Electrochemical and XPS Study of the Nickel ... - ACS Publications

Peter M. A. Sherwood. Department of Chemistry ..... M.T. Pham , M.F. Maitz , H. Reuther , E. Richter , W. Matz , A. Muecklich , F. Prokert. Journal of...
1 downloads 0 Views 241KB Size
Download PDF
Articles Anal. Chem. 1996, 68, 3330-3337

Electrochemical and XPS Study of the Nickel-Titanium Electrode Surface Pifang F. Luo† and Theodore Kuwana*

Department of Chemistry and the Center for Bioanalytical Research, University of Kansas, Lawrence, Kansas 66045 Dilip K. Paul

Department of Chemistry, Pittsburg State University, Pittsburg, Kansas 66762 Peter M. A. Sherwood

Department of Chemistry, Kansas State University, Manhattan, Kansas 66506

A Ni-Ti alloy with a 50:50 atomic composition has shown exceptional properties as a fixed potential LCEC detector for carbohydrates and related substances. It exhibited excellent sensitivity and superior long-term stability compared to pure Ni. A study was therefore undertaken by means of cyclic voltammetry (CV), scanning electron microscopy (SEM), and X-ray photoelectron spectroscopy (XPS) to understand the role of Ti and the respective surface oxides of Ni and Ti in the catalytic stability of the detector. CV results showed that Ti is initially oxidized, most likely to TiO2 in 0.1 M NaOH solution. The oxidation of Ni to nickel(II) oxide also occurs at potentials close to that of Ti. At higher potentials in the range of +0.4 to +0.5 V vs Ag/AgCl reference, nickel(II) oxide undergoes further oxidation to the Ni(III) oxidation state. This state is responsible for the catalysis of carbohydrates, amino acids and other biosubstances. When Ni-Ti and Ni are repetitively CV cycled in the potential range of 0.0 to +0.6 V, a second wave appears at more negative potentials during the reverse cathodic scan for Ni but not for NiTi. SEM images of these two electrodes in the oxidized form show the Ni-Ti surface remains smoother in appearance. This smoothness is consistent with the fact that the thickness of the surface “oxide” layer increases less rapidly, as Ni-Ti is repetitively CV cycled, compared to pure Ni. XPS results for the nature of the surface oxides are consistent with oxidized Ti as TiO2, Ni(II) predominantly as Ni(OH)2, and Ni(III) possibly as NiOOH. Possible reasons for Ti stabilizing the Ni-Ti alloy as a LCEC detector are discussed.

† Current address: Kansas Department of Agriculture, Division of Laboratories, Topeka, KS 66606.

3330

Analytical Chemistry, Vol. 68, No. 19, October 1, 1996

The Ni-Ti electrode has shown exceptional properties as an amperometric LCEC detector.1 It can be used at a constant applied potential for the catalytic oxidation of carbohydrates and related substances with superior sensitivity and selectivity over currently employed refractive index, conductivity, UV/visible, fluorescence, and pulse amperometric detectors.2,3 It has exhibited excellent stability and reproducibility for more than 40 days under continuous operation. It is also compatible with a variety of mobile phases including those containing inorganic modifiers, such as borate, and organic ones, such as acetonitrile. Previous cyclic voltammetry (CV) studies in these laboratories1 revealed that the voltammetric waves of the Ni-Ti electrode were similar to those of pure Ni with the characteristic oxidation of Ni(II) to Ni(III) at potentials of ∼0.50 V and its reduction back to Ni(II) on the reverse scan at ∼0.40 V in 0.1 M NaOH. Both electrodes exhibited electrocatalytic properties toward the oxidation of carbohydrates. However, the CV peaks for Ni-Ti were much smaller than for pure Ni. Upon continuous CV cycling, the voltammetric waves of Ni-Ti became stable and reproducible, with the CV waves exhibiting shapes that were characteristic of surface-confined species. To date, little has been mentioned about the surface properties of these species in the context of their role to provide a stable and reproducible highly sensitive LCEC detector. Several papers on the surface characterization of Ni-Ti alloys by electrochemical methods,4-12 X-ray photoelectron spectroscopy,5,11 X-ray diffraction,9 and optical microscopy9 have previously (1) Luo, P. F.; Kuwana, T. Anal. Chem. 1994, 66, 2775. (2) Shaw, P. E. CRC Handbook of Sugar Separations in Foods by HPLC; CRC Press: Boca Raton, FL, 1988. (3) Johnson, D. C.; LaCourse, W. R. Anal. Chem. 1990, 62, 589A. (4) Vracar, L.; Conway, B. E. J. Electroanal Chem. 1990, 277, 253. (5) Machida, K.; Enyo, M.; Adachi, G.; Shiokawa, J. Electrochim. Acta 1984, 29, 807. (6) Venkatesan, V. K.; Pattabiraman, R.; Indira, C. J.; Jayaraman, T. R.; Udupa, H. V. K. Proc. Interdiscip. Meet. Hydrogen Met. 1980, 270. S0003-2700(96)00236-3 CCC: $12.00

© 1996 American Chemical Society

been reported. These papers were focused mainly on the metal hydride formation and reoxidation processes4-7 involved in water electrolysis and secondary batteries. Some papers reported on the anodic process of the Ni-Ti electrode in the oxide potential region, studying the structural and electrical properties,9 the passivation process,8 the electrocatalytic oxidation of methanol,10 and the photoelectrochemical properties.11,12 However, most of these studies were carried out in either neutral or acidic media, not in alkaline solutions, and none of them dealt with the properties responsible for long-term stability and reproducibility of the electrode as an LCEC detector. In this paper, we present CV results showing that the titanium oxide initially formed prevents the Ni-Ti surface from undergoing extensive, rapid oxidation. These results are also supported by findings from other surface analytical techniques, such as scanning electron microscopy (SEM) and X-ray photoelectron spectroscopy (XPS). EXPERIMENTAL SECTION Reagents. The high-purity Ni (99.98%, 1.0 mm diameter) and Ti (99.99%, 1.0 mm diameter) metal wires and the shape-memory Ni-Ti (55:45; 1:1 atomic ratio, 0.8 mm diameter) alloy wires used in this study were purchased from Goodfellow (Malvern, PA). Sodium hydroxide and related electrolytes were prepared from electrolytic grade sodium hydroxide (Fisher) and the corresponding analytical pure reagents dissolved in NANOpure water. Electrodes. The working electrodes for CV experiments were made by embedding the metal or alloy wires (Goodfellow) into Teflon shrinkable tubing which were then sealed by heating in the oven at 300 °C for 2 min. The electrodes were polished successively with emery sand paper grit of sizes 150, 240, 300, 600, and 000 and finished to a mirror surface with alumina powder of 1.0, 0.3, and 0.05 µm suspended in water on a microcloth pad (Buehler Ltd., Lake Bluff, IL). The resultant electrodes were viewed with an optical microscope (×45) to check the cleanliness and flatness of surfaces. Finally, the electrodes were washed thoroughly with NANOpure water before use. The samples for SEM and XPS experiments were prepared from the same metal or alloy wires by polishing the wire surfaces following procedures similar to those discussed above. The electrode wires were then treated by cycling the electrode potential in 0.1 M NaOH in order to examine the oxidized surfaces. About 1 cm length of each wire was mounted into the sample holder for SEM measurements. Four pieces of wires, each 5 cm in length, were used in order to increase the surface area for XPS measurements. They were held together side by side by wrapping the ends with aluminum foil. One end was mounted onto the sample holder, which was then transferred into the UHV chamber of the XPS for analysis. Apparatus. CV experiments were performed with a Cypress Systems (Lawrence, KS) Model CS-1090 computer-controlled potentiostat. A three-electrode cell was used with a Ag/AgCl (3 M NaCl) as a reference and a platinum wire as a counter electrode. (7) Zhao, J.; Li, G. Rare Met. 1993, 12, 30. (8) Paleolog, E. N.; Fedotova, A. Z.; Derjagina, O. G.; Tomashov, N. D. J. Electrochem. Soc. 1978, 125, 1410. (9) Chougule, V. B.; Pawar, S. H. Mater. Res. Bull. 1983, 18, 1361. (10) Manoharan, R.; Goodenough, J. B. J. Mater. Chem. 1992, 2, 875. (11) Salvador, P.; Gutierrez, C.; Goodenough, J. B. J. Appl. Phys. 1982, 53, 7003. (12) Gutierrez, C.; Salvador, P.; Goodenough, J. B. J. Electroanal. Chem. 1982, 134, 325.

Figure 1. Traces a and b representing the first and second cyclic voltammograms obtained, respectively, for Ni-Ti, Ni, and Ti electrodes in 0.1 M NaOH. Scan rate 100 mV/s.

The SEM photomicrographs were obtained with a Hitachi Model S-570 electron microscope. The XPS spectra were collected on an AEI (Kratos) ES 200 B X-ray photoelectron spectrometer with a base pressure of about ∼10-8 Torr. The spectrometer was operated in the fixed retardation ratio (FRR) mode (ratio 1:23) using Mg KR X-ray radiation (240 W). Analyzer resolution was of the same order as the X-ray fine width (0.7 V). Typical data collection times were 12-14 h for the valence band and 1-3 h for each of the core regions. The spectrometer energy scale was calibrated using copper, and separation between photoelectron peaks was generated by Mg and Al KR X-rays and an argon ion etched copper plate according to the ASTM standard. The peak positions were referenced with respect to the C1s peak at 284.6 eV obtained from the trace hydrocarbon contaminants on the samples. The binding energy of the Ti3p was taken as 39 eV for calibration purposes in the valence band region. Curve fitting was performed with a nonlinear least-squares curve fitting program with a Gaussian/Lorentzian function. The Gaussian/Lorentzian mix was taken as 0.5 for all fitted peaks except for the nickel metal peak, which was taken as 0.8 with an exponential tail to represent the conduction band interaction in the metal. RESULTS AND DISCUSSION Cyclic Voltammetry. In order to improve our understanding of the role of Ti in stabilizing the Ni-Ti alloy electrode surface, attention has been focused on the initial oxidative process and the growth of the anodic oxide layer. Figure 1 shows the first and second CV scans over a wide potential region (-1.1 to +0.6 V) for Ni-Ti alloy (top), pure Ni (middle), and pure Ti (bottom) electrodes in 0.1 M NaOH solution. All three electrodes are seen to exhibit an initial oxidative wave at negative potentials, after which a significant level of current remains into the positive potential region. For Ni and Ni-Ti, a second oxidative wave is seen in the potential range of +0.4 to +0.5 V. On scan reversal at +0.6 V, a cathodic wave appears with a peak potential near +0.4 V. No other peaks are observed in the cathodic direction until a potential of -0.9 V is reached. On the second and Analytical Chemistry, Vol. 68, No. 19, October 1, 1996

3331

successive scans thereafter, the initial oxidative wave at negative potentials is absent or greatly diminished, and only the anodic wave at ∼+0.5 V is observed. The initial oxidative waves at negative potentials of less than -0.6 V are due to the oxidation of Ni(0) to nickel(II) oxide in the case of Ni and Ni-Ti, and most likely Ti(0) to titanium(IV) oxide, as Ti02, for Ti metal. It is difficult to prevent some oxide formation on the surface of these metal electrodes, even after polishing to expose a fresh surface. The potential and the peak heights of the initial oxidative waves are dependent on the amount of this initial oxide, especially in the case of Ti. To minimize oxide formation, the electrodes were polished and rinsed with water and then immediately transferred to the 0.1 M NaOH solution, followed by the application of a negative potential. The second oxidative wave for Ni and Ni-Ti, previously ascribed to the oxidation of NiO and Ni(OH)2 to NiOOH, catalyzes the oxidation of carbohydrates in the amperometric LCEC detection.1 Titanium and titanium oxide play no significant role in this catalysis since no catalytic current is found with pure Ti over a potential range of up to +0.6 V, irrespective of the number of prior CV scans. However, the presence of Ti helps to stabilize and lower the background current of Ni-Ti in comparison to pure Ni electrode for LCEC applications.1 There are subtle but significant differences in the electrochemistry of Ni-Ti compared to pure Ni in 0.1 M NaOH. For example, the oxidation of Ti and Ni occurs at nearly the same potential during the first cycle. In addition to the characteristic peak for Ni oxidation, the current shoulder at -0.8 V and the broad anodic current in the potential range between -0.4 and 0.3 V can be attributed to the oxidation of Ti. Since the oxidation of Ti is an irreversible process, a significant oxidative current was observed only in the first anodic scan, and none in the reverse and subsequent scans. The extent of Ni oxidation and dissolution was greatly diminished for the alloy, as can be seen by comparing the geometric area and peak currents of CVs for Ni-Ti (top) with those for Ni (middle) in Figure 1. The current density for Ni was found to be ∼3 times higher than for Ni-Ti (note the scale differences for current for CVs of NiTi and Ni). Furthermore, the peak potentials for oxidation to Ni(II) and Ni(III) were shifted to more positive potential by ∼30 mV. During the second CV scan, the alloy electrode exhibited features due to both Ni and Ti. The Ni(II)/Ni(III) redox wave appeared at a potential above 0.3 V and a flat baseline at potentials below 0.2 V, suggesting that the surface TiO2 acts as a “protective” overlayer to help inhibit the extensive oxidation of Ni. Previous studies on the anodic oxidation of Ti10, ,15-18 reported that, although the oxidation involved Ti (II) and Ti(III) oxide/ hydroxide intermediates, most of the titanium oxide was in the Ti(IV) oxidation state. Their studies also suggested that the growth of the oxide film took place via the slow diffusion of Ti atom into the interfacial region of the Helmholtz layer10 and that the thickness of the titanium oxide layer was linearly proportional to the maximum applied anodic potential. It has been reported19 (13) ASTM, E902-93: in 1994 Annual Book of ASTM standards; American Society for Testing and Materials: Philadelphia, 1994, Vol. 03.06. (14) Sherwood, P. M. A. In Practical Surface Analysis; Vol. 1. Auger and X-ray Photoelectron Spectroscopy, 2nd ed.; Briggs, D., Seah, M. P., Eds.; Wiley; Chichester, U.K., 1990, Appendix 3. (15) Luo, P.; Zhang , F.; Baldwin, R. P. Anal. Chim. Acta 1991, 244,169. (16) Marioli, J. M.; Luo, P. F.; Kuwana, T. Anal. Chim. Acta, in press. (17) Pankuch, M.; Bell, R.; Melendres, C. A. Electrochim. Acta 1995, 38, 2777. (18) Scrocco, M. Chem. Phys. Lett. 1979, 61, 453.

3332 Analytical Chemistry, Vol. 68, No. 19, October 1, 1996

Figure 2. Cyclic voltammograms obtained at Ni and Ni-Ti electrodes in 0.1 M NaOH. Scan rate 100 mV/s.

that the thickness of a titanium oxide film grew by ∼2 nm/V,19 and our present CV studies are consistent with these results. Furthermore, a plateau-shaped current, directly proportional to the CV scan rate, was observed at potentials up to 2.0 V, suggesting that oxygen evolution did not occur in this region. The reason for this lack of oxygen evolution is most likely due to formation of a semiconductor oxide layer, across which there is a significant potential drop. Therefore, the actual potential applied to the electrode/solution interface may be much lower than that required for oxygen evolution, although the oxidation of Ti continued. In the case of an anodic CV scan of the Ni electrode at potentials below 0.3 V, the current due to the oxidation of Ni(0) to Ni(II) is much less after the first scan. In addition, the Ni electrode exhibits current due to hydrogen evolution at potentials below -1.0 V during the first and subsequent CV scans. However, in the case of Ni-Ti and pure Ti, hydrogen evolution occurs between -1.0 and -1.1 V only during the first scan and none thereafter. This observation is consistent with the Ni-Ti electrode being covered with a layer of TiO2 that inhibits hydrogen evolution. Since the Ni-Ti electrode exhibited only a flat baseline at potentials below 0.2 V in an alkaline solution after the first cycle, our CV experiments for examining the stabilities of Ni and NiTi were focused on the potential region between 0 and 0.65 V, where the redox reaction of Ni(II)/Ni(III) occurred. The potential was CV cycled continuously up to 4000 scans. Selected CV waves up to 1000 scans for Ni and 2000 scans for Ni-Ti are shown in Figure 2. The anodic wave, due to the nickel(III) oxide on the surface of Ni-Ti, is seen to grow more slowly than on pure Ni (note that the current scale for Ni is 4 times larger than for NiTi). As the CV scans are continued, the anodic peak current for the Ni-Ti electrode is seen to plateau near a steady state level. Another significant difference, as noted earlier, is that a second reductive wave with an Epc of +0.31 V appears after some 40-50 repetitive scans for Ni but not for Ni-Ti, even after 2000 scans. In Figure 3, the charge for the anodic and cathodic waves is plotted as a function of the number of CV scans. For Ni-Ti, the anodic charge, Qa, increases rapidly until ∼200 cycles and then (19) Aramata, A. K.; Toyoshima, I. J. Electroanal. Chem. 1982, 135, 11.

Figure 3. Electrochemical charge, Q of the anodic and cathodic CV waves are plotted vs the number of CV cycles for Ni and Ni-Ti electrodes. Qa is the charge for the anodic wave; Qc1 is the charge under the first cathodic wave on scan reversal; and Qc2 is the charge under the second cathodic wave, which appears only for Ni. Table 1. Electrochemical Charge, Q, as a Function of CV Scan for Ni and Ni-Ti Electrodes nickela scan no. 1 2 10 50 100 200 300 450 600 1000 2000 a

Qa

Qc1

6.44 8.36 12.23 15.20 18.77 21.58 24.91 30.80 39.40

5.06 6.40 8.26 10.88 11.78 12.52 12.57 12.58 13.00 13.20

Ni-Tia

Qc2

Qc1 + Qc2

Qa

Qc1

2.01 3.33 6.25 9.40 12.33 17.80 26.20

5.06 6.40 8.26 12.89 15.11 18.77 21.97 24.91 30.80 39.40

1.06 1.03 1.71 2.85 3.61

1.04 1.23 1.87 2.86 3.67

5.74

5.57

8.61 8.34

8.61 8.23

Charge in microcoulombs.

gradually levels off as scans continue. The scan limits were 0.0 and +0.65 V. The symmetric shapes of the anodic and cathodic CV peaks and the equivalence of Qa to Qc are consistent with a redox process that is surface-confined. The Epa and Epc values varied from 0.49 to 0.51 V, and 0.41 to 0.42 V, respectively, for the first to the 2000th CV cycle (scan range of 0 to 0.65 V at scan rate of 100 mV/s). For Ni, Qa grew at a much faster rate than for Ni-Ti, as a function of the CV scans (see Figure 3). It also continued to grow as the CV cycles increased. Another major difference is that Qc1, the charge under the first cathodic peak, leveled off after ∼200 cycles, while the second cathodic peak, which appeared after ∼20-30 cycles, grew at a rate nearly proportional to Qa. However, the sum of Qc1 and Qc2 was equal to the anodic charge, Qa. This charge conservation between the anodic and cathodic CV waves for both Ni and Ni-Ti is consistent with a Ni(III)/Ni(II) redox process that involves only insoluble surface oxides. The Qa and Qc data are tabulated in Table 1. The presence of two reductive waves for pure Ni suggests that there is more than one phase of nickel oxide formed, whereas such is not the case with Ni-Ti. Also, the limited thickness of the oxide film with Ni-Ti may indicate that this film acts as a barrier to the penetration of water and ions so that extensive film growth is limited. The stable film may be one principal reason

for the reproducibility and longevity found with Ni-Ti compared to pure Ni in their application as LCEC detectors. Scanning Electron Microscopy. To understand the surface morphology of the anodic oxide film formed on Ni-Ti and Ni in alkaline solution, the surface was examined at different stages of oxidation by SEM. The SEM micrographs taken for Ni and NiTi before electrochemical oxidation showed no difference in surface morphology. The micrographs taken after the electrodes were repetitively CV cycled between the potentials of 0 and 0.65 V in 0.1 M NaOH, are shown in Figure 4 (at ×600) and Figure 5 (at ×10000). For pure Ni, the micrographs were taken after 50 and 1000 CV cycles; whereas for Ni-Ti, they were taken after 2000 scans. At a magnification of ×600, the resolution of the SEM images (Figure 4) is not sufficiently high to reveal any microstructural change at the Ni surface caused by the continuous oxidation. However, the scratch marks from the polishing and the white spots from the residual polishing powder, alumina (Al2O3), adhering onto the surface can be easily seen in the images. For the Ni-Ti surface, these surface marks and white spots are not obvious. These minor differences in the SEM images may be related to a higher mechanical strength of the Ni-Ti alloy over pure Ni. Upon further magnification (×10000 shown in Figure 5) of the central portion of the micrographs shown in Figure 4, the surface striation and white spots from the polishing powder become clearly visible. These white spots have an average size of 0.30.5 µm, which matches the particle size of the polishing alumina powder. A closer look at the surface images in Figure 5 indicates that Ni has a highly porous structure after 1000 CV cycles, whereas the surface of Ni-Ti appears to be smooth and flat without any indication of surface porosity, even after 2000 cycles. Thus, the images in the SEM micrographs for Ni-Ti resemble those for Ni after just 50 CV scans, where only the first cathodic peak (at 0.40 V) was present. These observations suggest that the two cathodic waves for Ni(III)/Ni(II) may correspond to two different structures: the first peak corresponds to a smooth wellordered oxide layer, while the second corresponds to a disordered, porous structure. A recent scanning tunneling microscopic (STM) study of Ni in 1M NaOH by Bard et al.20 indicated that, as the potential was increased through the passive region, a rhombic structure that formed became distorted at higher potentials to form a quasi-hexagonal structure. X-ray Photoelectron Spectroscopy. X-ray photoelectron spectra were recorded for Ni and Ni-Ti before and after CV cycling between the potentials of 0.0 and +0.60 V for 100 times in 0.1 M NaOH solution. The overall XPS spectra are not shown, as they do not add to the information provided by the higher resolution spectra. The Ni2p3/2 and the O1s regions are shown in parts A and B of Figure 6, respectively. The Ti2p region (for the Ni-Ti alloys only) is shown in Figure 7 with the valence band region in Figure 8. These spectra can be explained by an appropriate model for the surface chemistry, especially when the greater surface sensitivity of the higher binding energy (and thus the lower kinetic energy and the smaller escape depth electrons) regions are considered. In the case of Ni after 100 CV cycles, the spectra indicate the presence of a thick oxidation layer (3040 Å). This layer consists mostly of hydroxide [Ni(OH)2 and (20) Yau, S.-L.; Fan, F.-R. F.; Moffa, T. P.; Bard, A. J. J. Phys. Chem. 1994, 98, 5493-5499.

Analytical Chemistry, Vol. 68, No. 19, October 1, 1996

3333

Figure 4. Scanning electron microscopic images (magnification: ×600) obtained at a Ni and a Ni-Ti surface after CV cycled. Top left: Ni, after 50 cycles. Top right: Ni, after 1000 cycles. Bottom: Ni-Ti, after 2000 cycles. CV conditions are the same as in Figure 2.

possibly oxyhydroxide]. With Ni-Ti, the surface is significantly different between the polished and cleaned Ni-Ti (spectra in b and b′ in Figure 6A and B, respectively) and the alloy after 100 CV cycles (spectra c and c′ in Figure 6A and B, respectively). The polished and cleaned alloy shows, as expected, significant amounts of Ni and Ti metal. However, since Ti is a very reactive metal, a significant amount of oxidized Ti is on the surface, and the spectral features are dominated by this oxide. The less reactive Ni is present in the surface region largely in the form of the metal, although some hydroxide is observed in the Ni2p region. The XPS curves in Figure 6A are the Ni2p spectra indicating the presence of three different types of Ni species, corresponding to metal [Ni(0)], nickel(II) oxide as NiO, and nickel(II) hydroxide, as Ni(OH)2, in agreement with literature values and reference compounds. The metal [Ni(0)] peak is found at 852 eV with a satellite at 858.9 eV,21-24 and the NiO peak is found at 854.2 eV with satellite features at 855.7 and 860.7 eV.21-24 The Ni(OH)2 peak is found at 856.1 with satellite at 862.9 eV.24 (21) Ansell, R. O.; Dickinson, T.; Povey, A. F.; Sherwood, P. M. A. J. Electroanal. Chem. Interfacial Electrochem. 1979, 10, 69-77.

3334

Analytical Chemistry, Vol. 68, No. 19, October 1, 1996

Spectra in Figure 6B show XPS of O1s, indicating the presence of three overlapping peaks at 530.1, 532.1, and 533.4 eV corresponding to an O2- species,21-26 OH- species, and adsorbed water,21-24,26 respectively. Figure 7 shows the Ti2p region indicating peaks at 455.1 (2p3/2) and 461.3 eV (2p1/2) corresponding to Ti metal with a spin orbit splitting of 6.2 eV, whereas the presence of 461.0 (2p3/2) and 465.2 eV (2p1/2) peaks indicates the observation of TiO2 with a spin orbit splitting of 5.8 eV.18 Spectra a-c in Figure 6A show the curve-fitted Ni2p region and their deconvoluted features. Notice that the statistics degrade in going from the oxidized nickel to the alloy. This results because of the much lower nickel content of the alloy. The 100 times cycled nickel electrode (spectrum a) shows a high surface concentration of oxidized nickel species, indicated by OH- and (22) Arfelli, M.; Ingo, G. M.; Mattogno, G.; Beccaria, A. M. Surf. Interface Anal. 1990, 16, 299-303. (23) Marcus, P.; Olefjord, I.; Surf. Interface Anal. 1982, 4, 29-33. (24) Fontaine, R.; Feve, L.; Buvat, J. P.; Schoeller, C.; Caillat, R. J. Microsc. Spectrosc. Electron. 1989, 14, 453. (25) Langell, M. A.; Furstenal, R. P. Appl. Surf. Sci. 1980, 26, 445-460. (26) Moroney, L. M.; Smart, P. S. C.; Roberts, M. W. J. Chem. Soc., Faraday Trans. l 1983, 79, 1769-1778.

Figure 5. Scanning electron micrographs of Ni and Ni-Ti surfaces (magnification, ×10000 corresponding to central areas shown in Figure 4).

O) in the figures, with a low-intensity nickel metal peak indicated by M in the figures. The region marked S consists of overlapping satellite features from the metal and the oxide. The weak features to the lowest binding energy arise from K3,4 satellite features of the X-radiation used. The fit shown in Figure 6A for spectrum a shows about twice as much OH- species as O2- species, but the curve fitting of the region is complex, so that the fit shown should be regarded as one of a number of possible fitting combinations to this complex region (no curve-fitting approach provides a unique solution). The cleaned alloy in Figure 6A, spectrum b, shows a large amount of metallic Ni and a small feature corresponding to Ni(OH)2. The 100 times cycled alloy has a Ni2p region that is similar to that of the cycled Ni, but with poorer statistics. We have shown the spectrum after a binomial smoothing, repeated 20 times with the smoothing interval of about the fwhm above the untreated data, to help compare the data. The corresponding O1s regions show a difference between Ni and Ni-Ti for the 100 times cycled cases. In Ni the surface is largely hydroxide, with a very weak shoulder due to oxide at a higher binding energy. The cleaned and cycled alloy shows a clear two-peak structure that can be fitted to a peak due to oxide (530 eV), consistent with the normal values for such species.21-26

There is no evidence of any other oxygen-containing species. The alloy spectra gave high-quality agreement. All the O1s spectra have a Tougaard background removed from them.27 The fit to the 100 times cycled Ni metal spectra can include a weak oxide feature, but this gives a poorer quality fit. The reason for this difference can be readily understood because the less reactive Ni forms nickel hydroxide, but the much more reactive Ti metal has a protective layer of TiO2 on it, even after cleaning (Figure 6b). This oxide layer is retained after the 100-cycle treatment, though the relative amount of hydroxide with respect to oxide increases. The Ti2p region for the alloy shows that the oxidation layer on the metal is increased in thickness by the 100 times cycling process. Thus Figure 7b illustrates how the cleaned sample shows some Ti metal at 454 and 460 eV, but this spectrum is dominated by TiO2 (at 460 and 464 eV) the only species present in the surface region after cycling (Figure 7a). The s features arises from K3,4 satellites of the X-radiation used. The valence band region (Figure 8) confirms the features seen in the other regions. Thus, an intense O2s feature is seen around (27) Touggard, S.; Sigmund, P. Phys. Rev. B. 1982, B25, 4452.

Analytical Chemistry, Vol. 68, No. 19, October 1, 1996

3335

Figure 6. Core XPS (A) Ni2p3/2 (B) Ols spectra of Ni and Ni-Ti under different conditions: (a) and (a′) Ni metal after CV cycled 100 times; (b) and (b′) polished and cleaned Ni-Ti; (c) and (c′); Ni-Ti alloy after CV cycled 100 times. S refers to the overlapping satellite region.

Figure 7. Core XPS Ti2p spectra of Ni and Ni-Ti under different conditions: (a) Ni-Ti alloy after CV cycled 100 times; (b) *polished and cleaned Ni-Ti. S refers to the overlapping satellite region.

24 eV, though in the case of the titanium alloy, a K3,4 satellite feature arises from the intense Ti2p region at 39 eV, appearing on the high binding energy shoulder (indicated by 1). Notice that the relative amount of metal to oxide increases in the Ti2p region (Figure 8b) in going from Figure 7 to Figure 8, reflecting the electrons with high kinetic energy (and thus greater escape depth) ejected from the valence band surface of the alloy. This 3336 Analytical Chemistry, Vol. 68, No. 19, October 1, 1996

Figure 8. Valence band XPS for Ni and Ni-Ti under different conditions: (a) Ni metal after CV cycled 100 times; (b) polished and cleaned Ni-Ti; and (c) Ni-Ti alloy after CV cycled 100 times.

proves, as expected, that the oxide is on the surface of the alloy. Region 3 is the outer valence band region for TiO2. Figure 8c shows a typical TiO2 spectrum with all the features that are expected, based on a calculation of the spectrum (e.g., see ref 28). Comparison of spectra in parts b and c of Figure 8 (the cleaned and cycled alloy) shows that the cleaned alloy has two new features. One is a metal shoulder in the Ti2p region, and the other is a new feature near the band edge shown as 5. This consists of the outer valence band region of Ni and Ti. The cycled Ni shows feature 5 being very weak. Feature 4 is characteristic of oxidized Ni with a corresponding satellite (absent in the metal) shown as feature 2. Both 4 and 5 are largely Ni3d in character. It is clear from all these data that the alloy consists of a surface that is rich in TiO2, with only a small amount of oxidized Ni species in this region. This preferential oxidation of Ti over Ni is consistent with the very high negative free energy for the formation of TiO2, which is ∼203.8 kcal/mol, in contrast to NiO, which is only -51.8 kcal/mol. Overall, the XPS study reveals that the polished Ni-Ti surface is covered with a thin film of TiO2, which thickens over time as the potential is cycled. This causes the subsurface layer to be predominantly Ni, which may be responsible for contributing to the stable catalytic current, as has been observed in the Ni-TiO2 system, and named as a strong metal-support interaction.29-31 Although our present XPS study does not indicate any interaction between Ti or TiOx and Ni, there (28) Wang, T.; Sherwood, P. M. A. Chem. Mater. 1995, 7, 1031. (29) Kao, C. C.; Tsai, S. C.; Bahl, M. K.; Chung; Y. W.; Lo, W. J. Surf. Sci. 1985, 95, 1.

are numerous literature reports regarding the possibility of such an electronic interaction during preparation of heterogeneous catalysts.32 The XPS results do not rule out interaction between Ti, the oxide, and Ni. The nickel oxide layer clearly has considerable hydroxide, and this may be in the form of nickel oxyhydroxide (β or γ). The final oxide/hydroxide composition may be affected by alloying. There are small differences in XPS core data for such compounds,33 but the Ni2p region is too complex to conclusively identify them. There are also some differences between Ni(OH)2 (30) Badyal, J. P. S.; Gellman, A. J.; Judd, R. W.; Lambert, R. M. Catal. Lett. 1988, 1, 41. (31) Haller, G. L.; Resasco, D. E. Adv. Catal. 1989, 36, 173. (32) Gonzalez-Elipe, A. R.; Fernandez, A.; Espinos, J. P.; Munuera, G. J. Catal. 1991, 131, 51. (33) Dickinson, A. T.; Povey, A. F.; Sherwood, P. M. A. J. Chem. Soc., Faraday Trans. 1 1977, 73, 327. (34) Sherwood, P. M. A., unpublished results.

and NiOOH in the valence band,34 but these differences are small and difficult to distinguish in the presence of the titanium features (Ti2p and its radiation satellites). ACKNOWLEDGMENT This work was partially supported by a grant from the Center for BioAnalytical Research, as funded by the Kansas Technology Enterprise Corp. D.K.P. acknowledges the financial support by the NSF MACRO-ROA program.

Received for review March 11, 1996. Accepted July 12, 1996.X AC960236E X

Abstract published in Advance ACS Abstracts, August 15, 1996.

Analytical Chemistry, Vol. 68, No. 19, October 1, 1996

3337