The electrochemical activity of boron-doped polycrystalline diamond

Roger DeClements and, Greg M. Swain, , Tim Dallas and, Mark W. Holtz, , Robert D. Herrick II and, John L. Stickney. Electrochemical and Surface Struct...
0 downloads 0 Views 2MB Size
Download PDF
Anal. Chem. 1993, 65, 345-351

945

The Electrochemical Activity of Boron-Doped Polycrystalline Diamond Thin Film Electrodes Greg M. Swain’ Department of Chemical Engineering, Auburn University, Auburn, Alabama 36849

Rajeshuni Ramesham Electrical Engineering Department, Alabama Microelectronics Science and Technology Center, Auburn University, Auburn, Alabama 36849

The electrochemical activity of “as grown” boron-doped polycrystallinediamond thin film electrodes has been studled using cyclic voitammetry, chronoamperometry, and ac Impedance without external iliumlnatlon. The reslstlvltyof these materials after doping is ca. 10 ohm-cm. The diamond electrodes possess a low doubk layer capacitance and a relatively highpolarizationresistancetoward surface oxidation. Slow electrode kinetics are observed for Fe(CN)(13-’Cat the “as grown” surface; however, the electrode response was observed to be extremely stable over a 2-month period as compared with freshly polished glassy carbon. The resuits suggest that diamondelectrodes may have suitable properties as an electrode material for use in electroanalysis.

INTRODUCTION Diamond is an extremely hard crystalline form of carbon and is considered an excellent material for many applications due to its unusual physical and chemical properties such as high electrical resistivity, high thermal conductivity, high corrosion resistance, low coefficient of friction, chemical inertness, and optical tran~parency.’-~These are important properties for several technical applications such as highpower electronicdevices, coatingsfor cutting tools,and electrooptical devices. Undoped and impurity-free synthetic diamond is an insulator with a resistivity of >lo8 ohm-cm and a band gap energy of 5.4-5.7 eV. Structurally diamond is a cubic lattice constructed from sp3-hybridized tetrahedrally arranged carbon atoms with each carbon atom bonded to four neighbors. The stacking sequence is ABCABC with every third layer plane identical. This structure is fundamentally different from that of graphite which consists of layers of condensed polyaromatic sp2-hybridized rings with each carbon atom bonded to three neighbors, and the empty p-orbital forms a T band. The stacking sequence is ABAB with every other layer plane identical. From an application perspective, the wide variability in the electrical properties of synthetic diamond produced by various dopingprocesses is an attractive feature of this material. The electrical resistivity of microwave plasma-assisted chemically vapor deposited (PACVD) polycrystalline dia-

* Present address: Tohoku University, Faculty of Engineering, Department of Engineering Science, Aoba Sendai 980 Japan. (1) Field, J. E. The Properties of Diamond; Academic Press, Inc.: London, 1979. (2) Bachmann, P. K.; Messier, R. Chem. Eng. News 1989, May 15,24. (3) Ramesham, R.; Roppel, T.; Ellis, C.; Loo, B. H. J . Electrochem. SOC.1991,138, 2981. (4) h e s h a m , R.; Roppel, T.; Ellis, C.; Jaworske, D. A.; Baugh, W. J . Mater. Res. 1991, 6, 1278. 0003-2700/93/0365-0345$04.00/0

mond films depends on the hydrogen content.5 Hydrogen incorporation occurs during the film preparation since the presence of atomic hydrogen in the plasma is necessary to promote diamond formation over graphitization. Landstrass and Ravi697 have shown that the resistivity of CVD diamond thin films can be reduced to a range of ca. 102-108 ohm-cm. The relatively low resistivity was suggested to result from hydrogen passivation of defect sites. They also showed that the resistivity can be significantly increased by annealing at high temperature in a nitrogen atmosphere. These results clearly demonstrate the active role of hydrogen in the conduction mechanism of undoped CVD diamond films. Ramesham et al.3 have recently reported that the resistivity of CVD polycrystalline diamond thin films can be decreased to ca. 10ohm-cm after boron doping. Boron acts as an electron acceptor due to an electron deficiency in its outer shell giving the diamond electrodes p-type semiconducting properties by lowering the Fermi level due to the increased number of charge carriers.8 Okano et al.9Jo have reported that the resistivity of CVD diamond thin films can be reduced to ca. 0.01 ohm-cm after boron doping at a B/C ratio of ca. lo00 ppm. Boron doping of sp2-hybridized carbon materials has also been shown to produce more oxidation resistant materials.11-13 The abilityto prepare low-resistivitydiamond thin films stimulated the present research effort to characterize the electrochemical activity of this material. There are few known reports describingthe electrochemical properties of diamond films. Pleskov et al.14 have reported the photoelectrochemical properties of undoped polycrystalline diamond thin films. Fujishima et al.15 have recently reported the preliminary photoelectrochemical characterization results of boron-doped polycrystalline diamond thin films. This paper describes some preliminary results from the study of the electrochemical activity of yas grown” borondoped polycrystalline diamond thin film semiconducting electrodes without external illumination or surface pretreat(5) Celii, F. G.; Purdes, A. J.; Gnade, B. E.; Weathers, D. L. In New Diamond Science and Technology; Proc. of Znd Intl. Conf. on New Diamond Science and Technology; Messier, R., Glass, J. T., Butler, J. E., Roy, R., Eds.; Mater. Res. SOC.Intl. Conf. Proc. Series, Pittsburgh, PA, 1991; pp 631-636. (6) Landstrass, M. I.; Ravi, K. V. Appl. Phys. Lett. 1989, 55, 1391. (7) Landstrass, M. I.; Ravi, K. V. Appl. Phys. Lett. 1989, 55, 975. (8) Jones, L. E.; Thrower, P. A. Carbon 1991,29, 251. (9) Okano, K.; Naruki, H.; Akiba, Y.; Kurosu, T.; Iida, M.; Hirose, Y. Jpn. J . Appl. Phys. 1988,27, L173. (10) Okano, K.; Naruki, H.; Akiba, Y.; Kurosu, T.; Iida, M.; Hirose, Y.; Nakamura, T. Jpn. J . Appl. Phys. 1989,28, 1066. (11) McKee, D. W. Carbon 1986,24,737. (12) McKee, D. W.; Spiro, C. L.; Lamby, E. J. Carbon 1984,22,507. (13) Allardice, D. J.; Walker, P. L. Carbon 1970,8,375. (14) Pleskov, Y.; Sakharova, A.; Krotova, M.; Bouilov, L.; Spitsyn, B. J . ElectroanaL Chem. Interfacial Electrochem. 1987, 228, 19. (15) Patel, K.; Hashimoto, K.; Fujishima, A. J . Electrochem. SOC.Jpn. (Denki Kagaku), submitted 1992. 0 1993 American Chemical Soclety

340

ANALYTICAL CHEMISTRY. VOL. 65, NO. 4. FEBRUARY 15, 1993

ment. ScanningelectronmiQoscopyandRamans~~copy were used tocharacterizethe topography and carhon structure of the *as grown" diamond surface, respectively. Cyclic voltammetry and ac impedance measurements were used to memure the double layer capacitance, Cd, the uncompensated resistance, R,, and polarization resistance, R,, of the electrodes in different aqueous electrolytes. Cyclicvoltammetry and chronoamperometry were used to examine the electrochemical response for ferrilferrocyanide in KCI solution at "as grown" thin film electrodes. Theuse of this new electrode material is very intriguing hecause of the possibility that its study may provide new insights on electrochemistry at carhon electrodes. In addition, this electrode material may prove very useful for electroanalysis in harsh chemical environments.

EXPERIMENTAL SECTION The CMgrowth processofpolycrystallinediamondthinfilms has previously been desc~ibed.~,~ The diamond thin films were grown on a roughened n-type silicon substrate after a thorough chemical cleaning. .Boron doping and diamond growth were carried out in situ using a commercialhigh-pressure microwave plasma-assisted chemical vapor deposition system (ASTEX, Cambridge,MA). The growth was carried out at ca. 925 "C and a system pressure of ca. 45 Torr. Ultrapure methane and hydrogen were used at flow rates of 3.6 and 500 SCCM, respectively. Growth was carried out using a microwave power of ea. 1150 W. The growth rate was ca. 0.5-1 pm/h, and a continuousfilmcouldeasilybe achievedafter24 h. The diamond films used in this work were grown on 1-em2silicon substrates atafilmthicknessof 12-18pm. TheB/Cratiowasca. 1100ppm. The diamond thin films used in this work were annealed under a nitrogen atmosphere at ea. 425 "C for times in excess of 20 h and are referred to in the text as -as grown" films. The slow scan cyclic voltammetry and chronoamperometry were performed using an Omni 90 analog potentiostat and a Linseis LY 1600X-Y recorder (CypressSystems, Inc., Lawrence, KS). Both the time constant of the potentiostat and recorder were set at 200 ms. The equilibrium potentials were measured using a digital voltmeter (DVM). The ac impedance measurements were performed using a PAR Model 273 potentiostatl galvanostat (Princeton Applied Research, Princeton, NJ). The measurements were made in the frequency range of 0.01-5 Hz in the Fourier transform mode. The experimental apparatus did not allow for higher frequencies to he examined. An ac amplitude of 10 mV was used and the data were averaged over 10 cycles. Optical microscopy was performed with a Nikon SMZ-U microscope. The Raman spectroscopy was performed with the 514-nm line of a Spectra Physics Model 2020 argon ion laser excitation source. The electrochemicalmeasurementswere carried out in a single. compartment glass cell. The diamond filmlsilicon substrate electrode was clamped to the bottom of a smooth glass joint attachedtotheelectrochemicalceUwithahole0.4cmindiameter for exposure of the electrode surface to the electrolyte solution (geometric area 0.126 cm2). A gasket prepared with silicone adhesive was used to seal the diamond surface to the glass. Electrical contact was made both in the plane of the analyzed surface and on the backside of the silicon substrate by attaching a small piece of nickel foil with silver epoxy. The nickel foil contact was isolated from the areaexposed to the solution hy the siliconegasket. Similar electrode responses for ferrocyanideand equilibrium potentials were obtained using both contacts, indicating that the conductivity in the plane of the surface is the same as through the bulk material. Platinum gauze served as the counter electrode. The counter electrodegeometric area was overtwice thatoftheworkingelectrodesothat thecellimpedance was dominated by the capacitance of the working electrode. A commerical Ag/AgCl (3 M KCI) was used as the reference electrode, and all potentials are reported versus this reference. The measurements were made at room temperature (ca. 2&23 "C). The glass carbon (GC-20 Tokai, Ltd.) was polished with 0.05pm alumina on a felt polishing pad. The electrode was then rinsed with astream of ultrapure water (deionizedand distilled),

annealed diamond Blectrode "as grown"

3 x

Y

.2 ~

e .-A

-2 E

E m

6

d

Fbure 2. grown".

Raman Shift, cm-1 Raman SpaCtrurn 01 an annealed diamond elem& as "as

ultrasonicated in 2-propanolfor 3 min, and rinsed with ultrapure water prior to use. The geometric area was 0.071 cm2. Reagenbgrade potassium chloride,sodium hydroxide, sodium nitrate, sulfuric acid (Fisher Scientific), and potassium ferrocyanide (Aldrich Chemical) were used without further purification. All solutions were prepared fresh daily using ultrapure water. The solutions were deoxygenated with a vigorous purge of He or N2 for 10-15 min prior to analysis.

RESULTS AND DISCUSSION Figure 1 shows a scanning electron micrograph of the topography of an annealed diamond electrode *as grown". Similar images were obtained from the surface of unannealed electrodes. The surface is well faceted with a large degree of (111) plane exposed. Evidence for cracks in the film or extensive crystallite pitting is not observed. Also, amorphous regions associated with the presence of graphitic carhon are not ohserved. The film appears to be continuous over the entire substrate. The nominal crystallite size is ca. 1-10 pm. Figure 2 shows a Raman spectrum from an annealed diamond electrode "as grown". The intense and narrow Ramanline ohservedat 1334cm-'ischaracteristicofdiamond reflectinga highdegreeofsp3bondinginthe fihn.3.'.16 Similar spectra were obtained from the surface of unannealed electrodes. Thecharacterizationof diamond filmsby Raman spectroscopy is quite developed, and the polarization properties and temperature dependence of the spectra have been studied in detail by Solin and Ramdas.I7 The spectrum also (16)KniEht, D. S.;White, W.B. J. Mater. Sei. 1989,4,385. (17)Solin, S. A,: Ramdas, A. K. Phys. Re". B 1970, 1, 1681.

ANALYTICAL CHEMISTRY, VOL. 65, NO. 4, FEBRUARY 15, 1993

0

20

40

60

80

Time, hours Flgure 3. Eiectrlcai resistivity with respect to total treatment time for a diamond electrode (A) “as grown”, (B) annealed in nitrogen at 425 OC, (C) borondoped, (D) after additional boron doping of D, (E) borondoped sample annealed In nitrogen at 425 O C , and (F) borondoped sample exposed to a hydrogen microwave plasma.

shows some scattering intensity in the 1540-cm-lregion. This reflects the incorporation of some graphitic or amorphous carbon impurities into the diamond film during the growth process. The relative amount of these impurities would appear to be small given that the scattering cross section for diamond is ca. 9 X 10-7 cm-l/sr whereas the scattering cross section for graphite is 500 X cm-’/sr.l6 The Raman spectrum provides only semiquantitative information; however, a rough calculation indicates that the sp2-hybridized carbon represents perhaps only ca. 1-5% of the total carbon content based on the scattering coefficients and assuming a linear relationship exists between the spectral intensity and the amount of the carbon type present. The exposure of the boron-doped diamond electrodes to a series of repetitive hydrogen microwave plasma and heat treatments in nitrogen produced little change in the Raman spectrum, suggesting that the carbon surface microstructure is unaffected by such treatments.3 Figure 3 shows a plot of the electrical resistivity of an annealed diamond electrode“asgrown” as a function of plasma treatment with respect to total treatment time. The resistivity was measured with a conventional two-point probe (tungsten) method. Point A corresponds to the resistivity of the “as grown” undoped hydrogenated diamond thin f i i which is on the order of ca. 102 ohm-cm. Point B indicates that the resistivity can be increased by 6 orders of magnitude after annealing at 425 O C in a nitrogen ambient. The resistivity increases because of the desorption of hydrogen from the film or a change in its position in the lattice during the thermal treatment. Boron doping reduces the resistivity of the film to ca. 10 ohm-cm as shown in point C. It can also be seen in points D-E that annealing in nitrogen has little influence on the resistivity of boron-doped polycrystalline diamond thin film which is opposite the behavior exhibited by undoped diamond films (points A-B). The resistivity of the film is slightly decreased after exposure to a hydrogen microwave plasma as shown in point F. Similar data for these electrodes has been reported pre~iously.~ Figure 4 shows a cyclic voltammogram of an annealed diamond electrode “as grown” in 0.1 M KC1 at a scan rate of 100 mV/s. The voltammogram became reproducible after the fifth scan and is relatively featureless in the potential region between -500 and 700 mV. The voltammetric charge becomes noticeably smaller at potentials between 0 and -350 mV, reflecting a decreased double-layer capacitance presumably due to the formation of a space charge region within the semiconductor. Some type of surface faradaic process is evidenced by the peak current at ca. 850 mV just prior to the apparent onset of oxygen evolution. Differential pulse

947

-”

a -2 -3

l -500

l

1

1

l 0

l

l

1

l

l

I

500

I

I

l

l 1000

Potentlal mV vs. Ag/AgCI Flguro 4. Cycllc voltammogram at an annealed dlamond electrode “as grown” In 0.1 M KCI. Scan rate = 100 mV/s. Reference = Ag/AgCl.

Table I. Equilibrium Potentials and Anodic and Cathodic Potential Limits Observed at an Annealed Diamond “As Grown” and a Glassy Carbon Electrode. equilibrium anodic cathodic solution potential (mV) limit (mV) limit (mV) Annealed Diamond 0.1 M HzS04 +59 +1100 -200 0.1 M KCI +133 +900 -500 0.1 M NaN03 +236 +1100 -600 0.1 M NaOH +125 +I00 -400 Glassy Carbon 0.1 M H2S04 +398 +lo00 -lo00 0.1 M KCl +160 +lo00 -lo00 0.1 M NaN03 +225 +lo00 -400 0.1 M NaOH +171 +800 -lo00 Values obtainedby cyclicvoltammetry at a scan rate of 10 mV/s. voltammetricmeasurements,data not reportad here, indicate three surface redox processes at the “as grown” electrodes in each of these electrolytesat potentials of ca. 250,500, and 850 mV. Table I shows examples of equilibrium potentials along with anodic and cathodic potential limits determined from cyclic voltammetry for an annealed diamond electrode in four different aqueous electrolyte solutions. Similar data for polished glassy carbon are shown for comparison. The equilibrium potentials observed at the diamond electrode are similar to those observed at the polished glassy carbon in all the solutionsexcept HzS04. The equilibrium potentials for both electrodes showed considerable variability from experiment to experiment (MOOmV) presumably due to the surface cleanlinessand the presence of chemisorbed oxygen.14 However, the potentials observed at the diamond electrode in HzSO4 were consistently more negative than potentials observed in the other electrolytes. Without any surface pretreatment, the equilibrium potentials observed at the diamond electrodes exhibited a strong dependence on the electrode’s past history. Stable equilibrium potentials were usually reached within 15min of soaking in all of the solutions examined. The anodic potential limits at the annealed diamond electrode prior to the apparent onset of oxygen evolution in the aqueous electrolytes are similar to the limits observed at glassy carbon. Similar experiments performed using unannealed diamond electrodes, data not shown here, indicated similar anodic potential limits. The cathodic potential limits at the annealed diamond electrodes,however, are significantly

348

ANALYTICAL CHEMISTRY, VOL. 65, NO. 4, FEBRUARY 15, 1993

Table 11. Values*of R,,,4,and c d l of an Annealed Diamond 'As Grown" and a Glassy Carbon Electrode Obtained from Linear Regression Curve Fitting of Plots of Re(Z) vs -IM(Z)w solution R,, (ohm) R, (ohm) c d l (farads) 0.1 M 0.1 M KC1 0.1 M NaN03 0.1 M NaOH

Annealed Diamond 1.4 X lo6 3.1 X lo1 1.3 X lo6 3.1 X lo1 2.1 X lo6 2.4 X l o 7 1.5 X lo6 2.3 X lo1

4.6 x 4.3 x 4.8 x 5.0 x

10-7 10-7 10-7

0.1 M 0.1 M KCl 0.1 M NaN03 0.1 M NaOH

Glassy Carbon 1.9 x 4.3 x 104 2.2 x 105 9.6 x 8.9 x lo4 3.9 X 1.0 x 105 5.8 x

1.2 x 8.7 X 1.4 x 1.1x

10-5 lo+ 10-5 10-5

105 105 lo5 105

A

10-7

"Values obtained a t the equilibrium potential using an ac amplitude of 10 mV. Data was averaged over 10 cycles.

less negative compared with the limits observed at glassy carbon. Even though the solutions were deoxygenated, a substantial amount of the cathodic current maybe associated with the reduction of chemisorbed oxygen14 in addition to reductionof the solvent. Also, the cathodiccurrent may result from the reduction of impurities on the surface as the elelctrodes were not pretreated prior to use. If the current is due to hydrogen evolution, then the low overvoltage required may indicate that the diamond electrodes would be a very efficient photocathode for the generation of hydrogen. Fujishima et al. have reached a similar conclusion based on results obtained using illuminated boron-doped diamond thin film e1e~trodes.l~ This issue is currently being examined in more detail. It is interesting to note that the unannealed diamond electrodes showed cathodic potential limits similar to those of glassy carbon in each of the electrolytes. AC impedance analysis between 0.01 and 5 Hz was used to examine the frequency response of the annealed diamond electrode in different aqueous electrolytes. Table I1 summarizes the measured uncompensated and polarization resistance along with the double-layer capacitance in the different electrolyte solutions. The measurementswere made at the equilibrium potential. Plots of the real impedance, Re(Z), versus the imaginary impedance times the frequency, -Im(Z)u, were used to estimate the resistance and capacitance as shown in Figure 5. The values for R,,, R,, and c d l can be determined from such plots using the following relations:

and The total electrode impedance is given by Z, = [Re(Z)' + Im(Z)2]1/2 where Re(Z) is the real component of the complex impedance associated with the resistance in the electrochemicalcell and Im(Z) is the imaginary component associated with the capacitance. It can be seen from the data in Table I1 that the overall resistance (R,, + R,) of the diamond electrodes is larger than that of glassy carbon. The R,, values at the diamond electrodes are similar in all of the examined Solutions and are ca. 2 orders of magnitude larger than the values observed at glassy carbon. The R,, contains contributions from the solution resistance, any nonuniform current distributions at

4

L

1400

h

y N cz

\

4oo]

I

R,

' 0 ,

150

I

I

~

I

/

'

I

/

1

.I..

'

-

2 00 2 50 - ~ m ( Z ) t w * lo6 (ohm-sec-')

3 00

Flgurr 5. Plot of Re(Z) vs -Im(Z)w obtained at an annealed diemond electrode "as grown" In 0.1 M KCI determlned from ac Impedance

analysis at the equilibrium potentiel. Values obtalned at frequencles between 0.01 and 5 Hz. ac amplitude = 10 mV. Data averaged over 10 cycles. the working electrode resulting from improper cell design, and the ohmic resistance of the electrode and associated contacts. The higher R,, values are believed to result, in part, from the higher resistance of the semiconducting diamond in addition to the resistance associated with the ohmic contact. The magnitude of the R,, values observed at glassy carbon may also reflect resistance associated with the ohmic contact. Most importantly, the R , values are ca. 1-2 orders of magnitude higher at the diamond electrode than at glassy carbon, reflecting the increased oxidation resistance of the diamond surface. It is important to point out that, without the impedance data at the high frequencies (up to 10kHz), it is very difficult to accurately determine the values for R , and R , from the plots. The estimated values for R,, at both the diamond and glassy carbon electrodes are excessively large due mainly to the inavailability of the high frequency data where the overall impedance is dominated by Rut. In actuality, since the values for R , determined by this method are based on the values of R,, and the later are excessively high; the true values for R, are likely significantly larger than data shown. Highfrequency analysis of the diamond electrodes is currently in progress to more accurately determine these values. Nevertheless, the data reported here are useful for comparative purposes. The Cdl values are ca. 1-2 orders of magnitude lower at the diamond electrode than at glassy carbon. The lower capacitance may result from the existance of a space charge region within the near surface of the semiconductor.'4,'s,~~zlThe overall capacitance of a semiconductor electrode surface is the s u m of the space charge capacitance, C,,, t,he compact layer capacitance, Campact,and the diffuse layer capacitance, CDiffuse,in series shown by the following equation: The smallest capacitance will be the dominant term and, in (18)Randin, J.; Yeager, E. J. Electrochem. SOC.1971, 118,711. (19) Randin, J.; Yeager, E. J. Electroanal. Chem. Interfacial Electrochem. 1972, 32, 257. (20) Bard, A. J.;Faulkner, L. R. ElectrochemicalMethods; John Wiley & Sons: New York, 1980,Chapter 14. (21) Kohl,P. A.; Bard, A. 3. J . Electrochem. SOC.1979, 126,59.

ANALYTICAL CHEMISTRY, VOL. 65, NO. 4, FEBRUARY 15, 1993

Table 111. A Comparison of the Double-Layer Capacitance of an Annealed Diamond Electrode ‘As Grown” Obtained by AC Impedance Analysis and Cyclic Voltammetry solution 0.1M HzS04 0.1 M KC1 0.1 M NaN03 0.1 M NaOH

Cdi”

(rrF/cm2)

3.7 3.4 3.8 3.9

45

349

t

C ~ I(rrF/cm2) * 7.1 4.5 6.3 4.8

a Values obtained by ac impedance analysis at the equilibrium potential. Values obtained by cyclic voltammetry at a scan rate of 5 mV/s a t a potential of +50 mV. Values are normalized to the geometric area = 0.126cm2.

*

Table IV. Apparent Exchange Current Densities at an Annealed Diamond “As Grown” and a Glassy Carbon Electrode Determined by AC Impedance Analysis at the Equilibrium Potential. solution

Zo(A/cm2) Annealed Diamond

0.1 M 0.1 M KCl 0.1 M NaN03 0.1 M NaOH

1.3X 1.3 X 1.7 X 1.7 x 10-8 Glassy Carbon

0.1 M 0.1 M KCl 0.1 M NaN03 0.1M NaOH

3.9 x 7.6 x 1.9 x 1.2 x

10” 10-7 10” 10-6

a Exchange currents normalized to the geometric area of the electrodes (diamond = 0.126 cm2 and glassy carbon = 0.071 cm2).A value of n = 2 is assumed.

cases where a relatively large excess electrolyte is used, the overall capacitance is dominated by the capacitance of the space charge region. The low capacitance (i.e. low electrochemical noise) exhibited by the diamond electrode may prove to be very useful, especially in trace analyte electroanalytical detection schemes. A similar conclusion was reached in a communication by Iwaki et aLZ2where they reported dark current measurements at dielectric diamond treated by ionic implantation. However,the conducting surface formed during the implantation treatment was amorphous carbon and not sp3-hybridized diamond. The differential double-layer capacitance values of the diamond electrode in different solutions, measured by both cyclic voltammetry and ac impedance and normalized to the geometric area, are summarized in Table 111. The data show that the two different measurement techniques produce similar values with those obtained by cyclic voltammetrybeing slightly larger. The slightly larger values may result from the slow movement of charge in microcracks and fissures between the crystallites as the overall time scale of the slow scan voltammetry was longer than that of the impedance measurements. Assuming that the polarization resistance reflects mainly the resistance of the surface toward oxidation, the apparent exchange current density for the oxidation reaction can be determined from the following relation:

Io = RT/nR,,F where Io is the apparent exchange current density, n is the number of electrons transferred during the oxidationreaction, and R, is the polarization resistance. R, T, and F have their usual meanings. The data are summarized in Table IV. The number of electrons transferred during the corrosion reaction (22)Iwaki, M.; Sato, S.; Takohashi, K.; Sakairi, H. Nucl. Znstrum. Methods 1983,209,1129.

-400

0

400

Potential mV vs. Ag/AgCi Flguro 6. Cyclic voltammogram for 1 mM ferrocyanide in 0.1 M KCI at an annealeddiamond electrode “asgrown”. Scan rate = 10mV/s. Reference = AgIAgCI.

step is assumed to be 2 based on the literature.23-26 Quinonelike carbon-oxygenfunctionalities which undergoa 2-electron transfer are known to exist at the oxidized glassy carbon surfa~e.23-~~ The data indicate that the apparent exchange current densities observed at the diamond electrodes are ca. 1-2 orders of magnitude lower than the values observed at glassy carbon. The lower apparent exchange current densities reflect slower electrode kinetics for the surface oxidation reaction at the diamond surface. However, the actual exchangecurrent densities are likely significantly lower than the values reported due to the fact that R, is underestimated for the reasons stated previously. The electrochemical activity of an annealed diamond electrode was examined using 1 mM F ~ ( C N ) B ~in - / 0.1 ~- M KC1. Figure 6 shows a cyclic voltammogram obtained after 10 scans. The Upis ca. 325 mV, reflecting slow electrode kinetics.=~~~-~ The voltammetricresponse is likely influenced by the ohmic resistance of the diamond electrode. However, it is most interesting to note that this response was obtained without pretreatment of the yas grown” diamond electrode surface. The Upvalueremained relatively unchanged during the 2-month exposure of the electrode to atmospheric conditions. Based on this observation, the diamond surface appears to be immune to the rapid deactivation processes that adversely affect the electrode response of glassy carbon. For comparison, the AE, in the same solution for a freshly polished glassy carbon electrode had increased in excess of 300 mV within 2 h after exposure to the solution. The slow electrode kinetics observed at the diamond thin film surface are also influenced by the surface cleanliness. We are presently exploring the use of pretreatments designed to clean the diamond surface which may result in a more active surface. Good linearity in the i, vs Y ~ plot / ~was observed at scan rates between 10and 100mV/s, consistent with the mass transport dominated by planar diffusion. Mass transport control by planar diffusion is also evidenced by the “near zero” intercept of i vs t-1P plot for the oxidation of Fe(CN)o3-l4-at the diamond electrode shown in Figure 7. Good linearity is observed with a correlation coefficientgreater than 0.990. The electrochemicallyactive are, determined from (23)Fagan, D.T.; Hu, I. F.; Kuwana, T. Anal. Chem. 1985,57,2759. (24)Swain, G.M.; Britton, W. E., unpublished results. (25)Hu, I.; Karweik, D. H.; Kuwana, T. J. Electroanal. Chem. Interfacial Electrochem. 1985,188, 59 and references within. (26)Laser,D.; Ariel, M. J.Electroanal. Chem.ZnterfacialElectrochem, 1974,52,291. (27)McCreery, R.L. InElectroanalytical Chemistry;Bard, A. J.,Ed.; 1991;V O ~17, . pp 221-374. (28)Bowling, R.;Packard, R.; McCreery, R. L. J. Am. Chem. SOC. 1989,I l l , 1217 and references within. (29)Bowling, R.J.; Packard, R. T.; McCreery, R. L. Langmuir 1989, 5 , 683.

350

ANALYTICAL CHEMISTRY, VOL. 65, NO. 4, FEBRUARY 15, 1993

’i

exposed to HC104 solution completely lost much of their conductivitywithin a 2-day period after testing, during which time the electrodes were both soaked in solution and exposed to air. The conductivity could not be regained by polishing, soxhlet extraction in cyclohexane, or ultrasonication in 2-propanol. We are currently unable to explain this mysterious observation. One possibility is that the acid slowly creeps through the cracks and fissures in the diamond film, not visible with SEM,and attacks the silicon substrate. The acid attack would certainly result in the formation of insulating Si02 which may also cause the separation of the diamond film from the substrate surface.

rn

/

A

/

///

1

D

-

7.63 x 10-6 cm2/sec

-

slope 17.9 pA-sec”’ Electrcchemically active area

-

CONCLUSIONS 0.119 cm2

The boron-doped polycrystalline diamond thin film electrodes appear to have some very interesting properties. In the present work, the electrochemical properties of the “as 0 grown”annealed diamond electrodeswere characterized using 02 0.4 ?6 0.8 lJ,2)1 2 14 16 cyclic voltammetry, chronoamperometry,and ac impedance. Time (sec This material has a low double-layer capacitance perhaps Figure 7. Plot of i vs rl’zfor the oxidation of 1 mM ferrocyanide in due, in part, to the presence of a space charge region. Yeager 0.1 M KCI at an annealed diamond electrode “as grown”. Currents et al.18J9have explained the similarly low capacitanceobserved are background corrected. Potential step from 0 to 800 mV. at the basal plane of HOPG as resulting from the presence of a space charge region within the electrode material. The capacitance is also lower due to the absence of significant amounts of surface carbon-oxygen functionalities. The diamond electrodes have a high polarization resistance 80 reflecting the surface’s resistance toward oxidation. While the “as grown” diamond surface exhibits slow electrode kinetics, based on the cyclic voltammetric response for Fe(CN)63-/4-,the electrode response does appear to be very stable over a 2-month period during exposure to both the solution and atmosphere as compared with freshly polished glassy carbon. A significantly more reversible cyclic voltammetric response for Fe(CN)e3-l4-in HClO4 was observed after polishing the electrode. However, the electrodes mysteriously lost their conductivity within 2 days after testing. -120 The preliminary results described here indicate that the l l l l l i l l l l l l l l l low-resistivity boron-doped polycrystalline diamond thin 300 440 580 720 860 1000 films may have possibilities for use as a working electrode Potentlal mV vs. RHE material in electroanalysis. Further studies are needed to Figure 8. Cyclic voltammogram for 1 mM ferrocyanide In 0.1 M HCIO, fully characterize the properties of this material. This paper obtained at an annealed diamond electrode after polishing. Scan rate presents data obtained from one electrode; however, three = 100 mV/s. Reference = RHE (0.1 M HCIO,). different “as grown” annealed electrodes were examined in the voltammetry and impedance experiments with reprothe slope, was found to be 0.119 cm2 for the oxidation. This ducible results. The structural and electronic properties of area compared with the geometric area indicates that a the “as grown” diamond films have been previously observed significant fraction of the electrode surface is participating to be very reproducible,394 and there is no reason to expect in the electrochemical reaction. Figure 8 shows a cyclic voltammogram for F ~ ( C N ) G ~in- / ~ - that the electrochemicalproperties should be any different. The low double-layer capacitance, the chemical inertness of 0.1 M HC104obtained at an annealed diamond electrode after the surface, and the stable electrode response are all attractive polishing with 0.05-pm alumina and ultrasonication in features of this material, especially for use in highly sensitive isopropanol. A hEpof ca. 104 mV is observed and reflects electroanalytical detection schemes. The low overvoltage significantly improved electrode kinetics.23*27-29 The AEP required for the apparent evolution of hydrogen would appear values of ca. 100 mV were observed during the first 10-15 min to make this a useful cathode material. However, further of potential cycling. Long-term cycling using ferrilferrocywork is needed to identify the source of the cathodic current anide in acid solution is not possible due to the compound’s which may result from impurities. instability. The diamond electrodes have dark discolored regions on the surface after exposureto solution and potential One interesting issue to address is the nature of the conduction mechanismthrough the material, specifically the during electroanalysis. The light gray color of the “as grown” diamond can be reobtained after polishing. The polished role of the amorphous carbon. It appears from the Raman diamond surface appears to be hydrophilic as water completely results that the amorphous carbon content is on the order of wets the surface. The most reversible electrode response for 1-5 % . It must be remembered, however, that the Raman Fe(CN)63-/4-at the polished electrodes has been obtained in spectra reflects both surface and bulk properties as the HC104. The testing of three different annealed electrodes sampling depth is assumed to be on the order of 200 A. We are presently performing angle-resolved measurements to produced similar results. It is interesting that the electrode response for this redox couple in NaF’, KC1, KBr, and KI is determine the spatial distribution of the amorphous carbon significantly less reversible at the polished electrodes. Perat the surface and near the surface of the electrode. A firm haps the most interesting observation is that the electrodes conclusion regarding the nature of the conductionmechanism h

t

V

ANALYTICAL CHEMISTRY, VOL. 65, NO. 4, FEBRUARY 15, 1993

cannot be reached based on the data obtained thus far. It appears that both the amorphous carbon and the diamond surface are involved in the electron-transfer process at the untreated “as grown” films. The fact that a quasi reversible electrode response for ferri/ferrocyanide is observed at the semiconducting electrode is not unexpected giving the highly doped nature of the electrodes as well as the presence of the amorphous carbon. Consequently, a continuum of so called “surface states” must exist between the valence and conduction bands in diamond in order to explain the observed currents. Further insights as to the effect of the amorphous carbon should be provided by studies of the electrochemical behavior of electrodes with little or no amorphous carbon present. We are currently modifying our growth conditions to produce higher quality diamond films. Finally, it must be pointed out that the diamond surface is not electrochemically inert as an expected p-type semiconductor electrode response for ferri/ferrocyanide and hydroquinone are consistently

951

observed at undoped diamond electrodes that have been chemically etched to remove the amorphouscarbon from the surface. This issue will be discussed in further detail in a separate paper describing the photoelectrochemical response of the electrodes.30 Our present efforts are focused on determining the flat band potential, estimating the interfacial capacitance, and examining the photoelectrochemical response of these electrodes in different aqueous electrolytes.30 We are also studying ways of pretreating the diamond surface to make it more active toward a variety of aqueous and organic redox analytes.

ACKNOWLEDGMENT The Space Power Institute at Auburn University is greatfully acknowledged for providing financial support for this research.

RECEIVED for review March 12,1992. Accepted November (30)Swain, G.M.;Rameeham, R.,submitted.

4,1992.