Langmuir 1986,2, 37-42
37
Electrochemical Oxide Formation on Titanium and Titanium Hydride (Deuteride) Thin-Film Surfaces M. C. Burrellt and N. R. Armstrong* University of Arizona, Department of Chemistry, Tucson, Arizona 85721 Received January 4,1985. I n Final Form: August 19,1985 The electrochemical formation of oxides on the surfaces of titanium and titanium hydride (deuteride) thin films has been studied. The thin films were prepared in an ultrahigh vacuum environment and transferred in this environment into a cell where they are exposed to atmospheric pressure high purity argon and then to the aqueous electrolyte. Control experiments were conducted with Auger electron spectroscopy (AES) and quartz-crystal microgravimetry (QCM) to quantitate the oxide that would form on these surfaces prior to immersion in the electrolyte. The onset potential for further oxide formation was displaced from the formal potential for the Ti3+/Tioredox couple by ca. +0.5 V. The current/voltage curves showed that the rate of oxide formation was linear with applied potential (39-45 A/V), consistent with a constant field growth mechanism which has been postulated for the gas-phaseoxidation of this type of active metal. The onset potential for oxidation in the current/voltage curves is as close to the formal potential for active dissolution of the metal as is possible to obtain with the present apparatus, without resorting to continuous mechanical abrasion of the metal surface.
Introduction The electrochemical formation of oxide layers on active metals such as titanium has been the subject of interest for several years.'-' Most attention has been focused upon the formation of oxide layers that exceed 100 A in thickness, for two reasons: (a) these are the thicknesses that are desired to effect passivation of the metal surface, and (b) even if thinner oxides are of greater interest, the study of the initial oxidation of the clean metal to thicknesses of less than 100 A is hampered by the extreme reactivity of the metal to solution and atmospheric components. A review of the electrochemistry of titanium and its oxides has recently been published by Kelly.' In electrolyte solutions, the oxidation of titanium appears to take on some of the same characteristics as its oxidation in the gas phase which was explored in the previous paper? The oxidation process occurs through at least two different regimes, involving the formation of the first 1-10 molecular layers of oxide and the subsequent growth of the stoichiometric oxide layer beyond that point.1q3t4*6*g The formation of the first few layers of oxide on the titanium surface has been the focus of studies by Beck3 and later Gottesfeld and co-workers4 who were able to obtain the current voltage responses for "clean" titanium surfaces. The metal was mounted in a rotating electrode configuration which allowed its continual mechanical abrasion while suspended in the electrolyte solution. Under those circumstances, the clean surface underwent active dissolution a t the potential expected for the Ti/Ti3+ redox couple (4.8 V vs. Ag/ AgCl). At more positive potentials, the surface oxide was formed. Hubbard and co-workers have recently demonstrated that passive film formation can be studied in a controlled manner on single-crystal stainless steel surfaces-in a fashion similar to previous studies on noble metal electrodes.lOJ1 Even though the extent of reaction is several molecular layers on these active metal surfaces, surface characterization techniques can still be used to enhance the understanding of segregation of the metal components of the alloy, oxide formation, anion incorporation into the films, and attack of the passive film by aggressive anions.
* Author to whom correspondence should be addressed.
'Present Address: General Electric Co., Corporate Research & Development, Schenectady, NY 12301.
Our own previous studies have focused upon the electrochemical activity of a titanium surface as prepared by thin film vacuum deposition technologies.2 Films of both the pure metal (Ti) and the fully hydrided or deuterided metal (TiHl.g) were investigated. Both materials were found to possess an oxide layer after their formation and transfer to the electrochemical environment under argon. This initial oxide layer displaced the onset for oxidation of the surface some 1.1 V from the Ti/Ti3+ formal potential. In addition, a t more positive potentials it was possible to electrochemically remove hydrogen from the hydrided films. The potentials necessary to carry out this reaction were dictated by the extent of previous surface oxide formation. It would be clearly beneficial to study the electrochemistries of the clean active metal surface or a t least a surface having a controlled (and already characterized) thin oxide film present, across which further surface oxidation occurs. Methods for the preparation of well-characterized electrode surfaces have been documented by several aut h o r ~ . ' ~ - ' These ~ all have in common the coupling of an ultrahigh vacuum (UHV) environment, to an electrochemical chamber, by a transfer mechanism which hope(1)Kelly, E. J. In "Modem Aspects of Electrochemistry";Bockris, J. 0. M., Conway, B. E., White, R. E., Eds.; Plenum Press: New York, 1982; pp 319-424. (2)Armstrong, N.R.; Quinn, R. K. Surf.Sci. 1977,67,451.Quinn, R. K., Armstrong, N. R. J. Electrochem. SOC.1978,125,1790. (3)Beck, T.R. Electrochim. Acta 1973,18,815. (4)Laser, D.;Yaniv, M.; Gottesfeld, S. J. Electrochem. SOC.,1978,125, 1537. (5)McAleer, J. F.;Peter, L. M. Faraday Discuss., Chem. SOC.1980, No. 70, 1. (6)Muller, R.; Wittmer, M.; Stucki, S. J. Electrochem. SOC.1981,128, 1537. (7)Blondeau, G.;Froelkcher,M.;Froment, M.; Hugot-Leboff,A. Thin Solid Film 1976,38,261. (8)Burrell, M. C.; Armstrong, N. R. Lcmgmuir, preceding paper in this issue. (9)Vetter, K. J. Electrochim. Acta 1971,16,1923. (10)Harrington, D. A.; Wieckowski, A.; Rosasco, S. D.; Schardt, B. C.; Sabita, B. N.; Hubbard, A. T. Corros. Sci., in press. (11)Harrinpton, D. A.; Wieckowski, A.; Rosasco, S. D.; Salaita, G. N.; Hubbard, A. T. Langmir 1985,1,232. (12)Hubbard, A. T.Acc. Chem. Res. 1980,13, 177 and references therein. (13)Wagner, F. T.;Ross, P. N., Jr. J. Electroanal. Chem. 1983,150, 141. Ross, P. N., Jr. Surf.Sci. 1981,102,463. (14)Yeager, E. J. Electrochem. SOC.1981,128, 160'2.
0743-7463/s6/2402-0037~01.50/00 1986 American Chemical Society
38 Langmuir, Vol. 2, No. 1, 1986
Figure 1. Schematic view of the Auger spectrometer, coupled to the UHV transfer system, which allowed for movement of the Ti film from the main chamber, where its production and characterization were carried out, to the electrochemical cell, where immersion and further oxidation were conducted. The function of the various transfer rods and gate valves are explained in the text. fully does not allow for the change in surface composition to occur prior to immersion,. An active metal will not be amenable to the same surface protection schemes of some of the noble m e t a l ~ . ' ~ J It ~ is necessary to simulate the conditions that will prevail when the active metal sees the atmosphere above the electrochemical cell, just prior to its immersion. These gas-phase reactions alone are of interest because they are analogous to the events that are expected when bulk Ti is instantaneously exposed to an atmospheric environment by virtue of a fracture or "stress crack" in the material. The previous study in this seriesa has indicated that such a simulation is possible for a freshly prepared, atomically clean Ti surface. Auger electron spectroscopy (AES) and quartz-crystal microgravimetry (QCM)techniques were used in this study as well to produce electrode surfaces of controlled composition, so that the extent of the initially formed oxide layer, and its effect on subsequent electrochemical oxide formation, could be studied.
Burrell and Armstrong
Aluminum holder 114'quic k-
connect
l
b Electrolyte Figure 2. Enlarged schematic of the reaction chamber where the electrochemical experiments were conducted. After isolation of gate valve G3,the sample was rotated into position, the chamber brought to atmospheric pressure in high purity argon, and the cell raised into position through gate valve G4,followed by immersion into electrolyte. The reverse process was carried out for the return of the sample to the spectrometer.
except the short tab that connected the block to the outside electrical contact. A fresh titanium film was then evaporatedonto one face of this foil to create the electrochemically active surface. The approsimate area of active metal was controlled by immersing only the freshly evaporated region into the electrolyte (see below). Alternatively, the passive oxide on the front surface of this foil was argon ion sputtered (with a beam diameter exceeding the dimension of the front surface of the film) to expose the clean active metal. Since only one surface was exposed to the ion gun, the back surface of the foil remained electrochemicallyinactive. After production of each type of surface the composition was assayed by AES by using the data reduction proceudres outlined in an earlier publication.18 An electron beam of the same energy Experimental Section as the Auger transitions of interest was scattered from the surface Thin films of titanium and its hydride or deuteride were for each spectrum. A deconvolution procedure then was used to prepared in the UHV chamber as described p r e v i o u ~ l y . An ~ * ~ ~ ~ remove ~~ the energy loss contributions to each Auger spectrum, evaporation source was used to produce clean fiis of the metal making quantitation of the relative atomic ratios feasible. with thicknesses that varied between 100 and 2000 A. For several To carry out the electrochemicalexperiments, the sample was of the control experiments, the evaporations were carried out on then transferred from the central carousel in the main chamber a gold-plated QCM substrate so that subsequent exposures to to a UHV-transfer systemlgvia the magnetically coupled rotary water vapor and atmospheric oxygen sources could be gravimerod, R3 in Figure 1. After isolation of the main chamber via gate trically monitored. The sensitivity of the QCM was ca. f 1 Hz valve G1,the sample was then transferred onto another rod, R1, at the 10-MHz resonance frequency of the aytal, which translates which facilitated transfer to the reaction chamber via rod R2. This into ca. 4 ng/cm2 mass sensitivity-submonolayer sensitivity for reaction chamber (a standard six-way cross, 1.5-in. i.d. ports) could uptake of elements such as oxygen and titanium. be isolated from the rest of the vacuum system via gate valves Deuterium instead of hydrogen was used to form the hydrided G3 and G2. The entire transfer system was turbomolecular metal, because of the increased mass sensitivity afforded in the pumped (Balzers TSP 170)and had a base pressure of torr. QCM studies for uptake and release of hydrogen. Hydrided or Figure 2 shows the details of the reaction chamber, similar in deuterided films were found to be chemically and electrochemdesign to those of Hubbard and co-workers.12 With the sample ically equivalent.2J6 The electrode substrates were prepared by in position on rod R2 (G3closed) the chamber was backfilled with evaporation in the main UHV chamber (Figure 1) (base pressyre, ultrahigh purity argon to atmospheric pressure. At this point the torr) on a titanium metal foil which was suspended from sample was pceitioned adjacent to the reference electrode (Ag wire a large aluminum block. This Ti foil had been previously cleaned pseudoreference electrode, AgRE) and counter electrode (Pt) and then anodized to create a thick insulating oxide on all surfaces which were permanently attached to a multipath, current feedthrough in the side of the chamber. The drift in the AgRE potential was determined to be less than *lo mV during the c o w (15) Stickney, J. L.; Rosasco, S. D.; Schardt, B. C.; Hubbard,A. T. J. of these experiments. Rod R2 was rotated so that the sample Phys. Chem. 1984,88, 251. (16)Burrell, M. C. Ph.D. Dissertation, University of Arizona, Tucson, 1984. (17) Burrell, M. C.; Armstrong, N. R. J. Vac. Sci. Technol., A 1983, 1, 1831.
(18) Burrell, M. C.; Armstrong, N. R. Appl. Surf. Sci. 1983, 17, 53. (19) Nebesny, K.; Armstrong, N. R. J. Vac. Sci. Technol., in press.
Langmuir, Vol. 2, No. 1, 1986 39
Electrochemical Oxide Formation
(1) (2)
(3) (4) (5) (6) (7)
Table I. QCM Frequency Shifts Observed during Reactions of Ti and TiDz Surfaces" surface reactant A f , Hz ng/.cm2 480 clean Ti 0, in atmospheric pressure argon 110 1060 240 O2in atmospheric pressure argon TiD2 130 30 HzO (vap) (2 X lo-' torr, 30 min) clean Ti TiO, (from 3) TiO, (from 4) TiD, TiO, (from 1 + 2)
HzO (vap) (1 X Oz (3 x torr) HzO (vap) (2 X HzO (vap) (5 X
torr, 15 min)
torr, 30 min) torr, 130 min)
30 150 105 8
130 650 460 35
oxide thickness, 9.9 22 2.7 5.4 (total) 20 (total) 9.5 0.7
A
"Assuming roughness = 3, density of TiOz = 4 g/cm3.
surface pointed down and gate valve G4 opened to allow 'the electrochemicalcell to be moved upward from below. Once the cell surrounded the three electrodes,electrolytewas overpressured into the cell from the external reservoir. The solution level was raised until the active area of the sample foil was just immersed. At this point the potential scans or the potentiometricexperimenta were initiated. All voltammetric experiments were conducted with a three-electrodepotentiostat of conventional design. Open-circuit potentials were measured with a high-input-impedancepotentiometer. Once the electrochemical experiments were completed, the solution level was lowered, the cell removed, gate valve G4 closed, and the chamber pumped to ca. torr with a LN,-cooled molecular sieve sorption pump. Valve G3 was then opened and the sample returned to the analysis chamber under UHV conditions in less than 5 min. The electrolyte used in these experiments was 0.1 N HC1 produced from triply distilled water and electronic grade HC1. Control AES experiments were conducted which showed that no detectable chloride was incorporated into the Ti/TiO, surfaces after exposure to electrolyte. HC1 was efficiently desorbed from these surfaces during the pump-down cycles.
Results a n d Discussion Electrochemistry of Ti Films with Uncontrolled S u r f a c e Oxides. Figure 3 shows the current/voltage curves for several titanium film electrodes in contact with aqueous acidic electrolytes. Figure 3a is the current/ voltage response for a titanium film prepared by evaporation in a separate vacuum chamber with a base pressure of lo-' torr and then transferred under argon to a drybox environment and then subsequently to the electrochemical cell (as in ref 2). During these early experiments some exposure to atmosphere was unavoidable during each of these transfer stages resulting in surface oxides of uncontrolled thickness and this is reflected in the current/ voltage response. The onset for current flow typically occurs a t ca. +0.2 to, +0.3 V (point 1, Figure 3a). The onset potential for oxide formation of this type of film was displaced positively by ca. 1.1V from the formal potential of the Ti/Ti3+ redox couple (-0.8 V shown as point 3 in Figure 3b). This onset could be further positively displaced by prolonged atmospheric exposure at room temperature or elevated temperatures. Following the maximum in current flow, a steady current/voltage response was observed out to potentials of ca. +1.7 V. At this point a symmetric current spike was observed that was of variable intensity depending upon the method by which the Ti film was prepared. The exact cause for this current excursion was never determined but has been observed in other Ti thin-film electrochemi~tries~ and reappears in the electrochemistry of the clean Ti films as discussed below. In the potential region beyond 2.0-2.5 V (not shown) the current began to increase corresponding to the takeover of a different oxidation mechanism, i.e., the formation of the first 25-100 %, to oxide has been exceeded. Our previous surface analysis studies of such T i films showed that the surface consisted of a stoichiometric layer of T i 0 2 over a region of substoichiometric oxides (TiO, Ti203,etc.) of undetermined thickness.2 We hypothesized
00
-
.20 c;;'/Rcc Figure 3. Current/voltage curves for the electrochemical oxidation of titanium thin films. (a) Current/voltage activity for a titanium thin film produced by electron beam evaporation at torr and transferred under argon a chamber base pressure of to the electrochemical environment.2 (b) Current/voltage activity for titanium and titanium hydride (dashed line) thin films produced by evaporation in UHV and then transferred to the electrochemical cell shown in Figures 1and 2. Point 1 denotes the approximate onset potential for oxidation of the first Ti thin films: point 2 shows the onset potential for films produced under UHV conditions, point 3 denotes the formal potential for the Ti3+/Tio redox couple and the onset for oxidative currents for mechanically abraded surfaces: and point 4 shows the onset for oxidation of a Ti film prepared in UHV and then exposed to atmospheric pressure O2for 1 h. (c) Electrochemically grown oxide thickness as a function of applied potential estimated from the combined microgravimetric and coulometric studies above. VOltSJ5
that the current observed beyond the onset potential and negative of ca. 1.8 V was due to the oxidation of this suboxide layer. The propagation of the oxide at the Ti02/Ti interface was presumed to occur a t potentials positive to that point. As discussed below, the studies of truely clean Ti thin-film oxidation in the UHV environment, with microgravimetric and AES assay, lead us now to a different viewpoint.8 Preparation and Characterization of Ti Films with Controlled Oxide Thicknesses. An atomically clean active metal surface will react with trace oxygen in the inert gas used to backfill the UHV/electrochemical chamber. Oxygen concentrations of less than part-per-million levels will still form oxides of reasonable thickness. Even if the active metal could be maintained in an oxide-free state, contact with water above the electrolyte solutions immediately prior to immersion would cause some surface oxide formation.16 It is feasible, however, to expose the clean surface on a QCM to these potential oxidants (02, H20) prior to contact with the electrolyte, in a way that allows for prediction of the surface composition of the Ti electrodes a t the instant of immersion (Table I).
40 Langmuir, Vol. 2, No. 1, 1986
When a typical clean titanium film (22000A thick on the QCM substrate) was exposed to the argon used to backfill t h e chamber in the electrochemical chamber, a total frequency shift of 110 Hz was observed, corresponding to the adsorption of 484 ng/cm2 of foreign material by the surface (no. 1, Table I). Subsequent AES examination of the surface showed only titanium (LMM) and oxygen (KLL) peaks, indicating that the surface had reacted with residual O2 in the argon to form a thin oxide layer. The thickness of this oxide was similar to those formed previously on clean Ti surfaces exposed to ca. 1 X torr 02.8It is not unreasonable that the O2 partial pressure during the atmospheric pressure argon exposure would be about the same (i.e., 0.1 ppm 02).It is conceivable that the extent of this oxide formation could be further reduced by increased purification of the argon used in backfilling the chamber. Nevertheless, exposure to H 2 0 vapor above the electrolyte would still cause oxidation of the surface (see below). Titanium films that had been converted to TiD2,when exposed to the atmospheric pressure argon, produced a 240-Hz frequency shift (no. 2, Table I), consistent with our earlier findings that titanium deuteride surfaces are less resistant to oxidation than Ti at higher O2partial pressures (>10-5-104 torr).20 It should be noted that in some cases, a t argon pressures of 1 torr or greater, there was a frequency shift observed in these QCM measurements due only to the increased pressure (damping of the oscillation). The total frequency shifts due to adsorbed gas were determined by the difference in the QCM frequency before the exposure and after the exposure when the system had been returned to vacuum of loW4 torr or better. This was generally a small correction (less than 10%). It was also of interest to examine what reactions occurred when clean Ti or TiD, surfaces were exposed to H 2 0 vapor, since this would also occur just prior to immersion of the electrode into the electrolyte. Exposure of a typical clean Ti film to H 2 0 vapor a t 2 X lo-' torr for 30 min resulted in a 30-Hz frequency shift (no. 3, Table I), indicating that some H 2 0 was adsorbed and perhaps resulted in oxide formation. Further exposure at 1 X torr for 15 min resulted in an additional 30-Hz shift (no. 4, Table I). (Larger H 2 0 pressures were detrimental to our UHV system and were intentionally avoided). The AES spectra for the titanium films exposed to atmospheric pressure argon and to H 2 0 vapor are shown in Figure 4. The spectrum of the argon-exposed surface (Figure 4b) shows that an oxide layer has been formed. Our previous studies show that thin oxides such as those in the present case consist of a thin layer of Ti02 over titanium of lower oxidation states. The spectrum of the H20-exposed Ti film (Figure 4a) shows a much smaller oxygen (KLL) peak, which is likely due to an oxide layer thinner than the Auger escape depth. Changes in the Ti (LMV) peak structures (370-470 eV) suggest oxide formation. Subsequent exposure of this same H20-exposed surface to O2 at 3 X loW5 torr resulted in a 150-Hz shift in the QCM frequency (no. 5, Table I), indicating that the H,O-oxidized surface was no more resistant to oxidation by O2 than one with a previous 02-formed oxide of similar thickness. A TiD2 film surface exposed to H 2 0 vapor at 2 X lo-' torr reacted approximately 3 times faster than the clean Ti surface; during a 30 min exposure, the QCM frequency shift observed was 105 Hz (no. 6, Table I). The AES spectrum indicated stoichiometric oxide formation had occurred. The higher degree of reactivity of the deuteride (20) Burrell, M. C.; Armstrong, N. R. Surf. Sci.,in press.
Burrell and Armstrong
300
375
450
ELECTRON
K I NET IC
525
ENERGY,
600 eV
Figure 4. Auger spectra for Ti thin films surfaces oxidized to different extents. (a) The spectrum obtained after exposure to the fresh Ti film to H20vapor at partial pressure of 2 X lo-' torr (30 min), followed by 1 X torr for 15 min. (b) The spectrum obtained by exposure of the fresh Ti film to atmospheric pressure high-purity argon, which had a residual O2 level of ca. 0.1 ppm.
(compared to Ti) toward H 2 0 vapor is comparable to the increased reactivity toward oxidation in 02, which may in part be due to the higher porosity of these deuterided films.20 Clean Ti films were also reacted with the atmospheric pressure argon and subsequently were exposed to HzO vapor. Exposure of the clean Ti surface to the backfill argon, as described above, resulted in a 110-Hz QCM frequency shift. When the oxidized surface was subsequently exposed to H 2 0 vapor at 5 X torr, a limiting frequency shift of 8 Hz was noted (no. 7, Table I). This is attributed to adsorption of less than one monolayer of H 2 0 on the oxide surface, probably forming hydroxyl groups,B1but essentially no increase in oxide thickness. A notable decrease in the Ti3+ d d peak in the electron energy loss spectra (ELS) was observed after H 2 0 exposure. As determined previously, very thin oxides such as those formed when clean titanium films are exposed to the backfill argon consist of Ti02 along with some lower oxidation states. Subsequent exposure to H 2 0 apparently results in complete oxidation of these lower oxides to TiOP We therefore conclude that the surface of a Ti film, prior to insertion into the electrolyte solution in this reaction chamber, is covered by a thin layer (10-15 A) of stoichiometric Ti02. Anodic Polarization of Ti Films with Controlled Oxide Thickness. A titanium film that was exposed only to the argon atmosphere within the isolated electrochemical chamber was next subjected to anodic polarization in 0.1 N HCl. The equilibrium potential (at open circuit) of the electrode after immersion into the electrolyte was -0.30 V vs. Ag/AgCl, and the potential was linearly scanned positive from this value (5 mV/s), producing the voltammogram shown in Figure 3b. We believe the differences in current densities between voltammograms in Figure 3, parts a and b, to be mainly due to differences in surface roughness between the two types of Ti films. In earlier work (Figure 3a) the Ti films were quite smooth with roughness factors, R, near 1.0 (as estimated from capacitance studies and electron microscopy2). In this study, the evaporated films were rougher, with R 3 (as estimated from electron microscopy and electron spectroscopic +
(21) DePauw, E.;Marian, J. J.Phys. Chem. 1981,85,3550. Sham, T. K.; Lazarus, M. S. Chem. Phys. Lett. 1979, 68,426.
Electrochemical Oxide Formation methods).EJ6 The uncertainty in the immersed electrode area of these studies (0.25 f 0.05 cm2) was also larger than in previous studies. The voltammograms; if normalized to maximum currents, however, are comparable. Several features are denoted along each current/potential curve. The onset potential, a t ca. -0.3 V (point 2) is sensitive to the amount of oxide present on the surface prior to polarization. The anodic current/voltage curve previously observed for Ti film electrodes (in 1N HC104)was nearly identical with that observed in this study, with the exception that the onset potential was seen a t ca. +0.3 V vs. Ag/AgCl (point 1, Figure 3a). These prior films had more air-formed oxide on the surface before anodization, such that a higher applied potential was required to initiate further oxidation. This effect is further discussed below. Gottesfeld and co-workers4employed an in situ abrasion technique to obtain clean Ti surfaces in an electrochemical cell and observed the onset of a net anodic current a t -0.8 V (point 3 in Figure 3b). At this potential, a clean Ti surface is actively dissolving to form Ti(II1) ions in solution, and part of the anodic current could be due to this reaction. Scanning from this potential to +1.2 V resulted in current/voltage curve similar to that observed in these studies in the same potential region. Simultaneous ellipsometry was used by these researchers to show that the oxide thickness increased linearly with applied potential in this passive potential region and that oxide formation was occurring with 100% current efficiency. The anodic peak at ca. +1.6 to +1.8 V in Figure 3a,b2 has also been observed by others5 during anodic oxidation of Ti thin f i b s but is generally not observed during anodic polarization of “clean” surfaces prepared by mechanical abrasion in s ~ l u t i o n .The ~ potential at which this peak is observed seems to be independent of prior surface oxide treatment but the origins of its presence are still not clear. At potentials in excess of about +3.0 V (not shown), the anodic current rises to values in excess of 1 mA/cm2.2J6 A change in the oxide formation reaction, accompanied by film breakdown or restructuring, occurs at potentials positive of this value. The present study has been intentionally restricted to oxide growth below this potential, where the oxide layers are still relatively thin (