here are very similar and it is hard to make conclusions
about differences in this property based only on our calculation results. Therefore, more proper
calculations, for example reactive force field molecular dynamics that will include both grain and
grain boundary simulations are needed to get better insights in the problem we studied.
Figure 7. The energy versus Odonor-H distance of doped zirconates with different fixed Odonor-Oacceptor
distance.
141
Conclusions
Doped BaCeO3 oxides have the highest total proton conductivity among the proton-conducting
perovskite oxides, but in the CO2 containing environment these materials react with CO2 to form
barium carbonates and ceria. To improve their chemical stability, mixtures with some barium
zirconates might be used to reach a compromise between stability and conductivity. Our DFT
calculation simulate the lattice constants close to the experimental values. We discussed possible
proton transfer route for Ba8X7(OH)1MO23 (X=Ce or Zr, M=Y, Gd, Dy), and find that the proton
transfer via path 5 is not preferable because of its high initial state energy. According to our
calculation, inter-octahedra proton transfer paths 4 and 5 were substituted by two successive intraoctahedron proton transfers.
For fixed donor oxygen−acceptor oxygen distance calculations, the Gd-doped barium cerium
oxides have the slightly lower proton transfer (0.02~0.03 eV) barrier than Y- and Dy-doped barium
cerium oxides, while among the zirconate analoges, Y-doped BaZrO3 has the lower energy barrier
(0.02~0.03 eV) than the other two. The doped zirconates have the barrier value similar to the
experimental value of the activation energy when the distance between the donor oxygen and
acceptor oxygen is ~2.69 Å. Generally, from our calculations the doped cerates and zirconates have
similar bulk proton diffusion barriers and more comprehensive work including the grain boundary
proton diffusion simulations by ReaxFF reactive dynamics will be trained based on our QM results.
142
BIBLIOGRAPHY
[1] Frey, H. C.; Zhu, Y. Environ. Sci. Technol. 2006, 40, 1693-1699.
[2] Gomez, M. A.; Griffin, M. A.; Jindal, S.; Rule, K. D.; Cooper, V. R. J. Chem. Phys. 2005,
123, 094703.
[3] Schober, T. Solid State Ionics 2003, 162, 277-281.
[4] Gomez, M. A.; Chunduru, M.; Chigweshe, L.; Foster, L.; Fensin, S.; Fletcher, K. M.;
Fernandez, L. E. J. Chem. Phys. 2010, 132, 214709.
[5] Guan, J.; Dorris, S.; Balachandran, U.; Liu, M. Solid State Ionics 1997, 100, 45-52.
[6] Zhu, Z.; Sun, W.; Yan, L.; Liu, W.; Liu, W. Int. J. Hydrogen Energy 2011, 36, 6337-6342.
[7] Iwahara, H. Solid State Ionics 1996, 86, 9-15.
[8] Iwahara, H.; Asakura, Y.; Katahira, K.; Tanaka, M. Solid State Ionics 2004, 168, 299-310.
[9] Iwahara, H. Solid State Ionics 1995, 77, 289-298.
[10] Katahira, K.; Kohchi, Y.; Shimura, T.; Iwahara, H. Solid State Ionics 2000, 138, 91-98.
[11] Gomez, M. A.; Chunduru, M.; Chigweshe, L.; Foster, L.; Fensin, S.; Fletcher, K. M.;
Fernandez, L. E. J. Chem. Phys. 2010, 132, 214709.
[12] Ryu, K. H.; Haile, S. M. Solid State Ionics 1999, 125, 355-367.
[13] Fabbri, E.; D'Epifanio, A.; Di Bartolomeo, E.; Licoccia, S.; Traversa, E. Solid State Ionics
2008, 179, 558-564.
143
[14] Tao, S.; Irvine, J. T. Adv. Mater. 2006, 18, 1581-1584.
[15] Xie, K.; Yan, R.; Liu, X. J. Alloys Compd. 2009, 479, L40-L42.
[16] Bohn, H. G.; Schober, T. J. Am. Ceram. Soc. 2000, 83, 768-772.
[17] Merinov, B.; Goddard III, W. A. J. Chem. Phys. 2009, 130, 194707. 214709.
[18] Sundell, P. G.; Björketun, M. E.; Wahnström, G. Phys. Rev. B 2006, 73, 104112.
[19] Zając, W.; Rusinek, D.; Zheng, K.; Molenda, J. Cent. Eur. J. Chem. 2013, 11, 471-484.
[20] Hermet, J.; Torrent, M.; Bottin, F.; Dezanneau, G.; Geneste, G. Phys. Rev. B 2013, 87, 104303.
[21] Kreuer, K.; Schönherr, E.; Maier, J. Solid State Ionics 1994, 70, 278-284.
[22] Iwahara, H.; Uchida, H.; Ogaki, K.; Nagato, H. J. Electrochem. Soc. 1991, 138, 295-299.
[23] Wang, W.; Liu, J.; Li, Y.; Wang, H.; Zhang, F.; Ma, G. Solid State Ionics 2010, 181, 667-671.
[24] Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865-3868.
[25] Perdew, J. P.; Zunger, A. Phys. Rev. B 1981, 23, 5048-5079.
[26] Ceperley, D. M.; Alder, B. J. Phys. Rev. Lett. 1980, 45, 566-569.
[27] Kresse, G.; Hafner, J. Physical Review B 1993, 47, 558.
[28] Kresse, G.; Furthmüller, J. Comput. Mater. Sci. 1996, 6, 15-50.
[29] Kresse, G.; Furthmüller, J. Phys. Rev. B 1996, 54, 11169.
[30] Kresse, G.; Joubert, D. Phys. Rev. B 1999, 59, 1758.
[31] Knight, K. Solid State Ionics 1994, 74, 109-117.
144
[32] Mills, G.; Jonsson, H. Phys. Rev. Lett. 1994, 72, 1124-1127.
[33] Mills, G.; Jonsson, H.; Schenter, G. K. Surf. Sci. 1995, 324, 305-337.
AAA