Long-Term Stable Ti/BDD Electrode Fabricated with HFCVD Method Using Two-Stage Substrate Temperature

Transcription

1 /2007/ /D657/5/$20.00 The Electrochemical Society Long-Term Stable Ti/BDD Electrode Fabricated with HFCVD Method Using Two-Stage Substrate Temperature Liang Guo a and Guohua Chen a,b,z a Program of Environmental Engineering, and b Department of Chemical Engineering, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong D657 Boron-doped diamond-film-coated titanium Ti/BDD has become increasingly attractive because of the combined properties of these two unique materials. The challenge for the composite material is stability, especially when it is used as an electrode. In order to meet this challenge, a two-temperature 2-T -stage hot-filament chemical vapor deposition HFCVD method was employed in this study. The accelerated working lifetime was significantly increased to 804 h for the 2-T electrode, compared with 244 h for the diamond-film electrode fabricated under the one-temperature 1-T -stage method. With the characterization of micro-raman, X-ray diffraction, and cross-sectional scanning electron microscopy, a multilayer of Ti/TiC/ diamond amorphous carbon / diamond could be found in the 2-T sample and the structure of Ti/TiC/diamond in the 1-T sample. There was less void space observed in the interlayer of the 2-T sample. The multilayered compact structure played an important role in improving the adhesion of diamond film to the titanium substrate, which in turn increased the electrode working lifetime by over two times The Electrochemical Society. DOI: / All rights reserved. Manuscript submitted March 29, 2007; revised manuscript received July 10, Available electronically October 16, High-quality diamond films have become an attractive electrode material for electro-oxidation and electroanalysis due to high gasevolution overpotential and low background noise. An important factor to consider with diamond-film electrodes is the substrate material. The substrate should be conductive and inert against corrosion and anodic dissolution. Titanium is a preferred choice as substrate due to its relatively low price, high anticorrosion, and quick repassivation behaviors. 1,2 Diamond-coated Ti Ti/BDD has attracted much attention, especially as an anode, 2-4 because of the combined advantages of these two unique materials. Due to the high carbon affinity of Ti, titanium carbide TiC was formed prior to diamond deposition during the chemical vapor deposition CVD process. The situation of the TiC interlayer can be influenced by experimental conditions, especially the substrate temperature. It can be as thick as 150 m 5 or too thin to detect, 6 and it can be porous or with a dense crystal shape. A thick and porous TiC interlayer would reduce adhesion of the diamond film to the substrate. Moreover, the stability of the diamond films on titanium substrate is directly affected by the high thermal stress that resides in the diamond film and the substrate surface. 7 This residual stress is resulted during the cooling process from high reaction temperature to ambient temperature due to the large thermal-expansioncoefficient difference between the diamond and the substrate. This stress can cause severe film delamination. 8,9 It is, therefore, not an easy job to deposit high-quality diamond films on Ti with good adhesion. A silicon interlayer was previously applied to coat on titanium substrate in our group as a pretreatment method. It has been found that accelerated lifetime of the Ti/Si/BDD electrode has been successfully prolonged to 320 h 10 from 264 h for Ti/BDD fabricated under constant-temperature hot-filament CVD HFCVD. 11 In this work, the two-temperature-stage modified HFCVD method was adopted to manipulate the properties of the interlayers between the diamond film and the titanium substrate in order to further improve the working lifetime of the composition material as an electrode. The working lives of the electrodes were quantified by the accelerated life test and the result was compared with that of other samples. In addition, physical and electrochemical characterization was performed on the fabricated electrodes to understand the mechanism for the enhancement of the service life. z kechengh@ust.hk Experimental Diamond-film deposition. BDD films were deposited on a Ti disk using HFCVD. Ultrapure hydrogen and methane gases were the main precursor reactants. Trimethyl borate B OCH 3 3 was employed as the boron doping source. Dimethyl methane CH 2 OCH 3 2 was added into the gas phase to promote the diamond nucleation and growth process and improve diamond-filmelectrode stability. 3,11 The coil-shaped filament was made with tantalum wire. Substrate temperature and filament temperature were monitored by a thermal couple and an infrared thermometer, respectively. A detailed description of the HFCVD system can be found elsewhere. 12 Ti substrate was first ultrasonically rubbed for 30 min in diamond paste before it was cleaned in deionized water for 5 min in order to remove the adhered diamond debris from the surface. According to our experience, 12 the optimal conditions for diamond deposition on Ti substrate with constant substrate temperature are: CH 4 in H 2 ratio 1%, H 2 gas flow rate 300 sccm, T sub = 750 C, T fil = 2200 C, chamber pressure 40 mbar, deposition time 10 h. The samples obtained under these conditions are referred to as 1-T samples. The highly stable diamond-film electrode was fabricated under the two-stage temperature method, in which the starting substrate temperature was selected as 650 C and the deposition time was 5 h then was raised to 750 C to continue the deposition for another 5 h. This sample is referred to as the 2-T sample in the subsequent content. Diamond-film characterization. The surface morphology and cross-sectional view were characterized with scanning electron microscopy SEM JEOL A high-magnification view of diamond film was checked with transmission electron microscopy TEM, JEOL-2010F, Japan. X-ray diffraction XRD analysis was carried out with a Philips PW1830 scanning 2 from 30 to 90. Micro-Raman spectroscopy was performed using a Renishaw RM1000 micro-raman system with 514 nm line of an argon ion laser, and the laser incident power was 20 mw. Raman peaks centered at 1332 and 1550 cm 1 were attributed to sp 3 -bonded diamond crystal and sp 2 - and sp 3 -hybridized amorphous carbon. It has been reported that the Raman signal is 50 times more sensitive to the nondiamond carbon than to the crystalline diamond. 13 The percentage of diamond carbon in the total carbon content can be quantified with a purity index PI, calculated with the following equation I D PI = I D + I C /50 where I D and I C are the integrated peak intensity, calculated with a mixed Gaussian Lorentzian function, of diamond and nondiamond carbon phase, respectively. Electrochemical characterization. Electrochemical measurements were carried out in a conventional three-electrode cell with an Autolab PGSTAT100 system. The BDD electrode, namely, the

2 D658 Journal of The Electrochemical Society, D657-D working electrode WE, was pressed against a Viton O-ring in a poly tetrafluoroethylene PTFE -made electrode holder. The geometric working area of the WE was cm 2. The Ag/AgCl KCI sat. reference electrode RE and Pt wire counter electrode CE were used for all tests. The working area of the CE is much larger than that of the WE. All experiments were conducted under room temperature. Cyclic voltammogram CV analysis was conducted in 0.5 M Na 2 SO 4 solution and redox Fe CN 6 3 /4 solution, respectively. The electrode stability was studied using accelerated life test, which was galvanostatic polarization with a current density of 1 A/cm 2 in 3 M H 2 SO 4 solution. All chemicals were analytical grade and used without further purification. Sodium sulfate 0.5 M, Riedel-Dehan, sulfuric acid 3 M,BDH, and 10 mm of potassium ferrocyanide and potassium ferricyanide Sigma with supporting electrolyte of 0.5 M sodium sulfate Riedel-Dehan were prepared with ultrapure deionzed water 17.9 M cm from an E-Pure water purification system Barnstead. All solutions were deoxygenated by purging with nitrogen gas for 15 min prior to each experiment. Results and Discussion Ti/BDD fabrication. Cross-sectional SEM pictures Fig. 1a-c show the morphology of the TiC interlayer after 1 h of deposition under substrate temperatures of 750 and 650 C. Figure 1b is the enlarged SEM image of the marked area in Fig. 1a. At 750 C, the TiC layer is rough and porous Fig. 1b, with a thickness of over 50 m. At 650 C Fig. 1c, no clear interface between TiC and Ti can be found, and the TiC layer is not as porous as that at 750 C. After 4 h of deposition, accumulated carbon clusters on the TiC layer grew significantly large and were converted into diamond crystals. The highly magnified SEM picture Fig. 2a shows the growth process of diamond crystals from amorphous carbon a-c. Disordered carbon clusters increasingly changed into the ordered phase with clear-cut crystalline shapes. The diamond-film TEM image, Fig. 2b, shows that a nanocrystalline diamond with a diameter of about 5 nm was produced in the a-c phase, which indicates the a-c was converted into diamond crystals under the effect of atomic hydrogen. The TEM image is consistent with SEM results. A similar result was reported by Singh, 14 that is, diamond crystallites were formed as the result of direct transformation of diamondlike carbon into diamonds under the continuous bombardment of atomic hydrogen and a supply of energy at the surface. Figure 3a shows the XRD profile of the diamond film deposited under different substrate temperatures of 650 and 750 C, respectively. Titanium carbide was formed as the interlayer between the diamond layer and the titanium substrate. Under 750 C, the TiC peaks were much stronger than under 650 C, so the quantity of the TiC phase became much larger. As shown by cross-sectional SEM images, with the increase in substrate temperature, the TiC layer became thicker and rougher. A rough and porous interlayer could reduce the adhesion of the diamond film to the substrate. Therefore, a carbide interlayer formed under relatively high temperatures such as 750 C on the Ti substrate could reduce the adhesion of the following deposited diamond film, compared with low substrate temperatures. As for the diamond-film electrode, the crystal quality of the superficial diamond layer has great influence on the electrochemical properties. A high-quality electrode requires high diamond-crystal quality. Figure 3b shows the Raman spectrum, which can give information about the influence of substrate temperature on diamond purity and crystallinity. Under 650 C, the deposited diamond possessed lower crystallinity and mixed with larger amounts of graphite and amorphous carbon compared with under 750 C, due to the large full width at half-maximum fwhm of the diamond peak and high peak intensity for the nondiamond carbon phase. Stable diamond-film electrode requires high superficial diamond quality and high adhesion to the substrate. Under high substrate Figure 1. Micromorphologies of TiC interlayers formed under different temperatures a and b 750 C and c 650 C. temperature, high superficial diamond quality can be obtained, but the adhesion is reduced due to the formation of a rough and porous carbide layer. Comparatively, under low substrate temperature, high adhesion can be obtained with the formation of a dense carbide interlayer, but the diamond-crystal quality is reduced. In other words, the problem with high substrate temperature is low stability of the diamond electrode, and the problem with low substrate temperature is the low electrochemical performance. The aim of this work is to fabricate a high-quality top diamond layer with good adhesion to the substrate. Our two-stage temperature HFCVD method integrates the advantages of the high- and low-substrate-temperature process. The high-quality top diamond layer is fabricated under the high substrate temperature, and the adhesive interlayer is made under the low-substrate-temperature process. Characterization of diamond film. Figure 4 shows the topsurface morphology of 1-T and 2-T samples. The diamond films of these two samples both have fully covered the substrate but show different morphologies. The 1-T sample shows clear-cut faceted

3 D659 Figure 3. a XRD diagram and b Raman spectrum of diamond film deposited under 650 C and 750 C substrate temperature for 5 h. Figure 2. a A SEM image of diamond nucleation and b an HRTEM top-view image of a diamond film: A amorphous carbon and B diamond crystal. crystals with triangular and rectangular habits, with the particle diameters being in the range of 1 7 m. The XRD profile shows a 111 preferential orientation. In contrast, the 2-T sample displays a cubooctahedral facet, with the crystal size being smaller, in the range of 1 2 m, but the number of particles being larger. XRD shows a 220 orientation for the 2-T sample. With the micro-raman spectrum Fig. 5, the fwhm value for the diamond peak of 1-T and 2-T samples are measured as 14.6 and 18.5 cm 1, respectively. Due to much smaller crystal size and more grain boundaries, the crystallinity of the 2-T sample becomes lower compared with the 1-T sample. Moreover, the PI indices indicate that the diamond-carbon percentage in the total carbon content is 98.4 and 98% for the 1-T and 2-T samples, respectively. The slightly higher sp 2 nondiamondcarbon concentration in the 2-T sample results from the larger boundary area. Figure 6 shows the CV curves for the 1-T and 2-T BDD samples as electrodes in 0.5 M Na 2 SO 4 solution at a potential scan rate of 100 mv/s. The working potential window ±500 A/cm 2 geometric for 1-T and 2-T samples is 3.8 V V and 3.7 V V, respectively. These large potential-window values indicate high-quality polycrystalline BDD films. Comparatively, due to the smaller sizes and a slightly different facet of the diamond crystals, and a bit higher graphite content in the film, the 2-T BDD electrode shows a little bit smaller potential-window value. The CV curves for 1-T and 2-T diamond-film electrodes are shown in Fig. 7a. The peak potential difference for the 2-T sample is relatively smaller than that of the 1-T electrode, due to the lower resistance of the 2-T sample 10 compared with that of the 1-T sample about 100. Figure 7b indicates a linear relationship between E pc E 0 and ln v for the Fe CN 6 3 /4 redox in 0.5 M Na 2 SO 4 solution on the 1-T and 2-T BDD electrodes. This redox reaction shows quasi-reversible behavior on the electrodes. From the relationship of E pc E 0 ln v, the reduction reaction electrontransfer coefficient and electron-transfer rate constant k 0 can be calculated. For the 1-T sample electrode and k o are and cm/s, respectively, while for the 2-T sample electrode, and k o are and cm/s. The higher fraction of sp 2 carbon in the 2-T diamond film improves the values of the kinetic parameters. The calculated values of and k 0 on the BDD electrodes are consistent with those reported by Duo 15 and Ramesham. 16 Electrode stability. Diamond-film delamination is believed to be the main reason for electrode failure. It is common to find a sharp increase in the voltage when the diamond film delaminates significantly. Because voltage change can be easily checked from the electric-voltage spectrum, the time of the sudden increase in the cell voltage is recorded as the end of the accelerated working life. The acceleration is accomplished using high current density and low ph of the electrolyte. By employing this method, it is found that the accelerated working life is significantly increased to 804 h for the 2-T sample, compared with 244 h for the 1-T sample. Over twotimes improvement was achieved. On the electrodes after accelerated life test, a much larger number of locations where diamond film split away or cracked can be observed on the 1-T sample than on the 2-T sample. Figure 8 shows the surface morphology and the cross-sectional view of the failed electrodes. For the 1-T sample, cracks propagated through the dia-

4 D660 Journal of The Electrochemical Society, D657-D Figure 6. CV curves of 1-T solid line and 2-T dashed line BDD samples as electrodes. porous TiC layer is deposited. With the 1-T deposition method, a diamond layer is directly deposited on this porous substrate. Void and gap can be formed between the diamond and the carbide layer. Comparatively, under a low temperature of 650 C, the thin and solid TiC interlayer leads to a smaller amount of void between TiC and the growing diamond layer. Furthermore, the buffer mixture layer of Figure 4. Surface morphology of a 1-T and b 2-T BDD samples. mond crystals, resulting from the high residual stress in the film, and the exposed substrate shows a rough and porous morphology. According to the cross-sectional view, some void space can be clearly seen between the diamond film and the substrate. For the 2-T sample, the substrate material is much more compact, and no clear void space can be found at the interface. The micro-raman spectrum Fig. 3b shows a layer of diamond mixed with amorphous carbon formed under 650 C. After the first stage of 2-T deposition process, the deposited mixture layer works as the substrate for the following process. Combining the results of micro-raman and XRD analyses and the cross-sectional SEM images, it can be proposed that a multilayer structure of Ti/TiC/ diamond + amorphous carbon /diamond is formed for the 2-T sample. For the 1-T sample, the structure is Ti/TiC/diamond. Figure 9 illustrates the cross-sectional multilayer model for these two samples. Under high substrate temperature, 750 C, a thick and Figure 5. Raman spectrum of 1-T solid line and 2-T dashed line BDD samples. Figure 7. a CV curves for 1-T dashed line and 2-T solid line electrodes, and b relationship between E pc E o value and ln v on 1-T and 2-T samples for the 10 mm redox Fe CN 6 3 /4.

5 D661 Figure 8. Surface morphology and cross-sectional view of the a and b 1-T and c andd 2-T samples after accelerated life tests as electrodes. Figure 9. Multilayer model of a 1-T and b 2-T samples. diamond with amorphous carbon in the 2-T sample plays an effective role in reducing the lattice mismatch with the diamond layer. Combination of these two advantages can effectively help improve the adhesion of diamond film onto Ti substrate and the electrode stability. As we have mentioned, diamond-film delamination is the main reason for electrode failure during operation. The film-delamination process is complex, which is due to the combined effect of internal performance, including residual stress and nonuniform electrochemical properties, and external factors, especially high current loading. Here are three related possibilities. First, due to the nanopore in the as-grown diamond film, it is possible for the severe corrosive electrolyte to penetrate the diamond film and reach the substrate of the TiC phase. Due to the low anticorrosion performance of TiC compared with diamond, under high anodic loading 10,000 ma/cm 2, the TiC layer was dissolved and caused debonding with the upper diamond film. 17 Second, in the strong acidic solution 3 MH 2 SO 4, high anodic potential can induce diamondfilm corrosion, especially the boundary area with low crystallinity. Moreover, high residual stress in the film, mainly compress stress, can make defects, such as dislocations, in the diamond phase, which shows poor electrochemical stability compared with the crystallized phase. Corrosion at the grain boundary and defects can reduce the stability of the diamond film. Third, high residual stress is the main reason for film delamination not only to the as-grown sample but also to the diamond-film electrode. Under high anodic-current loading, huge amounts of gas bubbles were produced on the electrode surface. The bubble production can cause the seed for the release of stress, and delamination of the film is induced simultaneously. Conclusion High-quality BDD film has been successfully deposited on titanium substrate by HFCVD using two-stage substrate temperature. The 2-T sample displayed a cubo-octahedral facet with a crystal size in the range of 1 to 2 µm, much smaller than the crystal size of the 1-T sample. Due to the larger boundary area in the 2-T sample, a little bit higher nondiamond carbon content of 2% was found, compared with 1.6% for the 1-T sample. Therefore, the working potential window for the 2-T BDD electrode was 3.7 V, slightly lower in comparison with 3.8 V for the 1-T BDD electrode. The stability of the diamond electrode was significantly increased with the newly developed method. The accelerated life tests showed the working lifetime of the 2-T electrode reached 804 h compared with 244 h of the 1-T electrode. For the 2-T sample, the multilayer structure of Ti/TiC/ diamond + amorphous carbon /diamond played an important role in increasing the adhesion of diamond film on to the substrate, which in turn effectively improved the stability of the electrode. Acknowledgments Financial support from the Hong Kong Research Grant Council under project no. HKUST is gratefully acknowledged. Hong Kong University of Science and Technology assisted in meeting the publication costs of this article. References 1. F. Beck, H. Krohn, W. Kaiser, M. Fryda, C. P. Klages, and L. Schafer, Electrochim. Acta, 44, A. V. Diniz, N. G. Ferreira, E. J. Corat, and V. J. Trava-Airoldi, Diamond Relat. Mater., 12, X. M. Chen, G. H. Chen, F. R. Gao, and P. L. Yue, Environ. Sci. Technol., 37, X. M. Chen, F. R. Gao, and G. H. Chen, J. Appl. Electrochem., 35, Y. Q. Fu, H. J. Du, and C. Q. Sun, Thin Solid Films, 424, J. W. Ager and M. D. Drory, Phys. Rev. B, 48, S. S. Park and J. Y. Lee, J. Appl. Phys., 69, L. Chandra, M. Chhowalla, G. A. J. Amaratunga, and T. W. Clyne, Diamond Relat. Mater., 5, M. D. Drory and J. W. Hutchinson, Science, 263, Y. Tian, X. Chen, C. Shang, and G. Chen, J. Electrochem. Soc., 153, J X. Chen and G. Chen, J. Electrochem. Soc., 151, B L. Guo and G. Chen, Diamond Relat. Mater. 16, D. S. Knight and W. B. White, J. Mater. Res., 4, J. Singh, J. Mater. Sci., 29, I. Duo, C. Levy-Clement, A. Fujishima, and C. Comninellis, J. Appl. Electrochem., 34, R. Ramesham, Thin Solid Films, 339, X. Chen, Ph.D Thesis, The Hong Kong University of Science and Technology, Hong Kong 2002.