Understanding the Interplay Between Iron Adsorption, 
Surface Reconstruction, and Electrocatalytic Oxygen 

Evolution by Transition Metal (Hydr)oxides 
 

 

by 

  

Liam P. Twight 
 

A dissertation accepted and approved in partial fulfillment of the 

requirements for the degree of 

Doctor of Philosophy 

In Chemistry 

 

 

 

Dissertation Committee:  

Carl Brozek, Chair  

Shannon Boettcher, Advisor  

Ramesh Jasti, Core Member  

Erin Moore, Institutional Representative  

 

University of Oregon  

 

Fall 2024 

 

 

 



2 
 

 

 

 

 

 

 

 

© 2024 Liam P. Twight 

This work is openly licensed via CC BY 4.0. 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

https://creativecommons.org/licenses/by/4.0/


3 
 

DISSERTATION ABSTRACT 

Liam P. Twight 

Doctor of Philosophy in Chemistry 

Title: Understanding the Interplay Between Iron Adsorption, Surface Reconstruction, and 

Electrocatalytic Oxygen Evolution by Transition Metal (Hydr)oxide Electrocatalysts  
 Renewable electricity-driven alkaline water electrolysis is poised to be a key 

technology for reducing the CO2 emissions from combustion of fossil fuels in industrial 

sectors like aviation and fertilizer manufacturing. One of the largest costs of operating 

water electrolyzers is that of the electricity needed to drive the liberation of hydrogen and 

oxygen gases. Therefore, improving electrolyzer efficiency – lowering the electrical power 

needed per kilogram of H2 generated – is essential to their widespread deployment. The 

anodic oxygen evolution reaction (OER) is a process with intrinsically slow kinetics that 

makes large contributions to electrolyzer inefficiency. The kinetics of OER can be hastened 

with the proper choice of catalyst the best of which are based on Ni and Co oxides or 

hydroxides with a minority component of incorporated Fe. Decades of research involving 

Ni/Co/Fe OER catalysts have been done, but there is still debate about the nature of the 

active sites in these materials. The complexity of these systems is to blame; the formation 

and maintenance of high OER activity sites depends on highly dynamic processes 

involving surface iron site dissolution-redeposition and catalyst structural change both of 

which are likely functions of the electrolyte pH, iron concentration in electrolyte, applied 

electrical bias, and catalyst chemical composition. 

 This dissertation advances understanding of the nature of extremely high activity 

Fe active sites which form in-situ on nickel and cobalt hydroxide and lanthanum nickel 

oxide, three technologically promising catalysts. In Chapter I, I describe from a broad 

perspective the areas where high efficiency alkaline electrolyzers could serve to eliminate 

CO2 emissions and the role that OER catalysts will play in accomplishing this goal. In 

Chapter II, I report the results of fundamental investigations which reveal that Fe active 

sites are not of one, but two kinds: one that forms by surface adsorption of electrolyte Fe 

and another that substitutes for host metal atoms. In Chapter II, I describe the results of a 

deeper mechanistic investigation into the thermodynamic parameters that govern the 



4 
 

activity of these surface Fe sites. Chapter III extends the methods used to understand Fe-

sites on nickel hydroxides to an understudied perovskite oxide, lanthanum nickelate 

(LaNiO3). Together, these studies deepen our understanding of why Fe is a ubiquitous 

activator of OER catalysts and broaden the family of catalysts for which Fe activation is 

integral. As a result, new design principles for high performance alkaline electrolyzer 

anodes become evident. Catalysts should have a high density of sites for cooperative 

surface Fe site adsorption such that the activation energy and pre-exponential factors are 

optimized. Surface restructuring should be purposefully induced for those that need it for 

high OER activity to maximize active site formation. Where restructuring is required for 

high activity, surface chemical descriptors should be developed and utilized instead of bulk 

ones which may not directly connect to the relevant physical picture of catalysis. This 

dissertation contains previously published and un-published co-authored materials. 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 



5 
 

CURRICULUM VITAE 

NAME OF AUTHOR: Liam P. Twight 

GRADUATE AND UNDERGRADUATE SCHOOLS ATTENDED 

 University of Oregon, Eugene, OR 

 California State University, Long Beach, Long Beach CA 

DEGREES AWARDED 

 Doctor of Philosophy, Chemistry, University of Oregon 

 Master of Science, Chemistry, University of Oregon  

 Bachelor of Science, Chemistry, California State University, Long Beach 

AREAS OF SPECIAL INTEREST 

 Electrochemistry: electrocatalysis; water oxidation; electrolyzer devices 

 Materials science: characterization; surface chemistry; quantitative analysis 

PROFESSIONAL EXPERIENCE 

 Graduate research associate, 2019 - 2024 

 UO Inductively-coupled plasma mass spectrometry manager, 2020 - 2024 

 General chemistry lab instructor 2019 - 2020 

GRANTS, AWARDS, AND HONORS 

University of Oregon 

Renewable Energy Foundation: Sir Fraser Stoddart Scholarship, 2021 

NSF Graduate Research Fellowship Program: Honorable Mention, 2020 

UO Doctoral First Year Merit Award, 2019 

 

California State University, Long Beach   

Undergraduate Award for Outstanding Research, 2018 

James L. Jensen Undergraduate Research Fellowship, 2018 

Women in Philanthropy Undergraduate Research Scholarship, 2018 

American Chemical Society Award in Physical Chemistry, 2018 

CSULB Presidential Scholarship, 2015 

 



6 
 

PUBLICATIONS 

*denotes equal contributions 

1. Reactive surface Fe sites on nickel oxyhydroxide OER catalysts: coverage evolution, 
activity, and enthalpy-entropy compensation from temperature-dependent kinetics. 
Twight, L. Boetcher, S. W. in preparation for submission to J. Am. Chem. Soc. 2024. 

2. Trace Fe activates perovskite nickelate OER catalysts in alkaline media via redox-active 
surface Ni species formed during electrocatalysis. Twight, L., Boettcher, S. W. et al. J. 
Catalysis 432, 2024. 

3. Cooperative Fe sites on transition metal (oxy)hydroxides drive high oxygen evolution 
activity in base. Twight, L.*, Ou, Y.*, Boettcher, S. W. et al. Nat. Comm. 14, 7688, 2023. 

4. Purification of residual Ni and Co hydroxides from Fe-free alkaline electrolyte for 
electrocatalysis studies. Twight*, L., Liu, L.*, Fehrs, J., Ou, Y., Sun, D., Boettcher., S. W. 
ChemElectroChem. 9, 2022. 

5. Anode catalysts in anion-exchange-membrane electrolysis without supporting 
electrolyte: Conductivity, dynamics, and ionomer degradation. Krivina, R., Lindquist, G., 
Beaudoin, S., Stovall, N., Thompson, W. Twight, L., Marsh, D., Grzyb, J., Fabrizio, K., 
Hutchison, J., Boettcher. S. W. Adv. Mater. 34, 2022. 

6. Size-dependent properties of solution-processable conductive MOF nanocrystals. C. 
Marshall, J. Dvorak, L. Twight, L.    Chen, K. Kadota, A. Andreeva, A. Overland, T. 
Ericson, A. F. Cozzolino, C. Brozek. J. Am. Chem. Soc. (144) 2022. 

7. Hydrogen-evolution-reaction kinetics pH dependence: Is it covered? J. Mitchell, M. 
Shen, L. Twight, S. Boettcher. Chem. Catal. (2) 2022. 

8. Tunable band gaps in MUV-10(M): A family of photoredox-active MOFs with earth-
abundant open metal sites. K. 

9. Fabrizio, K. Lazarou, L. Payne, L. Twight, S. Golledge, C. Hendon, C. Brozek. J. Am. 
Chem. Soc. (143) 2021. 

10. Oxygen electrocatalysis on mixed-metal oxides/oxyhydroxides: From fundamentals to 
membrane electrolyzer technology. R. Krivina, Y. Ou, Q. Xu, L. Twight, N. Stovall, S. 
Boettcher. Acc. Mater. Res. (2) 2021. 

11. Improved Water Transport via Cationic Thin-Layers Enables High-Current Density 
Bipolar Membrane Electrolyzers. Sebastian Z. Oener, Liam P. Twight, Grace A. 
Lindquist, Shannon W. Boettcher. ACS Energy Lett. 6, 2020. 

12. Phase dependent encapsulation and release profile of ZIF-based biocomposites. F. 
Carraro, M. de J. Velásquez-Hernández, E. Astria, W. Liang, L. Twight, C. Parise, M. Ge, 
Z. Huang , R. Ricco, X. Zou, L. Villanova, C. O. Kappe, C. Doonan and P. Falcaro. Chem. 
Sci., (11) 2020 



7 
 

13. Gamma radiolysis of hydrophilic diglycolamide ligands in concentrated aqueous nitrate 
solution. G. P. Horne, A.Wilden, S. P. Mezyk, L. Twight, M. Hupert, A. Stärk, W. 
Verboom, B. J. Mincher, G. Modolo. Dalton Trans. (48) 2019. 

14. Radiolytic and hydrolytic degradation of the hydrophilic diglycolamides. A. Wilden, 
B. J. Mincher, S. P. Mezyk, L. Twight, K. M. Rosciolo-Johnson, C. A. Zarzana, M. E. 
Case, M. Hupert, A. Stärk, & G. Modolo. Solvent. Extr. Ion Exc., (36) 2018 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 



8 
 

ACKNOWLEDGEMENTS 

 

The work in Chapters II and IV was supported by the National Science Foundation 

Chemical Catalysis Program, Award #1955106. The ICP-MS instrument was funded by an 

NSF MRI Award #2117614. The work in Chapter III was supported by the National 

Science Foundation Chemical Catalysis Program, Award #2400195. The work in this 

dissertation benefited from contributions from many excellent collaborators. XAS 

measurements in collaboration with the Helmholtz-Zentrum Berlin für Materialien und 

Energie would not have happened without Marcel Risch and his team: Javier Villalobos, 

Joaquín Morales-Santelices, and Denis Antipin, whom I gratefully acknowledge. The 

computational aspects of Chapter II were performed by Bipasa Samanta and Santu Biswas 

under direction of Maytal Caspary Toroker at Technion-Israel Institute of Technology.  The 

extension of the perovskite study in Chapter IV would not have been possible without the 

preparation of key materials by Kunal Velinkar and Samji Samira of Eranda Nikolla’s 

group at University of Michigan.  Much of the characterization was enabled only through 

the expert training and advice provided by members of the UO CAMCOR staff, Steve 

Golledge, Kurt Langworthy, Josh Razink, and Valerie Brogden. 

 It is a simple fact that it takes a community to mint a new PhD student. I extend my 

sincere thanks to the leader of that community at UO, my advisor, Shannon Boettcher, who 

gave me the crucial opportunity to grow as an electrochemist in his lab, taught me how to 

think carefully and rigorously in the design and execution of basic research, and has been 

ever supportive of my intellectual goals. I thank all my lab colleagues and friends, who 

with their experience and unwavering willingness to help, were essential for me to make it 

to this finish line. Thank you especially Grace Lindquist, Jess Fehrs, Raina Krivina, Nicole 

Sagui, Nick D’antona, Aaron Kaufmann, Haokun Chen, Ana Konovalova, Minkyoung 

Kwak, and Olivia Traenkle for forming a wonderful lab community and the many 

memories we all made together that I will look back on fondly. Thank you also to my 

cohort members and friends with whom I have shared many needed moments of 

commiseration about the challenges of grad school and pleasant get-togethers filled with 

laughter, food, drinks, games, movies, and music. 



9 
 

Before the UO community there was a CSU Long Beach community. Stephen 

Mezyk, my undergraduate advisor deserves my special gratitude as the first person to see 

my potential as a scientist without whom I may not have decided to reach beyond the 

baccalaureate level. Thank you Mezyk lab members at the time, Brittany, Kylie, Jamie, 

Amir, Michael, Christian, and Landon for your friendship and mentorship. Thank you to 

the CSULB Presidential Scholarship Program who noticed the hard work I put into my 

academics in high school and gave me a scholarship that launched my entire endeavor in 

scholarship as it does for so many other deserving students.  

 Finally, thank you to the most important people in my life: my parents, sister, and 

partner. To my parents for giving me a start in life that required enormous effort on their 

part to provide, but is absolutely needed for a child to develop and tackle ambitious goals 

later in life. To my sister, for understanding me as only a sibling does. To Audrey, my 

partner, whose laughter, kindness, and love has changed my life.  

  

   

  

 

 

 

 

 

 

 

 

 

 

 



10 
 

TABLE OF CONTENTS 
I. THE ROLE OF OXYGEN EVOLUTION IN GREEN 
ELECTROLYSIS TECHNOLOGIES .............................................................................. 14 
II. COOPERATIVE FE SITES ON TRANSITION METAL (OXY)HYDROXIDES 
FOR HIGH OXYGEN EVOLUTION ACTIVITY .......................................................... 22 

Introduction ................................................................................................................... 22 
Methods ......................................................................................................................... 25 
Results and Discussion .................................................................................................. 29 
Conclusion ..................................................................................................................... 43 

III. TEMPERATURE-DEPENDENT KINETICS OF REACTIVE SURFACE FE 
SITES ON NICKEL OXYHYDROXIDE ........................................................................ 45 

Introduction ................................................................................................................... 45 
Methods and Materials. ................................................................................................. 47 
Results and Discussion. ................................................................................................. 49 
Conclusion ..................................................................................................................... 64 

IV. TRACE FE ACTIVATES PEROVSKITE NICKELATE OER 
CATALYSTS IN ALKALINE MEDIA VIA REDOX-ACTIVE SURFACE 
NI SPECIES FORMED DURING ELECTROCATALYSIS........................................... 65 

Introduction ................................................................................................................... 65 
Results and Discussion .................................................................................................. 68 
Conclusions ................................................................................................................... 83 
Methods and Materials .................................................................................................. 85 

V: SUMMARY AND OUTLOOK ON FUTURE RESEARCH ...................................... 89 
APPENDICES .................................................................................................................. 93 
A. CHAPTER II: SUPPLEMENTARY INFORMATION .............................................. 93 
B. CHAPTER III SUPPLEMENTARY INFORMATION ............................................ 126 
C. CHAPTER IV SUPPLEMENTARY INFORMATION ............................................ 132 
REFERENCES ............................................................................................................... 144 
 

 

 

 



11 
 

LIST OF FIGURES 
 

Figure 1. Generalized liquid alkaline electrolysis cell. ..................................................... 17 
Figure 2. Foreign cations interaction with oxyhydroxides via redox signatures. ............. 30 
Figure 3. Surface-confined Fe sites via chronoamperometric (CA) metal-ion-spiking. ... 34 
Figure 4. Intrinsic OER activity measured by turnover frequency 
(TOFFe) at η = 300 mV...................................................................................................... 35 
Figure 5. Fe versus Ni loading and TEM of Fe sites. ....................................................... 39 
Figure 6. Computational models and OER mechanisms for Fe sites on NiOOH. ............ 41 
Figure 7. Time to surface Fe-induced activation as a function of 
NiOxHy film thickness. ...................................................................................................... 50 
Figure 8. Surface Fe loading and Arrhenius data at 300 mV overpotential ..................... 53 
Figure 9. Temperature-dependent Tafel plots for Fe:NiOxHy. .......................................... 54 
Figure 10. Tafel parameters versus temperature for Fe:NiOxHy. ...................................... 55 
Figure 11. Activation energy and pre-exponential factor as a 
function of overpotential ................................................................................................... 59 
Figure 12. Plot of Arrhenius activation energy and ln(A) ................................................. 63 
Figure 13. Extended cycling of LaNiO3 in Fe-free KOH ................................................. 69 
Figure 14. EIS of LNO at 1.7 V vs RHE .......................................................................... 74 
Figure 15. Activation of LaNiO3 and La2NiO4 by intentionally introduced Fe. ............... 76 
Figure 16. Chronoamperometry with intermittent Fe-spikes ............................................ 78 
Figure 17. Identical-location TEM of a cluster of 214-LNO particles ............................. 79 
Figure 18. AFM of LaNiO3 epitaxial film before and after cycling ................................. 81 
Figure 19. X-ray photoelectron spectroscopy of O - 1s region of LaNiO3 ....................... 82 
Figure 20. Cyclic voltammetry showing effect of incorporated Fe .................................. 96 
Figure 21. In-situ mass change of freshly electrodeposited Ni/CoOxHy. .......................... 96 
Figure 22. Potential vs time deposition curves ................................................................. 97 
Figure 23. Fe spiking control on Pt substrate ................................................................... 97 
Figure 24. Chronoamperometry of Pt and NiOxHy ........................................................... 98 
Figure 25. Co2+ spiking voltammetry of NiOxHy. ............................................................. 98 
Figure 26. Chronoamperometry of Ni2+ spiking of CoOOH ............................................ 99 
Figure 27 The effect of cyclic voltammetry on foreign-cation incorporation .................. 99 
Figure 28. Mass loading of NiOxHy and incorporation of Fe.......................................... 100 
Figure 29. Mole of electrons per Ni loading ................................................................... 100 
Figure 30. Chronoamperometry of Fe-spiked NiOOH at 300 mV ................................. 101 
Figure 31. Chronoamperometry of Fe-spiked NiOOH at 350 mV ................................. 102 
Figure 32. SEM images of the surface of an electrodeposited NiOxHy .......................... 103 
Figure 33. TEM/EDX of cross-sections of a NiOxHy film .............................................. 103 
Figure 34. TEM-EDX images of NiOxHy electrodeposited directly onto a 
gold TEM grid................................................................................................................. 104 
Figure 35. Cycling of surface Fe films ........................................................................... 105 
Figure 36. Representative chronoamperometry (CA) tests of CoOOH .......................... 106 
Figure 37. Representative chronoamperometry (CA) tests of CoOOH .......................... 107 
Figure 38. Turnover frequencies of NiOOH and CoOOH .............................................. 108 



12 
 

Figure 39. Correlation between TOFFe and Fe mass loading.......................................... 109 
Figure 40. Impedance spectroscopy for NiOOH double-layer capacitance ................... 109 
Figure 41. NiOOH and CoOOH double-layer capacitance ............................................ 110 
Figure 42: In-situ XAS cell ............................................................................................. 111 
Figure 43. Chronoamperometry in XAS cell .................................................................. 111 
Figure 44. XANES at Fe-K-edge .................................................................................... 112 
Figure 45. Comparison of Fe K-edge to reference spectra ............................................. 113 
Figure 46. Fits of XAFS signal in k-space ...................................................................... 114 
Figure 47. Fe-spiking of Ni0.8Fe0.2OxHy.......................................................................... 116 
Figure 48. De-activation of Fe:NiOxHy in Fe-free KOH ................................................ 116 
Figure 49. Tafel slopes of CA Fe-spiked NiOOH. ......................................................... 117 
Figure 50. Computational structural model of NiOOH .................................................. 119 
Figure 51. Angular views of adsorption model for Fe-O-Fe dimer clusters ................... 120 
Figure 52. Gibbs free energy calculations slab model .................................................... 124 
Figure 53. Dimer and monomer reaction pathways ........................................................ 125 
Figure 54. Optimization of ICP-MS Fe detection ........................................................... 126 
Figure 55. Example of raw temperature-dependent chronoamperometry data ............... 127 
Figure 56. J-V plots of typical variable temperature OER currents. .............................. 127 
Figure 57. Tafel plots and fits ......................................................................................... 128 
Figure 58. Arrhenius plots .............................................................................................. 129 
Figure 59. Arrhenius parameters from TOFFe................................................................. 130 
Figure 60. PXRD comparison of literature and synthesized LaNiO3. ............................ 132 
Figure 61. SEM images of particulate LaNiO3. .............................................................. 133 
Figure 62. EDX spectrum of LaNiO3 ............................................................................. 133 
Figure 63. La2NiO4 rods PXRD and SEM ...................................................................... 134 
Figure 64. TEM/EDX images of La2NiO4 rod bundle .................................................... 134 
Figure 65. Adsorption isotherms/BET surface areas of 214-LNO and LNO. ................ 135 
Figure 66. Cyclic voltammetry of LaNiO3 in Co(OH)2 - cleaned 1.0 M KOH .............. 135 
Figure 67. CVs Fe-spiked LaCoO3 ................................................................................. 136 
Figure 68. Pt-coil Ni residue test .................................................................................... 136 
Figure 69. The effect of bubble accumulation ................................................................ 137 
Figure 70. LaNiO3 Ni redox wave after CA ................................................................... 137 
Figure 71. Ni redox wave after electrolyte exposure ...................................................... 138 
Figure 72. LNO Nyquist plot .......................................................................................... 138 
Figure 73. Redox active Ni from EIS ............................................................................. 139 
Figure 74. Cyclic voltammogram Fe-spiked epitaxial LNO .......................................... 139 
Figure 75. Overlays of Fe-spiked CVs for all catalysts .................................................. 140 
Figure 76. Fe-spiked LNO CV after 8 cycles. ................................................................ 141 
Figure 77. Selected area diffraction patterns for 214-LNO ............................................ 142 
Figure 78. O 1s XPS epitaxial LNO after KOH exposure .............................................. 143 
 

 

 



13 
 

 

LIST OF TABLES 
 

Table 1. Parameters of the data extraction and fits. ........................................................ 113 
Table 2. EXAFS fit results .............................................................................................. 115 
Table 3.  ΔG and theoretical reaction overpotential (ηth) for OER mechanisms* ........... 117 
Table 4. Supporting numerical data. ............................................................................... 120 
Table 5. Additional activity metric summary of investigated samples. .......................... 122 
Table 6. Oxidation states for Fe-O monomer and Fe-O-Fe dimer. ................................. 122 
Table 7. Free energies and theoretical overpotentials ..................................................... 123 
Table 8. Summary of Tafel data ..................................................................................... 129 
Table 9. SAED d-spacings .............................................................................................. 142 
Table 10. Post-CVs ICP-MS LNO and 214-LNO .......................................................... 143 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 
 

 



14 
 

I. THE ROLE OF OXYGEN EVOLUTION IN GREEN 
ELECTROLYSIS TECHNOLOGIES 

 
 It is broadly and unequivocally recognized among the scientific community that the 

rising emission of CO2 from fossil fuel combustion1 has increased average global land and 

sea temperatures2 via the greenhouse effect, a mechanism rooted in over a century of 

physics and chemistry research3-16. The 6th Assessment Review of the Intergovernmental 

Panel on Climate change (IPCC), authored by 743 international climate experts of its 

member nations, reports that this warming was ~1.1 °C compared to the period 1850-1900 

based on decades of collecting climatic observations and modeling. Far from benign, this 

global warming has, to a high degree of confidence, decreased food and water security, 

increased chance of extreme weather events, accelerated sea level rise and ocean 

acidification, and damaged substantially (in some cases irreversibly) terrestrial, aquatic, 

marine, and cryospheric habitats. The magnitude of these processes and the risk of low-

probability, irreversible, catastrophic turning points increases with every future increment 

of warming. Only “deep, rapid, and sustained” and in many cases “immediate” reduction 

of greenhouse gas emissions is required to keep global temperature rise below international 

targets of 1.5°C and 2°C thereby mitigating the worst climatic effects. The IPCC estimates 

that humanity has a carbon budget of about 1000 gigatons of CO2 remaining for 2024 

onward before warming of 2°C is reached (67% probability). For scale, this is equivalent 

to the amount emitted between the 30 year period between 1990 and 201917. 

 To eliminate CO2 emissions, a rapid transition must be made toward a net-zero CO2 

energy portfolio based on renewable sources. Currently, only 14% of global energy is 

supplied by renewables (solar, wind, biofuels, or hydropower), the large majority of which 

is from hydropower18. Thus a large gap exists between clean energy goals and clean energy 

deployment. There are many possible paths to overcome that gap19-21, any of which will be 

determined by current and future regulatory and technoeconomic conditions. In 2021, The 

International Energy Agency (IEA) published Net Zero by 2050, the first global timeline 

of targets by which the world might achieve a net-zero CO2 emission (NZE) reality. In the 

IEA model, annual global emissions fall in concert with increased integration of 

renewables from 37GtCO2 in 2023 to 21 GtCO2 by 2030 and then 0 GtCO2 by 2050. The 



15 
 

total energy consumption in 2050 consists of only 20% fossil fuels, these having been 

largely replaced with renewables and nuclear. The key mechanisms by which the NZE is 

attained are swift deployment of renewable energy technology, broad end-use 

electrification, increased energy efficiency, use of hydrogen and hydrogen based fuels, 

behavioral change by citizens, and policy instruments like CO2 pricing and termination of 

fossil fuel subsidies all of which would need to occur globally through international 

cooperation20. Tesla has also weighed in on the matter, publishing an influential 2023 

analysis of what a transition to a sustainable economy might look like. In its Master Plan: 

Part 3, the Tesla team prescribed 6 actions that should be taken, among these were 

repowering the existing grid with renewables, switching to higher efficiency electric 

vehicles and heat pumps, and sustainable synthesis of aviation and shipping fuel. Taken 

together and assuming accurate forecasting of material costs and future electricity demand, 

this transition could occur with 30 TW installed renewable power capacity (~10x the 2023 

total capacity)22, 10% of 2022 World GDP, $10 trillion dollars manufacturing investment 

(~36% of 2024 U.S. GDP and $4 trillion less than the fossil fuel only scenario )23, and 

0.21% land area required21. Unlike that of the IEA, the Tesla decarbonization path does not 

break down this target into a series of intermediate targets. Proposed pathways for U.S-

specific decarbonization echo the content of these global plans24. 

While direct electrification of CO2 emitting end uses is a prominent strategy for 

decarbonization, this is not a feasible strategy for every area of the economy. Parts of the 

industrial and transportation sectors20 will require a fuel source with energy density akin to 

that of fossil fuels. These are typically classified as “hard-to-abate” industries and include 

– with the percentage of global CO2 emissions given in parentheses – long-haul shipping25 

(2%)26, aviation27, 28 (3%)29, steel refining (7%)30, ammonia-based fertilizer (1.3%)31, 

cement (7%)32, and glass33 production (~0.2%)34. Taken together, the emissions from these 

industries constitute ~20% of the global total, a significant portion.  

Most decarbonization plans recommend hydrogen as a chemical fuel suitable for 

abating emissions from these industries because hydrogen can be produced without CO2 as 

a byproduct via direct electrolysis of water using renewable electricity20, 35. This “green” 

hydrogen can directly substitute unabated hydrogen in traditional applications in industry 



16 
 

and refining such as the Haber-Bosch process in which H2 is used to reduce N2 to 

ammonia36 and direct iron reduction37. In addition to direct substitution of fossil fuel-

derived hydrogen, green hydrogen can do so indirectly. For example, it has been used to 

reduce CO2 via the Fischer-Tropsch process to make CO, a molecule which can be 

electrochemically transformed to higher molecular weight organic products suitable for use 

as jet fuel27. Hydrogen could also serve to smooth the problematically intermittent supply 

profile of renewable electricity by diverting excess solar and wind electricity to an 

electrolyzer plant. The stored hydrogen from electrolysis could then be tapped when 

renewable electricity is relatively scarce by feeding it to a fuel cell plant, generating needed 

electricity with high efficiency38. Despite the proposed benefits of using green hydrogen 

for energy storage and distribution, they have not been realized at scale and are expected 

to have a minor impact in relation to applications in traditional industry and refining35. 

Meeting the demands of traditional industry with green hydrogen is a formidable 

challenge. In 2022, 95 Mt of hydrogen were consumed globally the overwhelming majority 

of which was used for heavy industry (53 Mt) and refining (41 Mt). Less than 0.1% of this 

hydrogen was produced by water electrolysis. Replacing 2022 levels of dedicated hydrogen 

production with green hydrogen is estimated to require 730-940 GW of installed 

electrolyzer plant capacity. Given only 700 MW of installed electrolysis capacity in 2022, 

a 1,040 – 1340-fold increase is needed to meet global hydrogen demand. The IEA estimates 

that capacity of 175 GW could be accomplished by 2030 based on announced projects and 

430 GW if early stage, unannounced projects are taken into account35. There are several 

obstacles to rapid deployment of electrolyzer capacity at hundreds of GW scale. Two 

notable ones are uncertainty about regulation and certification, and dearth of hydrogen 

distribution infrastructure, but chief among these is a high levelized cost of hydrogen 

(LCOH), the price per kg of hydrogen taking into account lifetime costs of a plant. LCOH 

is $3.4 -12.0/kg H2 when produced by water electrolysis versus only $1-3 kg H2 from 

methods using unabated fossil fuels.35, 39 LCOH is influence by multiple factors, but cost 

of electricity is the main driver for hydrogen produced using electrolyzers40 with additional 

variability attributed to choice of technology: proton exchange membrane (PEM), alkaline 

exchange membrane (AEM), or liquid alkaline (LA) electrolysis41. Decreasing the 

contribution of electricity cost to LCOH can be approached from two directions: decrease 



17 
 

the cost of electricity for renewables and/or increase the efficiency of electrolyzer 

electricity use. Accomplishing the former will require thoughtful efforts by policy makers 

to stimulate the build-out of transmission and generation infrastructure. The latter 

challenge motivates this dissertation as it can be addressed from a basic science and 

engineering standpoint. 

This thesis details fundamental electrochemical and material science investigations 

that may help lower the LCOH for liquid alkaline electrolyzers by improving electrolyzer 

efficiency. As of 2022 alkaline electrolyzers made up 60% of installed electrolysis 

capacity35, and are therefore likely to significantly contribute to hydrogen production in the 

future. A simple model of a typical liquid alkaline electrolyzer is depicted in Figure 1. 

 

Figure 1. Generalized liquid alkaline electrolysis cell.  
An applied voltage oxidizes hydroxide to oxygen gas at the anode and hydrogen evolution from water 
reduction at the cathode in a cell flooded with alkaline electrolyte, usually KOH.  

 

It consists of two electrodes, cathode and anode, in an alkaline electrolyte separated by a 

diaphragm to prevent electrode shorting and mitigate gas crossover. The electrochemical 



18 
 

half-reactions are the cathodic hydrogen evolution reaction (HER) and anodic oxygen 

evolution reaction (OER) given in the figure. Note the sum of these half-reactions sum to 

the total water splitting reaction. The half-reactions are driven by an applied voltage 

leading to a net current in the external circuit that is related to the stoichiometry of the 

chemical reaction. Using the Gibb’s free energies of formation for reactants and products 

for the water splitting reaction one finds that the total thermodynamic voltage required to 

drive these half-reactions is 1.23 V.42 In practice, the total voltage needed to drive an 

appreciable current density is higher because of additional contributions to the 

thermodynamic minimum from kinetic, ohmic, and mass transport phenomena.  

Kinetics contribute a significant overpotential to water electrolysis in an 

electrolyzer. The anodic oxygen evolution reaction (OER) is a 4-electron reaction while 

that of the cathodic hydrogen evolution reaction (HER) is a 2-electron reaction. The greater 

number of electrochemical steps in OER leads to kinetic sluggishness relative to HER, 

limiting the overall rate of reaction and contributing hundreds of millivolts of overpotential 

to the thermodynamic minimum for water electrolysis. Decades of electrochemical 

research43, 44 have been performed to improve OER kinetics with appropriate 

electrocatalysts that reduce the energetic barriers of rate-limiting intermediate steps 

associated with the reaction. Even so, losses in efficiency from slow OER kinetics are 

sizeable across electrolysis applications. This has been particularly well-illustrated by 

measurements using integrated reference electrodes which quantify the (often large) 

contribution of OER to the total cell potentials of alkaline exchange membrane45, proton 

exchange membrane46, and CO2 reduction electrolyzers47. 

A wide variety of materials have been tested as alkaline OER electrocatalysts 

including precious metal oxides, transition metal alloys, perovskite oxides, spinels and 

hydroxides, metal chalcogenides and metal pnictides.48 Using thin-film model systems 

where every deposited metal site participates in charge transfer, it has been found that 

materials with the highest alkaline OER activity are based on nickel and cobalt hydroxides 

and their derivatives.
49 Developing a comprehensive, molecular-level understanding of 

why these materials are the best alkaline OER catalysts is ongoing and a source of debate 

in the literature. One influential idea is based on the Sabatier principle – that the rate of 



19 
 

OER is fastest on those materials which bind reactants just strongly enough to allow the 

reaction to proceed, but not so strongly that the products cannot desorb and poison the 

catalytic site50. Since the binding of reaction intermediates occurs by bonding between 

intermediate and surface atom orbitals, it follows that the rate of OER should have a 

“volcano”-shaped dependency on periodic trends in electronic structure descriptors that 

characterize chemical bonding i.e. the reaction rate would be expected to maximize at 

descriptor values in the intermediate range51. For example, Trasatti obtained a volcano 

shape plot of OER overpotential versus enthalpy of lower to higher oxide transition (e.g. 

Ru2O3  RuO2), a proxy for the heat of adsorption of oxygenated intermediates of the 

lower oxide during OER.52 Shao-Horn and coworkers reported a volcano shaped 

relationship between OER potential at 50 µA cm-2
oxide

 (area normalized to the oxide surface 

area obtained form gas sorption measurements) and the occupancy of the antibonding eg 

orbital of the B-site transition metal site in a series of perovskite oxides53. A more detailed 

exposition of descriptors in OER catalysis can be found in Chapter IV. 

  These studies share a key feature: extraction of an electronic descriptor from DFT 

calculations, literature data, or measurement using the pristine, as-synthesized material 

structure as a model. The volcano plot is then formed using this descriptor as the 

independent variable. This approach requires the tacit assumption that the pristine catalyst 

structure accurately represents the surface structure. There is strong evidence that this is a 

poor assumption. At least two key phenomena complicate the conventional conclusions 

about OER activity trends obtained using bulk electronic descriptors: 

1) Iron impurity incorporation. Catalysts with activity in the highest percentiles 

almost always have incorporated iron either from impurities in the electrolyte 

or intentional inclusion during synthesis54 

2) Surface reconstruction. Across OER literature and material classes, in-situ 

reconstruction of as-prepared materials leads to formation of new surface 

structures the OER mechanism and activity of which is responsible for the 

observed reaction rate55 

It is not trivial to understand how each of these processes affects the electrochemical 

characteristics of a given catalyst. Iron, for example, is a common impurity in 



20 
 

commercially available KOH56. If steps are not taken to remove this impurity57, the iron 

can adsorb to the catalyst under study and increase the OER activity58, unbeknownst to the 

well-meaning experimentalist. Adventitious nickel impurities also present problems to 

unambiguous assignment of catalytic activity57, 59. When reconstruction occurs it can lead 

to very thin layers, as small as a few unit cells thick, or involve species that are 

heterogeneously distributed, which appear and disappear reversibly60-65. Identification of 

this kind of surface species may require exotic operando experimental methods not 

routinely available to most electrochemists. Adsorption of impurities or unidentified 

reconstruction during a study in which electronic descriptors are the focus could lead to 

unwitting rationalization of OER activity with catalyst features that are not relevant to the 

actual mechanism of catalysis. Conversely, if direct connection between OER activity and 

bulk electronic descriptors is obscured by impurities and reconstruction then in-depth study 

of the catalytic activity of sites formed by those very impurities and reconstruction is 

necessary to develop a physical model of catalysis by the highest activity materials. 

This dissertation addresses the connection between high OER activity of nickel and 

cobalt oxides, surface reconstruction, and impurity Fe species. In chapter II, introduction 

of Fe into an electrochemical cell during polarization of the electrode at OER potentials is 

used to prepare surface Fe sites which are distinct from bulk Fe sites prepared by co-

electrodeposition. This permitted measurement for the first time of the turnover frequency 

of surface and bulk Fe sites. It is shown that surface Fe sites are undercoordinated and have 

a more elongated nearest-neighbor environment than their bulk counterparts. They also 

exhibit cooperativity - the intrinsic activity per Fe site increased with Fe site loading. 

Density-functional theory calculations suggest this cooperativity functions via 

delocalization of oxidative charge among multiple Fe centers. This work was published in 

Nature Communications by myself as co-first author with Yinqing Ou and Bipasa Samanta. 

Contributing authors to the published version of this work are: Lu Liu, Santu Biswas, 

Jessica L. Fehrs, Nicole A. Sagui, Javier Villalobos, Joaquín Morales-Santelices, Denis 

Antipin, and. Marcel Risch. Corresponding authors were Maytal Caspary Toroker and 

Shannon W. Boettcher. 



21 
 

The ability to control Fe-site conformation enabled surface-Fe-specific temperature-

dependent kinetic studies. The results were analyzed using Butler-Volmer, Arrhenius, and 

absolute rate theory to provide mechanistic insight. Contrary to expectations based on the 

Butler-Volmer theory approach to analysis of these data, the Tafel slope is constant at 

temperatures from -5 to 70°C indicating an unexpected temperature-dependence of the 

transfer coefficient. Further, a compensation between the enthalpy and entropy of 

activation at various overpotentials was revealed. It is proposed this behavior is consistent 

with concepts relevant to inner sphere rather than outer sphere electron transfer such as the 

change of intermediate coverage and solvent structure with potential. This work was 

completed by myself and is unpublished as of the writing of this dissertation. 

Chapter IV provides new evidence that Fe-site driven OER is not restricted to 

oxyhydroxides and is likely a universal feature of catalysis by perovskite oxide nickelates, 

contrary to most literature on the subject. Observations were made of surface 

reconstruction-mediated Fe site formation on perovskite oxide LaNiO3 and Ruddlesden-

Popper oxide La2NiO4 by combining electrochemical and materials characterization 

methods. The amount of surface reconstruction is compared between these two seemingly 

disparate compounds and normalization of their OER activity to the quantity of 

reconstructed nickel suggests that the intrinsic activity of the reconstructed phase is similar 

if not identical, suggesting a convergence toward a thermodynamic minimum energy 

structure. An experimentally accessible surface chemical descriptor is proposed as an 

alternative to bulk electronic descriptors of OER activity. These results were published in 

Journal of Catalysis by myself as sole first author along with contributions from Ally 

Tonsberg, Samji Samira, Kunal Velinkar, Kora Dumpert, Yingqing Ou, Le Wang, and 

Eranda Nikolla. Shannon Boettcher served as corresponding author on the published 

version of this work 

 

 

 

 



22 
 

 

 

 

 

 

 

II. COOPERATIVE FE SITES ON TRANSITION METAL 
(OXY)HYDROXIDES FOR HIGH OXYGEN EVOLUTION 

ACTIVITY 
 

Yingqing Ou1,2ǂ, Liam P. Twight1ǂ, Bipasa Samanta3ǂ, Lu Liu2,5, Santu Biswas,3 Jessica 
L. Fehrs1, Nicole A. Sagui1, Javier Villalobos6, Joaquín Morales-Santelices6, Denis 
Antipin6, Marcel Risch6, Maytal Caspary Toroker3,4*, and Shannon W. Boettcher1* 

ǂ These authors contributed equally to the manuscript 

This chapter contains co-authored work published in Nature Communications in 2023. Reproduced 

with permission from [Nat. Comm. 2023, 14, 7688]66 The study was conceived and designed by 

Y.O., S.W.B, J.L.F, and L.P.T. S.W.B. directed the research project. Y.O. collected most of the 

electrochemical and ICP-MS experimental data. L.P.T. performed the materials characterization of 

the catalysts using electron microscopy and X-ray absorption spectroscopy and additional 

electrochemical stability and activity tests.; B.S. and S.B. completed the computational studies, 

directed by M.C.T.; Y.O., L.P.T., B.S., M.C.T. and S.W.B. analyzed the data and wrote the 

manuscript. L.P.T., N.S., J.V., J.M.S., D.A., and M.R. designed the in-situ XAS study, collected, 

and analyzed the XAS data. 

 
Introduction 

The electrolysis of water (2H2O → 2H2 + O2) to produce hydrogen fuel is critical 

for renewable-energy infrastructure67, 68. Even in optimized electrolyzers, the efficiency is 

reduced by the slow kinetics of the oxygen evolution reaction (OER, 4OH- → 2H2O + 4e- 

+ O2 in alkaline media)69, 70. Among the many OER catalysts studied over decades, Fe is 



23 
 

broadly pivotal in promoting the catalytic activity in alkaline media71-73. In particular, Ni 

and Co (oxy)hydroxides require Fe to achieve high activity, regardless of whether it is 

introduced intentionally or incidentally74, 75. Sustaining the high activity of these Fe-

containing (oxy)hydroxides requires soluble Fe species in the electrolyte due to the 

dynamic Fe exchange at the catalyst/electrolyte interface76-79. High-activity perovskite 

oxides also involve active Fe sites80. La1-xSrxCoO3 perovskites form CoOxHy layers in 

alkaline electrolyte with enhanced activity in the presence of trace-level Fe electrolyte 

species81. Ba0.5Sr0.5Co0.8Fe0.2O3 undergoes surface reconstruction and transforms to OER-

active Co/Fe oxyhydroxides82. In fact, Fe-containing (oxy)hydroxides are ubiquitous for 

OER-active materials including transition-metal oxides, sulfides, selenides, and 

phosphides83-85.  

Typically Ni-Fe and Co-Fe hydroxides adopt structures analogous to α-phase 

Ni(OH)2 and Co(OH)2, which consist of layers of [M(OH)6] octahedra with rotational 

disorder and water/ions intercalated into the interlayer space86-89. The Fe3+ is thought to 

substitute for the M2+ sites with the extra charge balanced by intercalated anions90. Prior to 

OER, Fe-doped α-M(OH)2 is oxidized to (nominally) γ-MOOH, accompanied by a 

contraction of the M-O bond length and interlayer distance, consistent with stronger M-O 

bonds upon formal cation oxidation91. Fe substitution induces a positive shift of the 

apparent M2+/M3+ redox potential, indicative of electronic interactions between Fe and host 

Ni or Co cations. The oxidation state of Fe during OER at the active site remains a point 

of discussion. Both Fe3+ and Fe4+ were found under OER conditions for co-deposited Ni-

Fe or Co-Fe (oxy)hydroxides, while in non-aqueous electrolyte Fe6+ was identified and 

invoked as the active intermediate92-95. Operando Mössbauer spectra show the oxidation 

of Fe3+ in Ni-Fe (oxy)hydroxide to Fe4+ during OER and that these Fe4+ species largely 

persist after the potential was decreased into a non-OER region96. Such Fe4+ species were 

hypothesized to arise from fully coordinated internal sites within the NiOOH that were too 

kinetically slow to catalyze OER. In contrast, the detected Fe4+ population in a CoFeOx 

film from a separate Mössbauer study correlated with OER activity, suggesting a central 

role of Fe4+ in catalysis97. One hypothesis to explain this discrepancy would be the presence 

of different populations of Fe cations, for example in the interior regions of the 

(oxy)hydroxide sheets versus under-coordinated surface or edge Fe species. 



24 
 

We previously found evidence for different Fe local environments within the host 

(oxy)hydroxides affecting OER activity77, 98. When cycled, both NiOxHy and CoOxHy 

adsorb Fe3+
 (intentionally added to the electrolyte) onto sites that were hypothesized to be 

easily accessible, i.e. at edges, corners, or defects in the two-dimensional (oxy)hydroxide 

structure. The Fe incorporated during the first voltammetry cycle dramatically enhanced 

the OER activity but had negligible influence on the host NiOxHy redox-peak position or 

size, indicating weak electronic interaction between these “surface” Fe and the majority of 

the Ni cations. Repeated voltammetric cycling increased the amount of Fe incorporated, up 

to a Fe:Ni ratio of ~ 0.25, and the Ni redox wave shifted positive indicating strong coupling 

between the additionally added Fe and the Ni cations. Yet this additional Fe caused only a 

small increase in OER activity. These data led to the hypothesis that at least two general 

types of Fe sites exist in the (oxy)hydroxides, i) OER-active surface sites from electrolyte 

Fe adsorption and ii) interior sites where Fe sits fully coordinated with bridging O(H) to 

neighboring M sites. Supporting this idea, NiFeOxHy with proposed surface-attached 

FeOOH nanoclusters has been reported to be more OER-active than benchmark Ni-Fe 

catalysts99, 100. The above findings illustrate that structural information or activity 

measurements for Fe-based sites collected from co-deposited (oxy)hydroxides are the 

weighted average of the multiple Fe environments which have very different properties. 

The precise intrinsic activity on the most-active “surface” Fe, and the associated key OER 

mechanism, remains unknown. 

Here we show how to confine the absorption of Fe to nominally surface sites where 

they have exceptionally high per-Fe activity yielding the catalysts Fe:NiOxHy and 

Fe:CoOxHy. We study how the intrinsic structural features and dynamics of NiOxHy and 

CoOxHy affect the incorporation. Fe added into the electrolyte during chronoamperometry 

under OER conditions (positive of the Ni2/3+ redox wave) yields electrolyte-adsorbed Fe 

species primarily on the surface (presumably at edge/defect sites) of Ni and Co 

(oxy)hydroxides. In contrast, potential cycling pure NiOxHy in the presence of Fe3+(aq.) or 

Co2+(aq.), yields more-homogeneous Ni(Fe)OxHy or Ni(Co)OxHy phases, respectively. 

Similarly, we find that it is more difficult to incorporate Fe3+ into the interior sites of 

CoOxHy compared to NiOxHy, consistent with the larger Co2+-O(H) bond strength 

compared to Ni2+-O(H)101 and higher sheet-morphology stability of CoOxHy
102

. With this 



25 
 

platform to control the location of the Fe-based active sites, we measure the intrinsic OER 

activity of surface Fe sites on Fe:NiOxHy and Fe:CoOxHy and compare these to activities 

with Fe at interior sites in NiOxHy. The turn-over frequency for OER, normalized to the 

number of Fe sites (TOFFe), increased linearly with the amount of absorbed surface added 

until a saturation limit, suggesting possible cooperativity of active sites, in sharp contrast 

to previous studies where the location of the Fe sites was uncontrolled79. We find an 

intrinsic TOFFe of ~40 s-1 at 350 mV for Fe:NiOxHy which is at least five times higher than 

for benchmark co-deposited phases73. This finding illustrates a cooperative effect between 

multiple Fe sites on the surface of NiOxHy and CoOxHy. Density-functional-theory  

(DFT) calculations suggest that this effect is derived from the ability of neighboring Fe 

atoms in a Fe-O-Fe cluster to share and stabilize positive charge during the oxidation of 

key intermediates, compared to a single FeOx supported on NiOxHy. These findings are 

important to understand and control structure in the design of higher-performance OER 

catalysts from earth-abundant metals for use in advanced electrolyzers69, 103 and 

photoelectrochemical systems. 

Methods 
 

Solution Preparation. Stock solutions of 0.1 M Ni(NO3)2·6H2O (Sigma Aldrich, 99.999% 

trace metal basis) and 0.1 M Co(NO3)2·6H2O (BeanTown Chemical, ≥ 99.999% trace metal 

basis) were separately prepared in 18.2 MΩ·cm water for the electrodeposition of NiOxHy 

and CoOxHy films. For the co-deposition of Ni(Fe)OxHy and Co(Fe)OxHy, FeCl2·4H2O 

(ACROS Organics, 99+%) was freshly dissolved in N2-purged 18.2 MΩ·cm water to 0.1 

M. Then 0.1 M Ni(NO3)2·6H2O or Co(NO3)2·6H2O was purged with N2 for ~20 min prior 

to the addition of 0.1 M aq. FeCl2·4H2O. Stock solutions of semiconductor grade 1.0 M 

KOH (Sigma Aldrich, 99.99% trace metal basis) and further-purified “Fe-free” 1.0 M KOH 

were used for electrochemical measurements. The electrolyte purification was conducted 

as reported previously74. Briefly, about 2 g of Ni(NO3) ·6H2O was dissolved in ~4 mL of 

18.2 MΩ·cm water in a 50 mL centrifuge tube and high purity Ni(OH)2 precipitated by 

rapid addition of ~20 mL of 1 M semiconductor grade KOH. The green Ni(OH)2 precipitate 

was washed by 2 rounds of addition of a 20 mL of water/ 2 mL of 1 M KOH solution. After 

the final wash, the supernatant was removed by centrifugation and the tube was filled with 



26 
 

semiconductor grade 1 M KOH. The tube was shaken vigorously to redisperse the  

precipitate and allowed to sit overnight during which time the Ni(OH)2 absorbs Fe 

impurities. All electrolyte used was pH 13.88 ± 0.06. The Fe-free KOH was recovered by 

centrifuging and then filtered using a 0.1 μm polyethylene sulfone (PES) syringe filter to 

remove residual Ni(OH)2 particulates. Stock solutions of 0.1 mM Fe(NO3)3·9H2O (Sigma 

Aldrich, ≥ 99.999% trace metal basis), 0.1 mM Ni(NO3)2·6H2O (Sigma Aldrich, 99.999% 

trace metal basis), and 0.1 mM Co(NO3)2·6H2O (BeanTown Chemical, ≥ 99.999% trace-

metal basis) were separately prepared for foreign-ion “spiking” experiments. To prevent 

the precipitation of Fe3+, the pH of the 0.1 mM aq. Fe(NO3)3·9H2O solution was adjusted 

to ~2 with HNO3.  

Film preparation. All electrodepositions were performed with a two-electrode 

configuration using pre-cleaned carbon cloth as the counter electrode. Prior to deposition, 

a bare substrate was cycled five times in Fe-free 1.0 M KOH to verify its cleanliness and 

induce hydrophilicity. Ni(Fe)OxHy was cathodically electrodeposited at -0.1 mA·cm-2 for 

120 s. Co(Fe)OxHy was electrodeposited at -2 mA·cm-2 for 8 s. The total metal-ion 

concentration in the electrodeposition bath was 0.1 M. For co-electrodeposited films, the 

Fe/Ni and Fe/Co ratios were adjusted by the Fe content in the deposition solution. Pt/Ti 

(50/10 nm) on glass slides, or Pt/Ti-coated quartz crystal microbalance (QCM) crystals (5 

MHz, Stanford Research Systems QCM 200), were used as the substrates for 

electrodeposition. The deposited composition and molar amount were determined by 

elemental analysis (see below). The mass loading of Ni in the NiOxHy film was ~ 2.7 μg·cm-

2. The mass of Co in the film of CoOxHy was ~5.0 μg·cm-2. The film mass loading and 

change during electrochemistry were determined by the frequency change of the QCM 

electrode using the Sauebrey equation (Δf = -Cf · Δm, where Δf is the measured frequency 

change of quartz crystal, Cf is the sensitivity factor with a value of 56.6 Hz cm2∙μg-1, and 

Δm is the mass change per unit area, μg∙cm-2).   

Electrochemical Characterization. Electrochemical measurements were made with a potentiostat 

(Bio-Logic SP300 or SP200) using a typical three-electrode setup. Different Hg/HgO reference 

electrodes (CH Instruments) were used for Fe-free and Fe-containing measurements. The standard 

electrode potential of the 1 M KOH Hg/HgO reference (E0 
Hg/HgO) was calibrated to be 0.094-0.099 V 

vs. NHE104. All recorded potentials were converted to the reversible hydrogen electrode (RHE) 



27 
 

scale by ERHE = EHg/HgO + E0 
Hg/HgO + 0.059∙pH, where EHg/HgO is the recorded electrode 

potential vs. Hg/HgO. A Pt coil was used as the counter electrode and pre-cleaned by aqua 

regia and 18.2 MΩ·cm water. Because alkaline media etches glass leaching Fe into the 

electrolyte, all measurements were made using inexpensive plastic (polymethylpentene) 

cells. The glass section of the substrate was covered by epoxy (Loctite, EA 9460) and hot 

glue (a commonly available hot-melt polymer adhesive). The overpotential (η) was 

calculated by the equation: η = ERHE  - 1.23 V - iRu, where Ru is uncompensated series 

resistance. Ru was determined by equating Ru to the minimum impedance between 10 kHz 

and 1 MHz, where the phase angle was closest to zero104. The double-layer capacitance 

(CDL) of the metal-oxyhydroxide films was determined by fitting the results of potentio-

electrochemical impedance spectroscopy (PEIS) measurements in a potential region where 

the electrocatalyst films are conductive105 over the frequency range of 0.1 Hz to 1 MHz.   

Experimental details for the XAS measurements are provided in the Supplementary 

Information. 

Metal-ion-spiking. In CA tests, catalyst films were first polarized in purified Fe-free 

semiconductor-grade 1.0 M KOH electrolyte for 3-5 min. Then 715 μL of 0.1 mM metal 

ion aq. solution was added dropwise into the 40 mL of KOH electrolyte to reach a 

concentration of added metal of 0.1 ppm (wt. Fe / wt. H2O). The low concentration of 0.1 

ppm was chosen to mitigate effects of insoluble Fe(OH)3 colloids or particles that form at 

higher Fe concentrations and obfuscate effects of the absorbed Fe species. We estimate 

from solubility constants106 that the solubility of Fe3+ is about 35 ppb in 1 M KOH and 

have historically used 0.1 ppm spiking amounts with no observation of precipitation. In 

voltammetry studies, the catalyst films were first measured in purified Fe-free or 

semiconductor grade 1.0 M KOH electrolyte for 2-4 cycles to get a stable electrochemical 

response. Then 715 μL of 0.1 mM metal-ion solution was added dropwise into the 40 mL 

KOH electrolyte. Magnetic stirring was used throughout the measurement to promote 

transport of the added metal ions and prevent bubble accumulation. After measurements, 

the electrode was removed and rinsed in 18.2 MΩ·cm water.  

Loading analysis and turnover frequency (TOF) calculation. Film mass loading of each 

element were determined by ICP-MS (iCAP-RQ Qnova Series, Thermo Fisher Scientific). 



28 
 

Calibration curves were prepared from third-party-certified reference solutions of the 

analyte of interest. Electrodes were immersed in 2 mL of 10 v/v% HNO3 for at least 24 h 

to dissolve the catalyst films. The HNO3 solution was then diluted with 2 mL 18.2 MΩ·cm 

water for ICP-MS analysis. To verify the full dissolution of catalysts into HNO3 solution, 

the substrates were rinsed by 18.2 MΩ·cm water and cycled in Fe-free KOH. The lack of 

Ni or Co redox peaks and similar OER activity to bare Pt/Ti substrate indicated the full 

dissolution of catalyst films104. To compare the intrinsic activity of Fe sites in different 

matrices, the TOFFe based on the total mass of Fe in the film was calculated at a constant 

overpotential104: 

TOF𝐹𝐹𝐹𝐹  =
current / 4F

(mol of Fe sites)
          (1)  

 

where F is faraday’s constant. The number of Fe sites was determined by ICP-MS. The 

OER current was recorded from real-time iRu compensated chronoamperometry tests. 

DFT calculations. Vienna ab initio Simulation Package (VASP)107-109 was used to perform 

the spin-polarized density functional theory (DFT) calculations with the DFT+U formalism 

of Duradev et. al107, 108, 110. For the effective modelling of DFT+U, U-J terms of 5.5 and 4 

for Ni111-114 and Fe115, 116 were used, respectively. All calculations were performed using 

the Perdew−Burke−Ernzerhof (PBE)117 exchange-correlation functional of the generalized 

gradient approximation (GGA). The projected augmented wave (PAW) potentials109, 118 

include the contribution of core electrons of each atom. An energy cut-off of 600 eV with 

k-point mesh of 1×1×1 was used for the entire calculation in accordance with the values 

reported in previous work (65,66). The structures were minimized with energy and force 

convergence criteria of 10-4 eV and -0.03 eV∙Å−1, respectively.  Gaussian smearing119 was 

used with symmetry imposition for all calculations. The geometries were relaxed with a 

conjugate gradient algorithm.120 

The overpotential (𝜂𝜂 ) of each reaction pathway is defined as the minimum potential 

required to make all reaction steps exothermic. Based on the calculations of the Gibbs free 

energy of each reaction step, the theoretical overpotential was calculated as follows121: 



29 
 

𝜂𝜂 = max (∆𝐺𝐺1,∆𝐺𝐺2,∆𝐺𝐺3,∆𝐺𝐺4,∆𝐺𝐺5) − 𝑉𝑉𝑂𝑂𝑂𝑂𝑂𝑂          (2) 

 

where ∆𝐺𝐺𝑖𝑖 is reaction energies of each step and  𝑉𝑉𝑂𝑂𝑂𝑂𝑂𝑂 is the equilibrium potential of water 

oxidation and its reported value is 1.23 eV. The potential energies obtained from the 

density functional theory calculations are converted to Gibbs free energies as detailed in 

the Supplementary Information. The overpotential changes were analyzed in terms of 

changes in the atomic oxidation states of Fe and Ni of each of the reaction intermediate. 

The oxidation state of Ni and Fe was inferred from the atomic magnetization observed in 

the output file. For Ni +2, +3 and +4 oxidation states are related to two, one and no singly 

occupied orbitals, respectively. For Fe +2, +3, +4 and +5 oxidation states are related to 

four, five, four and three singly occupied orbitals, respectively. The reduction in 

magnetization values observed results in increase of formal oxidation state. The Gibbs free 

energy reported here includes the zero-point energy correction.  

 

Results and Discussion 
 

Mechanisms of foreign-cation incorporation into NiOxHy and CoOxHy. We define two 

classes of sites where foreign ions can incorporate that contribute differently to the OER 

activity and electrochemical response. First, there are interior “bulk” sites where the cation 

is substituting either Ni or Co cations. These modify the electronic energies of the redox-

active host metal atoms (Ni or Co) and thus influence the peak position of the Ni or Co 

redox waves during voltammetry8, 122, 123. Second, surface sites where the cations are 

adsorbed onto, rather than substituted into, the host NiOxHy and CoOxHy porous structures 

(Figure 20) where the coordination by water and terminal hydroxyls makes them likely 

OER active sites77, 98. These seem to have little effect on redo- wave position, but generate 

most of the catalytic enhancement. To study these different sites, hydrated, electrolyte-

permeable NiOxHy and CoOxHy films were cathodically electrodeposited on Pt/Ti/glass 

substrates. Pt was used because it is OER inactive and has limited electronic interaction 

with any in-situ formed FeOxHy when Fe species are present in the electrolyte124. The 

porosity and thinness of the films ensured that almost all metal cations in the film are 



30 
 

electrochemically active as confirmed by electrochemical microbalance studies73, 104. SEM 

images of a representative NiOxHy film (Figure 32) illustrate the porosity and roughness of 

the film and cross sections (Figure 33) suggest a thickness between 5-20 nm. 

We use the redox response of Ni and Co cations to gain insight into the 

incorporation of foreign metal species from the electrolyte. Directly tracking the different 

Fe sites using electrochemistry is not possible as they do not provide a useful redox 

signature125; Fe3+ oxidation occurs at potentials within the OER regime and Fe3+ reduction 

occurs at potentials where the NiOxHy and CoOxHy host are electrically insulating. We 

expect that when foreign cations adsorb on the host they should provide a distinct wave 

separate from that of the host (if they are redox active), while when incorporated 

substitutionally in interior sites they will exert an electronic affect manifesting as a shift of 

the host redox-wave potential. 

 

Figure 2. Foreign cations interaction with oxyhydroxides via redox signatures. 

Cyclic voltammetry of Co-spiked NiOxHy (a,b) and of Ni-spiked CoOxHy (c,d) showing the evolution of 

redox features of the host metal hydroxide and phase formed by spiked ions. Inset numbers in (b) and (d) 

correspond to the cycle number, while the bottom cartoon illustrates schematically the process intentionally 

without atomic detail that is yet unknown. The data above was not iRu-compensated. 

 



31 
 

When 0.1 ppm Co2+ is added into the electrolyte during constant-potential 

polarization of NiOxHy we observed a new redox feature centered at 1.13 V vs. RHE 

(Figure 2a), ~50 mV positive of the same wave for as-deposited CoOxHy on Pt, which is 

thus assigned to Co(OH)2/CoOOH. During the CA, the OER current increased by a factor 

of five after Co2+ was added (Figure 24a), consisting with CoOxHy, which is more OER 

active122, absorbing on the NiOxHy.  

With subsequent voltammetry cycles (Figure 2b), the CoOxHy wave shifts to higher 

potential, and the host NiOxHy wave broadened (Figure 24). After 10 cycles the 

Ni(OH)2/NiOOH redox peaks also shifted negative by 11 mV (Figure 27a). By 25 cycles, 

the characteristic CoOxHy redox peaks have moved positive in potential and coalesced 

(Figure 2b) with that of NiOxHy. The loss of the independent CoOxHy redox peaks and the 

negative shift of the NiOxHy is consistent with the Co cations which were initially adsorbed 

in a phase separate from the host NiOxHy to disperse into the NiOxHy. The negative shift of 

the Ni redox wave with Co2+ incorporation is similar to that observed for co-deposited 

Ni(Co)OxHy films and indicates strong electronic coupling between dopant Co and host Ni 

sites126, 127. By analogy, similar processes are proposed when NiOxHy is cycled in the 

presence of solution Fe cations that have no apparent redox signature. 

In contrast to NiOxHy, cycling CoOxHy in Ni2+-spiked KOH yields a wave centered 

at ~1.32 V vs. RHE (Figure 2c), typical for the Ni(OH)2/NiOOH couple. The OER current 

during CA for CoOxHy is unaffected by spiking Ni2+ into the electrolyte (Figure 26a), 

consistent with NiOxHy being less OER-active than CoOxHy
122. The new Ni(OH)2/NiOOH 

wave that appears persists after cycling with only a slight negative shift of the peak 

potential (Figure 2d), while the Co(OH)2/CoOOH wave position is unaffected (Figure 27b). 

The lack of electronic interaction between foreign Ni cations on the CoOxHy host redox is 

consistent with Ni cations that persistently reside on the surface of CoOxHy as a separate 

NiOxHy phase despite cycling. The inability of CoOxHy to easily incorporate foreign 

cations, relative to NiOxHy, is consistent with its greater structural stability128 and its 

stronger Co-O(H) bonds relative to Ni-O(H)129, 130. Because the procedures used in these 

Co- and Ni-spiking experiments are identical to those used for Fe spiking in activity 

measurements below, we propose that Fe3+(aq.) adsorbs in a phase distinct from CoOxHy 



32 
 

and remains as such after cycling, but that Fe3+ incorporates into the internal sites when the 

host NiOxHy is electrochemically cycled. Evidence for separate phase formation by 

adsorbed electrolyte ions using this method has not before been previously demonstrated 

and is critical to understanding the origin of exceptional TOFFe described below. 

The different locations of the foreign cations, in or on, NiOxHy and CoOxHy films 

appear driven by structural changes during voltammetry. Using an electrochemical quartz 

crystal microbalance (EQCM), the mass change of NiOxHy and CoOxHy films upon cycling 

was monitored during voltammetry (Figure 21). During the positive oxidative scan of 

NiOxHy, the film mass increases at the potential where Ni(OH)2 oxidation is measured and 

continues into the OER region. The mass increase in the forward scan is ~ 0.33 μg ·cm-2, 

accounting for ~4.5% of total film mass (4.2 ± 0.2 g per mole e- passed). Mass loss is 

observed in the backward scan, particularly with the reduction of Ni species (~0.33 V vs. 

Hg/HgO). At the end of the cycle, the film mass returns to its initial value. For CoOxHy, a 

slight and irreversible mass gain (~0.2% of total film mass) is observed after the second 

voltammetry cycle, but no detectable mass change accompanies the oxidation of Co(OH)2.  

The mass gain with oxidation of Ni2+ is thought to originate from K+ and OH- 

intercalation along with release of H2O in the interlayer space of NiOxHy
131-133. Oxidation 

also causes contraction of the Ni-O bond and interlayer spacing, which introduces 

mechanical stress91, 133. Restructuring of Ni(OH)2 single-layer nanosheets into small 

nanoparticles was found by electrochemical atomic force microscopy (EC-AFM) 

suggesting that Ni(OH)2 undergoes a dissolution-redeposition process during cycling128, 

134. When Fe or other cations exist in the electrolyte, these cations can exchange for Ni 

leading to atomic-level mixing with the host Ni. In contrast, the voltammetry of CoOxHy 

does not involve obvious molecule/ion exchange based on EQCM and EC-AFM 

measurements and shows structural stability compared to NiOxHy
128, apparently making it 

difficult to incorporate foreign cations from the electrolyte into the structure interior.  

Confining Fe cations on the surface of NiOxHy and CoOxHy. Using the insights 

from the Ni- and Co-spiking experiments, we add Fe3+ during chronoamperometry to 

prepare Fe:NiOxHy and Fe:CoOxHy OER catalyst where Fe is surface-adsorbed. First, we 

applied ~1.55 V vs. RHE to NiOxHy (nominally NiOOH at this potential) in Fe-free 1.0 M 



33 
 

KOH for 3-5 min and recorded the (low) baseline OER activity. Then Fe3+ was added to a 

reach a concentration of 0.1 ppm (this low amount of Fe was selected to prevent bulk 

electrodeposition of nominally FeOOH we observed previously by electrochemical AFM 

at > 1 ppm51). A dramatic increase in OER current was immediately observed reaching a 

maximum after ~15 min, suggesting the fast adsorption of Fe species on NiOOH, likely 

limited by mass transport of Fe species to the electrode. The incorporation of foreign 

cations at fixed potentials in the OER regime, and positive of the nominal Ni2+/3+ redox 

wave, is important to limit the amount of host restructuring and associated intermixing that 

is driven by the redox transitions128, 134.  

After ~15 min the OER current reaches a stable maximum ~90× that of Fe-free 

NiOOH (without iR compensation) and ICP-MS analysis of the catalyst film shows ~5 at. 

% Fe (relative to Ni). The same measurements were performed with CoOxHy and the OER 

activity in Fe-free KOH was higher than NiOxHy, consistent with our previous reports122. 

Adding 0.1 ppm Fe3+ in the electrolyte resulted in a ~7-fold increase in OER current (Figure 

3c), a much smaller enhancement than for Fe:NiOxHy. The observed activity difference 

between Fe:NiOxHy and Fe:CoOxHy likely derives from different intrinsic activities of 

surface Fe on the two chemically different hosts (see below).  



34 
 

 

Figure 3. Surface-confined Fe sites via chronoamperometric (CA) metal-ion-spiking.  
CA measurements of (a) NiOOH  and (c) CoOOH at 1.55 V vs. RHE. After starting the measurement in 
purified Fe-free 1.0 M KOH electrolyte, aqueous Fe(NO3)3 was added to a concentration of 0.1 ppm. The 
first voltammetry cycle (red, 10 mV/s) after Fe-spiking CA measurements shows the dramatic effect of Fe 
incorporation on the OER activities of NiOxHy (b) and CoOxHy (d), but a minimal effect on the redox wave 
compared to the light green lines that show the initial voltammetry (cycle two) recorded in purified Fe-free 
1.0 M KOH. The grey curves illustrate the large effect on the redox wave position for NiOxHy but not for 
CoOxHy after cycling, but that the OER activity doesn’t further change much. The data is not iRu 
compensated. The cartoons illustrate schematically the process intentionally without atomic detail that is yet 
unknown. 

 

The first voltammetry cycle of Fe:NiOxHy after chronoamperometric (CA) Fe 

incorporation shows nearly identical redox response as Fe-free NiOxHy (Figure 3b), despite 

having dramatically enhanced OER activity. This implies that the Fe incorporated during 

CA is not yet electronically interacting with the majority of the host Ni metals. The 

Fe:NiOxHy was then cycled 20 times in the Fe-spiked KOH (Figure 3b, blue). The Ni wave 

shifts positively by 34 mV and the integrated peak area shrinks as the Fe incorporates into 

the oxyhydroxide structure to form Ni(Fe)OxHy (that also contains surface-absorbed Fe) 

with no change in OER activity. The positive shift of the NiOxHy wave is known to occur 



35 
 

upon mixing with Fe by co-deposition74. This data implies an intrinsic activity difference 

for surface and internal Fe-based sites. Similarly, the first CV cycle of Fe:CoOxHy after 

CA-spiking test shows almost the same redox peak position and area compared to CoOxHy, 

despite the enhanced OER activity (Figure 3d). Subsequent voltammetry on Fe:CoOxHy 

only slightly decreases the OER overpotential, with the redox waves almost unchanged. 

Bare Pt and Au substrates also absorb Fe increasing OER activity124, but the Fe-spiked 

electrolyte barely affected the activity of control Pt electrodes under CA conditions here 

(Figure 23). Thus the enhanced activity observed for NiOxHy or CoOxHy loaded on Pt is 

derived from Ni(Fe)OxHy or Co(Fe)OxHy, with the Fe-based active sites proposed to be all 

absorbed on the “surface” of the host oxyhydroxide. 

 
Figure 4. Intrinsic OER activity measured by turnover frequency (TOFFe) at η = 300 mV.  
The TOFFe is calculated based on the mass of all Fe sites determined by ICP-MS of each dissolved film. (a) 
Correlation between TOFFe and Fe/Ni atomic ratio for surface-confined Fe generated by CA, as well as mixed 
systems from cycling or co-deposition. (b) Correlation between TOFFe and Fe/Co atomic ratio. (c) Correlation 
between the TOFFe of surface-Fe sites on NiOOH (red) and CoOOH (grey) and the adsorbed Fe mass loading 
normalized by the electrochemical surface area of host oxyhydroxide. (d) Tafel slopes of Fe:NiOxHy (red) 
and Fe:CoOxHy (grey) as a function of Fe concentration (10, 40, 70, 100, and 200 ppb) in 1.0 M KOH 
electrolyte. Tafel analysis was performed using constant current steps from 0.18 to 3.2 mA∙cm-2, with each 
step held for 3 min. The steps were then repeated in reverse order. Before Tafel analysis, constant potential 
OER in Fe-spiked electrolyte was performed until the maximum OER current was reached. All current values 



36 
 

used for TOFFe concentrations were iRu-compensated where Ru was in the range 15.1 ± 0.5 Ohms. Error bars 
represent one standard deviation from the average of triplicate measurements 

Intrinsic activity of Fe sites on metal (oxy)hydroxides. The OER turnover 

frequency (TOF) is defined as the total number of O2 molecules generated per active site, 

per unit time104. One important fundamental challenge is the identification and 

quantification of the true active sites to enable catalyst design. TOF for OER for these 

systems can be calculated in several ways104. The simplest is to use the total number of 

cations in the film (regardless of their location and chemistry, including Ni, Fe, Co, etc.) 

to calculate TOFtm which thus provides the average activity at all metal sites, simplifying 

the reality of many sites with a range of activities.  

We calculate a TOF based on the total number of Fe cations (TOFFe), including 

both the interior and surface sites discussed here, as Fe is essential at the active site135-137. 

Our results show that Fe incorporates at surface sites at constant OER potentials and mixes 

into the interior only after cycling. The number of surface-adsorbed Fe sites can be 

controlled by stopping the CA experiment at different times while Fe is accumulating on 

the NiOxHy or CoOxHy, thus providing the first known route to create and study specific 

types of Fe sites in these important materials. 

Figure 4a shows the TOFFe values at η = 300 mV (iRu corrected) for Fe:NiOOH and 

Ni(Fe)OOH. For Fe:NiOOH, TOFFe  increases nearly linearly with the amount of Fe 

absorbed. Initially the sample is ~0.7 at.% Fe on NiOxHy and the TOFFe  is 2.0 ± 0.3 s-1. 

The TOFFe increases until maximum Fe adsorption and OER current at 5.1 at.% Fe on 

NiOxHy and the incredibly high TOFFe of 10.4 ± 1.4 s-1 at η = 300 mV. This data shows that 

each Fe-based-site becomes more OER active as surface-absorbed Fe sites accumulate. 

After reaching a maximum OER current at constant potential in Fe-spiked KOH, 

the resulting Fe:NiOxHy was cycled (Figure 28) resulting in the absorption of additional 

Fe, including at internal sites, and thus a large decrease in TOFFe as Fe increases to ~ 20 

at. % (Figure 4a). These data are consistent with the hypothesis that the absorbed Fe at 

surface sites drives OER, while internal Fe sites are comparatively inactive. We also 

prepared mixed Ni(Fe)OxHy via co-electrodeposition, for which Fe is homogeneously 

substituted for Ni. TOFFe values for these co-deposited samples decrease with Fe content 



37 
 

and are substantially smaller than those obtained from Fe spiking, consistent with an 

activity difference between surface and internal Fe sites. TOFFe at η = 350 mV shows 

similar trends (Figure 38). The highest TOFFe of ca. 40 ± 2 s-1 at η = 350 mV is obtained at 

the maximum surface adsorption. Previous studies show that the co-deposited optimal 

Ni0.75Fe0.25OxHy has a TOFFe of ca. 9 s-1 at η = 350 mV122. On a TOFFe basis, the Fe:NiOxHy 

reported here has the highest activity among all alkaline OER catalysts138.  

The surface Fe sites also become more OER active with increasing number on the 

surface of CoOxHy (Figure 4b). At maximum adsorption, ~2.4% Fe is adsorbed on CoOxHy 

and the highest TOFFe at η = 300 mV is 0.71 ± 0.08 s-1. This value is smaller than the TOFFe 

in Fe:NiOxHy with similar absorbed surface Fe amount (~5 s-1 at η = 300 mV with ~2.5% 

Fe) showing a large intrinsic activity difference between Fe sites on CoOxHy versus 

NiOxHy. The TOFFe of co-electrodeposited Co(Fe)OxHy is, like the Ni system, also smaller 

than that of the surface Fe sites on CoOxHy and decreases with Fe content (due to a larger 

fraction of less-active internal Fe sites). Because NiOxHy and CoOxHy films have different 

mass loading and electrochemically active surface area (ECSA), the TOFFe values are 

shown as a function mass of adsorbed Fe (from ICPMS) normalized to geometric surface 

area and ECSA (Figure 38 c-d). 

In contrast to our results, Chung and Markovic et al. observed a linear relationship 

between OER activity and the amount of absorbed Fe on several transition-metal 

(oxy)hydroxides including NiOxHy, CoOxHy, Ni(Cu)OxHy, and Ni(Mn)OxHy by 

incorporation of Fe from electrolyte. This data was used to argue that each Fe site has 

similar activity and thus the improvement of OER catalysis mainly relies on increasing the 

number of Fe sites by increasing the absorption energy of the Fe on the host79. The 

difference between our work and this study is that the nature of the Fe sites was not 

previously controlled. The higher Fe at. % of > 18% from their experiments indicate that 

Fe was not restricted to edge and defect sites and thus the number of active surface sites, 

versus fully coordinated interior ones, was unknown and uncontrolled and so the intrinsic 

activity of the two could not be separated nor cooperative effects between Fe sites 

discovered.  



38 
 

The above TOFFe calculations demonstrate that the intrinsic OER activity of Fe sites 

is dependent on the local configuration: i) surface Fe sites have much higher OER activity 

than bulk sites, ii) the intrinsic activity of surface Fe sites is affected by the interface with 

the host material, as exemplified by the difference in TOFFe  on NiOxHy and CoOxHy at a 

fixed Fe loading, and iii) the surface Fe sites show increasing  activity on a per-site basis 

with the extent of surface Fe accumulation.  

The emergence of surface-absorbed-Fe-site cooperativity can be explained simply. 

As more Fe is adsorbed, for example, the likelihood of two or more Fe sites being located 

adjacent to each other increases. This may create overall a catalytic site with favorable 

electronics for adsorbate formation and evolution (i.e. to be neither too strong nor too weak 

via the Sabatier Principle). There is precedent in the literature for clusters of different sizes 

to display different catalytic activity. For example, smaller clusters of CoOx were found to 

be more active for OER per Co site139 and size-dependent catalysis by metal clusters is a 

well-known phenomenon which, in general, arises from optimal electronic structure, and 

hence absorption energies, at a particular size.140, 141 This, however, is but one possible 

explanation for the data, and other mechanisms invoking interfacial Ni-FeOx cooperation142 

could also be at play, especially considering that the intrinsic activity of Fe:NiOxHy is 

greater than that of Fe:CoOxHy at a comparable Fe loading. Related catalyst/support-type 

interactions of this kind are known for FeOxHy thin films on Au substrates whose activity 

is much higher than when they are deposited on Pt125.  

Electrokinetic analysis of cooperative Fe sites. The Tafel slopes of Fe:NiOxHy 

and Fe:CoOxHy decrease with increased Fe concentration in the electrolyte (Figure 4d), 

which suggests Fe-absorption process that modulates OER mechanism and not just number 

of active sites76, 79. At low Fe concentration of 10 ppb, surface-Fe Fe:NiOxHy and 

Fe:CoOxHy have Tafel slopes of ~59 and ~62 mV dec-1 respectively, suggesting a similar 

OER mechanism when small amounts of Fe are adsorbed, perhaps as isolated sites. Tafel 

slopes of ~33 and 32 mV dec-1 are obtained with 100 and 200 ppb Fe3+ in the electrolyte 

(after the CA experiment until the maximum OER current results), like co-deposited 

Ni0.75Fe0.25OxHy. This result suggests that a similar mechanism is operative, likely due to 

the presence of cooperative surface Fe as the competent catalyst species in both. In 



39 
 

comparison, (surface-Fe) Fe:CoOxHy has Tafel slopes of ~44 and 42 mV dec-1 with 100 

and 200 ppb Fe3+ in the electrolyte, slightly higher than those for co-electrodeposited Co1-

xFexOxHy (x = 0.33~0.79 and Tafel slopes of 26~39 mV dec-1) films73. In both Fe-free and 

100-ppb-Fe electrolyte, Tafel slopes of Fe:NiOOH increase as a function of cycling (Figure 

49) with the increase faster for the Fe-free electrolyte, consistent with both the desorption 

of surface Fe into the electrolyte for the Fe-free case, and scrambling of the surface Fe with 

internal sites in the Fe-spiked electrolyte upon potential cycling. Further detailed 

electrokinetic analysis on a broader set of materials derived from these methods would be 

useful to understand how the proposed Fe-based local structures control apparent reaction 

pathways143.  

 

Figure 5. Fe versus Ni loading and TEM of Fe sites. 
 Plot (left panel) of the change in Fe loading by incorporation during CA and with CA plus 10 subsequent 
CV cycles versus the total Ni mass loading and film capacitance. All elemental data was obtained with ICP-
MS measurement of the dissolved films. The inset depicts how the mol or at % changed for each Fe 
incorporation technique from the lowest to highest Ni loading. While Fe incorporates at roughly constant mol 
% when the Ni and Fe species are allowed to mix by CV, the mol % of Fe incorporated by CA decreases 
substantially as the Ni loading is increased. The right panel TEM-EDX images of NiOxHy electrodeposited 
directly onto a gold TEM grid after Fe was adsorbed during chronoamperometry at 1.55 V vs RHE. While 
we observe localized Fe signal, it is not possible with this technique to distinguish between the surface 
absorbed and internal/bulk sites nor to quantify accurately the size of the Fe clusters. 

Structural characterization. TEM with EDX mapping reveals the proposed Fe-

oxo clusters must be small (Figure 5 and 34), likely of molecular dimensions, and 

consistent with electrochemical and ICP-MS data. Still, EDX is limited when low 

concentrations are being measured. We thus also quantified the amount of Fe incorporated 

in NiOxHy during CA in 0.1 ppm Fe as a function of NiOxHy mass loading (Figure 4) using 

ICP-MS to understand how Fe is incorporated. Both the integrated charge in the Ni redox 



40 
 

wave and the double-layer capacitance (CDL, measured in the oxidized state) of the films 

increased linearly with mass, showing the films are electrolyte-permeable and fully 

electron-accessible. The amount of Fe, ~10 at. %, incorporated by 10 voltammetry cycles 

was constant with mass loading up to the maximum loading studied of ~100 nmol∙cm-2 Ni, 

consistent with cycling incorporating Fe throughout the NiOxHy. In contrast, the amount of 

incorporated Fe from constant-potential CA decreases from ~ 8 at. %  to 3 at. % as the Ni 

loading is increased from 20 to 100 nmol∙cm-2. This is consistent with FeOx primarily 

absorbing on edge or defect sites. As NiOxHy loading is increased, both new nanosheets 

form and existing nanosheets grow, thus decreasing the amount of edge-site-area per mass. 

Pair-distribution-analysis (PDF) also shows the size of coherent-scattering Ni(Fe)OxHy 

domains increases with mass loading144. From the previous PDF data, we estimate that if 

all the edges of each coherently scattering domain, containing roughly 100-200 Ni cations, 

were decorated with single row of absorbed FeOx, this would correspond roughly 20 at. % 

Fe. The 3-8 at. % observed here indicates that absorption of Fe is not uniform along all 

edges and that all the edges are not covered.  

We next investigated the local structure of Fe adsorbed during CA on NiOxHy using 

operando XAS. The Fe3+ spiking experiments and EDX analysis suggest Fe species 

adsorbed during CA are at surface sites as low-nuclearity clusters of FeOxHy, which grow 

with time under positive polarization in the Fe-spiked solution. If so, we expect the Fe-O 

bond lengths of these species to more-resemble Fe oxyhydroxide and that subsequent 

cyclic voltammetry would lead to a contraction of the Fe-O bond due to incorporation into 

the NiOOH host (which has a shorter M-O bond length). This reasoning is based on the 

XAS measurements on co-deposited NiFeOxHy by Bell and coworkers91 where the Fe-O 

bond length measured under OER conditions had values that increased with the amount of 

Fe in the NiFeOxHy with the pure FeOxHy having the longest bonds. XAS measurements at 

the Fe K-edge were performed during polarization at 0.68 V vs Hg/HgO immediately after 

spiking with CA and again after cycling (Figure 44). The current increased immediately 

after Fe spiking and again after cycling (Figure 43) though these could not be normalized 

to the amount of Fe incorporated in this experiment. The similarity of the shape and 

position of the XANES edges of our sample and Fe2O3 corroborates that Fe is 

predominantly in a 3+ oxidation state (Figure 45); minor differences in shape may arise 



41 
 

from difference in crystal structure. From the EXAFS data fitting (Figure 46 and Table 1), 

we find that before cycling the Fe-O bond length was 1.953 ± 0.004 Å and after cycling it 

contracted to 1.939 ± 0.011 Å; consistent with the hypothesis of Fe moving from Fe-rich 

surface absorbed structures to internal sites. The determined coordination numbers (CN) 

of the first and second shell of near 5 for later stages of Fe adsorption are consistent with 

an Fe-oxo cluster of a molecular dimensions larger than a dimer adsorbed on the surface 

(CN 3) but smaller than an extended crystal (CN 6), for which also further Fe-M shells 

should have been resolved. The error bounds on the EXAFS data, however, precludes a 

definitive conclusion from this XAS data alone, and further study with XAS and 

complementary operando techniques is needed to elucidate in detail the nature of the 

cooperative Fe sites formed by adsorption under OER conditions. 

 

 
Figure 6. Computational models and OER mechanisms for Fe sites on NiOOH. 
 (a,b) Schematic of adsorbed active sites on the NiOOH (01�5) surface: (a) the isolated “Fe-O” case and (b) 
dimer “Fe-O-Fe” case, along with the mechanisms chosen for investigation. Values along the reaction 
pathways are the theoretical overpotential in units of eV for each step. (c) Summary of the total theoretical 
overpotentials for each depicted pathway including the effect on the overpotential of a single Ni substitution 



42 
 

at an Fe in the isolated Fe-O monomer or Fe-O-Fe dimer. Models are shown at larger scale in Supplementary 
Figure 52. 

 

 

Understanding OER Mechanisms via DFT calculations. To test how Fe 

cooperative interactions affect the OER mechanism and overpotential we built two model 

systems: i) Fe cations separated and each coordinated by four hydroxides, one water, and 

one oxo bonded to the surface Ni denoted as “isolated Fe-O” species (Fig. 6a, top, note 

there are no Fe-O-Fe linkages), and ii) two adsorbed Fe cations directly bonded by a bridge 

oxygen ligand as a “Fe-O-Fe dimer” (Fig. 6b, top). These two models simulate cooperation 

of neighboring surface-absorbed Fe-O species as a model active site, which may also be 

found on larger clusters. The structure of the adsorbed clusters at different view angles are 

shown in Figure 51. To build the surface, the bulk NiOOH structure was cleaved at the 01�5 

plane due to its known activity145-149. The 01�5-oriented unit cell is then multiplied in 2×1×1 

direction to have Fe concentration of 9%, similar to the experimental values for maximum 

TOFFe. The bottom part of Figure 6 shows elementary reactions steps and energy changes 

(eV) in the various mechanisms considered for the Fe-O and Fe-O-Fe modified surfaces. 

The theoretical overpotential (𝜂𝜂th) is defined as the voltage needed for all the reactions 

steps to have negative Gibbs free energies and are summarized in Figure 6c and 

Supplementary Table 3.  

Water oxidation at the axially coordinated water (axial pathway) is the common 

mechanism for both Fe-O-Fe dimers and isolated Fe-O species. Based on the calculated 

ΔG for each step in the axial mechanism (Table 7), the Fe-O-Fe dimer has a 𝜂𝜂th of 0.50 V, 

while for isolated Fe-O it is 0.96 V. Critically, both the Fe cations in the Fe-O-Fe dimer 

change their oxidation state from 3 to 4 during the step with the largest ΔG (Table 6), while 

for the isolated Fe-O only the Fe cation at the active site changes its oxidation state. To test 

whether the lower overpotential requires specifically two adjacent absorbed Fe cations, one 

of the Fe was replaced with Ni. In the Fe-O-Ni dimer, Fe changes its formal oxidation state 

from 4 to 5 while the Ni oxidation state is unchanged, yielding 𝜂𝜂th = 0.89 V. For the 

isolated Fe-O species, Ni substitution at one of the Fe atoms reduces 𝜂𝜂th only from 0.96 V 

to 0.80 V. To check whether the lowered overpotential upon Fe dimerization is facet 



43 
 

specific, the (001) surface of NiOOH (0Figure 50) was also modeled150. Like the (01�5) 

surface, the Fe-O-Fe dimer on the (001) surface has roughly half the 𝜂𝜂th compared to the 

isolated Fe-O. The oxidation-state changes are similar as well with both Fe3+ cations in the 

Fe-O-Fe dimer oxidized to Fe4+ during the potential-determining step. In general, the 

investigation of the axial pathway for both types of surfaces showed that isolated 

monomers tend to have higher OER overpotentials than dimers (when all surface-attached 

group atoms are Fe), in agreement with experimental TOFFe trends. Beyond dimers, larger 

surface Fe-oxo clusters, as compatible with EXAFS analysis, may further stabilize the 

potential-determining intermediate by spreading oxidative charge over multiple Fe sites. 

We note that the actual experimental system likely includes various more-complex surface 

geometries with more absorbed Fe cations than the simple dimer considered in the 

calculation. 

For the Fe-O-Fe dimer on the 01�5 surface other possible mechanisms were also 

investigated (Figure 6b). The axial and equatorial pathways proceed through the same 

intermediates except that they take place at two different positions; both have similar 𝜂𝜂th 

of 0.50 and 0.47 V and in both cases the Fe oxidation state changes from 3 to 4. The 

equatorial pathway on the isolated Fe-O was also assessed and found impractical due to 

the instability of the water-coordinated structure. The insertion pathway shares the same 

first step that has the largest free-energy change with the axial pathway and thus both 

mechanisms have similar 𝜂𝜂th. The key feature of the bridge mechanism that sets it apart 

from the others is that the bridge oxo group attached to two Fe and one Ni. When the 

bridging oxo group is deprotonated, all of these surrounding metal atoms are oxidized; the 

two Fe and one Ni increase their oxidation state from +3 to +4 (Table 7). Consequently, 

the bridge mechanism exhibits the lowest overpotential of all mechanisms studied for the 

Fe-O-Fe dimers with  𝜂𝜂th of 0.33 eV.  

Conclusion 
 

We studied the incorporation and OER-activation by foreign electrolyte ions into 

electrodeposited NiOxHy and CoOxHy films and discovered that under controlled oxidative 

conditions the incorporation can be limited to surface sites. In the case of Fe on NiOxHy, 



44 
 

we used this approach to demonstrate both record intrinsic OER activity (at optimal 

surface-Fe loading), and a fundamental picture emphasizing cooperative effects between 

multiple Fe sites that share oxidative charge. The computational calculations show that 

new low overpotential pathways for OER are possible through synergistic interaction of Fe 

multiple Fe species and host metal atoms whereby oxidative charge can be favorably 

delocalized and stabilized. The substitution of one Fe by Ni in a model dimer increases the 

overpotential from 0.33 to 0.55 eV, likely because only one of the three atoms changes its 

oxidation state, implying that Fe-O-Fe motifs play a key role in charge delocalization. A 

similar investigation, but for a cobalt host could explain the discrepancy between TOFFe of 

Fe:NiOxHy and Fe:CoOxHy at fixed Fe loading, namely, Co may not be as effective at 

sharing oxidative charge with surface Fe-O-Fe. This insight was accomplished through 

careful electrochemical and analytical techniques—in large part the tools of material 

characterization (SEM/TEM/XAS) fail to provide these insights because of the disordered 

nature of the NiOxHy support and low Fe loadings associated with the record-TOFFe surface 

FeOx species. 

This work thus provides insight into the exceptional OER activity of Fe-based 

mixed-metal oxyhydroxide catalysts from which design principles emerge that are 

important for advanced alkaline water electrolyzers (operating in hot concentrated basic 

electrolyte), alkaline membrane electrolyzers (that use a solid-ionomer electrolyte and pure 

water or dilute soluble electrolyte salts), and for broad classes of photoelectrochemical 

systems. Maximal OER activity is not only a function of the Fe content in NiOxHy and 

CoOxHy
79, but also exactly where that Fe is located and how it interacts. Because the Fe 

sites are dynamic, depositing from electrolyte but also simultaneously dissolving76-79, 

advanced electrolyzers should engineer the electrolyte “impurity” levels and/or the local 

dissolution/re-deposition environment such that the highest-activity cooperative Fe sites 

can be continually maintained during operation. These sites should be optimally supported 

on NiOxHy-based oxyhydroxides surfaces that provide electrical interconnection and 

nanostructure, but for operational stability and during, e.g., potential changes with on/off 

cycles, structurally more-robust CoOxHy (or other oxides) can be alloyed151 with the host 

NiOxHy and supported on stable porous-transport layers.  



45 
 

In the next chapter, the method of CA Fe-spiking to selectively make surface Fe 

sites is leveraged for temperature-dependent kinetic measurements yielding parameters 

which describe the activation barriers for the rate-determining step in the OER mechanism 

on this high activity sites. 

 

 

 

 

 

 

III. TEMPERATURE-DEPENDENT KINETICS OF REACTIVE 
SURFACE FE SITES ON NICKEL OXYHYDROXIDE  

 

Liam P. Twight, Shannon W. Boettcher 

This chapter contains unpublished work which has been prepared for submission to Journal of the 

American Chemical Society in 2024. The study was conceived and designed by L.P.T and S.W.B. 

S.W.B oversaw the project. All experimental design, measurement, and data analysis was 

performed by L.P.T. The manuscript was written by L.P.T with editorial assistance from S.W.B. 

 

Introduction 
 

Nickel-iron hydroxides (NizFe1-zOxHy) are the highest intrinsically active alkaline 

OER catalysts identified by three-electrode measurements49, 152. Comparison between the 

OER activity of NiOxHy in electrolyte rigorously purified of iron and in electrolyte with 

intentionally added Fe (100 ppb) have shown that iron incorporation into the oxyhydroxide 

host is essential and without which the record-low reductions in the OER kinetic 

overpotential of OER are not observed58. Yet, understanding the mechanistic details which 

enable this high activity is far from trivial because Fe sites are dynamic - they actively 

dissolve and redeposit during OER79, the nickel hydroxide host exhibits potential-

dependent structures133, 153 that change the local chemical environment of adsorbed Fe 



46 
 

sites66, and NizFe1-zOxHy appears to form ubiquitously in-situ on the surface of crystalline 

solids like NiO152, LaNiO3 and La2NiO4,154 BaNiO3,155 and metal chalcogenides156 where 

the surviving pre-catalyst could affect the Fe active sites by electronic125 or epitaxy 

effects157.  

Temperature-dependent measurements can be used to characterize the evolution of 

kinetic barriers with site structure. An Arrhenius plot of the rate versus inverse temperature 

allows extraction of the activation energy of the reaction from the slope and of the pre-

exponential factor from the intercept which should be characteristic for a given site type158. 

Absolute rate theory can then be applied to decompose these parameters into one which 

contain the enthalpy and entropy of activation describing equilibrium between reactants 

and the activated complex. Further, if