Communication pubs.acs.org/JACS
Reactive Fe-Sites in Ni/Fe (Oxy)hydroxide Are Responsible for Exceptional Oxygen Electrocatalysis Activity Michaela Burke Stevens, Christina D. M. Trang, Lisa J. Enman, Jiang Deng, and Shannon W. Boettcher* Department of Chemistry and Biochemistry, University of Oregon, Eugene, Oregon 97403, United States S Supporting Information *
the films using surface-interrogation scanning electrochemical microscopy and Mössbauer spectroscopy, respectively. Despite the large difference in the proposed mechanism or active site and the Fe/Ni oxidation state, all the materials in the cited studies are similarly efficient OER catalysts. The various reports discussed above have all demonstrated changes in the electrochemical profile, e.g. Ni redox potential, redox peak size and shape, etc., as a function of Fe incorporation into the NiOxHy film. Theoretical/computational studies have also provided insights into the role of Fe.15 However, it remains unclear where the Fe incorporates into the film to result in highly active Fe−O, Ni−O, or Ni−O−Fe sites. It is also unclear if the changes in bulk electronic structure, i.e., as interrogated by measurements of conductivity, redox peak position, and average e− per Ni in the redox wave, are directly responsible for the increased activity or simply a byproduct of the Fe incorporation unrelated to the enhanced catalytic activity. Answering these questions is important to develop a mechanistic picture of OER and a structural picture of probable active sites. Here we report the characterization of highly OER-active Ni(Fe) (oxy)hydroxides that are nearly identical in reduction potential, conductivity, and electrochemical accessibility to their Fe-free parent materials (that have low catalytic activity). Furthermore, we describe two Ni(Fe)OxHy films that have incorporated Fe differently and have significantly different electronic/electrochemical properties, while still maintaining high activity. These results indicate the presence of different types of active sites within Ni(Fe) (oxy)hydroxides and emphasize the role of the local geometric structure on activity as opposed to bulk electrochemical behavior. An electrolyte permeable (i.e., disordered/porous) Ni(OH)2 film was cathodically deposited from an aqueous Ni(NO3)2 solution onto a conductive substrate (Au or Pt).16 As Ni(OH)2 is cycled, a sharp precatalytic redox feature is visible (Figure S1). This redox wave is typically associated with the nominal 2+ → 3+ oxidation of Ni(OH)2 to NiOOH; however, as has been previously shown, this redox feature typically accounts for between 1.3 and 1.7 e− per Ni.17,18 The extra 0.3−0.7 e− per Ni could be due to further oxidation of the Ni3+ to an average Ni∼3.5+ with possible Ni4+ sites and/or to the oxidation of oxygen in the lattice (i.e., anion redox).19 Due to the dynamic changes in oxidation state, hydration level, counterion content during electrochemical cycling or catalysis, we typically refer to
ABSTRACT: Fe is a critical component of record-activity Ni/Fe (oxy)hydroxide (Ni(Fe)OxHy) oxygen evolution reaction (OER) catalysts, yet its precise role remains unclear. We report evidence for different types of Fe species within Ni(Fe)OxHy those that are rapidly incorporated into the Ni oxyhydroxide from Fe cations in solution (and that are likely at edges or defects) and are responsible for the enhanced OER activity, and those substituting for bulk Ni that modulate the observed Ni voltammetry. These results suggest that the exceptional OER activity of Ni(Fe)OxHy does not depend on Fe in the bulk or on average electrochemical properties of the Ni cations measured by voltammetry, and instead emphasize the role of the local structure.
U
nderstanding the oxygen evolution half reaction (OER, in alkaline media: 4OH− → 2H2O + 4e− + O2) is important for an array of energy-conversion technologies.1,2 Fe-doped Niand Co-based oxide/oxyhydroxide catalysts are among the most-active OER catalysts in alkaline media.3,4 Fe increases the intrinsic activity of both Co and Ni (oxy)hydroxides by roughly 30- and 1000-fold, respectively.3,5−8 For electrodeposited Ni or Co (oxy)hydroxide, Fe can be stabilized in a solid solution (i.e., it does not form phase segregated islands) apparently up to ∼25 and ∼50%, respectively.6,9 Further, Klaus et al. have shown that active Ni(Fe)OOH phases form regardless of catalyst deposition method.10 Despite the similarity of phases formed under a wide variety of preparation conditions, there have been many different mechanistic hypotheses regarding the role that Fe plays in (oxy)hydroxide OER activity.9,11,12 Li et al. have hypothesized that increased Ni−O hybridization as a result of the Lewis acidity of Fe3+ resulted in a more-active material.11 Gorlin et al. coupled quasi-in-situ X-ray absorption spectroscopy (XAS) and differential electrochemical mass spectrometry and reported that Ni remains in a 2+ oxidation state; they observe no evidence for increased hybridization as a function of Fe.12 Friebel et al. show that the Fe−O bond length in Ni(Fe)OxHy is different from the Fe−O bond length in FeOxHy and that the overpotential on this Fe−O site was lower than the overpotential on the Ni−O sites.9 It is similarly debated what the oxidation state of Ni and Fe are during catalysis and what role this plays in determining the OER activity. Friebel et al.9 and Gorlin et al.12 see no change in the Fe oxidation state (Fe3+) during the OER; however, Ahn and Bard13 and Chen et al.14 have shown evidence for Fe4+ in © 2017 American Chemical Society
Received: July 8, 2017 Published: August 8, 2017 11361
DOI: 10.1021/jacs.7b07117 J. Am. Chem. Soc. 2017, 139, 11361−11364
Communication
Journal of the American Chemical Society
overpotential at 1 mA cm−2 decreases only ∼20 mV between cycles 5 and 100 and >50% of that change occurs in the first 10 cycles after Fe-spiking (Figure 1, cycles 5−15, orange and green). Electron microscopy shows that the film morphology does not dramatically change as a function of cycling (Figure S3). The inconsistency between observed changes in activity and electrochemical profile as a function of Fe incorporation suggests that there is initial Fe incorporation at easily accessible edge/defect sites, forming Ni−O−Fe species that dramatically enhance catalysis, but that this has minimal effect on the majority of the Ni redox species (Figure 1a inset). In fact, this large effect on the catalysis is evidence that the catalysis is not particularly influenced by Fe in the NiOxHy bulk (i.e., Fe sitting on a Ni site fully coordinated by bridging O(H) with other metal cations) or by the bulk electronic structure. This is consistent with previous suggestions of active Fe edge sites.20 To probe the effect of incorporating Fe species, the effective electrical conductivity of Fe-spiked NiOxHy was measured as a function of applied potential before and after the addition of Fe(NO3)3 at 1 mM to the initially Fe-free electrolyte. Figure 2
these (oxy)hydroxide catalysts as MOxHy (where M is the metal cation). During the first four voltammetry cycles under Fe-free conditions the NiOxHy anodic peak potential (Epa) shifts cathodically 12−15 mV and the anodic peak integrated intensity decreases by 2−5%. From cycle 5−104 the Epa and e− per Ni do not change significantly (Figure S1). When Fe(NO3)3 at 1 mM is spiked into the alkaline solution (Figure 1), there is an immediate increase in activity shown by
Figure 1. Cyclic voltammetry demonstrating the role of Fe incorporation into NiOxHy from electrolyte solution. (A) NiOxHy cycled (10 mV s−1) initially in Fe-free aq. 1 M KOH (red, cycle 1−4) then moved to a 1 M KOH solution with 1 mM Fe(NO3)3 (cycle 5− 104). The inset depicts a possible schematic of Fe incorporation into a NiOxHy platelet as a function of cycling in Fe saturated solution. From the cyclic voltammetry, (B) the percent change in e− per Ni (triangles) and (C) the change in overpotential (mV) at 1 mA cm−2 (open circles) and position of Epa (mV) (squares) as a function of cycling relative to the last cycle in Fe-free conditions. Fe content listed is from ICP-OES analysis on samples cycled in identical conditions for 2 (orange), 10 (green), 30 (blue), and 100 cycles (violet) after Fespiking. The error bars are based on samples in triplicate with ∼3−5 μg cm−2 loading.
Figure 2. Steady-state effective conductivity of NiOxHy on an interdigitated array (IDA) electrode as a function of Fe incorporation from 1 mM Fe(NO3)3 in a 1 M aq. KOH solution. The effective conductivity was measured before Fe spiking (pre-Fe), immediately after Fe spiking (1), and after cycling (2−3). The film was held at each step for 2 min. Cyclic voltammetry is overlaid. There were two cycles (10 mV s−1) between each steady-state conductivity measurement; the second of each series is shown.
shows no correlation between conductivity and activity; the activity increases dramatically with Fe incorporation while there is only a small decrease in the effective conductivity. The decrease as a function of cycling is also seen in the Fe-free NiOxHy (Figure S4).7 As the film is cycled, the onset potential for conductivity shifts with the shift in Epa indicating a change in bulk electronic structure. This result is consistent with the edge and defect sites dominating the catalytic response of the system while the bulk electronic properties dominate the electrical transport properties. This result is different from coelectrodeposited Ni−Fe (oxy)hydroxides where the addition of Fe increased the conductivity; the difference is likely explained by a higher degree of crystallinity for the Ni−Fe relative to the pure Ni oxy(hydroxides) when both are electrodeposited under similar conditions.7 Small amounts of Fe are also known to induce large OER activity changes on Au substrates likely due to FeOxHy sites
the 130−150 mV decrease in overpotential. This dramatic change in activity is accompanied by a minimal increase in electrochemical accessibility to Ni (2−4%, as measured by the integrated charge in the redox wave) and anodic shift in the Ni Epa (2−5 mV). During 100 cycles in Fe3+ spiked solution, Epa shifts a maximum of 30 mV and the peak integration shrinks to reach 30% less e− per Ni than the Fe-Free NiOxHy. A similar trend is seen with the cathodic peak size and potential (Epc) (Figure S2). After electrochemical analysis, the film has 23− 25% Fe as determined by ICP-OES analysis (Table S1). Despite the large changes in apparent redox wave and peak potential of the Ni species, there is only a minimal increase in activity after the initial cycle in the Fe-spiked solution. The 11362
DOI: 10.1021/jacs.7b07117 J. Am. Chem. Soc. 2017, 139, 11361−11364
Communication
Journal of the American Chemical Society
Within 2 cycles in the Fe-spiked base, the NiOxHy film picks up an average of 11% Fe (the example in Figure 3a has 10% Fe) and the activity increases by ∼150-fold; however, there is only an ∼10 mV anodic shift in Epa and ∼2% difference in e− per Ni in the redox wave. The Epa for the codeposited Ni(Fe)OxHy sample with similar Fe content (Figure 3a) is ∼26 mV anodic of the Fe-free NiOxHy wave and has ∼7% more e− per Ni. The Fe-spiked sample has twice the TOFtm (Figure 3c) than the codeposited Ni(Fe)OxHy. After 100 voltammetry cycles in the Fe-spiked solution, the film incorporates ∼25% Fe. Epa is, however, ∼55 mV cathodic of that observed for a codeposited Ni(Fe)OxHy film with similar Fe content (Figure 3b). The e− per Ni is ∼30% smaller for the Fe-spiked sample compared to the codeposited one. The activity of the two, however, is more similar than for the lower Fe concentration pair. For the codeposited 30% Fe sample, the cathodic and anodic peaks are relatively narrow (and split by Epa − Epc = 80 mV) indicating a sample that has homogeneously incorporated the Fe into the bulk of the Ni lattice. This is consistent with the literature in which ∼25% Fe is soluble in NiOOH.9 In contrast, the 25% Fespiked sample has two distinct reduction and oxidation peaks yielding broader redox features; this suggests that Fe is inhomogeneously distributed in the bulk and on the edges of the NiOxHy in the Fe-spiked sample, thus inhomogeneously affecting the Ni redox wave position. The data discussed above demonstrate (1) a Fe−NiOxHy catalyst nearly identical to its parent NiOxHy in terms of Ni redox behavior but is ∼100-fold more active and (2) a spiked Fe-NiOxHy and codeposited Ni(Fe)OxHy that have similar Fe content and activities, but large differences in e− per Ni, Epa, and redox wave shape. This supports a view in which FeOxHy species initially incorporate into NiOxHy from solution at surface/edge/defect sites, before moving into the bulk of the NiOxHy nanosheet samples and significantly affecting the Ni redox behavior. Co-deposited Ni(Fe)OxHy films appear to incorporate Fe more uniformly during synthesis. These results indicate that the Ni−Fe activity is more related to the local active-site environment than the bulk electronic structure and there are Fe-sites in different local environments (i.e., edge versus bulk). Future studies should therefore search for different Fe geometries directly in samples such as these, for example using EXAFS at the Fe K-edge or infrared specroscopy.8 In conclusion, upon the introduction of Fe into the NiOxHy from solution the electronic properties, reduction potential, redox peak size, peak shape, etc. of the Fe−NiOxHy film change independently of the OER activity. Fe may initially incorporate at edges/defects and further incorporate into the “bulk” of the NiOxHy nanosheets (with Fe also remaining at the edge/defect sites). The e− per Ni in the redox wave does slightly increase with Fe addition, but the sample stays active even as the e− per Ni decreases with continued Fe incorporation in the bulk of the NiOxHy. The electrical conductivity of Fe−NiOxHy decays with cycling in Fe-containing solution even as the sample becomes more active. Finally, the reduction potential (as monitored by the Epa) does not shift to a potential expected of a codeposited sample with similar Fe content. These experiments are all consistent with the view that the “bulk” (oxy)hydroxide electronic structure is not the main factor influencing OER activity. These results may help explain the recent conflicting hypotheses on the role of Fe in Ni-based OER catalysts. The deposition method and electrochemical conditioning appear to
integrated with Au−O surface species.21,22 Fe-spiked control experiments on blank Au and Pt electrodes (Figure S5) demonstrate that although the activities of both “bare” electrodes are significantly impacted by Fe spiking, the majority of activity enhancement discussed here occurs on the NiOxHy rather than the substrate. However, fundamentally the Feactivated Au−O electrodes may be catalyzing the OER in a similar fashion to the Fe-activated NiOxHy with both utilizing surface (in the case of Au−O) or edge (in the case of NiOxHy) Fe sites for the enhanced activity. It is possible that the difference in active surface area between the NiOxHy film and the Au surface is largely responsible for the difference in overpotential, although efforts to normalize and extract the intrinsic activity suggest the Fe-activated NiOxHy is the faster catalyst.22 With continued cycling of NiOxHy in 1 M aq. KOH with 1 mM Fe(NO3)3 (Figure 1) the Epa shifts significantly anodically. It is known that the reduction potential of nominally Ni2+/3+ shifts anodically with increasing Fe content when deposited as a mixed (oxy)hydroxide.7,23,24 This shift in reduction potential has been associated with a bulk change in the electronic structure of the material and can be used to determine if Fe has incorporated into the Ni(Fe)OxHy film. Figure 3a and 3b compare an Fe-free NiOxHy film, a spiked Fe−NiOxHy film cycled 2 and 100 times, and a codeposited Ni(Fe)OxHy film with similar Fe content. The OER activity (Figure 3c) is based on the total-metal turnover frequency (TOFtm) or amount of O2 produced per metal cation per second.
Figure 3. Cyclic voltammetry (10 mV s−1) of Fe-free NiOxHy film (red) compared to (A) spiked Ni(Fe)OxHy cycled 2 (orange, 10% Fe) and (B) 100 (violet, 25% Fe) times in 1 mM Fe(NO3)3 and codeposited Ni(Fe)OxHy films with similar Fe content (green 8% Fe and teal 30% Fe, respectively). (C) TOFtm at η = 300 mV based on total metal content using the average of the forward and reverse sweep from panel (A). (D) e− per Ni obtained by integration of the oxidation wave. Metal content and ratios were determined after electrochemical analyses using ICP-OES. Open and closed symbols are for codeposited and Fe-spiked samples, respectively. Error bars are based on samples measured in triplicate. 11363
DOI: 10.1021/jacs.7b07117 J. Am. Chem. Soc. 2017, 139, 11361−11364
Communication
Journal of the American Chemical Society
(16) Stevens, M. B.; Enman, L. J.; Batchellor, A. S.; Cosby, M. R.; Vise, A. E.; Trang, C. D. M.; Boettcher, S. W. Chem. Mater. 2017, 29, 120−140. (17) Capehart, T. W.; Corrigan, D. A.; Conell, R. S.; Pandya, K. I.; Hoffman, R. W. Appl. Phys. Lett. 1991, 58, 865−867. (18) Corrigan, D. A.; Conell, R. S.; Fierro, C. A.; Scherson, D. A. J. Phys. Chem. 1987, 91, 5009−5011. (19) Grimaud, A.; Hong, W. T.; Shao-Horn, Y.; Tarascon, J.-M. Nat. Mater. 2016, 15, 121−126. (20) Hunter, B. M.; Hieringer, W.; Winkler, J.; Gray, H. B.; Müller, A. M. Energy Environ. Sci. 2016, 9, 1734−1743. (21) Klaus, S.; Trotochaud, L.; Cheng, M. J.; Head-Gordon, M.; Bell, A. T. ChemElectroChem 2016, 3, 66−73. (22) Zou, S.; Burke, M. S.; Kast, M. G.; Fan, J.; Danilovic, N.; Boettcher, S. W. Chem. Mater. 2015, 27, 8011−8020. (23) Louie, M. W.; Bell, A. T. J. Am. Chem. Soc. 2013, 135, 12329− 12337. (24) Corrigan, D. A. J. Electrochem. Soc. 1987, 134, 377−384.
be directly responsible for the Fe location and local environment within the film, but do not change the catalyst activity in all cases. Depending on how the catalyst is deposited, it might have a higher density of Fe in edge/defect sites versus bulk sites. This could determine the electrochemical profile of the material, and even the average oxidation state of the Ni and Fe sites, but leave the overall activity of the film unaffected.
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.7b07117. Experimental methods and additional voltammetry and materials characterization data (PDF)
■
AUTHOR INFORMATION
Corresponding Author
*
[email protected] ORCID
Shannon W. Boettcher: 0000-0001-8971-9123 Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This work was supported by the National Science Foundation under Grant CHE-1566348. J.D. acknowledges support from Zhejiang University and C.D.M.T. from the University of Oregon Presidential Undergraduate Research Scholarship.
■
REFERENCES
(1) Burke, M. S.; Enman, L. J.; Batchellor, A. S.; Zou, S.; Boettcher, S. W. Chem. Mater. 2015, 27, 7549−7558. (2) Hunter, B. M.; Gray, H. B.; Müller, A. M. Chem. Rev. 2016, 116, 14120−14136. (3) Trotochaud, L.; Ranney, J. K.; Williams, K. N.; Boettcher, S. W. J. Am. Chem. Soc. 2012, 134, 17253−17261. (4) McCrory, C. C. L.; Jung, S.; Peters, J. C.; Jaramillo, T. F. J. Am. Chem. Soc. 2013, 135, 16977−16987. (5) Burke, M. S.; Zou, S.; Enman, L. J.; Kellon, J. E.; Gabor, C. A.; Pledger, E.; Boettcher, S. W. J. Phys. Chem. Lett. 2015, 6, 3737−3742. (6) Burke, M. S.; Kast, M. G.; Trotochaud, L.; Smith, A. M.; Boettcher, S. W. J. Am. Chem. Soc. 2015, 137, 3638−3648. (7) Trotochaud, L.; Young, S. L.; Ranney, J. K.; Boettcher, S. W. J. Am. Chem. Soc. 2014, 136, 6744−6753. (8) Hunter, B. M.; Blakemore, J. D.; Deimund, M.; Gray, H. B.; Winkler, J. R.; Muller, A. M. J. Am. Chem. Soc. 2014, 136, 13118− 13121. (9) Friebel, D.; Louie, M. W.; Bajdich, M.; Sanwald, K. E.; Cai, Y.; Wise, A. M.; Cheng, M.-J.; Sokaras, D.; Weng, T.-C.; Alonso-Mori, R.; et al. J. Am. Chem. Soc. 2015, 137, 1305−1313. (10) Klaus, S.; Louie, M. W.; Trotochaud, L.; Bell, A. T. J. Phys. Chem. C 2015, 119, 18303−18330. (11) Li, N.; Bediako, D. K.; Hadt, R. G.; Hayes, D.; Kempa, T. J.; von Cube, F.; Bell, D. C.; Chen, L. X.; Nocera, D. G. Proc. Natl. Acad. Sci. U. S. A. 2017, 114, 1486−1491. (12) Gorlin, M.; Chernev, P.; De Araujo, J. F.; Reier, T.; Dresp, S.; Paul, B.; Krahnert, R.; Dau, H.; Strasser, P. J. Am. Chem. Soc. 2016, 138, 5603−5614. (13) Ahn, H. S.; Bard, A. J. J. Am. Chem. Soc. 2016, 138, 313−318. (14) Chen, J. Y. C.; Dang, L.; Liang, H.; Bi, W.; Gerken, J. B.; Jin, S.; Alp, E. E.; Stahl, S. S. J. Am. Chem. Soc. 2015, 137, 15090−15093. (15) Fidelsky, V.; Toroker, M. C. Phys. Chem. Chem. Phys. 2017, 19, 7491−7497. 11364
DOI: 10.1021/jacs.7b07117 J. Am. Chem. Soc. 2017, 139, 11361−11364
