Point Defect-Induced UV-C Absorption in Aluminum
Nitride Epitaxial Layers Grown on Sapphire Substrates by
Metal-Organic Chemical Vapor Deposition
Nadine Tillner, Christian Frankerl,* Felix Nippert, Matthew J. Davies, Christian Brandl,
Rainer Lösing, Martin Mandl, Hans-Jürgen Lugauer, Roland Zeisel, Axel Hoffmann,
Andreas Waag, and Marc Patrick Hoffmann
1. Introduction
Efficient light sources emitting in the UV-
C spectral range are required for a broad
range of medical applications, such as dis-
infection and water purification.
[1–3]
Widespread pathogenic microorganisms
are efficiently treated by exposure to UV-C
radiation,
[4–6]
most commonly generated
by high-pressure germicidal lamps which
are based on mercury and emit broad-
band UV-C radiation. Ongoing research
efforts on the ultra-wide-bandgap AlGaN
material system are anticipated to result
in the development of efficient, low-cost,
and environmentally friendly light emit-
ting diodes (LED), which are expected to
replacemercury-vaporlampsasthemain
UV-C emission source in the near future.
[7]
However, several fundamental limita-
tions of current AlGaN-based LED
devices must be overcome first, until their
electrical-to-UV-C power conversion effi-
ciency can compete with that of germicidal
lamps. The external quantum efficiency of
current devices still remains below 20%
[8]
for multiple reasons, such as poor doping
efficiency of p-AlGaN.
[9,10]
For this reason, p-GaN is often used
as a p-type contact layer, despite fully absorbing UV photons
N. Tillner, C. Frankerl, Dr. M. J. Davies, C. Brandl, Dr. R. Lösing,
Dr. M. Mandl, Dr. H.-J. Lugauer, Dr. R. Zeisel, Dr. M. P. Hoffmann
OSRAM Opto Semiconductors GmbH
Leibnizstraße 4, 93055 Regensburg, Germany
E-mail: Christian.Frankerl@osram-os.com
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/10.1002/pssb.202000278.
© 2020 The Authors. Published by Wiley-VCH GmbH. This is an open
access article under the terms of the Creative Commons Attribution
License, which permits use, distribution and reproduction in any
medium, provided the original work is properly cited.
Correction added on 19 August 2020, after first online publication: Projekt
Deal funding statement has been added.
DOI: 10.1002/pssb.202000278
N. Tillner, Prof. A. Waag
Institut für Halbleitertechnik
TU Braunschweig
Hans-Sommer-Straße 66, 38106 Braunschweig, Germany
C. Frankerl, Dr. F. Nippert, Prof. A. Hoffmann
Institut für Festkörperphysik
Technische Universität Berlin
Hardenbergstraße 36, 10623 Berlin, Germany
Prof. A. Waag
Epitaxy Competence Center
TU Braunschweig
Hans-Sommer-Straße 66, 38106 Braunschweig, Germany
Herein, the optical properties of aluminum nitride (AlN) epitaxial layers grown
on sapphire substrates by metal-organic chemical vapor deposition (MOCVD)
are reported. The structures investigated in this study are grown at highly
different degrees of supersaturation in the MOCVD process. In addition, both
pulsed and continuous growth conditions are employed and AlN is deposited
on nucleation layers favoring different polarities. The samples are investigated
by photoluminescence (PL), photoluminescence excitation (PLE), and
absorption spectroscopy and are found to vary significantly in absorption and
emission characteristics. Two distinct absorption bands in the UV-C spectral
range are observed and examined in greater detail, with either giving rise to
asignificant absorption coefficient of around 1000 cm
1
. The corresponding
defect transitions are identified by PL spectroscopy. Combined with secondary-
ion mass spectrometry (SIMS) measurements, these absorption bands are
allocated to the incorporation of carbon and oxygen impurities, depending
on the applied growth conditions. Furthermore, similarities with other epitaxial
growth techniques serving as basis for UV-C applications are highlighted.
These results are highly relevant for a better understanding of absorption issues
in AlN templates grown by various deposition techniques. In addition,
consequences for the growth of efficient UV-C devices by MOCVD on sapphire
substrates are outlined.
ORIGINAL PAPER
www.pss-b.com
Phys. Status Solidi B 2020,257, 2000278 2000278 (1 of 6) © 2020 The Authors. Published by Wiley-VCH GmbH
generated in the active area. UV LEDs with top emission there-
fore suffer from GaN absorption. As a consequence, flip-chip
concepts, meaning that the light is extracted through the back-
side of the structure, need to be utilized.
[9,11]
In this architec-
ture, however, the photons need to pass the whole AlN layer
between the sapphire substrate and the active region of the
UV LED. The optical quality of the template therefore is of
utmost importance.
Typically, AlN epitaxial layers grown on sapphire are used to
form a suitable template for UV-C LEDs due to their reasonable
lattice matching to Al-rich AlGaN layers and its large bandgap,
avoiding band–band- absorption.
[12]
On the downside, AlN epi-
taxial layers systematically suffer from a considerable concentra-
tion of unintentionally incorporated point defects.
[13]
Point
defects can introduce undesired absorption bands in the UV-
C spectral region which significantly deteriorate the light output,
as reported for AlN templates on different substrates, for native
AlN substrates grown by physical vapor transport (PVT) and AlN
grown by hydride vapor phase epitaxy (HVPE).
[14–16]
However,
the UV-C spectral region is of utmost relevance for medical
applications, as the maximum germicidal efficacy for several
pathogenic microorganisms
[4–6]
is reported at wavelengths close
to typical UV-C emission (250–280 nm). This emphasizes the
necessity of defect reduction and a thorough study of the absorp-
tion related to point defect incorporation in AlN epitaxial layers.
In this study, we examine the relation between absorption and
the introduction of point defects into AlN epitaxial layers grown by
metal-organic chemical vapor deposition (MOCVD) used for AlN
templates on sapphire substrates with varying growth conditions,
employing highly different degrees of supersaturation. The vapor
supersaturation describes the deviation from the thermodynamic
equilibrium during the AlN growth process and serves as a unified
parameter to describe the overall growth mode.
[17–19]
The individ-
ual process parameters, such as the growth temperature and the
V/III ratio, define thereby the supersaturation. For example, by
increasing the process temperature or lowering the V/III ratio,
the vapor supersaturation is decreased, whereas one of the most
effective ways to adapt the vapor supersaturation is the growth
temperature.
[18,19]
In addition, pulsed and continuous growth
modes and AlN layers with different polarities are investigated.
To analyze the samples, both photoluminescence (PL) and photo-
luminescence excitation (PLE), and absorption spectroscopy are
used as means of investigating the absorption and emission prop-
erties of AlN sample structures. In particular, two distinct absorp-
tion bands in the UV-C spectral range, in the vicinity of the desired
emission wavelength region, could be observed and their depen-
dencies on the used growth conditions and modes are studied.
The concentrations of the impurities incorporated into the sam-
ples are measured by secondary-ion mass spectrometry (SIMS)
and, in combination with the optical response of the structures,
an allocation of the absorption bands to the determined types
of point defects is proposed. These findings are compared to alter-
native deposition techniques for heteroepitaxial and homoepitaxial
AlN layers (hydride vapor phase epitaxy (HVPE) and physical
vapor transport (PVT)), which are all used in high-power UV-C
LEDs, despite their widely varying growth procedures. In addition,
this article might serve as a guideline to grow transparent AlN
epitaxial layers by MOCVD in the UV-C spectral range by carefully
adapting growth modes and kinetics.
2. Sample Structure and Measurement Methods
AlN epitaxial layers are deposited in a N
2
/H
2
atmosphere with
different temperatures and growth parameters on sapphire sub-
strates by MOCVD technique. Trimethylaluminum (TMAl) and
ammonia (NH
3
) were used as precursors. For all structures,
growth was initiated via a few nanometer thick AlN nucleation
layer and followed by an 1.4 μm thick high-growth-temperature
AlN layer. Four individual samples are grown and studied in this
study, with their properties summarized in Table 1. The first
sample (P) is grown under pulsed growth conditions, while
the other samples (C1, C2, C2*) use continuous growth condi-
tions. Sample P is grown at a temperature of 1330 C and a con-
tinuous metalorganic flow rate of 60 μmol min
1
in an NH
3
pulsed mode. Pulsing of NH
3
induces an Al-rich environment
supporting III-polar AlN deposition.
[9]
The pulsed sequence
consists of a 5 s simultaneous growth phase interrupted by a
3s NH
3
flow pause, generating a continuous equivalent V/III
ratio of 1216. The samples C1, C2, and C2* are grown under
continuous growth conditions with metalorganic flow rates of
140 μmol min
1
and V/III ratios of 38. The growth tempera-
tures vary from 1150 C for sample C1 to 1420 C for samples
C2 and C2*, with the only difference between the last two
samples being an alternating nucleation layer, which will be
described in more detail in the following main paragraph.
The detailed process conditions of each sample are additionally
unified under the supersaturation as the main growth control
parameter, which is estimated by the input parameters, to
describe the applied growth mode.
[19]
The supersaturation values
for the samples P, C1, and C2/C2* are 110
4
,110
7
, and
510
2
, respectively. As the order of magnitudes of the esti-
mated values are highly different, the supersaturations of the
samples P, C1, and C2/C2* are henceforth considered as
medium, high, and low, respectively.
Room temperature absorption measurements are performed
using a commercial Hamamatsu Quantaurus C9920-02 G sys-
tem. In this setup, the sample is placed inside a UV-capable inte-
grating sphere and exposed to light of a tunable wavelength
between 245 and 400 nm, originating from a 150 W xenon arc
lamp connected to a monochromator. After passing through
the sample, the residual light is collected and analyzed by a
Hamamatsu PMA-12, consisting of a second monochromator
coupled to a CCD detector unit. By adjusting the second
Table 1. Overview of the samples grown for this study. “þ”indicates
the continuous growth equivalent, while C2* uses a different nucleation
layer compared to sample C2 (see text). The supersaturation values
depending of the individual growth conditions are estimated and
compared relatively.
[19]
P C1 C2 C2*
Temperature [C] 1330 1150 1420 1420
Metalorganic flow rates
[μmol min
1
]
60 140 140 140
V/III ratio 1216
þ
38 38 38
Supersaturation Medium,
110
4
High,
110
7
Low,
510
2
Low,
510
2
www.advancedsciencenews.com www.pss-b.com
Phys. Status Solidi B 2020,257, 2000278 2000278 (2 of 6) © 2020 The Authors. Published by Wiley-VCH GmbH
monochromator to the original excitation wavelength, errors that
may arise through potential defect absorption and re-emission
(at lower energies due to Stokes shift) are excluded. As the size,
thickness, and surface roughness of all samples do not deviate
significantly, reflection coefficients are considered as equal.
Therefore, the acquired absorption spectra, generated by calcu-
lating the difference of the input and output light intensity at any
given wavelength, are believed to be very reliable.
For room temperature PL measurements, a quadrupled
Q-switched Nd:YAG laser (λ¼266 nm) is used as the excita-
tion source. Measurements are performed in a commercial
Nanometrics VERTEX-SM-230 PL mapping device. The excitation
source for the PLE measurements is a 450 W xenon arc lamp cou-
pled to a two-stage Acton SpectraPro 300i monochromator for
wavelength selection. Output light detection is realized by a
0.8 m focal length SPEX 1704 monochromator combined with
a Princeton Instruments liquid nitrogen–cooled UV-optimized
CCD camera. PLE measurements were performed at T¼5K
in a Janis ST-500 cryostat. Low-temperature PLE measurements
are preferred for reasons of a sufficient signal-to-noise ratio. SIMS
depth profiles were acquired utilizing a Cameca IMS WF mag-
netic sector instrument using a cesium primary ion beam.
3. Experimental Results and Discussion
In Figure 1, the room temperature absorption spectra of samples
P, C1, and C2 are shown. Sample P indicates a broad absorption
band ranging from below 245–315 nm wavelength with a distinct
peak around 265 nm, corresponding to a maximum absorption
coefficient of 1053 cm
1
. In contrast, the absorption spectra of
samples C1 and C2 are almost identical, exhibiting drastically
reduced absorption coefficients of 151 cm
1
(C1) and 168 cm
1
(C2) at an excitation wavelength of 265 nm. The absorption
spectrum of these samples is approximately constant across
the whole wavelength range studied, without revealing a distinct
peak. Measured net absorption at λ¼265 nm for the samples is
14% (P), 2% (C1), and 3% (C2), respectively. We would like
to emphasize here the similarity of the absorption spectra of sam-
ples C1 and C2 despite their significant difference in growth tem-
perature (1150 C vs 1420 C) and thus the vapor supersaturation
used during the MOCVD process. The absorption of the pulsed-
grown sample P, deposited at 1330 C and a medium supersatu-
ration, reveals in contrast a fundamentally different behavior.
As the process temperature and therefore the vapor supersatura-
tion of sample P were chosen to be in between those of samples
C1 and C2, the apparently deviating defect incorporation into the
epitaxially grown film must be a consequence of the NH
3
pulsed
growth mode rather than being solely growth-kinetic related. NH
3
pulsing, as used in sample P, is a popular growth technique and is
often used to achieve a very low vapor supersaturation or as a tech-
nique for multimode AlN growth to lower the threading disloca-
tion density (TDD).
[9]
Similar absorption coefficients in the range
of 1000 cm
1
have been reported for AlN epitaxial layers and
native AlN grown by MOCVD, HVPE, or PVT, which are prone
to high impurity incorporation originating primarily from oxygen
or carbon incorporation.
[14–16]
We emphasize here that a UV-C
absorption close to 265 nm overlaps with the desired emission
wavelength of UV-C LEDs used for disinfection and purification
purposes. This finding highlights the need for a more detailed
study of the defects accounting for the pronounced absorption
band, especially in sample P, which was grown in pulsed growth
mode, a method which recently gained popularity in the fabrica-
tion of AlN epitaxial layers.
For this reason, resonant PL measurements were performed
at room temperature, specifically pumping the absorption band
centered around 265 nm due to the choice of excitation source.
The obtained spectra are shown in Figure 2. Two peaks can be
identified in the PL spectrum of sample P, centered at 310 and
450 nm, as well as the second order peak of 310 nm at 620 nm.
Samples C1 and C2 reveal two distinct peaks that are slightly
shifted with respect to their positions in sample P, peaking at
325 and 470 nm, respectively. The PL spectra of samples C1
250 300 350 400
0
500
1000
P
C1
C2
Absorption coefficient [cm-1]
Wavelen
g
th [nm]
54.5 4 3.5
Energy [eV]
Figure 1. Room temperature absorption spectra of samples P, C1, and C2.
300 400 500 600 700
0.0
0.5
1.0
Normalized PL intensity [a.u.]
Wavelen
g
th [nm]
P
C1
C2
4.5 4 3.5 3 2.5 2
Energy [eV]
Figure 2. Room temperature PL emission spectra of samples P, C1, and
C2. The inset exemplarily shows the PLE spectra recorded at T¼5K of
sample P, integrated over the emission bands from 280–350 nm and from
400–500 nm, respectively, as indicated by the black arrows.
www.advancedsciencenews.com www.pss-b.com
Phys. Status Solidi B 2020,257, 2000278 2000278 (3 of 6) © 2020 The Authors. Published by Wiley-VCH GmbH
and C2 reveal emission peaks at identical energetic positions,
coinciding also in shape. This finding leaves no doubt about
the similar nature of the point defects incorporated. Here, we
cannot exclude that the specific concentration of point defects
introduced in samples C1 and C2 varies to some extent, although
it appears that the concentration in both samples is too low to
induce significant absorption. In addition, the clear energetic
shift of the emission peaks of samples C1 and C2 compared
to sample P indicates that if the same type of point defects is
responsible for the PL emissions in all samples, the defect con-
centration or their ratio in the latter sample is entirely different.
A similar effect, i.e., an energetic PL peak shift depending on the
defect concentration, was reported by Collazo et al. for a carbon-
related defect in AlN epitaxial layers grown by PVT and HVPE.
[14]
We have also performed low-temperature PLE measurements, as
exemplarily shown for sample P in the inset of Figure 2. The PLE
measurement confirms that the defect emissions centered
around 310 and 450 nm originate from the same defect(s) caus-
ing the absorption band at 265 nm (cf. Figure 1). The absorption
band at 265 nm in sample P may therefore be introduced either
by one type of impurity which is incorporated in two different
defect states, two different types of impurities, or even three
defects in donor–acceptor pair configuration. We note here that
the corresponding PLE response is similar for samples C1 and
C2, albeit of reduced absolute intensity, altogether indicating a
similar defect nature in all three samples, but heavily varying
in concentration and therefore absorption behavior depending
on the specific growth condition.
Until now, all three samples discussed in this manuscript
reveal comparable, smooth surfaces, as clearly observed using
differential interference contrast microscopy, indicating a pure
III-polar AlN growth. In contrast, sample C2* is grown with
the identical high-temperature growth and supersaturation con-
ditions used for sample C2, but on a different underlying nucle-
ation layer, specifically developed to favor a primarily mixed-polar
growth (both N- and Al-polar). As a result, the surface morphol-
ogy of sample C2* is altered and indicates mixed-polar growth
with 15% of the sample being defined by inversion domains.
Naturally, the nucleation layer also impacts the subsequent high-
temperature AlN deposition. Sample C2* can thus be regarded
as a test structure for AlN epitaxial layers utilizing lateral over-
growth of N-polar domains as means of TDD reduction.
[20]
As
the nucleation layer is found to modify the growth mode of
the subsequently grown AlN layer, it may also affect point defect
incorporation, e.g., native oxygen incorporation,
[21]
and therefore
the net absorption in the UV-C energetic range. To study this
effect, Figure 3 directly compares room temperature absorption
spectra of samples C2 and C2*, which differ only in the overall
polarity determined by the nucleation layer. Sample C2, as
already stated in Figure 1, shows an almost constant absorption
coefficient of around 150–200 cm
1
over the entire measurement
range without a distinct absorption band and a net absorption
of 2.5% at λ¼265 nm. In contrast, we find a severe increase in
UV-C absorption below 320 nm in sample C2*, implying a very
broad absorption band with a maximum at a wavelength signifi-
cantly below 245 nm, exceeding the measurement range of our
setup. Here, the absorption coefficient at λ¼265 nm is 927 cm
1
with a determined net absorption of 12.5%. The extracted absorp-
tion spectrum implies that one or several different types of point
defects are predominantly incorporated in sample C2* compared
to samples P, C1, and C2, favored by the difference in polarity of
the nucleation layer. In general, the emergence of inversion
domains during epitaxial growth drastically promotes the incor-
porating of oxygen;
[21]
thus, the interpretation of an oxygen-
related absorption band is justified.
To further investigate this change in absorption behavior, res-
onant PL measurements were conducted on sample C2*, pump-
ing the high-energy flank of the absorption band observed in
Figure 3, and compared to the corresponding PL spectra of sam-
ple C2. The result is shown in Figure 4. As already discussed in
Figure 2, the obtained PL spectrum of sample C2 reveals two
distinct peaks around 325 and 470 nm. In contrast, only one
distinct asymmetric peak centered at 380 nm is observed in
the emission spectrum of sample C2*. Therefore, the PL meas-
urements further verify the inherently different nature of the
incorporated point defects when altering the nucleation layer.
250 300 350 400
0
500
1000
1500
C2
C2*
Absorption coefficient [cm ]
Wavelen
g
th [nm]
54.5 4 3.5
Energy [eV]
Figure 3. Room temperature absorption spectra of samples C2 and C2*.
300400500600700
0.0
0.5
1.0
Normalized PL intensity [a.u.]
Wavelen
g
th [nm]
C2
C2*
4.5 4 3.5 3 2.5 2
Energy [eV]
Figure 4. Room temperature PL emission spectra of samples C2 and C2*.
www.advancedsciencenews.com www.pss-b.com
Phys. Status Solidi B 2020,257, 2000278 2000278 (4 of 6) © 2020 The Authors. Published by Wiley-VCH GmbH
To investigate the possible influence of impurity atoms jointly
responsible for the observed absorption and PL behavior, we per-
formed SIMS measurements on all samples discussed in this
manuscript. Using this method, we were able to reliably deter-
mine the respective concentrations of silicon, carbon, and oxy-
gen, which show the foreign atoms most commonly
incorporated during the growth process of AlN epitaxial layers
by MOCVD. The results are shown in Table 2, along with the
already discussed emission and absorption properties of the sam-
ples, as well as the degrees of supersaturation. We first note that
the concentration of Si remains substantially below 10
18
cm
3
for all samples and no clear correlation with the applied growth
conditions, the determined absorption coefficients, and the PL
emission spectra is found. Therefore, Si can be excluded as
the primary source of impurity absorption within the UV-C
region in all of the studied samples. The concentration of C is
found to be 2.0 10
19
cm
3
in sample P and only of the order
of low 10
17
cm
3
in samples C1, C2, and C2*. As the absorption
band centered around 265 nm was only found in sample P and
not observed in all other samples, it is unambiguously identified
as carbon-related. Our findings are supported by Alden et al. who
related the carbon-related defect luminescence at 320 and
460 nm to the absorption band at 265 nm in single crystal
AlN substrates grown by PVT.
[15]
In addition, Collazo et. al. also
showed a relation between the defect luminescence at 320 nm
and the absorption band at 265 nm for AlN grown by PVT and
HVPE.
[14]
We highlight the finding that the high incorporation of
carbon in our MOVCD-grown AlN layer in sample P is strikingly
similar to the incorporation of carbon in PVT-grown AlN native
substrates, despite the inherent differences in the growth
methods and kinetics of both techniques. Therefore, both
approaches serving as a basis of UV-C LEDs—the heteroepitaxial
AlN layer on foreign substrates and AlN native substrates for
homoepitaxy—suffer from a common defect behavior, affecting
the LED performance. Clearly, the near-identical carbon concen-
tration in samples C1 and C2 accounts for the similar absorption
and PL emission characteristics of the structures.
In general, the point defect concentration is found to increase
with decreasing vapor supersaturation in the growth process.
[22]
The vapor supersaturation is defined by the growth parameters,
such as the process temperature, the growth rate, and the V/III
ratio.
[17,18]
Sample P was grown in a medium supersaturation
condition with respect to sample C1 (high supersaturation)
and C2 (low supersaturation), as relatively estimated by the
growth temperature and V/III ratio.
[22]
The carbon impurity
concentration is highest in sample P, whereas the highest oxygen
impurity is measured for sample C2 (and C2*). Thus, NH
3
puls-
ing during growth promotes a significant increase in carbon-
related point defect concentration. The reason for the substantial
incorporation of carbon during pulsed growth is assumed to
be an effect of a decomposing SiC carrier at high process temper-
atures during AlN epitaxy, whereas pulsing gives the opportunity
for the impurities to be incorporated. The incorporation of car-
bon-based defects is more likely than silicon-related defects due
to the former’s lower point defect formation energy in AlN
layers.
[14,15]
A similar result in unintentional doping under con-
tinuously grown AlGaN-based layers due to SiC carrier decom-
position is reported by Jeschke et al.
[23]
A solution to this
unintentional silicon incorporation might be to switch to TaC
carriers, which have a higher temperature stability compared
to SiC, as also suggested by Jeschke et al.
[23]
Finally, the concentrations of oxygen are found to vary sub-
stantially across the samples but are the highest in samples
C2 and C2*. The low vapor supersaturation during the growth
of sample C2 leads to an increase in oxygen concentration,
but a clear maximum is observed in sample C2*. The increase
in O concentration is expected in sample C2*, as N-polar
domains in the introduced mixed-polar layer are known to favor
significant incorporation of oxygen.
[21]
We thus propose the
observed high-energetic absorption band peaking below 245 nm
in Figure 3 to be induced by an O-related defect state, induc-
ing significant absorption only for O concentrations above
10
18
cm
3
. The absorption coefficients determined for the high
absorbing AlN samples P and C2* deposited by MOCVD lie
around 1000 cm
1
and are therefore in good agreement with
reported values of AlN films grown by MOCVD with near-
identical impurity concentrations of oxygen and carbon.
[16]
The results are also consistent with alternative AlN deposition
techniques such as PVT and HVPE.
[14,15]
Thus, we propose that
the most frequently used deposition techniques for AlN all suffer
from similar absorption issues in the UV-C spectral region,
caused by carbon and oxygen incorporation.
We conclude our study by briefly discussing potential defect
configurations and transitions responsible for the absorption and
emission bands observed in Figure 1–4. We already determined
that the absorption band centered around 265 nm observed
in sample P is caused by carbon, typically introduced as a
substitute atom on a nitrogen site, i.e., C
N
.
[14]
The defect
luminescence observed in sample P and (slightly shifted in
energy position due to the lower carbon concentration) samples
Table 2. Impurity incorporation of silicon, carbon, and oxygen into the samples as determined by SIMS, along with a summary of the observed emission
and absorption properties. The supersaturations of the samples are also given to represent the applied growth mode.
P C1 C2 C2*
Si [cm
3
] 2.2 10
17
1.0 10
16
3.5 10
17
2.5 10
17
C [cm
3
] 2.0 10
19
2.0 10
17
1.8 10
17
3.0 10
17
O [cm
3
] 1.3 10
17
1.0 10
17
9.0 10
17
3.0 10
18
Supersaturation Medium, 110
4
High, 110
7
Low, 510
2
Low, 510
2
PL peak emission [nm] 310/450 325/470 325/470 380
Peak absorption [nm] 265 ––<245
www.advancedsciencenews.com www.pss-b.com
Phys. Status Solidi B 2020,257, 2000278 2000278 (5 of 6) © 2020 The Authors. Published by Wiley-VCH GmbH
C1 and C2 may be attributed to radiative transitions from the
conduction band (3.9 eV, CB !CN) and nitrogen vacancies
(2.8 eV, Vþ
N!CN) to the carbon substitute defects.
[24]
Concerning
sample C2*, the high-energy absorption band exceeding our
measurement range and peaking below 245 nm was identified
as oxygen-related, which may be incorporated in a multitude of
different defect states. Potential candidates for the PL emission
at 3.3 eV are transitions from an O-DX center to the valence band
(O DX !VB) and from O substituted on a nitrogen site to
an uncharged complex consisting of the same defect state
complemented by an aluminum vacancy (ON!VAl ONÞ.
[24]
4. Conclusions
In conclusion, we performed optical absorption and PL measure-
ments on AlN epitaxial layers grown by MOCVD on sapphire.
We found that, in general, low vapor supersaturation growth con-
ditions by increased process temperature or pulsed growth mode
significantly promotes the incorporation of point defects. In par-
ticular, NH
3
pulsing during epitaxial growth dramatically enhan-
ces the absorption coefficient in the UV-C spectral region. Two
different absorption bands were observed and the point defect
types responsible could unambiguously be identified. The sam-
ple grown using NH
3
pulsing exhibited a strong absorption band
centered at 265 nm, caused by enhanced carbon incorporation,
while samples grown with higher and even lower vapor supersat-
uration using a continuous growth mode contained a signifi-
cantly lower concentration of defects and showed only weak
absorption in the energy range studied. When switching the
AlN nucleation layer from III-polar to a more N-polar biased
growth (mixed-polar film), a different, high-energetic absorption
band peaking below 245 nm emerges. As inversion domains
themselves favor the incorporation of oxygen, the absorption
band could be allocated to oxygen, as further evidenced by
SIMS measurements. Therefore, heteroepitaxial MOCVD-grown
AlN epitaxial layers suffer from similar issues in terms of impu-
rity incorporation compared to their HVPE- and PVT-grown
counterparts, highlighting the possible compatibility of the
results obtained by experiments on either growth technique. Our
results in terms of point defect incorporation in heteroepitaxial-
grown AlN by MOCVD, depending on growth mode and
kinetics, allow for a crucial choice of AlN growth parameters
to provide highly transparent templates for UV-C LEDs proc-
essed in a flip-chip configuration by inhibiting occurring absorp-
tion bands at the desired emission wavelength.
Acknowledgements
N.T. and C.F. contributed equally to this work. A part of this work was funded
by the German Federal Ministry of Economic Affairs (Bundesministerium für
Wirtschaft und Energie) in the frame of the “Important Project of Common
European Interest (IPCEI) on Microelectronics”(16IPCEI623). Furthermore,
this work was supported by the German Federal Ministry of Education and
Research (BMBF) within the “Advanced UV for Life”project (03ZZ0134A), by
the German Science Foundation (DFG) within the Collaborative Research
Center 787 (CRC787) as well as under Germany’s Excellence Strategy labeled
EXC-2123 QuantumFrontiers 390837967. Open-access funding enabled and
organized by Projekt DEAL.
Conflict of Interest
The authors declare no conflict of interest.
Keywords
absorption, AlN, epitaxy, metal-organic chemical vapor deposition, point
defects
Received: May 22, 2020
Revised: July 16, 2020
Published online: August 18, 2020
[1] M. Stibich, J. Stachowiak, B. Tanner, M. Berkheiser, L. Moore, I. Raad,
R. F. Chemaly, Infect. Control Hosp. Epidemiol. 2011,32,3.
[2] G. Y. Lui, D. Roser, R. Corkish, N. J. Ashbolt, R. Stuetz, Sci. Total
Environ. 2016,553, 626.
[3] K. Song, M. Mohseni, F. Taghipour, Water Res. 2016,94, 341.
[4] T. Wang, S. J. MacGregor, J. G. Anderson, G. A. Woolsey, Water Res.
2005,39, 2921.
[5] G. Messina, S. Burgassi, D. Messina, V. Montagnani, G. Cevenini,
Am. J. Infect. Control 2015,43, 61.
[6] S. E. Beck, R. A. Rodriguez, M. A. Hawkins, T. M. Hargy,
T. C. Larason, K. G. Linden, Appl. Environ. Microbiol. 2016,82,5.
[7] A. Pandey, W. J. Shin, J. Gim, R. Hovden, Z. Mi, Photonics Res. 2020,8,3.
[8] T. Takano, T. Mino, J. Sakai, N. Noguchi, K. Tsubaki, H. Hirayama,
Appl. Phys. Express 2017,10, 031002.
[9] H. Hirayama, N. Maeda, S. Fujikawa, S. Toyoda, N. Kamata, Jpn. J.
Appl. Phys. 2014,53, 100209.
[10] P. Pampili, P. J. Parbrook, Mater. Sci. Semicond. Process. 2017,62, 180.
[11] Y. Nagasawa, A. Hirano, Appl. Sci. 2018,8, 1264.
[12] W. M. Yim, E. J. Stofko, P. J. Zanzucchi, J. I. Pankove, M. Ettenberg,
S. L. Gilbert, J. Appl. Phys. 1973,44, 292.
[13] T. Mattila, R. M. Niemen, Phys. Rev. B 1997,55, 15.
[14] R. Collazo, J. Xie, B. E. Gaddy, Z. Bryan, R. Kirste, M. Hoffmann,
R. Dalmau, B. Moody, Y. Kumagai, T. Nagashima, Y. Kubota,
T. Kinoshita, A. Koukitu, D. L. Irving, Z. Sitar, Appl. Phys. Lett.
2012,100, 191914.
[15] D. Alden, J. S. Harris, Z. Bryan, J. N. Baker, P. Reddy, S. Mita,
G. Callsen, A. Hoffmann, D. L. Irving, R. Collazo, Z. Sitar, Phys.
Rev. Appl. 2018,9, 054036.
[16] K. Nagata, H. Makino, T. Yamamoto, Y. Saito, H. Miki, Jpn. J. Appl.
Phys. 2019,58, SCCC29.
[17] W. K. Burton, N. Cabrera, F. C. Frank, Philos. Trans. R. Soc. A 1951,
243, 299.
[18] S. Mita, R. Collazo, A. Rice, R. F. Dalmau, Z. Sitar, J. Appl. Phys. 2008,
104, 013521.
[19] I. Bryan, Z. Bryan, S. Mita, A. Rice, J. Tweedie, R. Collazo, Z. Sitar,
J. Cryst. Growth 2016,438, 81.
[20] N. Susilo, S. Hagedorn, D. Jaeger, H. Miyake, U. Zeimer, C. Reich,
B. Neuschulz, L. Sulmoni, M. Guttmann, F. Mehnke, C. Kuhn,
T. Wernicke, M. Weyers, M. Kneissl, Appl. Phys. Lett. 2018,112,041110.
[21] R. Collazo, S. Mita, A. Rice, R. F. Dalmau, Z. Sitar, Appl. Phys. Lett.
2007,91, 212103.
[22] Z. Bryan, I. Bryan, J. Xie, S. Mita, Z. Sitar, R. Collazo, Appl. Phys. Lett.
2015,106, 142107.
[23] J. Jeschke, A. Knauer, M. Weyers, J. Cryst. Growth 2018,483, 297.
[24] T. Koppe, H. Hofsäss, U. Vetter, J. Lumin. 2016,178, 267.
www.advancedsciencenews.com www.pss-b.com
Phys. Status Solidi B 2020,257, 2000278 2000278 (6 of 6) © 2020 The Authors. Published by Wiley-VCH GmbH
