Ion beam induced luminescence
in natural diamond
P. A. Sullivan and FL A. Baragiola
Laboratory for Atomic and Surface Physics, Engineering Physics, University of Vkginia, Charlottesville,
Wrginia 22901
(Received 20 January 1994; accepted for publication 23 June 1994)
We used ion beam induced luminescence (IBIL) to study the evolution of damage during ion
implantation of natural type-IIa diamond with 30 keV carbon ions, for different sample temperatures
in the range 35-300 K. Blue band-Aluminescence decays with irradiation dose from an initial value
of -10e3 photons/e-h pair. This decay was modeled to obtain an effective damage cross section of
-40 A2, which is roughly independent of-temperature. We explored the use of IBIL as a tool to
characterize the processing of p-n junctions in natural type-IIb diamond and apply it to find
conditions that determine the ultraviolet response of light-emitting diodes.
1. INTRODUCTION
Optical communication, video display, memory storage,
and other optoelectronic technologies have need for efficient
light sources, like light-emitting diodes (LED).l More efficient short-wavelength LEDs are desirable, especially in the
ultraviolet, where they would allow higher storage densities
in optoelectronic memories. This use requires a semiconductor with a large band gap and high carrier mobilities, like
diamond, which has a band gap of -5.5 eV2 Diamond also
has many other unique properties, like high thermal conductivity and a small coefficient of thermal expansion, that
would make it very desirable for many semiconductor device
applications.
ITtyo main barriers for such a use exist at present: the
difficulty of synthetic growth of low-cost, high-quality material; and the difficulty in doping to obtain the desired electrical properties. The first barrier is being attacked by different techniques that attempt to grow single-crystal diamond
from the gas phase. The second barrier results from the fact
that diamond converts to graphite at the temperatures required to diffuse most useful impurities to electrically active
sites. Successful p layers have been produced by B ion
implantation3-s but it is found that heat treatment alone cannot anneal the radiation damage that accompanies the implantation technique. This damage, in the form of defect
complexes of interstitials, vacancies, and impurities, affects
properties such as the electrical conductivity and luminescence. On the other hand, ion implantation damage can show
n-type conductivity, ‘-lo which has led to the successful fabrication of light-emitting diodes.” The nature of this damage
is not well understood due to the difficulty of measuring and
characterizing defect configurations, but it must be controlled if ion implantation is to become a viable doping technique.
A useful probe for characterization of diamond and of
processing steps like ion implantation is the luminescence
resulting from electronic states in the band gap created by
defects. It can be induced by energetic particles or radiation
either directly by exciting a defect center, or indirectly by
exciting electron-hole pairs that later recombine radiatively
at the centers. There have been several studies on electronand photon-induced luminescence (cathodoluminescence and
photoluminescence) of unirradiated, irradiated, and irradiated
J. Appl. Phys. 76 (8), 15 October 1994
and then heat-treated diamond, as described in several
reviews.‘*-t4 The most intense luminescence band occurs in
the blue (-433 rmr), and was first found by Dean” on unirradiated diamond. He labeled it band-A emission and concluded from the width and the wavelength dependence of the
intensity decay rate that the light is produced by recombination of electrons from donor sites and holes from acceptor
sites. Other investigators determined the donor to be a nitrogen molecule, and the acceptor, a boron atom,‘” common
impurities in natural -diamond. However, recent investigations of cathodoluminescence from chemically vapordeposited diamond’7”8 and natural diamonds19 suggest that
the band-A is not due to donor-acceptor pair recombination
but to dislocations or vacancy clusters acting as recombination centers for electron-hole pairs. Additional luminescence
peaks can be induced by radiation damage, like the 5RL band
peaking around 315 nm in the near ultraviolet, tirst reported
by Wight” using cathodoluminescence. A feature around this
wavelength has also been found to be present in the electroluminescence spectra produced by applying an electrical
bias across ion implanted p-n junctions.”
Energetic ions incident on diamond can also induce luminescence through electronic excitations, in what is called
ion beam induced luminescence (IBIL).21 In the present
work we show how IBIL can be used as a new method to
monitor in situ the amount of damage produced during ion
implantation. We applied IBIL to investigate the role of implantation parameters on defect production and extract the
temperature dependence of the luminescence efficiency and
damage cross section for band-A centers. We also applied
IBIL to characterize ion implanted p-n junctions in the process of making LEDs. We found that IBIL spectra produced
by hydrogen ions on ion implanted and annealed samples
have the same general features found in electroluminescence
of LEDs, the near-UV and blue features mentioned above
and a less intense feature in the yellow peaking near 580 nm.
The correlation between IBIL and electroluminescence studies aIlowed us to identify implantation parameters that enhance the emission of ultraviolet light.
Il.-EXPERIMENTAL
METHODS
We used a crystal of type-IIa natural diamond (2.74
X1.65X8.33 mmj with large flat faces 8” off (110). Prepara-
0021-8979/94/78(8)/4847/6/$6.00
0 1994 American Institute of Physics
4847
Downloaded 20 Jul 2001 to 128.143.34.117. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
ION BEAM
b
,
GAS FLOW UMlTlNG
APERTURE (5 mm)
’
I
*
1
*
I
-
18
I
s
I
’
to
2500
UPAESSION CUP AND
C+ dose (lOI
--o-O.9
P ‘1
ions/cm’
....... . 2.8
4
--o--4.8
----6.8
______
-s-8.7
b
l
v
P.
0
250
300
350
:
,:’
400,
. ..-.
‘:,
450
L
500
550
600
Wavelength (nm)
FIG. 1. Schematic drawing of the target chamber.
tion of a diamond face for irradiation included a mechanical
polishing using a diamond wheel with 1400 diamond grit,
followed by a 15-min graphite etch in a boiling solution of
perchloric:nitric:sulphuric acids in a 1:4:3 ratio,” and finally,
an ultrasonic bath in de-ionized water.
The sample was placed in the target chamber (Fig. l),
maintained at a base pressure of 10m8Torr. W e clamped the
sample between a copper block and a nickel grid mask. Ion
bombardment ejected low-energy electrons from this grid
that neutralized the buildup of positive charge in the diamond. If the grid is not used, the diamond will charge to
several kilovolts until arcing takes place. This arcing can
produce pits that appear after heat treatment of the diamond.
A 1.1~mm aperture confined the ion beam entirely to the
diamond surface and was biased to prevent the escape of the
secondary electrons from the sample.
The irradiations were done with 1-2~4, 30-keV r2C ion
beams from the University of Virginia 150-keV heavy ion
accelerator. All experiments were done at 45” incidence, and
not aligned with a major axis to avoid channeling. IBIL was
observed with a spectrometer equipped with a 1200groove/mm grating blazed at 500 run, and working at a resolution of 25 nm with a Hamamatsu R1527 photomultiplier
detector. The spectra were taken for different ion beam doses
and sample temperatures in the range 35-300 K. After subtraction of electronic noise, the luminescence intensity was
normalized by dividing the adjusted value by the beam current.
Figure 2 shows examples of IBIL spectra taken during
carbon ion irradiation of a sample held at 100 K. The prominent blue band-A peak is due to intrinsic color centers in the
diamond samples, which are excited by the ion beam. Each
spectral curve was taken on a fresh spot, after different accumulated ion doses. It was found that the IBIL intensity
decayed rapidly with time and that irradiation by itself did
not introduce any new spectral features in our wavelength
range. The initial or zero dose value of the intensity of the
blue peak depended on the implantation temperature, but the
spectra did not change shape over the range from 35 to
300 K.
4848
J. Appl. Phys., Vol. 76, No. 8, 15 October 1994
FIG. 2. IBIL spectra taken during 30-keV C+ irradiation of type-Ha diamond at 100 K, for different accumulated Cc doses. The intensity of the
spectra decreased and the bandwidth remained constant with increasing
dose.
In another type of experiment we monitored the IBIL
intensity at the peak, 433 run, as a function of bombardment
dose. To take into account fluctuations of the beam current,
we normalized the count rate of the photomultiplier to the
ion beam current. Figure 3 shows that the IBIL decays rapidly with dose reaching a background level after a dose of a
few times 1Or4ions/cm2. The decrease in intensity is not due
to darkening produced by graphitization, which starts to be
noticeable at much higher doses.23
To study the effect of heat treatment on ion implanted
diamond, separate spots of the sample, maintained between
220 and 300 K, were irradiated with 30-keV carbon ions to
different doses. The ability to luminesce, which was destroyed after the high irradiation doses (Fig. 3), was restored
after heat treatment. W e probed implanted and heat-treated
samples using IBIL induced by a beam of -3 nA Hl at 30
keV. These ions have an excitation range of -1500 A or
about three times the depth of the C-implanted region,% so
lO'L--0
1
1
.
'
2
'
'
3
.
'
'
'
4
5
6
*
'
.7
8
Dose (10” cm-‘)
FIG. 3. Beam dose dependenceof the IBIL signal at 433 n m for 30-keV- Cc
on type-Ha diamond at 53 K. The solid line represents the fit to Eq. (6). The
signal, in photon counts per second, is normalized to the ion beam current.
P. A. Sullivan and R. A. Baragiola
Downloaded 20 Jul 2001 to 128.143.34.117. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
.o.
250
300
350
400
450
Wavelength
500
550
600
650
(nm)
FIG. 4. Luminescence spectra from diamond samples after carbon implantation and annealing at 1100 “C. The EL (electroluminescence) spectrum is
from a diode made from a type-IIb substrate, implanted with a combination
of 8, 14, and 26X10” Cc/cm2 at 30,50, and 80 keV, respectively. The IBIL
and CL. (cathodoluminescence) spectra were from a type-IIa diamond irradiated with 1.35 and O.55X1O1s C’/cm*, respectively. The intensity scale is
linear and the spectra are displaced vertically for ease of visualization.
the luminescence comes both from damaged and undamaged
regions of the sample. Hydrogen was used for this study
because it produces less damage than carbon, and so the
luminescence decayed more slowly with dose. Figure 4
shows that irradiation followed by heat treatment induces
two additional luminescence features besides the main A
band already present in the virgin samples: the ultraviolet
band discussed above and a less intense feature in the yellow
peaking near 580 nm. These features are thought to result
from two different luminescence centers related to defect
configurations.
We took hydrogen IBIL spectra at T&90 K from regions
of the sample implanted at 200-300 K with different C+
doses, and then annealed at 1100 “C. While we could not
compare absolute intensities due to variations in initial
sample conditions and alignment of the hydrogen beam with
the C+ implanted spot, we can still look at relative changes
in the spectra by normalizing them. This is done in Fig. 5,
which shows that the ratio of the near-UV to the blue peak
increases with increasing dose. The fact that the relative intensities of the UV and blue peaks change with C+ implantation dose supports the idea that the UV emission originates
from defect centers induced by the implantation-annealing
sequence.
To study electroluminescence, we made light-emitting
diodes guided by techniques developed by Prins,25 using
natural type-IIb diamond disks 0.25 mm thick with l-mmdiam faces. Prior to the irradiation the diamond was mechanically lapped, boiled in the etching solutionz2 to remove
any graphite, and cleansed in an ultrasonic bath of de-ionized
water. We made p ’ ’ contacts by implanting 35-keV boron
ions to a dose of 3X1016 cmw2 on samples held at 200 “C.
This implantation created a highly doped layer underneath a
defect-rich top layer. The irradiation was followed by a heat
treatment meant to collapse the defect-rich top layer and alJ. Appl. Phys., Vol. 76, No. 8, 15 October 1994
250
300
350
400
C+ dose (1O’5 ionsknY2)
450
Wavelength
500
550
600
1
650
(nm)
FIG. 5. IBIL spectra from 30 keV Hz on damaged, type-IIa diamond at 90
K. The sample was held between 200 and 300 K during the initial 30 keV
C+ implantation and then annealed at 1100 “C. The relative intensity of the
band in the near ultraviolet increases with increasing carbon dose.
low diffusion of vacancies out of the underlying boron-doped
layer. This was done by dropping the diamond into a quartz
furnace for 5 min at 1100 “C! in an argon atmosphere. The
diamond was then etched to remove the collapsed graphitic
top layer and leave a highly doped ohmic surface.
The other side of the diamond was irradiated with C+
ions at 30, 50, and 80 keV to create a relatively thick (-0.15
pm) “n-type” conducting layer. The doses at these energies
were kept in the proportion 4:7:13 to obtain a damaged layer
uniform in depth. The samples were then given heat treatment for 5 min at 1100 “C, a 20-min etch, and finally loloo-m-thick
Au coatings were sputter-deposited on each
side of the device.
The electroluminescence is emitted when the diodes are
forward-biased, and shows the same features as the IBIL
spectra on implanted and annealed samples, indicating that
the same luminescence sites are being accessed. We found
that the relative intensity of the spectral features depend sensitively on implantation..dose but not on implantation temperature. The relative intensity of the ne,ar-ultraviolet band
increases with increasing dose (Fig. 6); which correlated very
well with the H,f-excited IBIL spectra of Fig. 5 and also with
the electroluminescence investigations by Prims.”
III. MODEL FOR IBIL DECAY WITH DOSE
We modeled the decay of the band-A IBIL intensity with
dose to obtain information about the luminescence and radiation damage processes. We relate the observations’to the decay of the efficiency of scintillators, used routinely in nuclear
studies, as due to exposure to radiation. These scintillators
emit luminescence when struck by ionizing radiation and so,
together with a photomultiplier, can be used as nuclear particle detectors. Damage to the scintillator produced by prolonged exposure to radiation decreases- the luminescence
efficiency.“6 The decay of luminescence in organic scintillators with dose is often explained with the aid of a model by
P. A. Sullivan and R. A. Baragiola
4849
Downloaded 20 Jul 2001 to 128.143.34.117. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
1
-
I-
I.
1.
I,
I.,
No-N,=N,,
will not. Now, if c, and c, are transition rates
per unit concentration, then Rm=c,JV, and R,= c,.N, , SO
*
Total Cc dose (1016 ionskin*)
L=1i/(l
.z 0.8
(2)
where k=c,/c,
is called the transition rate ratio and
x=N,,JNr .
Radiation damage converts radiative into nonradiative
centers as
ii
5 0.8
111
B
2 0.4
dNr = - crdDN, ,
g
*
+c&v&$v~)=1~/(1+kx),
where D is the ion dose and cr an effective cross section for
damage of the radiative centers averaged over the path of the
ion.30 Integration gives
0.2
0.0
250
(3)
300
350
400
450
500
550
600
650
N,=No
Wavelength
exp(- CT-D).
(4)
Then,
FIG. 6. Normalized electroluminescence spectra from three different diodes.
Different total Cc doses at three energies were used in creating the n-type
layers. The intensity in the near-ultraviolet increases relative to that in the
blue with increasing implantation dose.
x=(No-N,)/N,=[No-No
L=I;(l+R,/R,).
(0
Let us denote by Na the total concentration of centers that
can radiate in the absence of radiation damage. Due to ion
bombardment, N, centers will be able to radiate and
4850
J. Appl. Phys., Vol. 76, No. 8, 15 October 1994
exp(-aD>]
64
or
x=exp(oD)-
Birks and Black.27 We use this model here since it gives the
simplest expression compatible with our observed dose
dependence.=
Carbon ions incident on diamond lose energy through
elastic, “nuclear” collisions, and inelastic interactions like
electron-hole pair excitation and direct excitations to states
in the band gab. These inelastic excitations can then relax by
nonradiative processes or produce luminescence. We make
the usual assumption that the production of damage is proportional to the nuclear energy deposited and the luminescence is proportional to the electronic energy de osited by
both the projectile and the recoiling target atoms. 8
Nonradiative electron-hole recombination (quenching)
can result from different processes, e.g., multiphonon, Auger,
and recombination with surface states. We denote the rate of
quenching as R, . Alternatively, an electron can be captured
(trapped) by a recombination center with a transition rate
R,. Once at the center, the electron can either decay nonradiatively with a rate R, if the center is in the presence of
radiation damage or radiatively with rate R, . We assume that
damage, local to a center and produced by the beam, does
not alter the capture probability of a recombination center
but does inhibit its ability to produce radiation either by creating a more efficient nonradiative path or by hindering the
formation of the radiative state. As the concentration of damage increases with dose, the fraction of luminescence events
decreases.
Consider the total number of electron hole pairs per unit
time, Ia, created by the incident beam. A fraction of these,
16, will be captured by either a recombination center in the
proximity of radiation damage or by a center which can radiate. The luminescence intensity per unit time is
exp(-crD)]/Na
1.
(5b)
From Eqs. (2) and 5(a):
L=La(l+k[exp(oD)-l]},
(6)
where Lo=L(D =0) is the intensity of luminescence at the
start of the irradiation.
We still need to determine the functional nature of Lo.
The fraction of electrons which decay via the recombination
level is determined by the rate ratio RJR, and the e-h pair
production rate. A derivation similar to that used to obtain L
in Eq. (2) can be used to obtain Lo:
L,=I,/(l+k’x’),
(7)
is another transition rate ratio and
where k’ =cq/c,
x’=N’/N,,
is the ratio of the total density of nonradiative
centers for conduction band electrons, N’, to the total density
of recombination centers, Na. The observed functional dependence indicates that the ratio N’/Na is not affected by
dose (i.e., the number of centers cannot be created or destroyed by the beam).
We use a combination of Eq. (6) and an additive constant
to model the measured dependence of IBIL on dose, L(D).
The constant accounts for background radiation from excited
sputtered and backscattered particles. The quantities Lo, k,
and u, obtained by fitting the model to the data, are displayed
in Fig. 4.
IV. DISCUSSION
Four parameters of interest can be extracted from the fit
of the model to our data of decay of luminescence with C+
dose: the transition rate ratio (k=cn/Cr), the effective damage cross section (a), the initial luminescence yield (Ye), and
the relative damage resistance, EIj2.
The ratio of the nonradiative to the radiative transition
rate, k, is found to vary between 90 and 310 and to increase
with temperature, as expected.31 The effective damage cross
sections obtained (averaged over the excitation path of the
ion) appear to be roughly independent of temperature below
P. A. Sullivan and R. A. Baragiola
Downloaded 20 Jul 2001 to 128.143.34.117. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
room temperature. This implies that annealing during irradiation is constant or is not important in this temperature regime. The value of the damage cross section obtained is
c-40 A2. An estimate of the vacancy production cross settion is a;=FIE,=
1.2 p, where F is the average nuclear
energy deposited per unit path length24 and Ed=43 eV is a
mean displacement energy for a carbon atom in a diamond
lattice.32 The relatively large value of the observed damage
cross section suggests that it is related to the extent of the
radiative state R=(40 w2/,rr)*“=3.6 k This value, similar to
the lattice spacing in diamond (3.57 A).>,is reasonable for the
radius of a radiative state centered at a vacancy or an impurity inside the lattice.
The initial intensity of light produced by a C’ beam is a
function of the ion beam current +i, and the term
(1 + k’x’)-l
represents the efficiency for nonradiative recombination of conduction band electrons. They are related
by the initial luminescence yield:
Yo=Lo~/$i=tZ~(
1 +k’X’),
(8)
where 4 is the ion charge and nr is the number of electron
hole pairs excited by each incident ion. The transition rate
ratio (for conduction band electrons), k’=cJc,
should increase (i.e., Lo should decrease) with diamond temperature
as previously discussed. The decrease in IBIL intensity is
consistent with the observations that the cathodoluminescence intensity from type-IIb33 or CVD34 diamond decreases
with temperature.
Some quantitative information about the relative efficiency of the radiative centers can be extracted from the
magnitude of Y,,. From Fig. 7 and an estimated light detection efficiency of -10e6, we derive Y,--1-4 photons/ion.
This has to be compared with nI, the number of e-h pairs
produced per incident ion. The total electronic energy dissipated by each 30-keV Cf projectile is -17.5 keV24,35while
the energy required to create an e-h pair can be estimated to
be 16.5 eV, or three times the band gap.% Therefore,
nl-1060 e-h/ion, so, initially, it is -260-1000 times more
likely for electrons in the conduction band to recombine with
a hole via a nonradiative process than through the radiative
process that produces the blue band-A light.
To determine the relative damage resistance of diamond
as a scintillator, we use Birks’ paramete?’ EI12, the energy
an incident particle dissipates per unit volume in reducing L
to L,/2. From Eq. (7) for 30-keV carbon incident on diamond,
L/L,,-l/(l+kcrD).
(9)
For LIL,=1/2,
-5.4x
1011 cm-2 3
(10)
where k and u are taken from the room temperature implantation. Using the total energy deposition per unit path length
averaged over the range of the carbon ions, F,-5 X lo9
eV/cm3. This
eVlcm,24 we obtain E 1,2=Fp1,2=2.7X102*
value, pertaining to our case where most damage is due to
elastic collisions, is of the, order of the highest values found
for ionization damage of organic scintillators.36
J. Appt. Phys., Vol. 76, No. 6, 15 October 1994
301
I
.
I
,
I
.
,
.
300 Y
./
200 -
/d
i
I
,
/
I
1
/f
100 -
0"0
s-d---N
508
' 100
8
' 150
1
" 200 1 250
"
Temperature
1 'I 350:
300
(K)
FIG. 7. IBIL parameters
from the model fit to Eq. (6). I,,, : normalized initial
luminescence (not corrected for photon collection efficiency), CT: effective
cross section for damage of the radiative center, and k: nonradiative to
radiative transition rate ratio. The lines are only meant to guide the eye.
V. CONCLUSIONS
The luminescence from type-IIa diamond excited by ion
bombardment is dominated by blue band-A emission near
433 nm with the intensity decaying with ion dose. This decay
was modeled assuming that the luminescence centers are deactivated by damage produced in their vicinity by the incident ions. The fit of the model to the data gives the relative
nonradiative transition rate of an electron at a damaged center and the effective damage cross section for a lattice atom.
Even in the absence of damage, diamond is an inefficient
scintillator, since the vast majority of the electron-hole pairs
produced recombine without light emission. However, its use
as a scintillator may be.desirable due to its relatively high
resistance to radiation damage.
IBIL can also be used to characterize the processing of
light-emitting diodes made by ion implantation followed by
rapid thermal annealing. Besides the A-band, the “n-type”
radiation damage region has two additional features in
cathodoluminescence, electroluminescence, and IBIL spectra, one in the near-ultraviolet and another, less intense, in
the yellow. The relative IBIL and electroluminescence intensities in the ultraviolet band increase with an increase in
carbon dose but do not appear to depend on the temperature
of the diamond between 35 and 300 K during this irradiation.
The existence and controllability of this feature make diamond a candidate for a solid-state ultraviolet light source.
The correlation between the spectra obtained by JBIL after
P. A. Sullivan and R. A. Baragiola
4651
Downloaded 20 Jul 2001 to 128.143.34.117. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp
annealing and the EL spectra for finished diodes suggests the
use of IBIL as an in situ characterization technique for intermediate steps in the processing of devices.
ACKNOWLEDGMENTS
We are grateful to J. W. Boring, R. E. Johnson, and
W. A. Jesser for useful discussions. We thank J. F. Prins for
lending us the diamond and for helpful suggestions. This
work was partially supported by the National Science Foundation through Grant No. DMR-9121272.
ISee, for example, S. Strite and H. Morkoc, J. Vat. Sci. Technol. B 10,
1237 (1992). aJ. E. Field, The Properties of Diamond (Academic, San Diego, 1979).
sV. S. Vavilov. M. I. Guseva. E. A. Konorova, V. V Krasnopevtsev, V. F.
Skrienko, and V. V. Tutov, Sov. Phys.-Solid.State 8, 1560 (1966).
4J. F. Prins, Phys. Rev. B 385576 (1988).
sJ. F. Prins, Nucl. Instrum. Methods B 59/60, 1387 (1991).
‘L. A. Davidson, S. Chou, J. F. Gibbons, and W. S. Johnson, Radiat. Eff. 7,
3.5 (1991).
7J. J. Hauser, Solid State Commun. 18, 789 (1976).
8J. J. Hauser, J. R. Patel, and 3. W. Rodgers, Appl. Phys. Lett. 30, 129
(1977).
‘J. F. Prim, Phys. Rev. B 31, 2472 (1985).
l”G. Braunstein, A. Talmi, R. Kalish, T. Bernstein, and R. Beserman, Radiat.
Eff. 48, 139 (1980).
rlJ. F. Prins, Appl. Phys. Lett. 41, 950 (1982).
“J. Walker, Rep. Prog. Phys. 42, 1642 (1979).
r3 G. Davies, Chemistry and Physics of Carbon, edited by P. L. Walker and
P. A. Thrower (Dekker, New York, 1977), p. 128.
r4A. T. Collins, Physica B 185, 284 (1974).
“P. J. Dean, Phys. Rev. 139, A588 (1965).
r6R. M. Chrenko, Phys. Rev. B 7, 4560 (1973).
4652
J. Appi. Phys., Vol. 76, No. 6, 15 October 1994
r7J K. Ruan, K Kobashi, and W. J. Choyke, Appl. Phys. Lett. 60, 3138
k992).
18J. K. Ruan and W. J. Choyke, Am. Ceram. Sot. Bull. 72, 139 (1993).
“J. F. Prins, Diamond and Relat. Mater. (to be published).
2oD. R. Wight, Ph.D. thesis, King’s College, University of London, 1968.
a1P. D. Townsend, Rep. Prog. Phys. 50, 501 (1987).
“J. F. Prins, Mater. Sci. Rep. 7, 271 (1992).
=X. K. Zhang, F. Fang, J. G. Guo, Y. F. Yao, and Z. Y. Zhou, Phys. Status
Solidi A 129, K97 (1992).
wDetermined by TRIM, a Monte Carlo simulation code [J. P. Biersack and L.
G. Haggermark, Nucl. Instrum. Methods 174, 257 (1980)]: We used version TRIM-9212 developed by J. Ziegler, IBM (unpublished).
=J. F. Prins, J. Phys. D 22, 1562 (1989).
%G. F. Knoll, Radiation Detection and Measurement (Wiley, New York,
1989), p. 225.
“J. B. Birks and F. A. Black, Proc. Phys. Sot. London Ser. A 64, 511
11951).
=H. B. Rosenstock and J. H. Schulman, J. Chem. Phys. 30, 116 (1959).
29M. Luntz and J. Figueroa, Phys. Rev. B 9, 87 (1974).
30Actually slightly less, since the ions do not produce significant electronic
excitations near the end of their range.
31J I Pankove, Optical Processes in Semiconductors (Dover, New York,
l-975).
32J. Koike, D. M. Parkin, and T. E. Mitchell, Appl. Phys. Lett. 60, 1450
(1992).
33N. Yamamoto, J. H. C. Spence, and D. Fathy, Philos. Mag. B 49, 609
(1984).
34H. Kawarada, Y. Yokota, Y. Mori, K. Nishimura, and A. Hiraki, Appl.
Phys. 67, 983 (1990).
35We do not include here the small electronic energy deposited by recoils.
We expect it to be smaller than the value of 3.4 keV calculated using TRIM
(Ref. 24), due to the large suppressing effect of the band gap to the energy
loss of slow particles.
a6J . B . Birks, The Theory and Practice ofScintillation Counting (Pergamon,
Oxford, 1964), p. 206.
P. A. Sullivan and R. A. Baragiola
Downloaded 20 Jul 2001 to 128.143.34.117. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/japo/japcr.jsp