Langmuir 2007, 23, 9911-9918
9911
Mechanism of Electrochemical Reduction of Hydrogen Peroxide on Copper in Acidic Sulfate Solutions Karen L. Stewart and Andrew A. Gewirth* Department of Chemistry, UniVersity of Illinois at Urbana-Champaign, Urbana, Illinois 61801 ReceiVed May 10, 2007. In Final Form: June 27, 2007 Hydrogen peroxide is a commonly used oxidizer component in chemical mechanical planarization slurries, used in the processing of Cu metallization in microelectronics applications. We studied the electrochemical reduction of hydrogen peroxide on Cu in 0.1 M H2SO4 solutions using methods including cyclic voltammetry, rotating disk electrode experiments, surface-enhanced Raman spectroscopy, and density functional theory (DFT) calculations. The spectroscopy reveals that the hydrogen peroxide molecule is reduced at negative potentials to form a Cu-OH surface species in acidic solutions, a result consistent with the insight from Tafel slope measurements. DFT calculations support the instability of peroxide relative to the surface-coordinated hydroxide on both Cu(111) and Cu(100) surfaces.
1. Introduction The interaction of hydrogen peroxide with metal surfaces is significant because of the role of H2O2 in processes as diverse as oxygen reduction, corrosion, and chemical mechanical planarization (CMP). In the CMP process, H2O2 is commonly used as an oxidizer in polishing slurries.1-4 During the CMP process, the copper surface is oxidized, and the subsequent removal of the surface layer is facilitated by abrasive particles.4 In the alkaline electrochemical environment, Cu(I) formation occurring as a result of surface exposure to an oxidizer is anticipated, but such is not the case in the low pH environment, where some CMP slurries operate. In the low pH CMP environment, sulfuric acid is the electrolyte of choice in nearly all applications.4 In the presence of glycine, peroxide is thought to decompose to form OH radical species, which have been indirectly monitored.3 However, the formation of a Cu(I) intermediate following peroxide exposure at low pH has never been observed. Additionally, H2O2 interaction with surfaces plays a role in the oxygen reduction process.5 On surfaces where the process has been evaluated, O2 reduction is thought to occur via a series mechanism, wherein the formation of superoxide and peroxide species should occur.5,6 Here,
O2 + e- + H+ f HOO•
(1)
HOO• + e- + H+ f H2O2
(2)
H2O2 + 2H+ + 2e- f 2 H2O
(3)
Reaction 1 is thought to be the rate-determining step (rds) in the process. * Author to whom correspondence should be addressed. E-mail:
[email protected]; phone: 217-333-8329; fax: 217-244-3186. (1) Deshpande, S.; Kuiry, S. C.; Klimov, M.; Obeng, Y.; Seal, S. J. Electrochem. Soc. 2004, 151, G788-G794. (2) Chang, S.-C.; Shieh, J.-M.; Lin, K.-C.; Dai, B.-T.; Wang, T.-C.; Chen, C.-F.; Feng, M.-S.; Li, Y.-H.; Lu, C.-P. J. Vac. Sci. Technol. B 2002, 20, 13111316. (3) Hariharaputhiran, M.; Zhang, J.; Ramarajan, S.; Keleher, J. J.; Li, Y.; Babu, S. V. J. Electrochem. Soc. 2000, 147, 3820-3826. (4) Oliver, M. R. Chemical-Mechanical Planarization of Semiconductor Materials; Springer: Berlin, 2004. (5) Adzic, R. Recent advances in the kinetics of oxygen reduction. In Electrocatalysis; Lipkowski, J., Ross, P. N., Eds.; Wiley-VCH, Inc.: New York, 1998; pp 197-242. (6) Li, X.; Gewirth, A. A. J. Am. Chem. Soc. 2005, 127, 5252-5260.
Thus, a detailed understanding of O2 reduction focuses not only on the interaction of O2 with electrode surfaces, but also on the role(s) of peroxide and superoxide intermediates and the mechanism of their reduction. A focus on peroxide is required, in particular, because peroxide is often found as an unwanted byproduct on surfaces that otherwise function as four electron O2 reducers. For example, peroxide formation on Pt, especially at potentials where underpotentially deposited (upd) hydrogen forms on the Pt surface, has long been observed.7 Understanding conditions for facile peroxide reduction activity can help in the search for better oxygen reduction catalysts. Finally, peroxide interaction with metal surfaces is important in many corrosion schemes. For example, peroxide produced during the radiolytic decomposition of water plays an important role in the corrosion of U, Cu, and related materials.8 There have been several studies of hydrogen peroxide interaction with Cu in high and intermediate pH solutions.9,10 In borax buffer (pH ) 9.2), the two electron reduction of peroxide is facilitated by a catalytic mechanism involving the Cu2O/CuO couple.9 Cu2O is reoxidized to CuO by peroxide at a potential where Cu2O is readily reduced. In Cl- solutions, Cu2O at the surface is dissolved and the redox catalytic mechanism is not observed.9 On Cu surfaces in solutions of intermediate pH, both Cu(I) and Cu(II) species are observed in ex situ measurements at relatively short times following peroxide introduction, but only Cu(II) is found at longer times.9 Surface oxides also play a role in the reduction of O2 on Cu in neutral media.11 Only a few studies have addressed oxygen and peroxide reduction in the acid electrochemical environment.12-14 Brisard and co-workers found that the oxygen reduction reaction in sulfuric acid is structure-sensitive, the origin of which was (7) Markovic, N. M.; Ross, P. N. Surf. Sci. Rep. 2002, 45, 117-229. (8) Shoesmith, D. W.; Hocking, W. H.; Ikeda, B. M.; King, F.; Noel, J. J.; Sunder, S. Can. J. Chem. 1997, 75, 1566-1584. (9) Vazquez, M. V.; Desanchez, S. R.; Calvo, E. J.; Schiffrin, D. J. J. Electroanal. Chem. 1994, 374, 179-187. (10) DeNardis, D.; Rosales-Yeomans, D.; Borucki, L.; Philipossian, A. Thin Solid Films 2006, 513, 311-318. (11) King, F.; Litke, C. D.; Tang, Y. J. Electroanal. Chem. 1995, 384, 105113. (12) Andersen, T. N.; Ghandehari, M. H.; Eyring, H. J. Electrochem. Soc. 1975, 122, 1580-1585. (13) Ghandehari, M. H.; Andersen, T. N.; Eyring, H. Corros. Sci. 1976, 16, 123-135. (14) Kokkinidis, G.; Jannakoudakis, D. J. Electroanal. Chem. Interfacial Electrochem. 1984, 162, 163-173.
10.1021/la7013557 CCC: $37.00 © 2007 American Chemical Society Published on Web 08/14/2007
9912 Langmuir, Vol. 23, No. 19, 2007
Stewart and Gewirth
suggested to be the relative strength of the sulfate-Cu interaction between different Cu faces.15 Additional work addressed the interaction of peroxide and oxygen with Cu in the presence of substantial HCl16 or Ce.17 None of these reports provide definitive identification of any intermediates formed during the reduction process. In this paper, we develop a model for the mechanism of peroxide reduction in an acidic environment using data from electrochemical measurements, surface-enhanced Raman spectroscopy (SERS), and density functional theory (DFT) calculations. 2. Experimental All solutions were made with high purity H2SO4 and H2O2 (J.T. Baker Ultrex II), high purity KCl (99.999%, Aldrich) and Millipore water (Milli-Q UV plus, Millipore, Inc., 18.2 MΩ cm). Cyclic voltammograms were obtained using a two-compartment glass electrochemical cell used with a Ag/AgCl reference electrode connected via a capillary salt bridge. All potentials are reported vs Ag/AgCl. The working electrode was either a polycrystalline, (111), or (100) Cu disk (Monocrystals, Inc.) with a diameter of 10 cm. Before each experiment, the crystal was mechanically polished to a level of 0.25 µm grit size diamond suspension (Metadi Supreme Diamond Suspension, Buehler). After polishing, Cu single crystals were suspended in a solution of phosphoric acid, sulfuric acid, water, (6.5:1:3 by volume), and copper ion and were electropolished at 2 V versus a Cu counter/reference electrode for approximately 2 min and rinsed with Millipore water. The copper electrode was placed in meniscus contact with the solution. Solutions were deoxygenated with Ar and maintained under Ar atmosphere throughout the experiments. Rotating disk electrode (RDE) data were obtained using a Pine model MSRX rotator equipped with a Kel-F collet to hold the copper crystal. Before each SERS experiment, the mechanically polished copper was electrochemically roughened in 0.1 M KCl solution as described previously.18 SERS experiments were done at room temperature using an in situ cell described previously.19 The electrochemical sample cell, based on designs in the literature, was made of Kel-F and glass. The Ag/AgCl reference electrode was connected to the cell via a capillary bridge, and the Au counter electrode was placed in the cell through a Teflon fitting. A quartz window was attached to the cell by a metal holder and sealed with an O-ring. The copper crystal working electrode was held onto a glass plunger with Teflon tape and pushed in close proximity to the window. A HeNe laser (632.8 nm) for Raman excitation was projected onto the sample at 45°. The system was allowed to equilibrate for 2 min at each potential before spectra were acquired. A 1200 grooves/nm grating dispersed radiation onto a cooled charge-coupled device (Andor). The typical acquisition time for SERS was 2 min. Calculations on periodic structures were carried out using the Cambridge serial total energy package (CASTEP) in MSICerius2.20,21 DFT calculations with the generalized gradient approximation with the functional of Perdew and Wang (GGA-PW91) were performed.22 Ultrasoft pseudopotentials were used to describe the electron-core interactions of Cu, O, and H. Valence states include (15) Brisard, G.; Bertrand, N.; Ross, P. N.; Markovic, N. M. J. Electroanal. Chem. 2000, 480, 219-224. (16) Smyrl, W. H.; Bell, B. T.; Atanasoski, R. T.; Glass, R. S. Mater. Res. Soc. Symp. Proc. 1987, 84, 591-601. (17) Molodov, A. I.; Markosyan, A. I.; Losev, V. V. Elektrokhimiya 1981, 17, 1131-1140. (18) Feng, Z. V.; Li, X.; Gewirth, A. A. J. Phys. Chem. B 2003, 107, 94159423. (19) Biggin, M. E. Ph.D. Thesis, University of Illinois at Urbana-Champaign, 2001. (20) Milman, V.; Winkler, B.; White, J. A.; Pickard, C. J.; Payne, M. C.; Akhmatskaya, E. V.; Nobes, R. H. Int. J. Quantum Chem. 2000, 77, 895-910. (21) Cerius2 Software, release 4.10; Molecular Simulations, Inc.: San Diego, CA, 2005. (22) Perdew, J. P.; Chevary, J. A.; Vosko, S. H.; Jackson, K. A.; Pederson, M. R.; Singh, D. J.; Fiolhais, C. Phys. ReV. B 1992, 46, 6671-6687.
Figure 1. Cyclic voltammograms obtained from Cu(poly) in a solution of 0.1 M H2SO4 (solid line), and Cu(poly) (dashed line), Cu(100) (dot-dashed line), and Cu(111) (dotted line) in 0.1 M H2SO4 + 60 mM H2O2 with a scan rate of 20 mV/s.
the 3d and 4s shells for Cu, 2s and 2p for O, and 1s for H. The electronic wave functions were expanded in a plane wave basis set with an energy cutoff of 400 eV. For the total energy calculations, a Monkhorst-Pack k-point sampling scheme was used with 25 k-points for Cu(111) and 16 k-points for Cu(100) supercells. In this work, Cu(111) and Cu(100) surfaces were modeled by a (2 × 2) unit cell consisting of three layers. The vacuum region between slabs was 10 Å. In all cases, surface modifications were applied only to one face of the slab. For H2O2 adsorption on the Cu surfaces, the topmost copper layer was allowed to relax during the adsorption process, whereas the other atoms in the bottom two layers were constrained in their position to mimic the bulk crystal. Calculations were performed using an SGI Origin 2000 computer within the School of Chemical Sciences at the University of Illinois.
3. Results 3.1. Electrochemical Behavior. 3.1.1. Cyclic Voltammetry. Figure 1 shows the cyclic voltammograms for Cu(poly) in Arpurged 0.1 M H2SO4 (solid line), and Cu(111) (dotted line), Cu(100) (dot-dashed line), and Cu(poly) (dashed line) in a solution of 0.1 M H2SO4 + 60 mM H2O2. In sulfuric acid without H2O2, the current is very low and increases on the anodic scan at ca. -0.1 V, where oxidation of the Cu surface begins. The cathodic current at negative potentials is likely due to the onset of hydrogen evolution. With the addition of H2O2, a large reduction current is observed, commencing at -0.10 V, which was the anodic limit of the scan. The current density from the Cu(111) surface was approximately a factor of 2 higher relative to either the Cu(100) or Cu(poly) surfaces. However, the origin of the increased current density cannot be attributed to changes in surface roughness for the different surfaces studied. In order to compare surface areas between Cu(poly), Cu(100), and Cu(111), we performed Pb underpotential deposition (upd) on the different surfaces. From coulometry associated with the Pb upd, and by assuming the Pb upd process consumes two electrons per Pb atom,23 we obtain the effective surface area. By using this method, we found that the effective area for our Cu(poly) sample was only 2.6% larger than that found for either Cu(100) or Cu(111). Additionally, the potential of maximum current on the Cu(111) face was shifted to more cathodic values relative to either the polycrystalline or (100) faces. Subsequent analyses (23) Brisard, G. M.; Zenati, E.; Gasteiger, H. A.; Markovic, N. M.; Ross, P. N. Langmuir 1995, 11, 2221-2230.
Electrochemical Reduction of H2O2 on Cu
Langmuir, Vol. 23, No. 19, 2007 9913
Figure 2. RDE voltammograms obtained from Cu(poly) + 0.1 M H2SO4 + 60 mM H2O2 in Ar-purged atmosphere. The rotation rate was varied from 100 to 2500 rpm, and the scan rate was 20 mV/s. The inset shows Koutecky-Levich plots obtained at different electrode potentials.
Figure 3. RDE voltammograms obtained from Cu(poly) + 0.1 M H2SO4 + 60 mM H2O2 + 1 mM KCl in Ar-purged atmosphere. The rotation rate was varied from 100 to 2500 rpm, and the scan rate was 20 mV/s. The inset shows Koutecky-Levich plots obtained at different electrode potentials.
were performed on the Cu(poly) surface, except where noted. The Cu(poly) surface is more relevant for studies of CMP. 3.1.2. Rotating Disk Electrode. Figure 2 shows the RDE voltammograms obtained from Cu(poly) in a solution containing 60 mM H2O2 + 0.1 M H2SO4 purged with Ar. The H2O2 reduction current begins around -0.1 V, consistent with the cyclic voltammetry. To ascertain the number of electrons associated with the H2O2 reduction process, the Koutecky-Levich plot was obtained and is shown as the inset in Figure 2. For each electrode potential, the inverse of the current, 1/i, has a linear relationship with the inverse square root of the rotation rate, ω-1/2, according to the Koutecky-Levich equation:24
1 1 1 ) + i ik 0.62nFAD 2/3ω1/2ν-1/6C 0 0
(1)
where ik is the limiting current, n is the number of electrons exchanged in the H2O2 reduction reaction, F is the Faraday constant, A is the electrode area (determined by analysis of Pb upd in separate measurements), D0 is the diffusion coefficient of H2O2 (1.3 × 10-5 cm2 s-1),25,26 ω is the rotation rate, ν is the kinematic viscosity (0.009 cm2 s-1),24 and C0 is the bulk concentration of H2O2 (60.0 × 10-6 mol cm-3). Fitting the data of the inset of Figure 2 to eq 1 yields n values ranging from 1.6 at -0.2 V to ca. 1.2 at -0.6 V. At higher pH, in 0.1 M borate, previous work on Cu(poly) found n ) 2 over a wide potential range.9 The lower n values here suggest the presence of a competitive process likely associated with the strongly adsorbing sulfate electrolyte used here. Figure 3 shows RDE voltammograms obtained from Cu(poly) in a solution containing 60 mM H2O2 + 0.1 M H2SO4 + 1 mM KCl. The currents associated with peroxide reduction in the presence of Cl- are somewhat smaller than those found in the Cl--free case. Koutecky-Levich plots in the inset of Figure 3 were used to obtain values for n and ik. In this case, the n values are small (
