Combinatorial screening of Pd-based quaternary electrocatalysts for oxygen reduction reaction in alkaline media

Abstract

The implementation of electrochemical systems such as fuel cells has been hindered by the slow development of low cost high activity catalysts. Here we examine the oxygen reduction reaction performance of a combinatorial Pd–Au–Ag–Ti thin film library using high-throughput screening and correlate the electrochemical behavior to the crystallographic properties. We find compositions of ca. 40–60 at% Pd and 30–35 at% Au exhibit both a low overpotential of close to the value of pure Pt as well as high current density. We also observe a volcano-like relationship between the overpotential and the solid formation strain. This study provides compositional guidance towards the future synthesis of nanostructured quaternary Pd–Au–Ag–Ti alloys and suggests the potential for broader application of high-throughput electrochemical characterization by means of an automatic scanning droplet cell.

---

Introduction

Fuel cells, highly efficient low/zero emission energy conversion devices, have been extensively investigated in recent decades due to high-energy demands, fossil fuel depletions, and environmental challenges throughout the world.^1,2 In particular, alkaline fuel cells (AFCs) have received significant attention due to their advantages over traditional proton exchange membrane fuel cells (PEMFCs) by providing faster oxygen reduction reaction (ORR) kinetics in a less corrosive catalyst environment (vs. acidic media).^3–13 Given the fact that the ORR is the key reaction at the cathode of a fuel cell, there have been considerable efforts to develop newer alkaline ORR electrocatalysts. In general, volcano-like plots are used to assist the design of newer materials for many electrochemical systems, among which the catalytic activity for mono- or bimetallic systems are correlated to the binding energies, the d-band center position, or the atomic coordination.^14,15

Previously, Nørskov et al. showed a volcano plot of the ORR activity of several binary metal surfaces versus the oxygen or OH binding energies and suggested that a combination of non-Pt metals might be close to and even surpass Pt with respect to ORR activity.¹⁶ Based on this observation, a rational choice for designing a new ORR material composition could thus start from selecting materials to match from each side of the volcano peak and study the ensemble effects in more complex multinary alloys. This approach, however, often turns into a serendipitous expedition, which neither accelerates the rapid development of new electrocatalyst materials nor provides guidance (i.e. generic design rules) for future discoveries. In addition, due to the modeling challenges associated with multinary systems there are no theoretical predictions available in literature for ORR activity of such systems. Without the guidance provided from simulation, it is neither easy to find an optimal composition nor practical to define a narrow chemical search space within multinary systems. Therefore, new approaches are needed so that multinary systems for specific chemical reactions can be examined in a more comprehensive way.

To this end, combinatorial materials science coupled with high-throughput characterization could be a viable approach to address these challenges. Since the first use in the 1960s for drug discovery,¹⁷ the combinatorial approach has expanded into materials synthesis and screening for a wide array of applications. Particularly in electrocatalysis, several researchers have recently developed high throughput electrocatalyst screening techniques, such as a scanning droplet cell (SDC),^18–20 multi-electrode based half-cells²¹ and full cells.²² Previous groups, including ours, have shown that these approaches can be used to study many facets of multinary systems, including variations of composition, crystal structure, and processing conditions.^18,23–26 Due to the advantage of scanning in a large compositional range,^20,27 SDC can be widely applied for the elucidation of specific properties of binary, ternary and quaternary material libraries.^18,25,26,28 However, SDC-based screening was seldom applied in ORR studies. Recently, we demonstrated, using SDC, that the ORR activity of thin Pt films sputtered on oxidized Ti/Nb surfaces is not significantly influenced by the substrate as long as the Pt film has a nominal thickness of >1 nm.²³ This time we will explore a non-Pt multinary system in alkaline ORR studies using a SDC as a primary screening tool.

From the perspective of experimental effort, cost savings, and reduced time for commercialization, it could be more beneficial to start from theory, narrow down the materials composition rapidly and then explore different methods of material fabrication. We illustrate this pathway including structural and electrochemical characterization in a “Gedankenexperiment” like materials funnel, where the SDC coupled with high throughput characterization is more efficient than a general electrocatalyst screening process (Scheme 1). Once an ideal material composition is determined, one can then tailor the catalytic structure for a target reaction and pursue a more in-depth structure and property characterization.²⁹ Through the years, various catalytic nanostructures have been demonstrated such as nanoprism,³ nanoporous sponge,⁴ bimetallic nanoparticle,⁸ nanothorn,¹⁰ core–shell nanostructures,^8–10,30 nanobrush¹¹ and nanoplate.¹² Previously, many Pd-based binary alloys were examined as ORR catalysts in alkaline solution such as Pd–Ti,³¹ Pd–Au³² and Pd–Ag.^5,30 Recently our group has shown that Ag–Pd core–shell structures³⁰ as well as metallic glass alloy structures^33–36 are active in the alkaline regime. Although these structural demonstrations show promise, the underlying development of new material compositions has been slow. For example, Lee et al. reported a ternary Pd/Au/Ag nanosponge,⁷ however compositional guidance from the literature was very limited.

Scheme 1 SDC coupled with compositional, structural and electrochemical characterization is more efficient than general electrocatalyst screening in narrowing down the materials composition.

In this paper, with the specific interest in applying SDC in the ORR studies with Pd-based multinary systems in alkaline media, we selected Au, Ag and Ti because: (1) they have previously been considered as good candidates for binary Pd-based alloy ORR catalyst;^30–32 (2) the structural similarity of Au and Ag could lower the complexity of the study of the multinary system;³⁷ and (3) Ti shows a much lower d-band center location than Au and Ag in the Pd binary alloy. In respect to solid solution alloying, Ti is also a good candidate material for Pd substitution as they have about the same ionic radius.³⁸ Here we show how high-throughput electrochemical characterization by means of SDC can contribute to the in-depth evaluation of a Pd-based quaternary system with analysis on the ORR on-set potential and current in correlation to composition and crystallographic properties.

Results and discussion

Compositional and structural properties

We show the compositional map of the entire Pd–Au–Ag–Ti thin film library (Fig. 1a). The composition variation in atomic percentage (at%) of the quaternary system is: Pd 21–71%, Au 15–65%, Ag 4–35%, and Ti 8–14%. This non-linear compositional distribution of the quaternary system of Pd–Au–Ag–Ti is caused by the orientation and the distance between the sputter targets and the base substrate as described previously.³⁹ The continuous thin-film library does not exhibit any color variation over the whole compositional gradient. From XRD patterns of the individual measurement areas of the ML, we observe that there are 5 to 6 diffraction peaks that shift systematically with compositional variation as indicated (Fig. S1†). The Ag–Pd–Au system is known to be a fcc solid solution with a low Ti solubility.⁴⁰ Upon addition of Ti, which has a low solubility,⁴⁰ we do not observe any evidence of a parasitic phase in this quaternary composition range from the XRD patterns.⁴¹ Based on these observations, we assume a solid solution of all elements over the entire materials library. We note that annealing of the library could lead to the formation of Ti-containing phases due to the far from equilibrium nature of the deposition process (magnetron sputtering onto substrate at room temperature). As there is no occurrence of other XRD peaks than those matching a single solid solution FCC phase, the only secondary phase that could exist is an amorphous matrix, which is therefore not detectable with the herein used XRD analysis.

Fig. 1 Color-coded property maps of the Pd–Au–Ag–Ti thin film library with the composition of Pd, Ag and Au measured by EDX vs. (a) Ti composition, (b) peak center and (c) peak width. (d) The peak center and height are plotted together with the Pd content as the color mapping.

With an increasing Pd content, the peaks shift to higher 2θ values, which indicates a contraction of the lattice (Fig. S1†). We illustrate that this trend becomes more pronounced when the peak center and width of the (111) peak are plotted over the compositional spread (Fig. 1b and c). For this plot the peaks were fitted by a simplified Cauchy curve:

Here x is the 2θ degree, y is the signal intensity, a is the peak height, b is the peak center, c is the peak width and d is a constant background. We show the peak center obtained from this fit (Fig. 1b). A trend along the Au–Pd gradient is visible for the peak shift. It is quite interesting to find in this quaternary system that the linear peak shift is in good agreement with Vegard's law, which is commonly used to describe the rule of mixture of lattice parameter of a solid solution of two constituents.⁴² This is shown by the plot of the peak center and height together with the Pd content (Fig. 1d). A linear trend with the Pd content for the peak position is visible with a lower peak height in the middle of the two quasi-equilibrium peak center values. The peak height is in fact lower; however, the integral peak intensity remains constant. From Vegard's law and the Williamson–Hall relationship,⁴³ the height of the peaks of a solid solution should become lower and the width should increase due to internal stress/strain with a compositional variation. We illustrate the case of peak broadening based on changes in the peak center (Fig. 1c). As the variation of the angle of the (111) is relatively small the width can qualitatively be discussed as the stress/strain in the system due to solid solution formation. Although according to the Debye–Scherrer equation, domain size can influence the width of a peak, it is not suitable in this case. Since the published liquidus temperature for the solid solution Pd–Ag–Au system⁴⁴ is varying from 1700 K to 1500 K (corresponding to a change in homologous temperature of about 15%), there should not be any doubled domain size at the apparently strained region than for the rest of the library. Therefore according to the variation of peak width, qualitatively the stresses are lowest near the quasi-binary Ag–Au (low Pd content) and Au–Pd (low Ag content) regions and highest within the quaternary composition space (Fig. 1c). We also show different representations with color-coded single element in correlation to the lattice constant from Vegard's law vs. the measured one to validate the correlation in Fig. S4.†

Evaluation of electrocatalytic activity

We show the open circuit potential (OCP) at each measurement area (Fig. 2a). Electrode materials, however, may not remain pure at the high thermodynamic ORR potential (1.23 V vs. NHE at standard conditions), but may undergo oxidation, which changes their surface properties. For example, the steady-state rest potential of a Pt electrode in O₂ saturated 1 M H₂SO₄ is 1.06 V vs. NHE, a mixed value of the thermodynamic potential of Pt oxidation (0.88 V vs. NHE) and O₂ reduction (1.23 V vs. NHE).⁴⁵ In general, it is reasonable to find an increase in the OCP with the addition of noble metals.⁴⁶ However we do not observe this in the quaternary system (Fig. 2a). For example, the OCP does not vary linearly with the Au content, showing instead a periodic trend. The peaks showed up around samples with 30, 40 and 50 at% Au display uniformly high OCP (∼0.96 V). The samples between the peaks display rather low OCP values (∼0.91 V). The OCP mapping reflects the stability of the alloys in 1 M KOH, and a higher OCP represents a higher starting point for ORR. The alloying effect may have a contribution to the trend but further studies are needed for a clear explanation.

Fig. 2 Color-coded property maps of the Pd–Au–Ag–Ti thin film library with the composition of Pd, Ag and Au measured by EDX vs. (a) OCP, (b) overpotential for ORR and (c) current extracted from chronoamperometry at 0.635 V (vs. RHE) measured by means of a SDC with an error bar of 5%.

We show the ORR overpotential potential extracted from the linear sweep voltammograms following the OCP of the quaternary thin-film library and find a volcano-like relationship between the overpotential and Au content (Fig. 2b). The 30–35 at% Au region appears to have the lowest overpotential for the ORR. Within this region, we reveal promising compositions with the lowest overpotential for the ORR, namely Pd₃₁Au₂₉Au₂₈Ti₁₂ (0.265 V), Pd₄₀Au₂₉Ag₂₀Ti₁₂ (0.267 V), Pd₃₅Au₃₀Ag₂₃Ti₁₂ (0.265 V), Pd₂₇Au₃₀Ag₃₀Ti₁₃ (0.269 V), which is very close to that of the pure Pt thin film (0.268 V, Fig. S2a† inlet) sputtered for comparison. The Pd : Au ratio is roughly 1 : 1 with one exception of Pd₄₀Au₂₉Ag₂₀Ti₁₂, which is 1.33 : 1. We note that the 1 : 1 Pd : Au ratio has been previously reported exhibiting a similar overpotential to Pt.⁶ Previously Oliveira et al. found for binary Pd–Ag electrocatalysts decreasing overpotentials with lower Ag content.⁵ We observe this same trend in the region above 50 at% Au, where the overpotential shows a decreasing trend with higher Ag content (Fig. 2b).

We reveal that from Pd 40% to Pd 52%, the overpotential increases with a decreasing current at 0.635 V vs. RHE (Fig. S2a†). However, we find a short-range order with a volcano relationship from Pd 23% to Pd 38%, and the lowest overpotential, with Pd 27%, 31%, and 38%. Fig. S2a† shows the plateau current of the compositions. Although the compositions around Au 40% did not achieve a low overpotential they all exhibit a similar current plateau of around −0.4 μA. Remarkably, in the binary Pd–Ag electrocatalyst system, there is a decreasing ORR current density with increasing Ag content,⁵ however, we notice that in the quaternary system high current densities around the 7 at% Ag region are found. This finding suggests that upon addition of Au and Ti, the relationship between current density and Ag content does not follow the same trend as shown in Pd–Ag binary system.

The current acquired with the SDC in this study is not comparable with RDE measurement since the mass transport is not controlled. We demonstrate chronoamperometry as a supplement to examine the activity for ORR (Fig. 2c). The following compositions have the highest current at 0.4 V vs. Ag/AgCl/3 M KCl: Pd₃₇Au₃₁Ag₂₀Ti₁₂ (−199.13 μA cm⁻²), Pd₃₇Au₃₉Ag₁₂Ti₁₂ (−192.15 μA cm⁻²), Pd₅₁Au₃₁Ag₇Ti₁₁ (−184.48 μA cm⁻²), and Pd₅₇Au₂₂Ag₁₀Ti₁₀ (−182.68 μA cm⁻²). Combining these observations with respect to high current density and the previous observations of low overpotentials, we suggest an ideal composition to be at about 40–60 at% Pd, 30–35 at% Au, 7–20 at% Ag and 12 at% Ti. Apparently, Au has a bigger influence on the ORR activity in this quaternary system than the other elements.

Based on this data abundance, we correlate our results from XRD and electrochemistry. Internal stress has previously been shown to drastically affect the electronic properties to the benefit of many electrochemical reactions.^47,48 We find an interesting relationship when we plot the overpotential over the (111) peak center and peak width/strain (Fig. 1d) as shown in Fig. 3. Here, a non-trivial trend is visible for a given center position of the (111) peak. The lowest overpotentials are always recorded at intermediate values for the peak width/strain. It has been shown that strain can induce significant changes in the electronic structure of catalysts,⁴⁷ which is in agreement with Fig. 3. However, there is a strong non-linearity with variation of the peak center position. The inset shows the overpotential over the peak width at a peak center position of 38.3° that hints at a volcano-like relationship. Especially when considering that the lowest overpotential is observed in the region of highest peak width/strain and in between the two quasi-equilibrium positions of the (111) peak at around 38.0° this correlation suggests further examination. We demonstrate an alternative visualization of this correlation is possible via the fact that a solid solution is present (Fig. 4). By plotting the lattice constant derived from Vegard's law and the lattice constant derived from XRD with the color-coded overpotential, we observe a similar trend that hints at a connection between strain and functional properties. We show that the lattice expands from a ‘equilibrium’ Vegard's value when moving rightward from a “Slope of 1” line (Fig. 4). For certain, small, lattice expansions the overpotential becomes higher. And similarly the peak width becomes larger for lattice contractions when moving leftward from a “Slope of 1” line. This changing of the peak broadness is shown in Fig. S6.† This finding indicates a possible route for optimization of ORR catalysts by finely adjusting the strain induced in a system through combinatorial alloying. Additionally, non-equilibrium strained alloys could be of interest as a design criterion for high-throughput computational investigations as performed at the Center for Next-Generation of Materials by Design: an Energy Frontier Research Center (CNGMD-EFRC).

Fig. 3 (111) peak width plotted over the (111) peak center as obtained from fitting a simplified Cauchy function. According to the Williams–Hall relation the width of the XRD peak at a given angle is related to the intrinsic strain in the system. In this plot, a trend is visible that for a given position of the (111) peak center the overpotential is always measured at the relative intermediate peak width. The inset shows this relation for the peak center position of about 38.3° where the relative lowest overpotential is measured at intermediate peak width/strain. For better readability outliers from fitting were removed manually.

Fig. 4 Correlation of Vegard's law and XRD derived lattice constants with color-coded overpotential potential values.

Conclusion

A Pd–Au–Ag–Ti thin film library (340 measurement areas) was investigated for potential high activity ORR candidates in the alkaline regime using a SDC. We reveal that the lowest overpotential region is shown to be around a quaternary composition region with 30 at% Au and >7 at% Ag. However, the highest current from chronoamperometry was observed to be around 40% Au with >7 at% Ag. A composition of Pd–Au–Ag–Ti of 4 : 3 : 2 : 1 is shown to have both promising overpotential and plateau current. The abundant data provided by SDC and the first study on quaternary Pd–Au–Ag–Ti thin film library could be applied in other fields of studies of electrocatalysts and provide guidance for a potential subsequent nanomaterial synthesis. A systematic volcano-like correlation between strain and overpotential is shown that could serve as a guideline for the optimization of new ORR catalysts.

Acknowledgements

The authors are grateful for support from NSF under Grant No. MRSEC DMR 1119826 (CRISP). A. D. T. also thanks the NSF-PECASE award (CBET-0954985). Facilities use was supported by YINQE, Brookhaven National Laboratory, and NSF MRSEC DMR 1119826 (CRISP). H. S. S. and K. S. acknowledge a PhD fellowship from the International Max Planck Research School for Surface and Interface Engineering (IMPRS-SurMat).

References

1. J. Sun, H. Ma, H. Jiang, L. Dang, Q. Lu and F. Gao, J. Mater. Chem. A, 2015, 3, 15882–15888.
2. M. Carmo, R. C. Sekol, S. Ding, G. Kumar, J. Schroers and A. D. Taylor, ACS Nano, 2011, 5, 2979–2983.
3. L. Xu, Z. Luo, Z. Fan, X. Zhang, C. Tan, H. Li, H. Zhang and C. Xue, Nanoscale, 2014, 6, 11738–11743.
4. J. Son, S. Cho, C. Lee, Y. Lee and J. H. Shim, Langmuir, 2014, 30, 3579–3588.
5. M. C. Oliveira, R. Rego, L. S. Fernandes and P. B. Tavares, J. Power Sources, 2011, 196, 6092–6098.
6. J.-J. Lv, S.-S. Li, A.-J. Wang, L.-P. Mei, J.-R. Chen and J.-J. Feng, Electrochim. Acta, 2014, 136, 521–528.
7. C.-L. Lee, Y.-C. Huang, L.-C. Kuo, J.-C. Oung and F.-C. Wu, Nanotechnology, 2006, 17, 2390–2395.
8. L. Kuai, B. Geng, S. Wang and Y. Sang, Chem.–Eur. J., 2012, 18, 9423–9429.
9. L. Jiang, A. Hsu, D. Chu and R. Chen, Electrochim. Acta, 2010, 55, 4506–4511.
10. G. Fu, Z. Liu, Y. Chen, J. Lin, Y. Tang and T. Lu, Nano Res., 2014, 7, 1205–1214.
11. C.-W. Chen, Y.-S. Hsieh, C.-C. Syu, H.-R. Chen and C.-L. Lee, J. Alloys Compd., 2013, 580, S359–S363.
12. C.-L. Lee, H.-P. Chiou, C.-M. Syu, C.-R. Liu, C.-C. Yang and C.-C. Syu, Int. J. Hydrogen Energy, 2011, 36, 12706–12714.
13. M. Carmo, G. Doubek, R. C. Sekol, M. Linardi and A. D. Taylor, J. Power Sources, 2013, 230, 169–175.
14. P. Quaino, F. Juarez, E. Santos and W. Schmickler, Beilstein J. Nanotechnol., 2014, 5, 846–854.
15. W. R. Fawcett, J. Am. Chem. Soc., 2000, 122, 6139.
16. J. K. Nørskov, J. Rossmeisl, A. Logadottir, L. Lindqvist, J. R. Kitchin, T. Bligaard and H. Jónsson, J. Phys. Chem. B, 2004, 108, 17886–17892.
17. R. B. Merrifield, J. Am. Chem. Soc., 1963, 85, 2149–2154.
18. H. S. Stein, R. Gutkowski, A. Siegel, W. Schuhmann and A. Ludwig, J. Mater. Chem. A, 2016, 4, 3148–3152.
19. J. A. Haber, C. Xiang, D. Guevarra, S. Jung, J. Jin and J. M. Gregoire, ChemElectroChem, 2014, 1, 524–528.
20. J. P. Kollender, A. I. Mardare and A. W. Hassel, Electrochim. Acta, 2015, 179, 32–37.
21. K. C. Neyerlin, G. Bugosh, R. Forgie, Z. Liu and P. Strasser, J. Electrochem. Soc., 2009, 156, B363.
22. R. Jiang and D. Chu, J. Electroanal. Chem., 2002, 527, 137–142.
23. D. Schaefer, C. C. Mardare, A. Savan, M. D. Sanchez, B. Mei, W. Xia, M. Muhler, A. Ludwig and W. Schuhmann, Anal. Chem., 2011, 83, 1916–1923.
24. M. L. Green, I. Takeuchi and J. R. Hattrick-Simpers, J. Appl. Phys., 2013, 113, 231101.
25. H. Stein, D. Naujoks, D. Grochla, C. Khare, R. Gutkowski, S. Grützke, W. Schuhmann and A. Ludwig, Phys. Status Solidi A, 2015, 212, 2798–2804.
26. K. Sliozberg, H. S. Stein, C. Khare, B. A. Parkinson, A. Ludwig and W. Schuhmann, ACS Appl. Mater. Interfaces, 2015, 7, 4883–4889.
27. K. Sliozberg, D. Schäfer, T. Erichsen, R. Meyer, C. Khare, A. Ludwig and W. Schuhmann, ChemSusChem, 2015, 8, 1270–1278.
28. J. Clausmeyer, J. Henig, W. Schuhmann and N. Plumere, ChemPhysChem, 2014, 15, 151–156.
29. M. Black, J. Cooper and P. McGinn, Meas. Sci. Technol., 2005, 16, 174–182.
30. R. C. Sekol, X. Li, P. Cohen, G. Doubek, M. Carmo and A. D. Taylor, Appl. Catal., B, 2013, 138–139, 285–293.
31. J. L. Fernández, V. Raghuveer, A. Manthiram and A. J. Bard, J. Am. Chem. Soc., 2005, 127, 13100–13101.
32. M. Nie, P. K. Shen and Z. Wei, J. Power Sources, 2007, 167, 69–73.
33. G. Doubek, R. C. Sekol, J. Li, W. H. Ryu, F. S. Gittleson, S. Nejati, E. Moy, C. Reid, M. Carmo, M. Linardi, P. Bordeenithikasem, E. Kinser, Y. Liu, X. Tong, C. O. Osuji, J. Schroers, S. Mukherjee and A. D. Taylor, Adv. Mater., 2016, 28, 1940–1949.
34. R. C. Sekol, M. Carmo, G. Kumar, F. Gittleson, G. Doubek, K. Sun, J. Schroers and A. D. Taylor, Int. J. Hydrogen Energy, 2013, 38, 11248–11255.
35. S. Mukherjee, R. C. Sekol, M. Carmo, E. I. Altman, A. D. Taylor and J. Schroers, Adv. Funct. Mater., 2013, 23, 2708–2713.
36. S. Mukherjee, M. Carmo, G. Kumar, R. C. Sekol, A. D. Taylor and J. Schroers, Electrochim. Acta, 2012, 74, 145–150.
37. Z. W. Lu, S. H. Wei and A. Zunger, Phys. Rev. B: Condens. Matter Mater. Phys., 1991, 44, 10470–10484.
38. R. D. Shannon, Acta Crystallogr., Sect. A: Cryst. Phys., Diffr., Theor. Gen. Crystallogr., 1976, 32, 751–767.
39. Y. Liu, J. Liu, S. Sohn, Y. Li, J. J. Cha and J. Schroers, Nat. Commun., 2015, 6, 7043.
40. H. Baker, ASM International Alloy Phase Diagram Committee and ASM International Handbook Committee, Alloy phase diagrams, ASM International, Materials Park, Ohio, 1992.
41. In Noble Metal Systems. Selected Systems from Ag–Al–Zn to Rh–Ru–Sc, ed. G. Effenberg and S. Ilyenko, Springer Berlin Heidelberg, Berlin, Heidelberg, 2006, pp. 1–7,  DOI:10.1007/10916070_6.
42. L. Vegard, Z. Phys., 1921, 5, 17–26.
43. S. Vives, E. Gaffet and C. Meunier, Mater. Sci. Eng., A, 2004, 366, 229–238.
44. A. Prince, in Ternary Evaluations, ed. G. Effenberg, MSI, Materials Science International Services GmbH, Stuttgart, 1988.
45. C. Song and J. Zhang, in PEM Fuel Cell Electrocatalysts and Catalyst Layers: Fundamentals and Applications, Springer London, London, 2008, 89–134,  DOI:10.1007/978-1-84800-936-3_2.
46. D. Wang, H. L. Xin, Y. Yu, H. Wang, E. Rus, D. A. Muller and H. D. Abruña, J. Am. Chem. Soc., 2010, 132, 17664–17666.
47. B. Yildiz, MRS Bull., 2014, 39, 147–156.
48. P. Strasser, S. Koh, T. Anniyev, J. Greeley, K. More, C. Yu, Z. Liu, S. Kaya, D. Nordlund, H. Ogasawara, M. F. Toney and A. Nilsson, Nat. Chem., 2010, 2, 454–460.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ta08088j

---