Water Oxidation Using a Cobalt Monolayer Prepared by

Article pubs.acs.org/Langmuir

Water Oxidation Using a Cobalt Monolayer Prepared by Underpotential Deposition David A. Marsh, Wenbo Yan, Yu Liu, John C. Hemminger, Reginald M. Penner,* and A.S. Borovik* Department of Chemistry, University of California-Irvine, 1102 Natural Sciences II, Irvine, California 92697, United States S Supporting Information *

ABSTRACT: Development of electrocatalysts for the conversion of water to dioxygen is important in a variety of chemical applications. Despite much research in this field, there are still several fundamental issues about the electrocatalysts that need to be resolved. Two such problems are that the catalyst mass loading on the electrode is subject to large uncertainties and the wetted surface area of the catalyst is often unknown and difficult to determine. To address these topics, a cobalt monolayer was prepared on a gold electrode by underpotential deposition and used to probe its efficiency for the oxidation of water. This electrocatalyst was characterized by atomic force microscopy, grazing-incidence Xray diffraction, and X-ray photoelectron spectroscopy at various potentials to determine if changes occur on the surface during catalysis. An enhancement of current was observed upon addition of PO43− ions, suggesting an effect from surface-bound ligands on the efficiency of water oxidation. At 500 mV overpotential, current densities of 0.20, 0.74, and 2.4 mA/cm2 for gold, cobalt, and cobalt in PO43− were observed. This approach thus provided electrocatalysts whose surface areas and activity can be accurately determined.

■

INTRODUCTION The development of electrocatalysts for water oxidation has attracted much attention in recent years, predominately because of the importance of this transformation in energy science. Systems based on cobalt have been shown to be particularly adept at this conversion and have been used extensively as electrocatalysts.1−3 In most prior work, the catalyst has been electrodeposited onto catalytically inefficient electrode surfaces.4,5 In such cases, the mass loading of the catalyst can be accurately measured using coulometry, but the surface area of the catalyst-covered electrode is subject to considerable uncertainty. It is difficult to determine the catalytically active surface area for cobalt atoms, and its oxides and hydroxides, because of uncertainty arising from roughening that occurs during the electrodeposition process.6−8 Double-layer capacitance measurements are often used to determine surface area but are known to significantly underestimate the actual surface area of a material, causing systematically larger current densities and results in higher apparent activities for the catalyst on the electrode surface.7,9 The problems with determining surface areas have been mitigated somewhat by using ultrathin films, which minimize the inaccuracy in using geometric area.10 However, the unavoidable uncertainty in the determination of the surface area of a catalyst-covered electrode renders the measurement of the catalyst activity inaccurate. The problems associated with surface area of the electrocatalyst can be circumvented by electrodepositing a single atomic layer of cobalt onto a gold surface that has an accurately known area. This process, called underpotential deposition (UPD), produces a conformal catalyst monolayer that retains the surface area of the parent gold surface. Because the surface © 2013 American Chemical Society

area of the gold electrode can be accurately measured by electrochemically forming and removing a single oxide monolayer (vida supra), the surface area of the cobalt monolayer can also be determined with a high degree of accuracy. To gain greater fundamental insight into the role of cobalt electrocatalysts in water oxidation, we have used UPD to prepare a single monolayer of cobalt on a gold substrate (denoted Co ML/Au). We reasoned that a single monolayer of catalyst would allow for complete characterization of the active species, including an accurate determination of mass loading and surface area of the electrocatalyst. This information enabled us to accurately determine the efficiency for water oxidation and to evaluate the speciation of the cobalt electrocatalyst during the oxidation process. Our investigation demonstrated that a single monolayer of cobalt can perform water oxidation at modest overpotential, and this oxidation is enhanced by the presence of phosphate ions.

■

EXPERIMENTAL SECTION

Physical Methods. All chemicals were purchased from SigmaAldrich and used as received unless otherwise noted. All electrochemical measurements were performed in a one-compartment threeelectrode cell using a Gamry Instruments Series-G 300 potentiostat. Pt mesh was used as a counter electrode, and saturated mercurous sulfate (MSE) was used as a reference electrode (E(MSE) = +0.640 V vs SHE). Grazing-incidence X-ray diffraction (GIXRD) involving the collection of in-plane and out-of-plane X-ray diffraction patterns was accomplished at room temperature using a Rigaku Smartlab X-ray Received: August 7, 2013 Revised: October 14, 2013 Published: October 18, 2013 14728

dx.doi.org/10.1021/la403038e | Langmuir 2013, 29, 14728−14732

Langmuir

Article

diffractometer with a combination of an X-ray source (Cu Kα = 0.154 nm, voltage/current, 40 kV/44 mA) and a parabolic multilayered mirror with a fixed incident angle of 0.5°. Atomic force microscopy (AFM) images and calculations were acquired using an Asylum Research MFP-3DTM AFM equipped with Olympus, AC160TS tips in laboratory air. For AFM, XRD, and X-ray photoelectron spectroscopy (XPS) measurements, gold films of 40 nm in thickness were thermally evaporated onto precleaned 1″ × 1″ squares of soda lime glass. A 1−2 nm thick chromium layer was deposited first to make better adhesion of gold on glass. XPS was accomplished using an ESCALAB MKII surface analysis instrument (VG Scientific) equipped with a twin anode X-ray source (Mg/Al) and a 150 mm hemispherical electron energy analyzer. XPS spectra were collected using a Al Kα source at 1486.6 eV in constant energy mode and a pass energy of 20 eV. All XP spectra were acquired at a base pressure of 5 × 10−10 Torr. We calibrate the binding energy of all spectra by Au 4f7/2 at 84.0 eV from gold substrate. Similar results are also obtained by calibration using C 1s peak of adventitious carbon set at 284.8 eV and Ag 3d of silver paint set at 386.3 eV. Using XPSPEAK software with Shirley-type background functions, we fit the Co 2p XP spectra with spin−orbit doublets (2p3/2 and 2p1/2) at a fixed intensity ratio (2:1). Measurement of Gold Surface Area. A stationary gold disk electrode (Ageo = 0.0314 cm2) was polished with diamond paste of sizes 1, 0.5, and then 0.25 μm, and sonicated in Milli-Q deionized water (2 × 30 s). The gold electrode was first cycled at 20 mV/s in 0.1 M H2SO4 (99.999%) for ≈15 cycles. The area of the gold surface was calculated by integrating the gold oxide reduction wave at ≈0.5 V vs MSE using the conversion 482 μC/cm2.11 This electrode was used immediately to prepare the Co monolayer. Electrodeposition of a Cobalt Monolayer on Gold (Co ML/ Au). Cobalt underpotential deposition followed the procedure of Mendoza-Huizar.12 Briefly, a solution containing 20 mM CoCl2 and 1 M NH4Cl was adjusted to pH 9.5 using concentrated KOH solution. Co was deposited by linear sweep voltammetric scan from 0 to −1.1 V vs MSE at a scan rate of 50 mV/s. The electrode was equilibrated at open circuit in this plating solution, removed, and rinsed with ultrapure H2O and then used immediately for water oxidation experiments. Water Oxidation. In 0.1 M KOH (pH 13), cyclic voltammograms for bare gold and cobalt UPD-modified gold were recorded in the range of −0.4 to 0.5 V vs MSE with a scan rate of 2 mV/s. For experiments with phosphate, a solution containing 0.1 M KOH and 0.3 M K3PO4 was prepared and adjusted to pH 13 using dilute H3PO4. Only the first voltammogram was used, to ensure that the surface area is unaltered, and the data shown is the average of three independent experiments. However, the Co ML/Au electrocatalyst was stable over 10 CV cycles and was active toward water oxidation for several hours (see Supporting Information for details).

roughness factor, r (r = Areal/Ageo), ranging from 1.2 to 2 was calculated. We shall demonstrate that the roughness factor estimated by AFM is unaffected by catalyst deposition because a single conformal catalyst monolayer is electrodeposited in this study. Thus, we have a direct and accurate measure of the electrode area for the catalyst-coated electrodea critically important parameter that has, to our knowledge, been missing in prior studies. Cobalt UPD on gold was characterized by two voltammetric waves at −0.34 and −0.59 V vs MSE (Figure 1). Bulk gold electrodes and gold electrodes prepared on chromium-coated glass showed identical voltammetry. The assignment of these two peaks is unresolved, but we speculate that the more positive wave, −0.34 V, could result from deposition on defects and step edges, whereas the more negative wave at −0.59 V correlates with deposition on terraces.14 The total charge associated with the two peaks, 0.413 ± 0.18 mC/cm2, is reproducible and correlates with the electrodeposition of ∼1 monolayer of cobalt.12 More precisely, 1.16 ± 0.5 Co MLs were deposited by this process. The presence on the gold surface of monolayer coverages of Co2+ and/or Co3+ oxides is also indicated by XPS, which we discuss separately below and in the Supporting Information. Characterization of Co ML/Au. The topography of the cobalt monolayer was determined using atomic force microscopy (AFM). AFM was also used to determine if the surface roughness was altered by the formation of Co ML/Au. The chemical state of the deposited cobalt was probed using GIXRD in combination with XPS. AFM images recorded before and after UPD of cobalt (Figure 2) show that the surface roughness of the gold electrode is not altered by cobalt deposition. Specifically, for a cleaned gold electrode, a root mean squared (RMS) roughness of 1.44 ± 0.06 nm was measured, whereas the RMS roughness of the same surface was 1.51 ± 0.17 nm after cobalt deposition. These results demonstrate that the electrode area determined for the gold electrode is also valid for the Co ML/Au surface. Grazing incidence X-ray diffraction patterns were acquired for Co ML/Au using both out-of-plane detection (GIXRD) and in-plane detection of the diffracted beam. The latter mode, called grazing-incidence in-plane X-ray diffraction (GIIXRD), has two advantages: (1) higher sensitivity for monolayers beyond that available using GIXRD because the X-ray beam path length within the monolayer is extended to more than 1 mm, and (2) the scattering vector is rotated into the plane of the surface, providing information on the in-plane periodicities present in the monolayer, rather than the out-of-place structure probed in a conventional GIXRD experiment. XRD patterns for both modes are dominated by reflections from the gold substrate at 2θ values of 38.2°, 44.3°, and 64.6° (Au, JCPDS #01-071-4073). In addition to the intense reflections from gold, GIIXRD patterns also reveal the presence of two cubic CoO structures with reflections at 36.6°, 42.6°, 61.5°, and 73.8° (JCPDS #01-75-0418) and weaker reflections at 34.3° and 57.5° (JCPDS #00-42-1300) (Figure 3). The formation of CoO occurs presumably because the experiments were done under aerobic conditions, which allow for oxidation of cobalt metal to Co(II) ions. Further evidence for a monolayer can be obtained by comparing the diffraction patterns detected from in-plane measurements and out-of-plane measurements. These reflections are not observed in the GIXRD patterns, indicating either that there is insufficient X-ray path length in this configuration to provide measurement X-ray signal, or that the periodicities

■

RESULTS AND DISCUSSION Preparation of Electrocatalysts. The process for preparing Co ML/Au is shown in Scheme 1. The surface Scheme 1. Method for Preparation of a Cobalt Monolayer on a Gold Electrode (Co ML/Au)

area of the gold substrate was determined by standard methods using the integrated charge of the cathodic wave for gold oxide reduction to gold metal.13 This is known to produce 482 μC/ cm2 of charge for polycrystalline gold, and this value can be used to determine the active surface area of the substrate, Areal.11 Using the geometric surface area of the electrode, Ageo, a 14729

dx.doi.org/10.1021/la403038e | Langmuir 2013, 29, 14728−14732

Langmuir

Article

Figure 1. (a) Cyclic voltammogram of a gold electrode in x M CoCl2 + 1 M NH4Cl at pH 9.5, x = 0 (black) and x = 0.02 (blue), 50 mV/s scan rate. (b) Expanded view of the peaks associated with underpotential deposition (b).

Figure 2. AFM images of a gold electrode after cleaning in H2SO4 (a) and after UPD of cobalt (b). Root mean squared roughness was 1.444 ± 0.06 nm (a) and 1.514 ± 0.17 nm (b).

mV (E° = −0.18 V vs MSE). For the gold electrode, a maximum current of 0.8 mA/cm2 was achieved at 640 mV overpotential; 1 mA/cm2 was not achievable at any overpotential. Our results indicate that by adding a monolayer of Co on Au, the water oxidation activity is enhanced by approximately three times in the potential range studied. Other cobalt-based thin films have been reported with current densities of 1 mA/cm2 at an overpotential of ∼390 mV.5,10 However, these current density values were based on estimated surface areas, either geometric areas or calculated from doublelayer capacitance measurements. Both of these methods will underestimate the active surface area and thus overestimate the current density. The influence of anions on electrocatalysis was also investigated. There was no dependence of the catalysis on anions including sulfate, carbonate, triflate, or borate; however, the rate of water oxidation was accelerated in the presence of phosphate ions. Specifically, in the presence of 0.3 M K3PO4 the overpotential associated with a current density of 1 mA/ cm2 was reduced from 510 to 450 mV. These observations are consistent with the findings of Nocera and Kanan, who found that a molecular cobalt species on an electrode surface in phosphate buffer (pH ∼ 7) also catalyzed water oxidation, achieving 1 mA/cm2 at an overpotential of 410 mV.4 Turnover frequency (TOF) facilitates a direct comparison of the activity of Co ML/Au with other cobalt-based electrocatalysts. As in previous work,15 a cobalt atom density of 1.112 × 1015 Co atoms/cm2equivalent to that of the [100] facet of cubic CoOwas used for the Co ML/Au surface. The TOF is

Figure 3. X-ray diffraction pattern of Co ML/Au, showing out-ofplane geometry (black) and in-plane geometry (blue).

corresponding to the reflections seen in the GIIXRD pattern do not exist along the surface normal. This result is qualitatively consistent with a monolayer of cobalt being deposited on the gold surface because only a single such layer will possess inplane ordering and a GIIXRD pattern with no out-of-plane diffraction probed by GIXRD. Water Oxidation. The activity of a Co ML/Au electrode for the oxidation of water was compared with gold in 0.1 M KOH solution (Figure 4). Because the atomic density of Co ML/Au is nearly identical to the gold electrode, the ratio of current density at every potential provides a direct comparison of the catalytic activity of the two surfaces. After deposition of Co ML/Au, an enhancement in the current for water oxidation was observed, reaching 1 mA/cm2 at an overpotential of 510 14730

dx.doi.org/10.1021/la403038e | Langmuir 2013, 29, 14728−14732

Langmuir

Article

Figure 4. (a) Linear sweep voltammograms at 2 mV/s showing water oxidation for a gold electrode (black), Co ML/Au in 0.1 M KOH (blue), and Co ML/Au in 0.1 M KOH + 0.3 M K3PO4 (red), with corresponding Tafel plots (b).

Table 1. Summary of Electrochemical Data for Water Oxidation current density (mA/cm2) at η = 500 mV TOF (mol O2/cm2·s) at η = 500 mV TOF (mol O2 /mol Co·s) at η = 500 mV Tafel slope (mV/dec)

Au 0.1 M KOH

Co ML/Au 0.1 M KOH

Co ML/Au 0.1 M KOH + 0.3 M K3PO4

0.20 ± 0.08 5.12 × 10−10 ± 1.9 × 10−10 0.28 ± 0.10 72 ± 7

0.74 ± 0.4 1.92 × 109 ± 9.7 × 109 1.0 ± 0.53 68 ± 8

2.4 ± 0.4 6.29 × 109 ± 1.1 × 109 3.4 ± 0.62 55 ± 5

Figure 5. (a) X-ray photoelectron spectra of Co ML/Au at various potentials (a) with the fitting for the Co 2p main peaks (blue), the Co 2p satellites (red), and the total fit (black). (b) From the ratio of the satellite/main peak, the ratio of Co2+/Co3+ was determined.18

ML/Au was very similar to the pure gold electrode, ∼70 mV/ dec. This was in the same range as spinel-based cobalt oxide electrocatalysts,2 suggesting a similar mechanism for water oxidation is involved. The Tafel slope for Co ML/Au in the presence of PO43− ions is lower, 55 mV/dec, corresponding to a higher activity for water oxidation. In prior work,16,17 the theoretical Tafel slope of 2RT/F (∼50 mV/dec) has been attributed to two possible rate-determining steps:

then defined as the number of moles of O2 produced per mole of cobalt per second. This results in TOF values of 0.28, 1.0, and 3.4 s−1 at 500 mV overpotential for Au, Co ML/Au, and Co ML/Au in the presence of PO43− ions, respectively. These values are lower than those calculated from other systems at 500 mV overpotential.5,10 The lower TOF values we report here are not necessarily more accurate than the estimates provided in prior studies. However, it is important to point out that the accuracy of the TOF value is no better than the accuracy of the surface area measurement for the catalytically active catalyst surface. This is where the approach we describe here has an advantage over those employed in previous studies. The Tafel slope confers information about the mechanism of the water oxidation process. The Tafel slope determined for Co

CoIII−OH → CoIV −OH + e−

(1)

CoIII−OH + OH− → CoIV −O + H 2O + e−

(2)

Reactions 1 and 2 are differentiated by the participation of OH− ions in (2), yet both reaction sequences have been implicated 14731

dx.doi.org/10.1021/la403038e | Langmuir 2013, 29, 14728−14732

Langmuir

■

as possibilities to afford the observed Tafel slope. However, the presence of high valent (IV, V, VI, etc.) cobalt centers within the catalyst, as proposed previously, is not supported by the ex situ XPS analysis of the Co ML/Au surface, as discussed in greater detail below. Characterization of the Electrocatalyst. X-ray photoelectron spectroscopy (XPS) was performed to determine the oxidation state of Co ML/Au at different potentials (Figure 5). The ratio of the intensity of the Co 2p photoelectron peak to its satellite peak can be used to estimate the oxidation state of cobalt.18 Using this approach, we observed only Co(II) ions in the initially prepared Co ML/Au, corroborating the in-plane GIXRD data (Figure 3) showing the presence of cubic CoO within the catalyst layer. XPS was also used to evaluate the chemical state of the Co ML/Au catalyst after water oxidation was carried out at selected potentials (Table S1 in the Supporting Information). In these experiments, the current was permitted to stabilize for 30 min, and the electrode was rinsed and rapidly transferred to the vacuum chamber for immediate XPS analysis. These data show that the fraction of Co(III) centers within the catalyst layer increases as the electrolysis is carried out at more positive potentials. For instance, whereas 20% of the Co centers have been converted to Co(III) at +0.3 V vs MSE, 60% are Co(III) at +0.5 V vs MSE. Accurate determination of the oxidation state of the electrocatalyst is important because some studies have implicated Co(IV) as an intermediate in cobalt-based systems that are able to oxidize water.16,19,20 In one case, an S = 1/2 signal was observed in small quantities by EPR spectroscopy, which was attributed to low-spin Co(IV).21 Even at high potentials in our system, we have no evidence for Co(IV) by XPS, however there has not been previous detection of Co(IV) by XPS in water oxidation catalysts. The XPS measurements indicate that the active catalyst in Co ML/Au is composed of a mixture of Co(II)/Co(III), with Co(III) centers likely to be involved in water oxidation since Co(III) is only observed at potentials where water oxidation occurs.

Article

ASSOCIATED CONTENT

S Supporting Information *

This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]; [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS XRD was performed at the Laboratory for Electron and X-ray Instrumentation (LEXI) at UC Irvine. A.S.B. and D.A.M. acknowledge UC Irvine for financial support. R.M.P. and W.Y. acknowledge financial support as part of the Nanostructures for Electrical Energy Storage, an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under Award #DESC0001160. Y.L. and J.C.H. acknowledge support from the U.S. Department of Energy, Office of Basic Energy Sciences, Division of Materials Sciences and Engineering under Award #DE-FG0296ER45576 to carry out the XPS analysis.

■

REFERENCES

(1) Artero, V.; Chavarot-Kerlidou, M.; Fontecave, M. Angew. Chem., Int. Ed. 2011, 50, 7238−7266. (2) Hamdani, M.; Singh, R. N.; Chartier, P. Int. J. Electrochem. Sci. 2010, 5, 556−577. (3) Lyons, M. E. G.; Brandon, M. P. Int. J. Electrochem. Sci. 2008, 3, 1425−1462. (4) Kanan, M. W.; Nocera, D. G. Science 2008, 321, 1072−1075. (5) Yeo, B. S.; Bell, A. T. J. Am. Chem. Soc. 2011, 133, 5587−5593. (6) Bicelli, L. P.; Bozzini, B.; Mele, C.; D’Urzo, L. Int. J. Electrochem. Sci. 2008, 3, 356−408. (7) Trasatti, S.; Petrii, O. A. Pure Appl. Chem. 1991, 63, 711−734. (8) Jarzabek, G.; Borkowska, Z. Electrochim. Acta 1997, 42, 2915− 2918. (9) Trasatti, S.; Petrii, O. A. J. Electroanal. Chem. 1992, 327, 353− 376. (10) Trotochaud, L.; Ranney, J. K.; Williams, K. N.; Boettcher, S. W. J. Am. Chem. Soc. 2012, 134, 17253−17261. (11) Hoogvliet, J. C.; Dijksma, M.; Kamp, B.; Van, B. W. P. Anal. Chem. 2000, 72, 2016−2021. (12) Mendoza-Huizar, L. H.; Robles, J.; Palomar-Pardave, M. J. Electroanal. Chem. Interfacial Electrochem. 2002, 521, 95−106. (13) Xiang, C.; Güell, A. G.; Brown, M. A.; Kim, J. Y.; Hemminger, J. C.; Penner, R. M. Nano Lett. 2008, 8, 3017−3022. (14) Flis-kabulska, I. J. Appl. Electrochem. 2005, 36, 131−137. (15) Chou, N. H.; Ross, P. N.; Bell, A. T.; Tilley, T. D. ChemSusChem 2011, 4, 1566−1569. (16) Bockris, J. O.; Otagawa, T. J. Phys. Chem. 1983, 87, 2960−2971. (17) Singh, R. N.; Hamdani, M.; Koenig, J.-F.; Poillerat, G.; Gautier, J. L.; Chartier, P. J. Appl. Electrochem. 1990, 20, 442−446. (18) Guse, K.; Papp, H. Fresenius J. Anal. Chem. 1993, 346, 84−91. (19) Lyons, M. E. G.; Brandon, M. P. J. Electroanal. Chem. 2010, 641, 119−130. (20) Boggio, R.; Carugati, A.; Trasatti, S. J. Appl. Electrochem. 1987, 17, 828−840. (21) McAlpin, J. G.; Surendranath, Y.; Dincǎ, M.; Stich, T. A.; Stoian, S. A.; Casey, W. H.; Nocera, D. G.; Britt, R. D. J. Am. Chem. Soc. 2010, 132, 6882−6883.

■

CONCLUSION Underpotential deposition provides a route to prepare electrocatalysts that consist of approximately one monolayer of a transition metal onto a noble metal substrate, which is gold in our case. One advantage of this method is that the active surface area of the electrocatalyst can be considered the same as its parent substrate, whose surface area can be accurately determined based on its known electrochemical redox reactions. Our results demonstrate that an electrocatalyst composed of a monolayer of Co loaded onto a Au electrode can increase the current density for water oxidation by a factor of 3 over the bare gold electrode. An additional factor of 3 increase in current density was observed in the presence of phosphate ions, suggesting that other species can affect the catalytic performance. Because the surface areas were determined to a high degree of accuracy, the enhancement in current density associated with water oxidation is directly correlated to the presence of cobalt within the electrocatalyst. This method could be extended to other transition metals that are compatible with UPD, such as nickel and copper, to directly compare their catalytic efficiencies without discrepancies caused by surface area effects. 14732

dx.doi.org/10.1021/la403038e | Langmuir 2013, 29, 14728−14732