Characterization and Electrochemical Responsiveness ofBoron-Doped Nanocrystalline Diamond Thin-Film

Electrodes

Yoshiyuki Show, Małgorzata A. Witek, Prerna Sonthalia, and Greg M. Swain*

Department of Chemistry, Michigan State University, East Lansing, Michigan 48824-1322

Received September 9, 2002. Revised Manuscript Received December 12, 2002

The deposition, characterization, and electrochemical responsiveness of boron-dopednanocrystalline diamond thin-film electrodes is reported. The films consist of clusters ofdiamond grains, ∼50-100 nm in diameter, and possess an rms surface roughness of 34 nmover a 5 × 5 µm2 area. The individual and randomly ordered diamond grains areapproximately 10-15 nm in diameter, as evidenced by TEM. The ∼4-µm-thick films weredeposited by microwave-assisted chemical vapor deposition (CVD) using a CH4/H2/Ar sourcegas mixture (1%/5%/95%). Under these conditions, C2, rather than CH3

•, appears to be thedominant nucleation and growth precursor. The nanocrystallinity is a result of a growthand nucleation mechanism discovered by Gruen, which involves the insertion of C2 carbondimer into C-H bonds on the growth surface (MRS Bull. 1998, 23, 32). The nanocrystallinemorphology results from a high renucleation rate. However, unlike previously reportednanocrystalline diamond thin films that have electrical properties dominated by the highfraction of π-bonded carbon atoms in the grain boundaries, the present films are doped withboron, either using B2H6 or a solid-state boron diffusion source, and the electrical propertiesappear to be dominated by the charge carriers in the diamond. The films were characterizedby scanning-electron microscopy, atomic-force microscopy, transmission-electron microscopy,visible-Raman spectroscopy, X-ray diffraction, boron-nuclear-reaction analysis, and cyclicvoltammetry, using Fe(CN)6

3-/4-, Ru(NH3)63+/2+, IrCl6

2-/3-, methyl viologen, Fe3+/2+, and 4-tert-butylcatechol. Analytical application of this advanced carbon electrode material for thedetection of trace metal ions is discussed.

Introduction

High-quality diamond films can be formed with twodifferent morphologies and microstructures, microcrys-talline and nanocrystalline. The distinction betweenthese two structures arises from the nominal grain size,which for microcrystalline films is >1 µm and fornanocrystalline films is ∼20 nm. Conventional micro-crystalline diamond CVD growth uses hydrocarbon-hydrogen (e.g., 1% CH4/99% H2) gas mixtures and it isknown under such growth conditions that hydrogenplays a number of critical roles.2-8 Among these arestabilization of the diamond lattice and removal of sp2-bonded carbon nuclei, when formed, due to preferentialgasification over sp3-bonded diamond. Gruen discoveredthat phase-pure nanocrystalline diamond can be grownfrom CH4/Ar gas mixtures containing very little or noadded hydrogen.1,6-10 There are two kinds of nanocrys-

talline diamond films often described. The first are filmsdeposited from high CH4/H2 (>3%) gas mixtures. Thesefilms have nanometer-sized features due to the high rateof nucleation, but are generally of low quality (so-called“dirty” diamond) with significant levels of secondarynucleation and sp2- bonded carbon impurity phases. Thesecond are films deposited from CH4/Ar (∼1%) gasmixtures. These films consist of randomly oriented,nanometer-sized grains of phase-pure diamond and aregenerally of higher quality. The grains are on the orderof 20 nm in diameter, and the grain boundaries consistof π-bonded, carbon atoms (ca. 2-4 atoms wide). Theπ-bonded grain boundaries have a profound effect on themechanical, electrical, and optical properties of theselatter films.

Our work involves nanocrystalline diamond thin filmsof the second type. The most remarkable difference infilms grown using hydrogen-poor Ar gas mixtures,compared with conventional hydrogen-rich mixtures, isthe nanocrystallinity of the former compared with themicrocrystallinity of the latter. The nanocrystallinity isa result of a growth and nucleation mechanism involv-ing the insertion of carbon dimer, C2, into surface C-Hbonds. The C2 addition is believed to occur by a two-

(1) Gruen, D. M. MRS Bull. 1998, 23, 32.(2) Angus, J. C.; Will, H. A.; Stankop, W. S. J. Appl. Phys. 1968,

39, 2915.(3) Hsu, W. L. J. Vac. Sci. Technol. 1988, A6, 1803.(4) Frenklach, M. J. Appl. Phys. 1989, 65, 5124.(5) Butler, J. E.; Windischmann, H. MRS Bull. 1998, 23, 22.(6) Zhou, D.; Gruen, D. M.; Qin, L. C.; McCauley, T. G.; Krauss, A.

R. J. Appl. Phys. 1998, 84, 1981.(7) Jiao, S.; Sumant A.; Kira, M. A.; Gruen, D. M.; Krauss, A. R.;

Auciello, O. J. Appl. Phys. 2001, 90, 118.(8) McCauley, T. G.; Gruen, D. M.; Krauss, A. R. Appl. Phys. Lett.

1998, 73, 1646.(9) Gruen, D. M. Annu. Rev. Mater. Sci. 1999, 29, 211.

(10) Bhattacharyya, S.; Auciello, O.; Birrell, J.; Carlisle, J. A.;Curtiss, L. A.; Goyette, A. N.; Gruen, D. M.; Krass, A. R.; Schlueter,J.; Sumant, A.; Zapol, P. Appl. Phys. Lett. 2001, 79, 1.

879Chem. Mater. 2003, 15, 879-888

10.1021/cm020927t CCC: $25.00 © 2003 American Chemical SocietyPublished on Web 01/30/2003

step growth mechanism.8 A C2 molecule approaches theunreconstructed monohydride surface and inserts intoa C-H bond. The C2 molecule then rotates to insert itsother carbon into a neighboring C-H bond on thesurface. A C2 molecule then inserts into an adjacentC-H bond, parallel to the newly inserted C2 dimer. Theoriginal state of the surface is recovered by the forma-tion of a bond between carbon atoms in the adjacentsurface dimers. Very high rates of heterogeneous nucle-ation are observed, on the order of 1010 cm-2, and theresulting films consist of randomly oriented, phase-purediamond grains with well-defined grain boundaries.9The smooth nanocrystalline films possess interestingmechanical, tribological, and electrical properties owingto the small grain size. For example, the films transitionfrom an electrically insulating to an electrically con-ducting material with a reduction in crystallite size fromthe micrometer to the nanometer level.10 This is largelydue to the presence of high-energy, high-angle grainboundaries that contain π-bonded carbon atoms (i.e., ahigh density of electronic states over a wide energy orpotential range). The grain boundaries are conductingbecause of π-states, and since their numbers vastlyincrease with decreasing crystallite size, the entire filmbecomes electrically conducting and functions as anelectrode material.11 Theoretical calculations suggestthat localized electronic states are introduced into theband gap of these films due to the presence of sp2-bonded dimers and sp3-hybridzed dangling bonds in thegrain boundaries.9 There is a lack of spatial connectivityamong the sp2-bonded carbon sites; therefore, the as-sociated gap states do not form an extended π-systembut rather are localized.

We recently reported on the structural and electro-chemical characterization of nitrogen-incorporated nano-crystalline diamond thin-film electrodes, deposited fromCH4/N2/Ar gas mixtures.11 The electrical conductivityof these nitrogen-containing films increases with in-creasing nitrogen content up to about 5%, and the filmsare generally more conductive than the early forms ofnanocrystalline diamond.11,12 Recent Hall Effect mea-surements (mobility and carrier concentration) for filmsdeposited with 10 and 20% N2 revealed carrier concen-trations of 2.0 × 1019 and 1.5 × 1020 cm-3, respectively.10

The room-temperature carrier mobilities were 5 and 10cm2/V‚s, respectively. A negative Hall coefficient indi-cated that electrons are the majority charge carrier. Anexplanation for the effect of nitrogen is that the impuritycauses microstructural changes within the grain bound-aries (i.e., increased π-bonding), resulting in an increasein the localized density of electronic states. Computa-tions indicate that the incorporation of nitrogen into thegrain boundaries is energetically favored by 3-5 eV oversubstitutional insertion into the grains.9

The electrical properties of the nitrogen-containingnanocrystalline diamond films are largely influenced bythe π-bonding in the grain boundaries.9-13 While theseelectrode materials possess good electrochemical behav-ior, much like those for high-quality microcrystalline

diamond films, their electrical response is stronglylinked to the physicochemical properties of the grainboundaries. Therefore, the electrochemical response canbe strongly influenced by changes in the π-bonded grainboundary atoms. This is particularly true during expo-sure to chemically harsh solutions that can cause theoxidative etching or disruption of the π-bonded grainboundary atoms. It would be better if the nanocrystal-line diamond films exhibited a through-grain conductionmechanism as a result of impurity incorporation, suchas boron doping. Such films should exhibit electricalconductivity that scales with the doping level and shouldhave electrochemical properties that are largely unaf-fected by changes in the physicochemical properties ofgrain boundaries.

We report on the characterization and electrochemicalresponsiveness of boron-doped nanocrystalline diamondthin-film electrodes. The boron doping imparts electricalconductivity to the entire film (i.e., through-grainconduction); hence, the electrical conductivity is nolonger dominated by the grain boundaries. The filmswere deposited from a 1% CH4/5% H2/94% Ar source gasmixture using 800 W of microwave power and a systempressure of 140 Torr. The continuous films were depos-ited for 2 h with a film thickness of about 4 µm and anapparent in-plane electrical resistivity of 0.2 Ω-cm. Thefilms were characterized by scanning electron micros-copy (SEM), atomic force microscopy (AFM), transmis-sion electron microscopy (TEM), X-ray diffraction (XRD),visible-Raman spectroscopy, and cyclic voltammetry,using Fe(CN)6

3-/4-, Ru(NH3)63+/2+, IrCl6

3-/4-, methylviologen, Fe3+/2+, and 4-tert-butylcatechol. Electroana-lytical application of the electrodes for the determinationof metal ions, Cd2+, Pb2+, Cu2+, and Ag+, via anodicstripping voltammetry is discussed. The results indicatethat (i) highly conducting nanocrystalline diamond thinfilms can be deposited in a short period of time (i.e., costsavings), (ii) the phase-pure, smooth films consist of∼20-nm diamond grains, (iii) the films have electricalconductivity that scales with the boron-doping level andis largely independent of the physicochemical propertiesof the grain boundaries, and (iv) the films possesselectrochemical properties similar to those observed forhigh-quality, hydrogen-terminated, microcrystalline dia-mond thin films.

Experimental Section

Electrode Fabrication. The boron-doped nanocrystallinediamond thin films were deposited on p-type Si(100) substrates(∼10-3 Ω-cm, Virginia Semiconductor Inc., Fredricksburg, VA)using a commercial microwave-assisted, chemical vapor depo-sition (CVD) system (1.5 kW, ASTeX Inc., Lowell, MA). Thesurface of the Si substrate was mechanically scratched on afelt polishing pad with 1-µm-diameter diamond powder (GESuperabrasives, Worthington, OH). The scratched substratewas then sequentially washed with ultrapure water, isopropylalcohol (IPA), acetone, IPA, and ultrapure water to removepolishing debris from the scratches. This is a critical step inthe pretreatment. The cleaned substrate was then placed inthe CVD reactor on a molybdenum substrate holder. Thescratching treatment enhances the nucleation of small dia-mond particles during the initial growth stage. This leads tothe formation of a thin and continuous nanocrystalline dia-mond film in a relatively short period of time. Ultrahigh purityCH4, Ar, and H2 (99.999%) were used as the source gases. Thegas flow rates were 1, 94, and 5 sccm, respectively. Themicrowave power and system pressure were maintained at 800

(11) Chen, Q.; Gruen, D. M.; Krauss, A. R.; Corrigan, T. D.; Witek,M.; Swain, G. M. J. Electrochem. Soc. 2001, 148, 44.

(12) Fausett, B.; Granger, M. C.; Hupert, M. L.; Wang, J.; Swain,G. M.; Gruen, D. M. Electroanal. 2000, 12, 7.

(13) Keblinski, P.; Wolf, D.; Phillpot, S. R.; Gleiter, H. J. Mater.Res. 1998, 13, 2077.

880 Chem. Mater., Vol. 15, No. 4, 2003 Show et al.

W and 140 Torr, respectively. The substrate temperature wasestimated, by an optical pyrometer, to be about 800 °C. Thedeposition time was 2 h and the resulting nanocrystallinediamond thin film was approximately 4-µm-thick, as estimatedfrom the weight change.

Boron doping of the nanocrystalline diamond was ac-complished using either a solid-state or gas-phase boronsource. A BoronPlus (GS 126, Techneglas, Inc., Perrysburg,OH) solid-state ceramic source was placed under the Sisubstrate on the molybdenum stage. Boron diffuses from theceramic matrix as B2O3 and is incorporated into the film duringgrowth. On the other hand, B2H6, diluted in H2, was used asa gas-phase source and was introduced into the source gasmixture. The concentration of B2H6 was 10 ppm.

The plasma was ignited and the growth was initiated withall gases flowing into the reactor. At the end of the depositionperiod, the CH4 flow was stopped and the Ar and H2 flowscontinued. The diamond film remained exposed to the H2/Arplasma for approximately 10 min at the deposition conditions.The Ar flow was then stopped and the substrate was cooledin an H2 plasma (i.e., atomic hydrogen) to an estimatedtemperature of <400 °C by slowly reducing the power andpressure over a 4-min period. Post growth annealing in atomichydrogen is essential for etching away adventitious nondia-mond carbon impurity, minimizing dangling bonds, and ensur-ing full hydrogen termination.

Material Characterization. The surface morphology ofthe boron-doped nanocrystalline diamond thin films wasinvestigated using scanning electron microscopy (SEM) (Hi-tachi, S-4500, Japan), atomic force microscopy (Nanoscope IIIa,Digital Instruments/Veeco Metrology Group, Santa Barbara,CA), and transmission electron microscopy (TEM) (JEM-4000EX, JEOL, Japan). The sample for the TEM investigationswas prepared by growing a thin nanocrystalline diamond filmfor 10 min on a Si substrate and dissolving the substrate usingan HF/HNO3 solution. The free-standing pieces of diamondwere then collected from solution on a copper TEM grid. Thebulk crystallinity and microstructure were characterized byX-ray diffraction (1.540 Å, X-pert MRD, Philips, Netherlands),and visible-Raman spectroscopy (Raman 2000, Chromex Inc.,Albuquerque, NM), respectively. The 2θ XRD measurementswere performed from 30 to 100°. The Raman spectrographconsisted of a diode-pumped, frequency-doubled Nd:YAG laser(532-nm excitation at 50 mW), a multichannel detector, and afocusing microscope with a 20× objective lens. The borondopant concentration was determined by boron nuclear-reaction analysis (Surface Characterization Facility, CaseWestern Reserve University). Calibration was performed witha piece of high-quality boron nitride. The carrier concentrationwas determined through a Hall measurement (Model HL5500,Bio-Rad Microscience, England) performed at room tempera-ture under a magnetic field of 0.492 T. A square van der Pauwarrangement using Ti contacts was employed. Water contactangles (First Ten Angstroms, Portsmonth, VA) of ca. 80° weremeasured and were stable with time, indicative of a hydro-phobic surface.

Electrochemistry. Most of the electrochemical measure-ments were made with a CYSY-2000 computerized poten-tiostat (Cypress System Inc., Lawrence, KS) using a single-compartment, three-electrode glass cell. The diamond workingelectrode was pressed against a Viton O-ring and clamped tothe bottom of the cell. Ohmic contact was made with analuminum plate after placing a bead of In/Ga alloy on thebackside of the scratched and cleaned substrate. A graphiterod was used as the counter electrode and a commercial Ag/AgCl electrode (3 M KCl) served as the reference (E°Ag/AgCl )-0.045 V vs SCE). The geometric area of the working electrodewas ca. 0.2 cm2. All measurements were performed at roomtemperature, ∼25 °C. The supporting electrolytes were 1.0 MKCl and 0.1 M HClO4. The concentrations of Fe(CN)6

3-/4-,Ru(NH3)6

2+/3+, 4-tert-butylcatechol, and Fe2+/3+ were 1.0 mM,IrCl6

2-/3- was 0.25 mM, and methyl viologen was 0.5 mM.The differential pulse voltammetry for the trace metal ion

analysis was performed on a Model 650A computerized po-tentiostat (CH Instruments, Austin, TX). All measurements

were made in solutions deoxygenated with N2 for at least 10min initially and then for 2 min after each anodic strippingsweep. The solutions were blanketed with the gas during allmeasurements and the electrochemical cell placed inside anelectrically grounded Faraday cage. The electrolyte solutionwas 0.1 M acetate buffer, pH 4.5. The commercial referenceelectrode was isolated from the main electrolyte solution, usinga cracked-glass tube (i.e., double junction). The tube was filledwith the acetate buffer solution. The anodic-stripping volta-mmetric measurements used a 3-min deposition time with nostirring and a 0.5-min “quiet” time prior to initiation of theanodic sweep. The differential pulse voltammetry settings wereas follows: 100-mV pulse height; 2-mV step height; 50-mspulse width; 35-ms sampling time; and a constant potentialof 600 mV for 120 s after completion of the anodic sweep tofully oxidize all metal deposits.

Chemicals. All solutions were prepared with ultrapurewater (>17 MΩ) from an E-pure purification system (Barn-stead). The KCl, NaCl, K4Fe(CN)6, Cl3Ru(NH3)6, K2IrCl6, 1,1′-dimethyl-4,4′-bipyridinum dichloride (methyl viologen), and4-tert-butylcatechol (Aldrich Chemical), Fe2(SO4)3‚6H2O (Mathe-son Coleman & Bell), HClO4 (ultrapure, Aldrich Chemical),and sodium hydroxide (Fisher Scientific) were reagent-gradequality, used as received. Silver nitrate (Fisher), cadmiumnitrate (Aldrich), cupric nitrate (Aldrich), lead nitrate (Ald-rich), and zinc nitrate (Aldrich) were all reagent-grade quality,used without additional purification. Acetate buffer (0.1 M)was prepared by mixing appropriate amounts of 99% sodiumacetate (Aldrich) and acetic acid (Aldrich). All solutions wereprepared fresh daily and purged with N2 (99.999%) for 10 minprior to any electrochemical measurement. All glassware wascleaned by a three-step procedure: ethanol/KOH bath, alconox/ultrapure water solution, and ultrapure water rinse.

Safety: B2H6 (0.1% diluted in hydrogen) is a hazardous gasthat should be appropriately contained within a vented gascabinet. All other reagents and chemicals can be used withroutine safety precautions.

Results and Discussion

Material Characterization. Figure 1 shows a top-view SEM image of a boron-doped nanocrystallinediamond thin film deposited for 2 h. The smooth anduniformly coated film is composed of nodular features∼50-100 nm in diameter. No voids or cracks were foundin the coating and the thickness was uniform over theentire substrate. AFM images were also acquired toprobe the film morphology. The root-mean-square sur-face roughness was found to be 34 nm over a 5 × 5 µm2

area and was independent of the growth time (i.e., filmthickness). The nodular features are actually clustersof individual diamond grains formed as a result of thehigh nucleation rate.

Figure 1. SEM image of a boron-doped nanocrystallinediamond thin film.

Boron-Doped Nanocrystalline Diamond Electrodes Chem. Mater., Vol. 15, No. 4, 2003 881

Figure 2 shows a TEM image for the film, which isrepresentative of several that were taken. The filmconsists of diamond grains with diameters from 10 to15 nm. Aggregates of these diamond grains form thenodular features seen in Figure 1. Moreover, eachdiamond grain is a single crystal and randomly orientedin the film. This conclusion is based on the appearanceof the lattice-fringe patterns and their orientationaldifferences from grain-to-grain. A rough estimate of thefringe spacing was made, even though these are nothigh-resolution images. The estimated spacing on atleast a couple of grains is ∼0.20 nm. This is close to the0.206-nm interplanar distance for diamond 111 planes.The transmission-electron diffraction (TED) patternfrom one of the grains, an insert in Figure 2, shows anintense spotted-ring pattern, indicative of atomicallyordered but randomly oriented grains. The diffractionpattern is indexed to the (111), (022), (113), (004), (133),(224), (115), and (333) planes of cubic diamond (ASTM6-0675).

Figure 3 shows an XRD spectrum for the film. Threebroad reflections are observed at 2θ values of 43.8, 75.5,and 91.5°. These reflections are assigned to the (111),(220), and (311) planes of the cubic diamond, respec-tively. The peaks are broader than they are for amicrocrystalline film of the same approximate thicknessdue to the 2 orders of magnitude, or more, smallerdiamond grain size in the nanocrystalline film. Table 1summarizes the calculated lattice spacings and the

relative peak intensities. Reference data for cubicdiamond powder (ASTM or PDF 6-0675) are alsopresented, for comparison. The diffraction data revealthat the bulk film structure is sp3-bonded diamond.

Figure 4 shows a visible-Raman spectrum for the film.This spectrum is characteristic of high-quality nano-crystalline diamond films. A single, sharp peak at 1332cm-1 in the Raman spectrum is frequently used as asignature of high-quality, single-crystal, or large-graineddiamond.13,14 The spectrum for the nanocrystalline filmis quite different. Broad peaks are seen at 1150, 1333,1470, and 1550 cm-1. The peak at 1333 cm-1, atop alarge background signal, is associated with the first-order phonon mode of sp3-bonded diamond. Due toresonance effects, the Raman cross-section scatteringcoefficient (visible excitation) for sp2-bonded carbon islarger (50×) than that for the sp3-bonded carbon, andthe scattering intensity for the former can often greatlyexceed that for the latter.15 The peak width (fwhm) forthe diamond line is much broader than that for micro-crystalline diamond, 140 versus 10 cm-1. This is becauseof the small grain size in the nanocrystalline film. Thereare two possible causes for the line broadening. Onepossiblity is the well-established confinement model.16

This model states that the smaller the domain size, thelarger the range of phonon modes (with different qvector and energy) that are allowed to participate in theRaman process. Hence, the line width results from thespread in phonon energy. Another, and more likely,explanation is phonon scattering by impurities anddefects (i.e., grain boundaries).16 The scattering eventshortens the lifetime of the phonons and thus broadensthe Raman line.

The peak at 1150 cm-1 is often used as a signaturefor high-quality nanocrystalline diamond.14 Prawer andco-workers, through the study of clean nanocrystallinediamond particles (∼5-nm diameter), have attributed

(14) Nemanich, R. J.; Glass, J. T.; Lucovsky, G.; Shroder, R. E. J.Vac. Sci. Technol. 1988, A6, 1783.

(15) Knight, D. S.; White, W. B. J. Mater. Res. 1989, 4, 385.(16) Bergman, L.; Nemanich, R. J. J. Appl. Phys. 1995, 78, 6709.

Figure 2. TEM image and TED pattern for a boron-dopednanocrystalline diamond thin film.

Figure 3. X-ray diffraction pattern for a boron-doped nano-crystalline diamond thin film.

Figure 4. Visible Raman spectrum for a boron-doped nano-crystalline diamond thin film. λex ) 532 nm. Laser power )50 mW. Integration time ) 5 s.

Table 1. XRD Peak Intensities and Calculated LatticeSpacing for a Boron-Doped Nanocrystalline Diamond

Thin Film

measured ASTM (6-0675)

planediffractionangle (deg)

fwhm(deg)

spacing(Å)

relativeintensity

spacing(Å)

relativeintensity

(111) 44 0.75 2.06 100 2.06 100(220) 75.5 0.65 1.26 23 1.26 25(311) 91.5 1.0 1.07 13 1.07 16

882 Chem. Mater., Vol. 15, No. 4, 2003 Show et al.

this peak to a surface phonon mode of diamond.17 Onthe other hand, Ferrari and Robertson have madearguments for this peak being associated with sp2-bonded carbon, specifically transpolyactelylene seg-ments at grain boundaries.18 Their assignment of sp2-rather than sp3-bonded carbon, as has often beenproposed,14 is based on the observations that the peakposition changes with excitation energy, the peakintensity decreases with increasing excitation energy,and the peak is always accompanied by another peakat ∼1450 cm-1, which behaves similarly with excitationenergy. We, therefore, tentatively assign the peaks at1470 and 1550 cm-1 to disordered sp2-bonded carbonin the grain boundaries. The 1550-cm-1 peak is down-shifted from the expected 1580-cm-1 position for graph-ite. This shift results because the sp2-bonded carbon isamorphous and is mixed with sp3-bonded carbon. It isimportant to note that, very likely, the sp2-bondedcarbon is confined to the grain boundaries of thenanocrystalline film, producing a network of 3- and4-fold coordinated carbon atoms.19 Additional researchis needed to confirm the origins of the 1150- and 1470-cm-1 bands.

The boron concentration in the film deposited with10 ppm B2H6 was determined to be 810 ppm (1.43 ×1020 B/cm3 by boron nuclear-reaction analysis. Prelimi-nary Hall measurements indicated the major carrier tobe the hole (positive Hall coefficient), and the carrierconcentration and conductivity to be 6.4 × 1017 cm-3, anumber that seems low, and 10 Ω-1cm-1, respectively.The carrier mobility was found to be 90.4 cm-2/V‚s. Morework is required to accurately determine the carrierconcentration, mobility, and carrier activation energy.

Electrochemical Responsiveness. Figure 5A showsa background cyclic voltammetric i-E curve for a boron-doped nanocrystalline diamond thin film in 1.0 M KCl.The curve is largely featureless over the potential rangeand is stable with cycling. There are no obvious peakspresent, associated with redox-active surface carbon-oxygen functionalities, although there is a small anodiccharge passed between 500 and 800 mV, just prior tothe onset of chlorine evolution.20,21 The curve shape issimilar to that for boron-doped microcrystalline dia-mond; however, the current magnitude of the former isslightly higher. For instance, at 250 mV, the anodiccurrent for the nanocrystalline film is 0.96 µA (4.8 µA/cm2), whereas the current for the microcrystalline filmsis about a factor of 1.5 less at 0.6 µA (3.0 µA/cm2). Thehigher current may be due to surface faradaic processesbecause of the increased fraction of exposed π-bondedcarbon in the grain boundaries. This carbon is a sourceof π states that contribute to the density of electronicstates in the material, increasing the capacitive com-ponent of the background current. Some of this sp2-bonded carbon could also be electrochemically active at

the potentials probed, giving rise to a faradaic compo-nent in the background current. In fact, the backgroundvoltammetric currents for both diamond electrodes aretoo high to be solely due to electric double-layer charging(∼10 µF/cm2). From the currents, one calculates appar-ent double-layer capacitances of 48 and 30 µF/cm2 forthe nanocrystalline and microcrystalline films, respec-tively. The higher apparent capacitance is due, in part,to a low estimate of the electrode area. All currents arenormalized to the geometric area of the films. However,both film types have roughness factors >1. Also, asmentioned above, some of the background current islikely due to surface faradaic processes (i.e., oxidationof the carbon in the grain boundaries and in the grains).Importantly, the current for both diamond electrodesis significantly less than that for freshly polished glassycarbon, 7-10 µA (35-50 µA/cm2). The low backgroundcurrent is a characteristic feature of diamond electrodesand leads to improved SBR in electroanalytical mea-surements.20,21

Figure 5B shows a background cyclic voltammetrici-E curve for the same boron-doped nanocrystallinediamond in 1.0 M KCl over a much wider potentialrange. The working potential window is about 3.1 V((100 µA or 500 µA/cm2) with a largely featurelessresponse between the potential limits. The anodiccurrent at 1200 mV is due to the oxidation of chlorideto chlorine. The cathodic current flowing at -1800 mVis due to the reduction of water to hydrogen. There is acurrent crossover at about -1700 mV, indicative of anactivation overpotential necessary for the initiation ofthe hydrogen-evolution reaction. This feature is often

(17) Prawer, S.; Nugent, K. W.; Jamieson, D. N.; Orwa, J. O.;Bursill, L. A.; Peng, J. L. Chem. Phys. Lett. 2000, 332, 93.

(18) Ferrari, A. C.; Robertson, J. Phys. Rev. B 2001, 63, 121405.(19) Gruen, D. M.; Krauss. A. R.; Zuiker, C. D.; Csencsits, R.;

Terminello, L. T., Carlisle, J. A.; Jimenez, I.; Sutherland, D. G. J.; Shu,D. K.; Tong, W.; Himpsel, F. J. Appl. Phys. Lett. 1996, 68, 1640.

(20) Granger, M. C.; Xu, J.; Strojek, J. W.; Swain, G. M. Anal. Chim.Acta 1999, 397, 145.

(21) Granger, M. C.; Witek, M.; Xu, J.; Wang, J.; Hupert, M.; Hanks,A.; Koppang, M. D.; Butler, J. E.; Lucazeau, G.; Mermoux, M.; Strojek,J. W.; Swain, G. M. Anal. Chem. 2000, 72, 3793.

Figure 5. Background cyclic voltammetric i-E curves in (A)1.0 M KCl (narrow potential range) and (B) 1.0 M KCl(working potential window) for a nanocrystalline boron-dopeddiamond thin film. Scan rate ) 100 mV/s. Electrode geometricarea ) 0.2 cm2.

Boron-Doped Nanocrystalline Diamond Electrodes Chem. Mater., Vol. 15, No. 4, 2003 883

observed for diamond films; however, the mechanisticcause of this overpotential is currently unknown. A lowbackground current and wide working potential window(3-3.5 V) are characteristic features of both high-quality, nitrogen-incorporated nanocrystalline and boron-doped microcrystalline diamond.11,20,21 The backgroundcyclic voltammetric data presented in Figure 5A,B arevery similar to what has been reported for high-qualityand well-characterized microcrystalline and ultra-nanocrystalline diamond.12,20-22

The electrochemical responsiveness of the boron-doped nanocrystalline diamond toward six redox sys-tems was investigated using cyclic voltammetry. Theinfluence of diamond’s physical, chemical, and electronicproperties on the electrode reaction kinetics and mech-anism for these systems has previously been discussed.21

Figure 6 shows cyclic voltammetric i-E curves for (A)Fe(CN)6

3-/4-, (B) Ru(NH3)62+/3+, (C) IrCl62-/3-, (D) meth-

yl viologen (MV+2/+) in 1 M KCl, (E) 4-tert-butylcatechol,and (F) Fe3+/2+ in 0.1 M HClO4 . The potential scan rate(ν) was 0.1 V/s. The E1/2 for these redox systems rangesfrom approximately +800 to -1100 mV, so they are veryuseful for probing the film’s electronic properties overa wide potential range. A summary of some of the cyclicvoltammetric data is provided in Table 2.

A reversible response is observed for Fe(CN)63-/4-

with a ∆Ep of 63 mV and an ipox/ip

red of 1.0. Fe(CN)63-/4-

is a surface-sensitive redox system on both glassy carbonand boron-doped microcrystalline diamond.21-23 Theelectrode reaction kinetics for this couple are stronglyinfluenced by the amount of exposed edge plane on sp2-bonded carbon, as well as the surface cleanliness.23

Granger et al. showed that surface carbon-oxygenfunctionalities on microcrystalline diamond significantlyinfluence ∆Ep with increasing oxygen content, causingan increase in the peak potential separation.24 A similareffect was also observed by Fujishima and co-workers.25

The inhibiting effect of the surface oxygen is reversedif the film is rehydrogenated in a hydrogen plasma.24

Apparently, the oxygen blocks a surface site that isinvolved in the reaction on the hydrogen-terminatedsurface. The small ∆Ep seen for the boron-doped nanoc-rystalline diamond film is indicative of a high level ofsurface cleanliness and low surface oxide coverage. Inother words, this result suggests the diamond surfaceis largely hydrogenated and clean surface sites exist forthis reaction.24

A reversible response is seen for Ru(NH3)63+/2+ with

a ∆Ep of 60 mV and an ipox/ip

red of 0.99. A reversibleresponse is also seen for IrCl6

2-/3- with a ∆Ep of 61 mVand an ip

ox/ipred of 0.98. The electrode-reaction kinetics

for both of these systems are relatively insensitive to

(22) Wang, J.; Swain, G. M.; Mermoux, M.; Lucazeau, G.; Zak, J.;Strojek, J. W. New Diamond Front. Carbon Technol. 1999, 9, 317.

(23) Chen, P.; McCreery, R. L. Anal. Chem. 1996, 68, 3958.(24) Granger, M. C.; Swain, G. M. J. Electrochem. Soc. 1999, 146,

4551.(25) Yagi, I.; Notsu, H.; Kondo, T.; Tryk. D. A.; Fujishima A. J.

Electroanal. Chem. 1999, 173, 473.

Figure 6. Cyclic voltammetric i-E curves for (A) 1.0 mM Fe(CN)63-/4-, (B) 1.0 mM Ru(NH3)6

2+/3+, (C) 0.25 mM IrCl62-/3-, (D)

0.50 mM methyl viologen (MV+2/+) in 1 M KCl, (E) 1.0 mM 4-tert-butylcatechol, and (F) 1.0 mM Fe2+/3+ in 0.1 M HClO4 for ananocrystalline boron-doped diamond thin film. Scan rate ) 100 mV/s. Electrode geometric area ) 0.2 cm2.

Table 2. Summary of Cyclic Voltammetric Data for a Boron-Doped Nanocrystalline Diamond Thin-Film Electrode

analyte ∆Ep (mV) Epox (mV) Ip

ox (µA) Ipox/Ip

red

0.5 mM MV2+/+/1 M KCl 60 -633 34.5 1.11 mM Ru(NH3)6

3+/2+/1 M KCl 59 -135 62.7 0.991 mM Fe(CN)6

3-/4-/1 M KCl 63 309 69.4 1.021 mM tert-butylcatechol/0.1 M HClO4 419 655 112.2 1.80.5 mM IrCl6

2-/3-/1 M KCl 61 823 22.7 0.981 mM Fe2+/3+/0.1 M HClO4 679 926 36.0 0.911 mM cadaverine/0.01 M BBpH10.6 irreversible 927 136.0 irreversible

884 Chem. Mater., Vol. 15, No. 4, 2003 Show et al.

the physicochemical properties of both sp2-bonded car-bon and diamond electrodes.21-23 The kinetics aremainly influenced by the density of electronic states atthe formal potentials for the couples. In other words,the nanocrystalline diamond electrode possesses a suf-ficient charge-carrier density at both -200 mV, apotential well negative of the apparent flatband poten-tial (ca. 500 mV vs Ag/AgCl) for a hydrogen-terminatedp-type (semiconducting) diamond, and 800 mV, a valuepositive of the flatband potential.26-28

A reversible response is seen for MV+2/+ with a ∆Epof 60 mV and an ip

ox/ipred of 1.1. Like Ru(NH3)6

2+/3+ andIrCl6

2-/3-, the electrode reaction kinetics for MV+2/+ arerelatively insensitive to the physicochemical propertiesof both sp2-bonded carbon and diamond electrodes.21,22,29

The kinetics are mainly influenced by the density ofelectronic states at the formal potential. In fact, goodresponsiveness was also observed for the MV+1/0 redoxcouple at an even more negative potential of -950 mV.The cathodic peak at -1025 mV is associated with thereduction of MV+• to MV0. The peak shape is consistentwith a diffusion-limited reaction. However, the corre-sponding oxidation peak at -1010 mV for the MV0 toMV+• transition does not have the shape expected for adiffusion-limited process but, rather, is sharp andnarrow, consistent with an oxidative desorption event.MV0 has limited solubility in aqueous media, anddepending on the solution conditions, MV concentration,and the electrode surface properties, MV0 can adsorband accumulate on the electrode surface.30 This is thecase presently. The sharp oxidation peak results fromthe oxidative desorption of surface-confined MV0.

The cyclic voltammetric i-E curves for 4-tert-butyl-catechol (t-BC) and Fe3+/2+ have much larger ∆Ep’s andmore asymmetric peak shapes. ∆Ep’s of 419 and 679 mVare observed for t-BC and Fe3+/2+, respectively. The ip

ox/ip

red ratios are 1.8 and 0.91 for t-BC and Fe3+/2+. Theasymmetry for t-BC is particularly evidentsthis is dueto the transfer coefficient being <0.5. The larger peakseparations, as compared to the other four redox ana-lytes, are due to more sluggish electrode reactionkinetics.21,22 The catechol redox reaction involves thetransfer of both electrons and protons, and the electrodekinetics are highly sensitive to the surface microstruc-ture, the presence of surface carbon-oxygen function-alities, and the surface cleanliness of sp2-bonded carbonelectrodes.23 Surface adsorption appears to lower thereorganization energy for this analyte. In other words,adsorption increases the reaction kinetics.31 It is sup-posed that the slow kinetics for diamond result from alack of adsorption on the sp3-bonded, hydrogen-termi-nated surface. Recent results have demonstrated a clearcorrelation between the fraction of exposed sp2-bondedcarbon in microcrystalline and nanocrystalline diamondfilms and the surface coverage of adsorbed catechols.32

The greater the coverage is, the smaller ∆Ep is.

The electrode kinetics for Fe3+/2+ are strongly influ-enced by the presence of surface carbon-oxygen func-tional groups, specifically carbonyl groups, which cata-lyze the reaction on sp2-bonded electrodes.33 Thehydrogen-terminated diamond surface is void of suchfunctionalities, so this is the postulated reason for thesluggish kinetics. Increasing the surface oxygen contenton diamond has been observed to slightly decrease the∆Ep.25

Figure 7 presents plots of ipox versus ν1/2 for the

different redox analytes. It can be seen that the oxida-tion peak current for each redox system varies linearlywith the scan rate1/2 with a near-zero y-axis intercept,indicative of reactions limited by semi-infinite lineardiffusion of reactants to the electrode surface. The quasi-reversible to reversible voltammetry for all the couplesindicates that the boron-doped nanocrystalline diamondhas a sufficient charge-carrier density over the widepotential range, a prerequisite for all redox reactions,to support a rapid electron transfer.

The electrical conductivity of previously reportednanocrystalline diamond thin-film electrodes, both asdeposited and as deposited with incorporated nitrogen,largely results from the π-bonding in the intercrystallinegrain boundaries.11 If this π-bonding is removed, thenthe localized density of electronic states is reduced andthe electrical conductivity decreases significantly. Hence,the electrochemical responsiveness of these films depends,to a great extent, on the chemical and electronic proper-ties of the grain boundaries. The electrochemical re-sponsiveness of the boron-doped nanocrystalline thinfilms should be much less influenced by the chemicaland electronic properties of the grain boundaries. Thediamond grains themselves should be highly conductingdue to the carriers provided by the substitutionallyinserted boron dopant atoms. To test this, an acidwashing and hydrogen plasma treatment was applied,which is very effective at oxidatively removing sp2-bonded carbon.21 The first step involved immersing thefilms in warm aqua regia for 30 min. The films werethen rinsed with ultrapure water. The second stepinvolved exposing the samples to a warm solution of

(26) Alehashem, S.; Chambers, F.; Strojek, J.; Swain, G. M.;Ramesham, R. Anal. Chem. 1995, 67, 2812.

(27) Fujishima, A.; Rao, T. N.; Tryk, D. A. Proc. Electrochem. Soc.2000, 99-32, 383.

(28) Pleskov, Y. V.; Mazin, V. M.; Evstefeeva, Y. E.; Varmin, V. P.;Teremetskaya, I. G.; Laptev, V. A. Electrochem. Solid State Lett. 2000,3, 141.

(29) Qiu, F.; Compton, R. G.; Marken, F.; Wilkins, S. J.; Goeting,C. H.; Foord, J. S. Anal. Chem. 2000, 72, 2362.

(30) Engelman, E. E.; Evans, D. H. Langmuir 1992, 8, 1637.(31) Hunt DuVall, S.; McCreery, R. L. Anal. Chem. 1999, 71, 4594.

(32) Cvacka, J.; Swain, G. M., unpublished results.(33) McDermott, C. A.; Kneten, K. R.; McCreery, R. L. J. Electro-

chem. Soc. 1993, 140, 2593.

Figure 7. Plots of ipox versus scan rate1/2 for Fe(CN)6

3-/4-,Ru(NH3)6

2+/3+, IrCl62-/3-, methyl viologen (MV+2/+), 4-tert-

butylcatechol, and Fe2+/3+ for a nanocrystalline boron-dopeddiamond thin film.

Boron-Doped Nanocrystalline Diamond Electrodes Chem. Mater., Vol. 15, No. 4, 2003 885

H2O2 (30%) for 30 min. This was followed by rinsingwith ultrapure water and then drying. The films werethen hydrogen-plasma-treated (microwave-assisted) for30 min to remove the surface oxides formed during theacid washing and to terminate the surface with hydro-gen.21,24 The hydrogen was introduced into the reactorat a flow rate of 200 sccm. The plasma power, systempressure, and temperature were 1 kW, 35 Torr, and 800°C, respectively.

The electrochemical response of two nanocrystallinefilms was evaluated before and after the acid washing/hydrogen-plasma treatment. One was a nanocrystallinefilm deposited in the presence of the boron dopantsource and the other was a nanocrystalline film depos-ited without any intentional boron dopant added. Theresponse for the boron-doped nanocrystalline film wasunaffected by the chemical/plasma treatment. A nearlyreversible response was observed for Fe(CN)6

3-/4- witha ∆Ep of 78 mV before and 72 mV after treatment. Anearly reversible response was also observed forRu(NH3)6

3+/2+, IrCl63-/4-, and methyl viologen with

∆Ep’s of 74, 71, and 59 mV, respectively, after treatment.The response of the nanocrystalline film, depositedwithout intentionally added boron, was altered by thetreatment. The ∆Ep for Ru(NH3)6

3+/2+ increased to 165mV after treatment. The ∆Ep’s for Fe(CN)63-/4-, IrCl63-/4-,and MV2+/+ all increased after treatment to 166, 83, and99 mV, respectively. A decrease in the electrical con-ductivity, due to a decrease in the π-bonded carbon inthe grain boundaries, causes the increase in ∆Ep. The∆Ep increases for this particular nanocrystalline filmwere not as dramatic as we have seen in tests of otherfilms. The reason is that this film, even though depos-ited with no intentionally added boron, was doped tosome extent because it was prepared in a reactor thatis regularly used for depositing boron-doped films. Infact, recent boron nuclear-reaction analysis measure-ments of an “undoped” film, deposited in the samereactor, revealed a doping level of ca. 50 ppm B/C.Previous tests have shown that the electrochemicalresponse of an undoped nanocrystalline film can almostbe completely inhibited after the same chemical/plasmatreatment. Even though not as dramatic, the presentresults suggest that the electronic properties of theboron-doped films are dominated by the acceptor con-centration in the diamond lattice rather than by thephysicochemical properties of the grain boundaries.

Electroanalysis. Two analytical applications usingthese electrodes were investigated. First, these elec-trodes were observed to be useful for the quantitativeelectrooxidation of aliphatic polyamines.34,35 Second,they were used for the detection of trace metal ions.Heavy metal contamination in the environment is agrowing concern. Anodic-stripping voltammetry is ef-fective for the multicomponent analysis of metal-containing solutions due to the technique’s high sensi-tivity and selectivity.42 Mercury and mercury-coated

electrodes function well as the working electrode, butthe metal is highly toxic and there is much interest infinding alternate, nontoxic electrode materials. Hydrogen-terminated diamond is one of these alternative elec-trodes that has very attractive features for this appli-cation:20 (i) a large overpotential for hydrogen evolution,(ii) an overpotential for the reduction of oxygen, (iii) astable surface chemistry and microstructure, (iv) achemically inert surface that resists fouling, and (v) nometal-diamond chemical interactions.

Metal deposition on diamond is an immensely com-plicated process and is not fully understood. First, aparticular metal under study must nucleate and growon the surface. The sites at which this occurs, as wellas the nucleation and growth mechanism for manymetals, is unknown. Copper deposits equally well onboth the facets and grain boundaries of highly boron-doped microcrystalline diamond.43 The metal depositsby an instantaneous nucleation and growth mechanismat low overpotentials and by progressive mechanism athigh overpotentials. Silver deposits on microcrystallinediamond have also been studied.44 At low overpotentials,an instantaneous nucleation and growth mechanism isobserved, whereas at high overpotentials, a progressivemechanism is operative. Zinc deposits on microcrystal-line diamond by a progressive mechanism at both lowand high overpotentials.43

Second, when multiple metals are deposited simul-taneously, as is the case in a real stripping voltammetricmeasurement, not only is their interaction with thediamond surface important but equally critical is theirinteraction with each other. There is a possibility ofintermetallic compounds or alloys forming, both whichwill affect the oxidation or stripping potential for each.When these heterogeneous deposits form, the oxidationof a particular metal can occur from different sites onthe diamond surface or from another metal surface.Oxidation from these multiple sites leads to peakbroadening due to a spread in reaction kinetics. Ideally,for this application, highly dispersed metal deposits oflow volume, without any intermetallic interactions, aredesired. Even with these complexities, we suppose thatdiamond may be a useful electrode for the determinationof trace metal ions via anodic-stripping voltamme-try.20,45,46

A summary of the individual metal ion detectionfigures of merit for the determination of Ag(I), Cu(II),Pb(II), Cd(II), and Zn(II) by anodic-stripping voltam-metry is given in Table 3. The preconcentration stepinvolved the application of -1200 mV for 3 min (nostirring). The oxidation peaks occurred at ca. 400, 110,-450, -710, and -1010 mV versus Ag/AgCl for 100 µMsolutions of Ag(I), Cu(II), Pb(II), Cd(II), and Zn(II) andshifted positive with increasing solution concentration(i.e., increasing surface coverage). The limit of quanti-tation for the individual metals is in the low ppb range.

(34) Koppang, M.; Witek, M.; Blau, J.; Swain, G. M. Anal. Chem.1999, 71, 1188.

(35) Witek, M. A.; Swain, G. M. Anal. Chim. Acta 2001, 440, 119.(36) Tabor C. W.; Tabor H. Annu. Rev. Biochem. 1984, 53, 749.(37) Pegg A. E. Cancer Res. 1988, 48, 759.(38) Marton L. J.; Feuerstein B. G. Pharm. Res. 1986, 3, 311.(39) Simon, P.; Lemacon, C. Anal. Chem. 1987, 59, 480.(40) Eerola, S.; Hinkkanen, R.; Lindfords, E.; Hirvi, T. J. Assoc.

Anal. Chem. 1993, 76, 575.

(41) LaCourse, W. R.; Johnson, D. C. Anal. Chem. 1990, 62, 589A.(42) Wang, J. Stripping Analysis: Principles, Instrumentation and

Applications; VCH Publishers: Weinheim, FRG, 1985.(43) Wang, J.; Swain, G. M., unpublished results.(44) Vinokur, N.; Miller, B.; Avyigal, Y.; Kalish, R. J. Electrochem.

Soc. 1999, 146, 125.(45) Manivannan, A.; Tryk, D.; Fujishima, A. Electrochem. Solid

State Lett. 1999, 2, 454.(46) Prado, C.; Wilkins, S. J.; Marken, F.; Compton, R. G. Elec-

troanal. 2002, 14, 262.

886 Chem. Mater., Vol. 15, No. 4, 2003 Show et al.

The limit of quantitation is highest for Zn, and this isdue, at least to some extent, to the hydrogen-evolutionreaction interfering with the coulometric efficiency ofthe metal deposition. Good response precision andstability (<4%) are seen for all five metals.

In another series of more “real world” experiments,solutions containing mixtures of Cd(II), Pb(II), andAg(I) were analyzed using the same experimental condi-tions. Figure 8 shows anodic-stripping voltammogramsfor metal ion concentrations ranging from 0.01 to 10 µM.The deposition was performed at -1000 mV for 3 min(no stirring). The peaks were identified based on thestripping peak potentials for the individual metals. Itcan be seen that the oxidation peak potentials shiftpositive and increase in width with increasing solutionconcentration (i.e., increasing surface coverage). Also,the Pb(II) stripping peak splits into two fractions. Themore negative component is apparent at higher solutionconcentrations, although the ratio of the two remainconstant with increasing coverage. One possible expla-nation for this split is metal stripping from two differentsites on the electrode. For example, the Pb may beoxidizing from a homogeneous deposit on the diamondsurface and heterogeneous sites where multiple metalshave co-deposited. Another possibility is the formationof an intermetallic compound. In a series of separateexperiments, it was observed that standard additionsof either Cu(II) or Ag(I) to a test solution of Pb(II)resulted in the formation of two Pb(II) stripping peaks.Therefore, it is supposed that the dual Pb(II)-strippingpeaks, observed presently, are caused by the formationof a Ag-Pb intermetallic compound.

It is also observed that the Cd(II)-stripping peakcharge is suppressed when co-deposited with Ag(I). Eventhough equimolar amounts of Pb(II) and Cd(II) arepresent in solution, the stripping charge for Cd(II) is

significantly less than that for Pb(II). It was observed,in a series of additional experiments, that the additionof Ag(I), Pb(II), or Cu(II) to a solution containing Cd(II)caused some supression of the Cd(II)-stripping peak,with the greatest suppression seen after the additionof Ag(I). In fact, in the presence of Cd(II), the strippingcharge for Ag(I) is enhanced, suggesting that some ofthe deposited Cd is stripping at the Ag oxidationpotential. Therefore, the suppression of the Cd(II) peakis attributed to the formation of an Ag-Cd intermetalliccompound.

Even with the complications associated with thesuspected intermetallic compound formation, the lineardynamic range extended from 0.01 to 100 µM for Cd(II)and Pb(II) with linear regression correlation coefficientsof 0.992 each, and from 0.001 to 50 µM for Ag(I) with alinear regression correlation coefficient of 0.994. Thelimit of quantitation for Ag(I), Pb(II), and Cd(II) in themixture was 0.01 µM or 1, 2, and 1 ppb respectively, atwhich the S/N ratio of the Cd(II) peak was just over 3:1.Ag(I) and Pb(II) still showed a good S/N (∼30:1) at thisconcentration.

ConclusionsBoron-doped nanocrystalline diamond thin films were

deposited by CVD from an CH4/H2/Ar source gas mix-ture. Boron doping was accomplished either using asolid-state diffusion source or B2H6 added to the sourcegas mixture. XRD revealed that the bulk crystal struc-ture of the deposited film is cubic diamond. TEMindicated the film consists of 10∼15 nm randomlyoriented but atomically ordered diamond grains. SEMshowed these grains form aggregates ∼50-100 nm insize. Large water contact angles indicated the surfaceis hydrophobic, consistent with a hydrogen surfacetermination.

The films exhibited a wide working potential window,a low voltammetric background current, and goodresponsiveness for Fe(CN)6

3-/4-, Ru(NH3)62+/3+, IrCl62-/3-,

and methyl viologen without any pretreatment. Thequasi-reversible voltammetry for all the couples indi-cates that the boron-doped nanocrystalline diamond hasa sufficient density of electronic states over the widepotential range to support rapid electron transfer. Moresluggish electrode kinetics were found for 4-tert-butyl-catechol and Fe3+/2+. The sluggish kinetics for theformer are attributed to weak surface adsorption, andthe latter, at least partially, to the absence of catalyzingsurface carbonyl functional groups. Boron-doped nano-crystalline films were found to be useful for the detectionof trace metal ions, such as Ag(I), Cu(II), Cd(II), Pb(II),and Zn(II). The detection of the metal ions in solutionmixtures is, however, complicated by intermetalliccompound formation.

Most importantly, the electrical conductivity of thefilms is not affected by changes in the physicochemial

Table 3. Summary of the Anodic-Stripping Voltammetric Data for Individual Metal Ions for a Boron-DopedNanocrystalline Diamond Thin-Film Electrode

parameterlinear dynamic

range (ppb) r2 > 0.99limit of quantitation

(ppb) (S/N >3)sensitivitiy

(µC/ppb)response precision

(%) (n ) 10)response stability

(%) (3-day test)

Ag (I) 0.1-10000 0.11 0.111 0.92 3.87Cu(II) 0.6-6400 0.64 0.045 1.88 4.87Pb(II) 2-20000 2.07 0.050 0.50 1.67Cd(II) 1.1-11000 1.12 0.099 1.30 3.37Zn(II) 6.5-6500 6.54 0.035 3.39 3.83

Figure 8. Differential pulse anodic-stripping voltammetriccurves for Ag(I), Pb(II), and Cd(II) for a boron-doped nano-crystalline diamond thin film in 0.1 M acetate buffer, pH 4.5.The metal ion concentrations are (a) 10, (b) 1, (c) 0.5, (d) 0.1,and (e) 0.01 µM. Preconcentration at -1000 mV for 3 min (nostirring).

Boron-Doped Nanocrystalline Diamond Electrodes Chem. Mater., Vol. 15, No. 4, 2003 887

properties of the grain boundaries, but rather is domi-nated by the charge carries within the diamond due tothe substitutionally inserted boron dopant atoms.

Acknowledgment. The authors would like to thankDr. T. Mitsui of NIMS, Drs. M. N. Gamo and T. Andoof CREST for performing the TEM and making the Hallmeasurements, Dr. A. Mcllwain of Case Western Re-serve University for the boron nuclear-reaction analysis,

and Mrs. K. Kojima, S. Watchi, T. Ishitobi, and Dr. T.Izumi of Tokai University for the XRD measurements.This work was supported by grants from the U.S.Department of Agriculture (2001-35102-10045) and theNational Science Foundation (CHE-0049090). The workwas also partially funded by the Center for Fundamen-tal Materials Research at Michigan State University.

CM020927T

888 Chem. Mater., Vol. 15, No. 4, 2003 Show et al.