ELECTROCHEMICAL PROPERTIES OF SULFUR-TREATED DIAMOND
Sally C. Eaton,*
Alfred B. Anderson, John C. Angus
Case Western Reserve University
Cleveland, OH 44106 USA
Yulia E. Evstefeeva, Yuri V. Pleskov
Frumkin Institute of Electrochemistry
Moscow, 117071 Russia
ABSTRACT
The electrochemical properties of sulfur-treated
diamond were explored. Growth of
diamond with H2S present in the source gases and treatment of
diamond in the absence of growth with a sulfur-containing plasma were
performed. In both cases, Mott-Schottky
analysis showed evidence of donor centers.
Measurements of open-circuit potentials in the presence of UV
irradiation and measurements of the thermoelectric effect also showed n-type
conductivity. The presence of sulfur
was confirmed by particle induced x-ray emission (PIXE). The donor activity of the sulfur may arise
from surface states or grain boundaries rather than sulfur incorporated
substitutionally into the bulk of the diamond.
Preliminary evidence indicates that boron may facilitate the attachment
of sulfur onto the growing diamond surface and hence aid its incorporation into
diamond.
Keywords: Diamond films; n-type conductivity; Sulfur-doping;
Mott-Schottky plots
INTRODUCTION
Boron has long
been used as a p-type dopant in diamond, but a viable n-type dopant remains
elusive. For n-type conductivity,
several substitutional impurities have been investigated. Nitrogen is a deep thermal donor at 1.6 eV
below the conduction band; phosphorus has a donor level about 0.6 eV below the
conduction band and a mobility of 100 cm2/V-s (ref. 1). Recently, sulfur has been reported to give
n-type conductivity (refs. 2,3).
However, other measurements have indicated that the samples contained
boron and were p-type (ref. 4). The
experiments described here were motivated by the possibility of co-doping
diamond with boron and sulfur to produce n-type conductivity.
EXPERIMENTAL
Growth experiments
In this work, diamond films were grown in an ASTeX
microwave reactor using H2S as the sulfur source. The methane concentration ranged from 0.1 -
0.4%; the S/C atomic ratio from 15 to 20,000 ppm. In some experiments, co-doping of sulfur and boron was attempted
by including trimethylboron (TMB) in the source gases. Single crystal diamond with {111}, {110},
{100} faceting and n-type silicon wafers were used as substrates. The gas flow was 200 sccm, the pressure was
25 torr, and the microwave power was 1020 W.
Substrate temperatures were measured using an optical pyrometer
(Williamson 8220C) and ranged from 700 to 750oC. The samples were analyzed for sulfur by
Secondary Ion Mass Spectroscopy (SIMS) and Particle-Induced X-ray Emission
(PIXE). Gold-capped titanium contacts
were annealed under vacuum up to 400oC for two hours to create ohmic
junctions when required.
Electrochemical measurements were conducted in a cell filled with 0.5 M
H2SO4 electrolyte solution, using a Ag/AgCl reference
electrode. Differential capacitance
measurements were conducted with an EG&G Instruments Potentiostat, Model
No. 283, and EG&G Instruments Frequency Response Detector, Model No. 1025.
The growth experiments proceeded in two distinct
stages. Prior to the first set of
growth experiments, the chamber was dismantled, cleaned, and reassembled. In the first set of experiments, no boron
was used in the feed gas, and it is believed that there was no significant
amount of boron in the reactor.
Analysis by SIMS of diamond grown in the reactor showed no increase in
boron concentration above background levels.
PIXE analysis of diamond samples grown under these conditions gave no
detectable sulfur-incorporation when H2S was included in the source
gas. Additionally, most samples were
too resistive for electrochemical measurements.
The second set of experiments began after boron (as
TMB) was introduced to the reactor. The
reactor was not disassembled and cleaned between these runs so there were low
levels of boron present even when TMB was not added to the source gases. Residual gas analysis showed the presence of
BH3 in runs done in the absence of TMB. In this second set of runs, i.e., after boron was introduced to
the reactor, sulfur could be observed by PIXE.
Sulfur was observed in diamond grown on {111}, {110}, and {100}
surfaces. Figure 1 shows a PIXE analysis for a {110} sample.
When TMB and H2S are both used in the feed
gases, the resulting diamond films are p-type.
The Mott-Schottky plots show a negative slope, indicative of p-type
carriers with a non-compensated acceptor concentration ranging from 1018
to 1021 cm-3. A representative example is shown in Figure 2, which has a flat-band potential at 0.43 V vs.
Ag/AgCl. This flat band potential is
about 1V more negative than is observed on diamond grown with no sulfur present
in the source gas. (See Figure 5.) If TMB was not added to the feed gas, then
the samples grown using H2S are generally n-type. The Mott-Schottky plots show a positive
slope, which denotes n-type conductivity.
The slopes indicated donor concentrations that ranged from 1014
to 1021 cm-3. A
representative Mott-Schottky plot of one of these samples is shown in Figure 3.
Figure 1. PIXE results: counts in arbitrary units versus energy in
keV. H-ion energy was 3 MeV with a 22.5o incident angle. The
sulfur peak occurs at 2.31 keV. The
substrate was a (110) diamond. Growth conditions were S/C atomic ratio in
the gas phase of 1250 ppm, a methane concentration of 0.2% and residual boron
in the reactor. Film thickness was 14
microns.
Figure 2. Mott-Schottky
plot, C-2 vs. E, from a (111) diamond surface.
Growth conditions were S/C atomic ratio in the gas phase of 2500 ppm, a
B/S atomic ratio of 0.4, and methane concentration of 0.2%. The negative slope indicates p-type
conductivity. The number
of acceptors, NA, is 5 x 1021 cm-3. The
electrolyte is 0.5 M H2SO4.
Figure 3. Mott-Schottky
plot, C-2 vs. E, from (110) diamond surface.
Growth conditions were S/C atomic ratio of 1250 ppm, methane
concentration of 0.2% and residual boron in the reactor. The positive slope indicates n-type
conductivity. The number
of donors, ND, is 4 x 1017 cm-3. The
electrolyte is 0.5 M H2SO4.
Plasma treatment experiments
Other experiments were performed in which diamond
substrates were treated in an H2/H2S plasma at 25 torr
for 12 hours without CH4 in the feed gas. There were some residual hydrocarbons in the chamber
however. These samples exhibited some
of the same electrical properties as samples in which diamond growth occurred
in the presence of H2S. For
example, a virgin diamond macle treated for 12 hours with a hydrogen plasma
containing 13 ppm H2S gave
an n-type Mott Schottky plot with a donor concentration of close to 1021
cm-3. This is shown in
Figure 4.
Figure
4. Mott Schottky plot, C-2 vs. E, from
a virgin diamond macle (111) surface treated with a H2/H2S
plasma containing 13 ppm H2S for 12 hours.
Treatment of boron-doped diamonds with a sulfur
containing plasma was also performed. A
Mott-Schottky plot from a typical heavily boron-doped diamond is shown in
Figure 4.
The slope indicates p-type conduction with an acceptor concentration of
2x1019 cm-3; the intercept gives a flat band potential of
about 1.3 V vs. Ag/AgCl. The macle of
Figure 4 was then treated under a H2/H2S plasma for 12
hours; the results are shown in Figure 5. The maximum at » 0.6 V vs. Ag/AgCl was reproducible and has been seen
in other samples. This type of behavior
has been observed in other systems by others and can arise from several causes:
the presence of rapid surface states located in the band gap (ref. 5) or the
presence of more than one donor or acceptor level.
Other Measurements
The response of the open-circuit photo-potential under
ultraviolet illumination at 254 nm was also tested and these experiments
confirmed the Mott-Schottky results.
Samples that showed n-type Mott-Schottky behavior had a negative shift
in the open circuit photo-potential and p-type samples had a positive shift as
expected.
Figure 5. Mott-Schottky plot, C-2 vs. E, from
a heavily B-doped diamond film grown on a (111) diamond macle surface. The number of acceptors, NA, is
2.3 x 1019 cm-3. The electrolyte is 0.5 M H2SO4.
Figure 6. Mott-Schottky plot of macle
from Figure 4 after treatment in a H2/H2S plasma
containing 13 ppm H2S for 12 hours.
The electrolyte is 0.5 M H2SO4.
Measurements of the sign of the thermoelectric effect
were also performed. These measurements
were done using a two-point probe in which one of the probes was heated several
tens of degrees above room temperature by a very small resistance heater wound
around the probe. Both probes were
tungsten. The sign of the
thermoelectric effect was determined by measuring the voltage deflection when
the heater was turned on. The system
was tested using known samples of n-type and p-type silicon and with pure
copper metal. For the n-type silicon
samples, the cool probe was negative relative to the warm probe; for p-type
samples the cool probe was positive relative to the warm probe. Copper gave no
detectable signal. All diamond samples
that showed n-type Mott Schottky plots also showed n-type thermoelectric
effect; all p-type diamond samples showed a p-type thermoelectric effect.
DISCUSSION OF RESULTS
The results can be summarized as follows. Sulfur incorporation, in or on diamond, was enhanced by the presence of boron. Secondly, n-type behavior was only observed when sulfur was used, either during growth or in a post-growth plasma treatment, and when only residual amounts of boron were present in the reactor. The observation of n-type conductivity following plasma treatment with sulfur in the absence of appreciable growth is indication that surface states or near-surface structures play a role in the observed n-type behavior. This does not preclude the possibility that donor centers may also arise from substitutional sulfur or sulfur at grain boundaries. Recent calculations by Albu (ref. 6) indicate that substitutional S and substitutional BS centers are deep donors, with levels that each lie about 1.5 eV below the conduction band.
The co-doping experiments with boron and sulfur were motivated by elementary strain energy considerations that suggest that the smaller boron atom may facilitate the incorporation of the larger sulfur atom into diamond. Also, gas-phase equilibrium calculations and Langmuir adsorption calculations show that the sulfur concentration on the growing surface is increased by the presence of boron. These calculations indicate that gas phase radical species such as BS and BS2 can increase the sulfur concentration on the surface by up to two orders of magnitude by bonding to diamond surface radicals.
In the second group of experiments, residual boron was present in the reactor even when TMB was not added to the source gases. The wide variation in the observed number of donors may be due to different levels of compensation of the sulfur by the residual boron. Two boron atoms are required to compensate doubly ionized sulfur; however, boron is more easily incorporated into diamond than sulfur.
It should be noted that in some cases the doping levels in the diamond may have been too high to meet the assumptions inherent in the Mott-Schottky analysis. Also, as noted, surface states can influence the results. Therefore, the values of the carrier concentrations obtained from the plots must be treated as approximations.
ACKNOWLEDGEMENTS
The support of the National Science Foundation (Grant
CHE98-16345) and the Civilian Research and Development Foundation (Grant RC1-2053) is gratefully acknowledged. Mr. Cliff Hayman designed and constructed
the thermoelectric device.
REFERENCES
1. Koizumi, S; Teraji, T.; and Kanda, H.: Phosphorus-doped chemical vapor deposition of diamond. Diam. and Relat. Mater. 9, 3-6, April-May 2000, 935-940.
2. Sakaguchi, I., et al.: Sulfur: A Donor Dopant for n-type Diamond Semiconductors. Phys. Rev. B. 60, 4, July 1999, 2139-2141.
3. Nishitani-Gamo, M., et al.: Homoepitaxial Diamond Growth with Sulfur-Doping by Microwave Plasma-Assisted Chemical Vapor Deposition. Thin Solid Films, 382, 1-2, Feb. 2001, 113-123.
4. Kalish,
R.,et al.: Is Sulfur a Donor in Diamond? App. Phys. Lett., 76, 6, Feb. 2000,
757-759.
5. Barrai, G.; Njanjo-Eyoki, F.; and Maximovitch, S.: Characterization of the Passive Layer and of Hydroxide Deposits of Nickel by Impedance Spectroscopy. Electrochim. Acta 40, 17, 1995, 2815-2828.
6. Albu, T.; Anderson, A. B.; and Angus,
J. C.: Extended Abstract, 2001 Joint International Meeting of the
Electrochemical Society and the International Society of Electrochemistry, San
Francisco, September 2-7, 2001