Carbon-based electrode materials (e.g., carbon fibers, glassy carbon, and graphite) are used in various electrochemical technologies, including electroanalysis, energy storage devices, and electrosynthesis. These materials have similar microstructures consisting of layers of condensed, six-membered aromatic rings with sp²-hybridized carbon atoms trigonally bonded to one another. The crystallite size and extent of microstructural order can vary from material to material (i.e., edge-to-basal plane ratio), which has important implications for the electron-transfer kinetics of a given redox analyte (1, 2).

The electrochemical performance of these materials has been studied in detail over the past three decades, and much has been learned about the structure-reactivity relationship (1, 2). The use of synthetic conductive and semiconductive diamond thin films in electrochemistry has only recently been reported (3-13). However, the relationship between the physical, chemical, and electronic properties of diamond and its electrochemical performance is not well understood and is the subject of much recent interest.

The objective of this Report is to acquaint the reader with the conditions required to grow highly conductive, boron-doped diamond thin films, as well as with some of the analytical methods used to characterize films, electrochemical properties of the films in aqueous media, and some applications. The discussion will be limited to the properties and performance of highly doped films (>10¹⁹ cm^-3), although it should be noted that the electrochemical response depends on the doping level. Diamond thin films can possess electronic properties ranging from those of an insulator at low doping levels, to those of a semiconductor at moderate levels, to those of a semimetal at high levels (14, 15).

Diamond chemistry
Diamond possesses several technologically important properties including extreme hardness, high electrical resistance, chemical inertness, high thermal conductivity, high electron and hole mobilities, and optical transparency (14, 15). The material is a wide bandgap semiconductor (E_g = 5.5 eV) and offers advantages for electronic applications under extreme environmental conditions.

Each carbon atom in diamond is tetrahedrally bonded to four other carbons using sp³-hybrid orbitals. Microstructurally, the atoms arrange themselves in stacked, six-membered rings, with each ring in a "chair" rather than a planar conformation. In boron-doped films, the boron impurity atoms substitute in place of some of the carbon atoms during film growth.

Diamond is one of nature's best insulators; but when doped with boron, the material possesses semimetal electronic properties, making it useful for electrochemical measurements. For example, synthetic diamond thin films grown using hot-filament or microwave-assisted chemical vapor deposition (CVD) can be doped to as high as 10,000 ppm B/C, resulting in films with resistivities <0.1 -cm.

Energetically, the boron dopant atoms (electron acceptors) form a band located ~0.35 eV above the valence band edge. At room temperature, some of the valence band electrons are thermally promoted to the boron acceptors, leaving free electrons in the dopant band and holes, or vacancies, in the valence band to support the flow of current. In addition, boron-doped diamond thin films commonly possess a rough, polycrystalline morphology with grain boundaries at the surface and a small-volume fraction of nondiamond carbon impurity. Consequently, the electrical conductivity of the film surface and the bulk is influenced by the boron-doping level, the grain boundaries, and the impurities.

Potentially lucrative electronic applications for synthetic diamond, such as amplifiers and diodes, have been largely unrealized because of the rough, polycrystalline morphology of most CVD-grown films and the ubiquitous presence of a nondiamond carbon impurity phase. These applications require low-defect, single-crystal films with only trace levels of the nondiamond carbon impurity. Single-crystal films can be grown homoepitaxially on expensive single-crystal diamond substrates and heteroepitaxially on cubic-boron nitride (c-BN), BeO, Si, and Ni single-crystal substrates (16, 17 ). The problem with homoepitaxial growth is that the films are not easily removed from the substrate. The problem with heteroepitaxial growth on foreign substrates is that the domain size of well-ordered films is limited.

Although the conditions required to grow diamond thin films are well established, much less is understood about the factors that actually influence the nucleation and growth on the substrate. Electrochemical methods of analysis offer the possibility of providing new insights into the surface and bulk electronic properties of the material. Worldwide interest in the use of diamond thin films in electroanalysis, electrosynthesis, corrosion protection, energy storage devices, and electrochemical-based toxic waste remediation has also emerged, driven by the prospects of this material's unique properties.

Growth and characterization
Diamond thin films can be grown from dilute mixtures of a hydrocarbon gas (e.g., methane) in hydrogen using one of several energy-assisted CVD methods, the most popular being hot-filament and microwave discharge (14, 15). In these processes, a carbon-containing gas is energetically activated to decompose the molecules into methyl radicals and atomic hydrogen if a methane-hydrogen source gas mixture is used.

The growth methods mainly differ in the manner in which the gas activation is accomplished. Typical growth conditions are C/H volume ratios of 0.5-2%, pressures of 10-100 torr, substrate temperatures of 800-1000 °C, and microwave powers of 1000-1300 W, or filament temperatures of ~2100 °C, depending on the method used. The film grows by nucleation at rates in the 0.1-1 µm/h range. For the substrates to be continuously coated with diamond, the nominal film thickness must be ~1 µm (14, 15). Interestingly, microwave CVD growth of nanocrystalline diamond using a unique fullerene-argon gas mixture produces continuous films at thicknesses as low as 30 nm (18). Substrate diameters of several inches can easily be coated in most modern reactors.

Boron doping is accomplished from the gas phase by mixing a boron-containing gas such as B₂H₆ with the source gases, or from the solid state by gasifying a piece of hexagonal-boron nitride (h-BN) (11). In the solid-state approach, diborane is produced by the interaction of atomic hydrogen with h-BN and is then incorporated into the gas flux to the substrate. Surprisingly, very little nitrogen is incorporated into the films using h-BN.

The substrate is pretreated by cleaning it with a series of solvents and "seeding" it with small diamond particles by polishing the substrate in diamond powder. The embedded particles serve as nucleation centers for film growth.

Atomic hydrogen plays an important role in all of the growth methods (14, 15). It prevents surface reconstruction from a saturated sp³-hybridized diamond microstructure to an unsaturated sp²-hybridized graphite microstructure; suppresses the formation of nondiamond carbon impurity; and abstracts hydrogen from hydrocarbon surface sites and gas-phase hydrocarbon species to form reactive radicals. Landrass and Ravi demonstrated the influential role lattice hydrogen plays in controlling the electrical resistivity by defect state passivation (19, 20).

Several analytical techniques are routinely used to characterize the morphological, optical, chemical, and electronic properties of diamond thin films (20). Figure 1 is a scanning electron micrograph of a boron-doped diamond thin film grown on p-type Si (100) by microwave-enhanced CVD using 1% C/H. The film is continuous over the entire substrate surface and possesses a polycrystalline morphology. Such image features are characteristic of most conventional CVD-grown films. The film is composed of sharp, well-faceted microcrystallites ranging in size from 0.5 to 3 µm, with many twinned and misoriented crystallites with no obvious preferential orientation. Numerous grain boundaries can also be seen.

FIGURE 1. Scanning electron micrograph of a boron-doped diamond thin film grown on p-Si (100) by microwave-enhanced CVD.

Typical Raman spectra for high-quality diamond and boron-doped diamond thin films feature one intense band at 1332 cm^-1 with a fwhm of ~10 cm^-1. This band is attributable to the first-order phonon mode for diamond (21-23). Weak scattering intensity in the region around 1520 cm^-1 can be caused by nondiamond carbon impurity incorporated into the film. These two bands are often superimposed on a broad but weak photoluminescent background when excited with visible light.

Polycrystalline diamond films commonly contain varying amounts of nondiamond carbon impurity, depending on the growth conditions. The nondiamond carbon likely consists of a mixture of sp³- and sp²-hybridized bonding, similar to diamondlike or amorphous carbon.The Raman cross-sectional scattering coefficient for diamond is 9 X 10^-7cm^-1/steradian and for graphite is 500 X 10^-7 cm^-1/steradian (22). Therefore, the technique is quite sensitive to the presence of nondiamond carbon impurity.

Scanning tunneling and atomic force microscopies are used to evaluate the film morphology and probe the local electronic properties. Powder X-ray diffraction analysis is used to investigate the preferential crystallite orientation of the films. Surface elemental composition is determined by Auger electron and X-ray photoelectron spectroscopies. Secondary ion MS is used to quantitate the boron dopant concentration and probe the spatial distribution of the dopant species over the surface and within the bulk.

Electrochemical properties
Several interesting and electroanalytically important electrochemical properties distinguish boron-doped diamond thin films from conventional carbon electrodes. The films exhibit voltammetric background currents and double-layer capacitances up to an order of magnitude lower than for glassy carbon (3-5, 10, 11). Figure 2a shows a cyclic voltammetric current versus potential curve, along with a double-layer capacitance versus potential plot for diamond.

FIGURE 2. Cyclic voltammetric current...

The voltammetric response in 0.1 M KCl is featureless within the working potential window and indicates that the diamond-electrolyte interface is almost ideally polarizable. The current between -1000 and 1000 mV is <10 µA (50 µA/cm²) and is stable with repeated sweeps. A small oxidation peak is observed on the forward sweep at ~1200 mV just prior to the onset of chlorine evolution. The origin of this peak is unknown but may be related to the electrochemical activity of nondiamond carbon impurity exposed at the grain boundaries (11).

No features appear on the reverse sweep before potentials negative of -1000 mV, at which point some cathodic current flows, presumably caused by the reduction of dissolved chlorine and/or oxygen in the solution. The cathodic current associated with hydrogen evolution commences at approximately -1750 mV. It is interesting that the residual background current is about an order of magnitude lower than for freshly polished glassy carbon of similar geometric area. This property leads to enhanced signal-to-background (S/B) ratios in voltammetric measurements for certain redox analytes, such as Fe(CN)₆^-3/-4, IrCl₆^-2/-3, and azide. The double-layer capacitance for this particular film in 1 M KCl ranges from 4 to 8 µF/cm² over a 2-V potential window. There is a general trend toward increasing capacitance with more positive potentials. This shape is characteristic for p-type semiconductor electrode-electrolyte interfaces (4).

These capacitance values are comparable in magnitude to those observed for the basal plane of highly oriented pyrolytic graphite (HOPG) (4, 24) and significantly lower than those for glassy carbon (4). By comparison, the double-layer capacitance for glassy carbon is 25-35 µF/cm² at these potentials. The capacitance versus potential profile shape and magnitude for diamond is largely independent of the electrolyte composition (0.1 M NaF, NaCl, NaBr, and NaI) and solution pH (3, 4, 9).

These two observations suggest that the specific adsorption of these electrolyte ions does not occur to any appreciable extent and that the surface is relatively void of ionizable surface carbon-oxygen functionalities. The low and stable background current and capacitance are attractive features of diamond for potentially improved S/B in electrochemical assays.

Three possible, and not necessarily unrelated, factors could explain the low background current and capacitance of diamond. First, the relative absence of electroactive carbon-oxygen functionalities on the hydrogen-terminated diamond surface results in a lower current. For example, our group recently reported that oxygen-free, hydrogenated glassy carbon exhibits background voltammetric currents that are a factor of 3 to 6 less than the freshly polished surface (7). Vacuum heat treatment of glassy carbon and carbon fibers has also been shown to remove electroactive carbon-oxygen functionalities, thus lowering the voltammetric background current and capacitance (25). Therefore, the absence of electroactive surface carbon-oxygen functionalities can explain some but not all of the decreased current and capacitance for diamond.

A second contributing factor may be a lower density of surface electronic states near the Fermi level caused by the semimetal-semiconductor nature of boron-doped diamond (4, 12). A lower surface charge carrier density at a given potential would lead to a reduced accumulation of counter-balancing ions and water dipoles on the solution side of the interface, thereby lowering the background current and capacitance. Similar reasoning has been invoked to explain the anomalously low background current and capacitance for the basal plane of HOPG (24).

A third possible contributing factor could be that the diamond thin-film surface is structured like an array of microelectrodes. In other words, perhaps the diamond surface has "electrochemically active" sites separated by more insulating regions - in much the same way that composite electrodes have very reactive regions of carbon separated by insulating regions of the Kel-F polymer support (26). Experimental data have not provided evidence to unequivocally distinguish between the second and third possible factors.

Another interesting electrochemical property is that the films exhibit a wide working potential window for solvent-electrolyte electrolysis in conventional aqueous media, which means that a large overpotential exists for the evolution of chlorine (7, 11), oxygen, and hydrogen (10, 11). Figure 2b is a comparison of the cyclic voltammetric current versus potential curves for glassy carbon and a boron-doped diamond thin film in 0.1 M KCl. The working potential window, defined as the potentials at which the anodic and cathodic currents reach 50 µA (250 µA/cm²), is 3.5 V for diamond and 2.5 V for glassy carbon. The anodic current for glassy carbon on the forward sweep increases sharply at 1000 mV as a result of chlorine evolution. An associated cathodic peak on the reverse sweep at 700 mV is caused by the reduction of dissolved chlorine. Clearly, sizable currents for these reactions are absent for diamond at these potentials. This figure also reveals the much lower voltammetric background current for diamond.

The reason for the large overpotentials for chlorine, oxygen, and hydrogen evolution has not been determined. One possible explanation is the absence of the requisite surface sites needed for the adsorption of reaction intermediates on diamond (8, 12). All three of these reactions proceed by mechanisms involving surface intermediates. The diamond surface is fairly inert and is primarily hydrogen terminated after growth. Angus et al. have shown that the overpotentials for hydrogen and oxygen evolution are directly related to the nondiamond carbon impurity content (11). The higher the fraction of nondiamond carbon present, the lower the overpotentials for both these reactions, presumably because these impurities provide surface sites for the intermediates. The larger overpotentials for diamond could possibly allow redox analytes with more positive and negative standard reduction potentials to be studied.

Another property is that untreated films exhibit reversible to quasi-reversible electron transfer kinetics for inorganic redox analytes such as Fe(CN)₆^-3/-4, Ru(NH₃)₆^+2/+3, and IrCl₆^-2/-3. In other words, boron-doped diamond thin films show good electrochemical activity without any pretreatment. Figures 3a-c show cyclic voltammetric current versus potential curves for these three analytes. This particular film was exposed to laboratory air for two weeks prior to these voltammetric measurements. This strong tendency to resist deactivation is another attribute of the material.

Well-defined redox waves are seen for all three of these compounds with E_p and I_p^ox/I_p^red values of 88 mV, 1.0; 72 mV, 1.0; and 65 mV, 1.0, respectively. In all cases, the I_p^ox values vary linearly with the analyte concentration from 1 mM down to 10 µM and linearly with the sweep rate^1/2 for scan rates from 100 to 500 mV/s.

These values indicate that the current is limited by semi-infinite linear diffusion of the reactant to the interfacial reaction zone and that the reaction does not involve any surface-confined species. Heterogeneous electron transfer rate constants for these three analytes typically range from 0.002 to 0.1 cm/s at untreated diamond. Clearly, analytically useful responses are observed for all three analytes.

The Fe(CN)₆^-3/-4 data are very interesting because the heterogeneous electron transfer rate for this analyte is extremely sensitive to the microstructure of conventional graphitic electrodes (1, 2). The higher the fraction of graphitic edge plane exposed, the larger the rate constant. The E_p values vary from 150 to 300 mV (100 mV/s) for the iron cyanide redox analyte at untreated glassy carbon exposed to the laboratory atmosphere for two weeks - the same length of exposure as this particular diamond thin film. These findings indicate that the boron-doped diamond thin films exhibit good activity for these three inorganic redox analytes and resist fouling and deactivation to a higher degree than does glassy carbon.

Figure 3d shows the cyclic voltammetric current versus potential curve for 0.1 mM hydroquinone in 0.1 M HClO₄ with diamond. The electrode reaction for this analyte involves proton and electron transfer - unlike the three previous analytes, which involved only electron transfer. The E_p for diamond is 560 mV and reflects the slow kinetics for this analyte. It has been shown that the electrode reaction kinetics for quinonelike compounds involving both proton and electron transfer are mediated by certain carbon-oxygen functionalities on the surface of graphitic electrodes (1, 2). The diamond surface is predominantly hydrogen terminated, so the requisite carbon-oxygen functionalities are absent. For this reason, slow electrode-reaction kinetics are observed for this class of analytes (hydroquinone, catechol, dopamine) at hydrogen-terminated diamond (4). Experiments are in progress to determine whether the kinetics can be altered by forming oxygen functional groups on the diamond surface via electrochemical anodization.

FIGURE 3. Cyclic voltammetric current...

Hydrogen-terminated diamond thin-film surfaces are relatively inert to impurity adsorption. Experiments have been conducted to quantitatively examine the adsorption of anthraquinone-2,6-disulfonate (AQDS). This molecule adsorbs strongly on glassy carbon and weakly on the basal plane of HOPG (27). The cyclic voltammetric E_p values were obtained by cycling the electrode potential until a pseudo steady-state response was observed in 0.01 mM AQDS + 0.1 M HClO₄. The E_p value of 200 mV for diamond is notably larger than the 48 mV value for glassy carbon (0.2 V/s). This difference is indicative of slow electrode-reaction kinetics, probably because of a lack of mediating carbon-oxygen functionalities on the diamond surface. The peak shapes were very different for these electrodes. For glassy carbon, the peaks were sharp and narrow, and the peak current varied linearly with the potential sweep rate, consistent with a surface-confined redox process. For diamond, the peaks were broad with a time^-1/2 decay of the current on the high potential side of the peaks. The peak current varied linearly with the potential sweep rate^1/2.

A more conclusive test for AQDS adsorption was made by chronocoulometric measurements in which the electrode surface was exposed to 1 mM AQDS + 0.1 M HClO₄ for 15 min. The solution was removed, and the cell was rinsed and then refilled with 0.1 M HClO₄. From the data, oxidation charge versus time^1/2 plots were constructed to determine the amount of AQDS adsorbed. The data show that the charge associated with the adsorbed AQDS for glassy carbon is more than a 1000-fold larger than the value for diamond (284 vs. <1 µC/cm²). The resistance of boron-doped diamond thin films to deactivation during exposure to air and adsorption of macromolecules may allow diamond to be used in electroanalytical measurements in which the analyte solution may contain organic contaminants that tend to adsorb and foul conventional carbon electrodes.

Applications
Sodium azide, which is currently being phased out as a propellant for airbags, is not normally found in water supplies. However, as current-model cars containing azide-based airbags are retired to salvage yards, the likelihood of azide contamination of groundwater will increase significantly. To the best of our knowledge, no EPA-recommended protocol exists for detecting azide, nor does a mandated ceiling exist limiting exposure. GC, ion chromatography, and capillary electrophoresis detect azide anion with limits in the high-ppb to low-ppm range.

However, azide anion is electrochemically active at carbon (diamond and graphite), platinum, and gold electrodes (28-30). Surprisingly, only a few investigations of the electrochemistry of this anion have been conducted. Figure 4 shows a linear-sweep voltammetric current versus potential curve for a boron-doped diamond thin film in 1 mM NaN₃ + 0.1 M phosphate buffer, pH 7.2. Diamond gives a peak-shaped faradaic response for the oxidation of azide superimposed on a low, stable background current. The oxidation peak potential is 1100 mV, and the background-corrected peak current is 57µA. Plots of the peak current versus the potential sweep rate^1/2 were linear with a near-origin intercept. Plots of the peak current versus the azide solution concentration were linear from 0.003 down to 10^-6 M, an analytically useful response for this analyte.

Figure 4. Linear-sweep voltammetric current versus potential curve for a boron-doped diamond thin-film electrode exposed to 1 mM N^-₃ in 0.1 M phosphate buffer at pH 7.2 showing total and background current.

The response for diamond remained unchanged even after weeks of exposure to laboratory air and was unaffected by F^-, Cl^-, NO₃^-, and AQDS in solution. By comparison, the faradaic signals for diamond and freshly polished glassy carbon were most often within 5% of each other, whereas the background currents were always a factor of 5 to 100 larger for glassy carbon. This means an enhanced S/B for diamond. In fact, S/B for diamond films is 38 to 50X larger than the ratio for glassy carbon at the 0.1 mM azide concentration level. The enhanced S/B leads to lower detection limits and results from the sizably lower residual background current for boron-doped diamond at the azide oxidation potentials.

Flow injection analysis (FIA) studies in the amperometric detection mode were performed with a diamond-based thin-layer flow cell (31). The measurements were made at a constant potential of 1.25 V and a flow rate of 1 mL/min. The linear dynamic range for diamond is 4 orders of magnitude and for glassy carbon, 3 orders of magnitude. The sensitivity is 33 nA/µM, similar for both electrodes. The detection limit for diamond is 0.3 ppb (S/N = 3) and is lower than glassy carbon by a factor of nearly 7. The peak height variability is also significantly less for diamond, ranging from 0.5-5%.

Response stability is another key aspect of diamond thin-film performance. During a 12-h period of continuous use, the background current for diamond changed by <5%, whereas the current for glassy carbon increased by >300%, because surface oxidation occurred at the detection potential of 1.20 V. The S/B for injections of 1 mM N₃^- changed by <5% for diamond and decreased by more than 50% for glassy carbon during this time period. Azide oxidation is an example of a reaction that can be sensitively and stably detected at boron-doped diamond thin films because of a resistance to surface oxidation and an increased oxygen evolution overpotential. These experimental results clearly indicate that diamond outperforms glassy carbon and that this diamond-based electrochemical assay is a viable method for low-level detection of azide in wastewater.

Another application is anodic stripping voltammetry (ASV) of heavy-metal ions in aqueous media. Previously, our group reported on the electrodeposition of Pt, Hg, and Pb adlayers on boron-doped diamond thin films (5). ASV data for the determination of Pb⁺², Cu⁺², and Cd⁺² ions using Hg-coated diamond thin-film electrodes have also been reported (5, 13).

Figure 5 shows linear-sweep voltammetric current versus potential curves for the oxidation (i.e., stripping) of Pb from a Hg-coated diamond thin film in the flow injection mode as a function of the injected Pb⁺² volume. The deposition and stripping sequences were made in a specially designed thin-layer flow cell with a volume of ~20 µL during continuous flow of the mobile phase. Mercury (0.35 µg/cm²) was electrodeposited on the diamond surface, followed by injecting different volumes of Pb⁺² solution at an applied potential of -1.0 V. After each volume injected, the potential was scanned in the positive direction at 10 mV/s.

Figure 5. Linear-sweep voltammetric current versus potential curves for Pb stripping from a Hg-coated diamond thin film in the continuous-flow mode for different volumes of injected 1 mM Pb⁺² solution. The eluent was 0.1 M acetate buffer at pH 4.4.

The data reveal that the peak current and charge increase with the total injected solution volume containing Pb⁺². The oxidation peak potential shifts in the positive direction from -450 to -400 mV, with increasing injected volume consistent with an increased surface coverage of Pb. As the Pb coverage increases, the stripping peaks also develop a shoulder on the positive-potential side of the main peak. The response of this particular Hg-coated film was stable for more than two weeks of continuous use. A large hydrogen evolution overpotential potential for diamond allows for the efficient electroreduction of Hg. FIA studies revealed a linear dynamic range over 5 orders of magnitude, a sensitivity of 94 nC/M, and a detection limit of 21 ppb (S/N >2). The detection limit for glassy carbon is 42 ppb.

The surface roughness of diamond (Figure 1) is supposed to play an influential role in improving the analytical performance of the detector. Preliminary scanning electron microscopy studies of the Hg-coated diamond surface indicated that the microcrystallites may influence the deposit size (i.e., volume) by restricting the deposit to the grain boundary regions. The deposits on the diamond surface were observed to be uniformly distributed and possessed a relatively narrow distribution of sizes, much more so than deposits on glassy carbon. It is also supposed that the roughness serves to shield the deposits from the shearing forces of the fluid flow, as little Hg was lost even after two weeks of use. Hg deposits on a planar surface, such as glassy carbon, are often subject to removal by the fluid flow or to coalescence with other deposits, thereby increasing the Hg volume. Both of these effects are undesirable and lead to diminished detector performance.

Conclusion
The use of boron-doped diamond thin films in electrochemistry is an exciting new field of research for this technologically important material. The field is relatively new, and little is currently known about the structure-reactivity relationship. As the electrochemical properties of conductive and semiconductive diamond thin films become better understood, new electrochemical applications will develop.

REFERENCES

ACKNOWLEDGEMENTS