Optical and Electrical Properties of GaMnN Films
Grown by Molecular Beam Epitaxy
A.Y. Polyakov, A.V.
Govorkov, N.B. Smirnov, N.Y. Pashkova
Institute of Rare Metals,
B.Tolmachevsky 5, Moscow,109017,Russia
G.T.Thaler, M.E.Overberg,
R.Frazier, C.R.Abernathy and S.J.Pearton
Department of Materials
Science and Engineering,University of Florida,Gainesville,FL,32611,USA
Jihyun Kim and F.Ren
Department of Chemical
Engineering,University of Florida,Gainesville,FL 32611,USA
ABSTRACT
Optical
absorption spectra, microcathodoluminescence (MCL) spectra and electrical
properties of GaMnN films grown by molecular beam epitaxy with Mn concentration
in the range of 3 atomic % to 10 atomic % were studied. Optical absorption and
MCL spectra show the presence of strong bands corresponding to the transition
from the Mn acceptors near EC-2 eV to the conduction band. The other
strong band observed in MCL measurements was the blue band peaked near 2.9 eV
and associated with the transition from the valence band to deep donors with a
level near Ec-0.5 eV. All GaMnN samples were shown to be lightly
n-type which suggests close self-compensation of the Mn acceptors by some
native defect donors.A plausible scenario is that such compensating donors
could be due to nitrogen vacancies and that the Ec-0.5 eV donor
defects are complexes between the Mn acceptors and the nitrogen vacancy donors.
INTRODUCTION
The doping of GaN with Mn has been of
great interest recently because it was predicted that GaMnN solid solutions
might be magnetic semiconductors with a Curie temperature higher than room
temperature(1,2) Recently it was indeed shown that GaN films doped
with Mn either by diffusion, ion implantation or in the process of growth by
molecular beam epitaxy (MBE) show Curie
temperatures close to or even above room temperature(3-6).
Successful growth of GaMnN solid solutions by metalorganic chemical vapor
deposition was reported although the magnetic properties of such samples were
not studied in detail(7,8). However, these GaMnN films were used to
obtain important information on the position of levels introduced by Mn into
GaN. It was shown that the Mn forms a deep acceptor level near Ev+1.4 eV (7)and
optical absorption and photoluminescence bands corresponding to transitions
from this level to both bands were detected.
It would be very interesting to perform similar studies on GaMnN films
grown by MBE and having well established magnetic properties. This was the
principal aim of the present paper.
EXPERIMENTAL
GaMnN films studied in this paper were
grown by MBE on sapphire substrates as discussed in detail previously(4,9)
The Mn concentration was measured by Auger spectroscopy and also from magnetic
susceptibility measurements. The thickness of all layers was near 0.4-0.5
μm and they were grown using thin low temperature GaN buffers. The
majority of the films studied in this paper were grown at 700 0C,
but some films were also grown at higher temperature of 750 0C. The
Mn concentration in the samples ranged from 3 atomic % to 10 atomic %. The
results of the magnetic properties measurements were already published in Refs
4 and 9. In short these measurements showed that the highest magnetization
signal was achieved for a Mn concentration of 3%, with the Curie temperature
going down for higher Mn concentrations because of the onset of Mn-Mn
antiferromagnetic interactions(1,2). X-ray analysis and transmission
electron microscopy TEM studies showed that MBE grown GaMnN samples were single
phase up to the Mn concentration of about 8% at our particular growth
temperatures.
In the present study, a set consisting
of an undoped GaN film, three GaMnN films with the Mn concentration of 3%, 5%
and 10% and a GaMnN film with Mn concentration of 3.5 % was studied. All films
but the 3.5% sample were grown at 700 0C. The 3.5% sample was grown
at 750 0C. On these samples we measured the room temperature optical
transmission using a Hitachi 330 UV-visible spectrophotometer,
microcathodoluminescence(MCL) spectra at 90K and 300K, the electrical resistivity
and mobility at room temperature and the temperature dependence of resistivity
by the van der Pauw method. We also carried out capacitance-voltage (C-V)
measurements at frequencies from 10 Hz to 10 MHz at temperatures up to 400K
using Au Schottky diodes prepared by vacuum evaporation through a shadow mask.
More detailed description of MCL measurements can be found e.g. in Refs 10 and
11. Other measurements were fairly standard.
RESULTS AND DISCUSSION
Fig. 1 presents room temperature
spectral dependence ofoptical absorption coefficients from the samples.These
were obtained from the transmission values using the spectral dependence of the
reflection coefficient (12,13).
The transmission spectrum of the control sample shows no features except
for the interference fringes and a very sharp bandedge. When converted to
absorption the spectral dependence of the absorption coefficient was quite
similar to the one described in literature (see e.g. Refs.12,13 ).Since the
samples were quite thin, only about 0.4 μm, one could easily get to very
high values of the absorption coefficient corresponding to the onset of
transitions to the L side minimum (12)
Introduction of Mn led to the
appearance of a very strong absorption band with a threshold near 1.9 eV that
has been previously associated with a transition from Mn acceptors to the
conduction band(7,8). The absorption coefficient in this band scales
with the Mn concentration in the samples,which supports the above attribution.
If the samples were highly resistive p-type with the Fermi level pinned near
the Mn level at about Ev+1.4 eV one would expect to see another band
with a threshold near 1.4 eV corresponding to the optical transition from the
valence band to the partially filled Mn acceptor. (7,8) The fact
that we don’t observe such a band indicates that the Fermi level is located in
the upper half of the bandgap and all the Mn acceptors are totally filled. This
is consistent with the results of electrical measurements to be reported below.
In addition to the strong absorption in
the 1.9 eV band we also saw a strong increase in the apparent values of optical
absorption deep within the conduction band . The nature of this effect has yet
to be understood. All studied GaMnN samples but the 10% Mn sample were single
phase according to x-ray diffraction results and transmission electron
microscopy. Thus the observed phenomena should not have come from increased
scattering. It seems factors like changed strain, modification of selectivity
rules in absorption due to sp-d hybridization and also possibly a formation of
resonant states deep within the conduction band (see below) should be
considered.It does not seem that likely that we are dealing with scattering
since the increase in absorption is observed in both single and multi-phase
samples.There is strain that would alter the relative positions of the extrema
and might affect the matrix elements of transitions,but this is impossible to
quantify without detailed theoretical calculations.
MCL spectra
We
start with the results of studies of the vicinity of the 2 eV photon energy
range where one expects to observe the radiative transition from the conduction
band to the Mn acceptor. Unfortunately, this is the region where luminescence
from Cr color centers in sapphire strongly interferes with the measurements (14)
particularly since the samples are very thin. However, when one compares
the 90K MCL spectra of the control sample and the two GaMnN samples with Mn
concentration of 3% and 5% (see Fig. 2) one can see that, in addition to the
well known broad band and a sharp peak attributed to Cr (14) in the
control sample, there appears in the Mn doped samples an additional MCL peak
near 1.9 eV. Note that the intensity of this 1.9 eV additional transition increased
with increasing Mn concentration, which supports the attribution of the
observed new MCL band to the transition from the conduction band to Mn
acceptors also detected in optical absorption.
We could not detect the 1.3 eV band
reportedly present in MOCVD grown p-type GaMnN samples and corresponding to the
transition from the partly filled Mn acceptors to the valence band (7,8).
This confirms again that in our samples the Fermi level is pinned high above
the level of Mn.
Consider now the spectra in the
near-bandedge region. Fig. 3 presents the 90K MCL spectra taken at high
excitation intensity for the control sample and the GaMnN samples with 3%, 5%
and 10% Mn. The control sample’s spectrum consisted of the bandedge line near
3.4 eV and a strong blue band centered near 3 eV. With addition of Mn the
intensity of the bandedge line strongly decreased in all samples, presumably
due to incorporation of a high density of deep recombination centers (an
increase in the resistivity of the GaMnN films compared to the control sample
could also be a factor here, see below). The relative intensity of the blue
band in the 3% and 5% samples was very similar and approximately the same as in
the control sample, although the position of the band was slightly shifted to
the red (2.9 eV instead of 3 eV). The shift in position of the blue band was
clearly discernable with our experimental resolution,making us believe we are
dealing with different centers.In addition,we saw clear differences in samples
doped with either Mn or Co,another transition metal impurity found to produce
interesting magnetic properties in GaN(15).In the 10% Mn sample the
blue band became dominant and again its position shifted slightly to the red
(the peak position near 2.9 eV). When
MCL spectra were taken with about 2 orders of magnitude lower excitation
intensity (Fig. 4) it could be seen that the magnitude of the bandedge band
steadily decreased while the relative intensity of the blue band became more
and more prominent with increasing the Mn concentration. The parameters of the
blue band in the Mn doped samples are very similar to those of heavily Mn
implanted samples. The detailed results of the latter experiments will be
discussed in a separate paper (15) but here we would like to point
out that these experiments suggest that the blue band observed in GaMnN samples
is different from that detected in undoped samples. In the former the band
seems most likely to come from recombination involving deep electron traps near
Ec-0.5 eV which could be attributed to formation of complexes between the Mn
acceptors and some point defects, such as nitrogen vacancies NV,
whereas in the latter the blue recombination band involves deep acceptors near
Ev+0.4 eV (see e.g. detailed discussion in Ref. 16 ).
It
is interesting that the MCL spectra of GaMnN samples were rather sensitive to
the growth conditions. All the GaMnN samples discussed so far were grown at 700
0C while one of the present set of samples, the sample with 3.5% Mn,
was grown at higher temperature of 750 0C. In Fig. 5 we compare the
MCL spectra taken on this sample at 90K at high and low excitation intensity
with the spectra taken under the same conditions on the 3% Mn sample. Although
the Mn concentrations are quite close, it can be seen that, first, the bandedge
intensity in the 3.5% sample grown at higher temperature is very much stronger.
Secondly, the relative intensity of the blue band is also very much higher in
the 3.5% sample. The higher bandedge luminescence intensity in the 3.5% sample
is at least in part due to much lower resistivity of this sample compared to
the 3% sample (see next section). At the same time there is no question that
higher growth temperature enhances the formation of deep defects responsible
for the blue luminescence, even though the 3.5% sample was shown to be
single-phase. It seems likely that under these growth conditions the density of
structural defects forming complexes with the Mn acceptors is strongly
promoted. Mind also that in both samples, the 3% Mn and the 3.5% Mn, the
intensity of the blue band is not saturated by increasing the excitation
intensity which suggests very high densities of corresponding defects (in the
control sample the blue band was easily saturated).
Electrical properties.
The room temperature resistivities of
the studied samples are summarized in Table I. The control sample showed a
resistivity of 18 Ω.cm with apparent electron mobility of about 30 cm2/V.s,
although this value measured on an undoped thin sample is not necessarily
meaningful because of the impact of mosaicity of the grown layer (see e.g. a
discussion in the review paper (17)). The activation energy of
conductivity Ea at temperatures from slightly below room temperature
to 400K was 60 meV (see Ea values in the fourth column of the
table). At lower temperatures the conductivity was of Mott type (18,19).
C-V measurements performed on Au Schottky diodes prepared on this sample gave
the concentration close to 1016 cm-3. The resistivity of
all GaMnN samples grown at 700 0C was considerably higher than the
resistivity of control sample. The activation energy of conductivity for
temperatures 250K-400K was close to 0.1 eV, at lower temperatures the
conductivity was of Mott type.An example of the temperature dependence of sheet
resistance for the 3 at.% Mn sample is shown in Arrhenius form in Figure 6. The
resistivity showed a minimum for the Mn concentration of 5%. C-V measurements
on Schottky diodes showed that all films were fully depleted up to the
temperature of 400K. For the 3.5% Mn sample grown at 750 0C the
resistivity was lower than for the undoped sample (2Ω.cm), the sample was
nominally n-type with the apparent mobility of 20 cm2/V.s, the
activation energy of resistivity at temperatures higher than 250K was quite low
(20 meV). All GaMnN samples showed about two orders of magnitude lower
photosensitivity than the undoped sample. Thermally stimulated current (TSC)
measurements after illumination of GaMnN samples at 85K for 15 minutes with
deuterium lamp did not produce any high concentrations of traps giving rise to
TSC peaks.
The above results are rather puzzling.
Low activation energy of resistivity and the n-type of resistivity in the 3.5%
Mn sample show that all our MBE grown GaMnN samples are n-type, with rather low
electron concentration not exceeding 1017 cm-3. Indeed,
the most shallow known acceptor impurities in GaN, the Mg and the Ca, show
activation energies of about 0.2 eV (12,13), i.e. much higher than
in our GaMnN specimens. Also, as discussed above, we don’t see optical bands in
absorption or MCL that are attributed to the transition from the partly filled
Mn acceptors near Ev+1.4 eV to the valence band, while such
transitions should be prominent in p-type material (7,8). At the same time our optical absorption
measurements show that the Mn acceptors are there and the measured very high
values of absorption coefficient suggest that, with any reasonable optical
cross section values, the density of the Mn acceptors should be in the 1019-1020
cm-3 range. Because the growth was by MBE, hydrogen passivation of
the Mn acceptors should not be an
important factor because of the very low density of hydrogen in MBE grown GaN
films (20). So one has to assume that during growth equally high
concentrations of compensating donors are introduced. The nature and the levels
position of these donors are not yet clear. The only centers detected in high
concentration in optical measurements are the ones giving rise to the blue
luminescence and attributed above to the Ec-0.5 eV donors. They,
however, cannot be the centers in question because if that were the case the
Fermi level would have been pinned somewhere between Ec-0.5 eV (the
donor level) and Ec-1.9 eV(the Mn acceptors). The density of
uncompensated portion of the 20 meV or the 100 meV donors found to actually pin
the Fermi level in the GaMnN MBE samples does not exceed at most 1017
cm-3 and is actually much lower in the samples grown at 700 0C.
With the density of acceptors being on the order of 1020
cm-3
, such close compensation is next to impossible by chance contamination by some
impurities coming e.g. from the Mn source during the MBE growth. Rather one
should talk about some mechanism of self-compensation in which the formation of
high densities of donors equal to the density of acceptors is provided by
formation of donor point defects. However, this should be a very fundamental
process and it is not obvious why a slight change in growth temperature should
produce the observed change between the type of dominant donors in the samples
grown at 700 0C (100 meV) and 750 C 0(20 meV). It seems
much more likely that these 100 meV or 20 meV donors are simply some chance
defect centers on which the Fermi level is pinned in the material where the Mn
acceptors are compensated by point defect donors. In that case the donors in
question should be even more shallow, e.g. forming resonance states within the
conduction band. Possible candidates in such a scenario would be the nitrogen
vacancies NV proposed to form high-lying resonant states in the
conduction band of GaN(21) although the theorists do diverge widely
on predicted energy levels and formation energies of such defects(22).
Formation of these NV and complexing of them with the Mn acceptors
would also explain the presence of high densities of Ec-0.5 eV
donors responsible for the strong blue band in MCL spectra. One also wonders if
the strong enhancement of absorption within the conduction band of GaMnN
observed in Fig. 1 could be attributed to such resonant states. Clearly much
more work is necessary to prove or disprove this explanation but on the surface
it seems plausible.
The decrease of photosensitivity and the
lack of strong TSC peaks in GaMnN samples seem to be both related to efficient
recombination of holes via the Mn acceptors as evidenced by the strong MCL band
near 1.9 eV (Fig. 2). In the samples with the Fermi level close to the
conduction band, as in our GaMnN films, one can see in TSC only the hole traps
(the electron traps can be recharged only if their level is above the Fermi
level, i.e. in our case if they are very shallow and thus not observable in the
studied temperature range). If all holes are captured by the dominant deep Mn
acceptors and then recombine with electrons one naturally does not see strong
signals from any hole traps.
Finally,we should point out that the low
carrier density in our samples probably precludes carrier-induced mechanisms as
being responsible for the ferromagnetism(1),although it may be
possible to have strongly localized carriers play a role in the observed
magnetic properties(23).There are numerous examples in the
literature of ferromagnetism being reported for Mn-doped materials with very
low carrier concentrations(6,7,24).
CONCLUSIONS
We have shown that in MBE grown GaMnN
films with high Curie temperatures, the Mn ions form deep acceptor centers with
a level near Ec-1.9 eV which is in agreement with previous findings
on MOCVD grown GaMnN films.(7,8) Together with these centers we
observed the formation of high densities of centers giving rise to strong bands
peaked near 2.9 eV in MCL spectra. These MCL bands are very similar to those
observed in Mn implanted GaMnN films where they could be associated with
formation of complexes between point defect donors, such as NV, and
the Mn acceptors which results in forming deep donors with level near Ec-0.5 eV (15).
The GaMnN films grown by MBE are shown to be of electron conductivity with low
concentration of electrons not exceeding at most 1017 cm-3.
This suggests that some mechanism of self-compensation involving forming of
donor type native defects in equal concentration to the density of Mn acceptors
is at work. The donors in question should be fairly shallow, shallower than 20
meV, which is consistent with behavior expected of the nitrogen vacancies. Thus
it is somewhat tentatively proposed that incorporation of high densities of Mn
in GaMnN triggers formation of equally high density of NV donors
thus leaving the material lightly n-type due to residual donor defects present
in the films and somewhat different for different growth conditions. Some of
these NV donors form complexes with the Mn acceptors producing the Ec-0.5
eV responsible for the blue band. The latter process is sensitive to the Mn
concentration (the density of the Ec-0.5 eV seems to increase with
the Mn density) and to the growth temperature (higher growth temperature seems
to produce higher densities of the Ec-0.5 eV donors if judged by the
corresponding MCL bands intensities).
ACKNOWLEDGEMENTS
The
work at IRM was supported in part by the grant from the Russian Foundation for
Basic Research (Grant #01-02-17230). The work at UF is partially supported by
NSF(CTS991173,DMR 0101438) and ARO. The authors would like to thank Mrs. E.F.
Astakhova for preparing the Schottky diodes used in this study.
REFERENCES
1T.Dietl,H.Ohno,F.Matsukura,J.Cibert
and D.Ferrand,Science 287,1019(2000).
2.T.Dietl,J.Appl.Phys.89 ,7437(2001).
3. M.K. Reed, N.A. El-Masry, H.H. Stadelmaier, M.K. Ritums, M.J. Reed, C.A. Parker, J.C. Roberts and S.M.Bedair,Appl.Phys.Lett.79 3473 (2001).
4.G.T. Thaler, M.E. Overberg, B. Gila, R. Frazier, C.R. Abernathy, S.J. Pearton, J.S. Lee, S.Y. Lee, Y.D. Park, Z.G. Khim, J. Kim and F. Ren, Appl. Phys. Lett. 80 May 27 (2002).
5.S. Sonoda, S. Shimizu, T. Sasaki, Y. Yamamoto and H. Hori, J. Cryst. Growth 237-239,1358(2002)
6.T.Sasaki,S.Sonoda,Y.Yamamoto,K.Suga,S.Shimizu,K.Kindo and H.Hori,J.Appl.Phys.91 7911 (2002).
7.R.Y.Korortov,J.M.Gregie and B.W.Wessels,Appl.Phys.Lett.80 1731(2002).
8.R.Y. Korotkov, J.M. Gregie and B.W. Wessels, Mat. Res Soc. Symp. Proc. 639, G3.7.1 (2001)
9.M.E.Overberg, C.R.Abernathy, S.J.Pearton, N.A.Theodoropoulou,K.T.McCarthy and A.F.Hebard,Appl.Phys.Lett.79 1312(2001).
10. A.Y. Polyakov, N.B. Smirnov, A.V. Govorkov, M.G. Mil’Vidskii, J.M. Redwing, M. Shin, M. Skrowronski, D.W. Greve and R.G. Wilson, Solid-State Electron. 42 627 (1998).
11.A.Y. Polyakov, N.B. Smirnov, A.S. Usikov, A.V. Govorkov and B.S. Pushniy, Solid-State Electron. 42 1959 (1998).
12.M.E.Levinshtein,S.L.Rumyantsev and M.S.Shur,Properties of Advanced Semiconductor Materials;GaN,AlN,InN,BN,SiC and SiGe(Wiley and Sons,NY,2001).
13.Properties of Group III Nitrides,ed. J.H.Edgar,EMISDataReviews,No.11( IEE,London,1994).
14.J. Baur, K. Maier, M. Kunzer, U. Kaufmann, J. Schneider, H. Amano, I. Akasaki, T. Detchprom and K. Hiramatsu, Appl. Phys. Lett. 64, 857 (1994).
15.A.Y.Polyakov,N.B.Smirnov,A.V.Govorkov,N.Y.Pahkova,J.Kim,F.Ren,M.E.Overberg,G.T.Thaler,C.R.Abernathy,S.J.Pearton and R.G.Wilson,J.Appl.Phys.(in press,August 15,2002 issue).
16.M.A. Reshchnikov, F. Shahedpour, R.Y. Korotkov, B.W. Wessels and M.P. Ulmer, J. Appl. Phys. 87, 3351 (2000)
17.H.Morkoc,Mat.Sci.Eng.R33 135(2001).
18.D.L.Rode, Low Field Electron Transport, in Semiconductors and Semimetals,Vol 10(Academic Press,NY,1975),pp1-89.
19.D.C.Look, Electrical Characterization of GaAs Materials and Devices (Wiley,NY,1989).
20.M.S.Shur and M.Asif Khan,in GaN and Related Materials II,ed.S.J.Pearton(Gordon and Breach,NY 2000).
21.D.W. Jenkins, J.D. Dow and M.-H. Tsai, J. Appl. Phys. 72, 4130 (1992)
22.J. Neugebauer and C.G. Van de Walle, Phys. Rev. B50, 8067 (1994).
23.A.C.Durst,R.N.Bhatt and P.A.Wolff,Phys.Rev.B65,235205 (2002).
24.S.Cho,S.Choi,G.-B.Cha,S.C.Hong,Y.Kim,Y.-J.Zhao,A.J.Freeman,J.B.,Ketterson,B.J.Kim,Y.C.Kim and B.-C.Choi,Phys.Rev.Lett.88,257203(2002)
Table I. Electrical parameters of the studied samples
|
Mn
concentration (at. %) |
Growth
temperature( 0 C) |
Resistivity
at 300K (Ω.cm) |
Ea
(meV) |
|
0 |
700 |
18 |
60 |
|
3 |
700 |
105 |
90 |
|
5 |
700 |
45 |
100 |
|
10 |
700 |
450 |
120 |
|
3.5 |
750 |
2 |
20 |
FIGURE CAPTIONS
Fig.
1. The optical absorption spectra taken on a control sample and three GaMnN
samples with Mn concentration of 3%, 5% and 10%,calculated using the reflection
coefficient spectra of GaN from Ref. 12,13
Fig.
2. 90K MCL spectra in the vicinity of 2 eV taken for the control sample and the
two GaMnN samples with Mn concentrations of 3% and 5% (all spectra were taken
in similar conditions);
Fig.
3. 90K MCL spectra taken at high excitation intensity for control sample and for
GaMnN samples with Mn concentration of 3%, 5% and 10% of Mn
Fig.
4. 90K MCL spectra taken for the GaMnN samples with 3%, 5% and 10% Mn at low
excitation intensity
Fig.
5. 90K MCL spectra taken for the 3% Mn (growth temperature of 700 0C)
and the 3.5% Mn (growth temperature of 750 0C) samples at low
excitation intensity (the lower set of curves) and high excitation intensity
(the upper set of curves).
Fig.6.Arrhenius
plot of sheet resistance of 3 at.% Mn GaMnN sample.
Fig. 1
Fig. 2.
Fig. 3.
Fig. 4.
Fig. 5.
