ETH Library
1950°C post implantation
annealing of Al+ implanted 4H-
SiC: Relevance of the annealing
time
Journal Article
Author(s):
Fedeli, Paolo; Gorni, Marco; Carnera, Alberto; Parisini, Antonella; Alfieri, Giovanni; Grossner, Ulrike ; Nipoti, Roberta
Publication date:
2016-01
Permanent link:
https://doi.org/10.3929/ethz-b-000119807
Rights / license:
Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International
Originally published in:
ECS journal of solid state science and technology 5(9), https://doi.org/10.1149/2.0361609jss
This page was generated automatically upon download from the ETH Zurich Research Collection.
For more information, please consult the Terms of use.
ECS Journal of Solid State
Science and Technology
OPEN ACCESS
1950°C Post Implantation Annealing of Al
+
Implanted 4H-SiC:
Relevance of the Annealing Time
To cite this article: P. Fedeli et al 2016 ECS J. Solid State Sci. Technol. 5 P534
View the article online for updates and enhancements.
This content was downloaded from IP address 195.176.113.190 on 20/03/2021 at 18:47
P534 ECS Journal of Solid State Science and Technology, 5 (9) P534-P539 (2016)
1950
◦
C Post Implantation Annealing of Al
+
Implanted 4H-SiC:
Relevance of the Annealing Time
P. Fedeli,
a
M. Gorni,
b
A. Carnera,
c
A. Parisini,
b
G. Alfieri,
d
U. Grossner,
e
and R. Nipoti
a,∗,z
a
CNR-IMM of Bologna, Bologna 40129, Italy
b
CNISM - Dipartimento di Fisica, Universit
`
a di Parma, Parma 43124, Italy
c
Dipartimento di Fisica e Astronomia “G. Galilei”, Universit
`
a degli Studi di Padova, Padova 35131, Italy
d
ABB Corporate Research, Baden, Daettwil 5405, Switzerland
e
ETH, Advanced Power Semiconductor Laboratory, Zurich 8092, Switzerland
Previous studies have shown that the electrical activation of a given implanted Al concentration in 4H-SiC increases with the
increasing of the post implantation annealing temperature up to 1950–2100
◦
C and different annealing times in the range 0.5–5 min.
This study shows that, at 1950
◦
C, the electrical activation of Al implanted in 4H-SiC increases with the increase of the annealing
time up to attain saturation for annealing times longer than 22 min. Samples were obtained from the same Al implanted HPSI 4H-SiC
wafer with an implanted Al concentration lower than the Al solubility limit in 4H-SiC at the annealing temperature of 1950
◦
C. The
annealing time was varied in the range 5–40 min.
© The Author(s) 2016. Published by ECS. This is an open access article distributed under the terms of the Creative Commons
Attribution Non-Commercial No Derivatives 4.0 License (CC BY-NC-ND, http://creativecommons.org/licenses/by-nc-nd/4.0/),
which permits non-commercial reuse, distribution, and reproduction in any medium, provided the original work is not changed in any
way and is properly cited. For permission for commercial reuse, please email: oa@electrochem.org. [DOI: 10.1149/2.0361609jss]
All rights reserved.
Manuscript submitted July 15, 2016; revised manuscript received August 1, 2016. Published August 12, 2016.
Nowadays, silicon carbide (SiC) components have gained a sig-
nificant position in the power electronic market among the low-loss,
high-power and high-frequency devices capable of operating in harsh
conditions.
1–4
Large area 4H-SiC epitaxial wafers, from 4 to 6 inches
diameter, are commercial available
3
and foundries offering SiC wafer
processing can be found.
5
Nevertheless, industrial and academic re-
search are still very active both on the side of the SiC material growth
and on that of the SiC wafer processing. This study deals with the SiC
processing and in particular with the electrical doping of SiC by ion
implantation.
Ion implantation is a consolidated technology for obtaining a de-
sired doping depth profile in selected regions of a semiconductor wafer
during the fabrication of planar electronic devices. In the case of SiC
devices, an alternative technology has been developed for obtaining
the doping of selected regions while preserving a planar wafer sur-
face. Such a technology is the embedded epitaxial growth.
6
Anyway,
the cost of this technology and the difficulty of its control are so high
that embedded epitaxial growth is limited to the fabrication of device
channel regions while for the device termination regions or for the
test of new doping profiles, ion implantation remains the preferred
technology.
The electrical doping of a crystalline semiconductor by ion implan-
tation requires a mandatory post implantation thermal treatment. This
treatment aims to recover the disorder in the semiconductor lattice
produced by the ion bombardment and to favor the contemporaneous
allocation of the implanted atoms in substitutional lattice sites where
they act as electrically active dopant. In the case of SiC, in particu-
lar for implanted Al in 4H-SiC, there are still open issues related to
the understanding of the post implantation annealing effects on both
implanted and unimplanted SiC materials. As an example, it can be
mentioned the fact that by increasing the post implantation annealing
temperature up to 1950–2100
◦
C, with different annealing times in the
range 0.5 −5 min, the electrical activation of Al implanted in 4H-SiC
is improved,
7–9
but, the role of the annealing duration at such high
temperatures has never been studied. While for annealing tempera-
tures below 1800
◦
C, such a role has been studied and contradictory
results have been obtained.
10–13
Therefore, further investigations ap-
pear as necessary for understanding the effect of the duration of the
post implantation annealing on SiC.
It is worthwhile to recall that, in principle, the maximum expected
electrical activation for a given implanted species in a semiconductor
∗
Electrochemical Society Member.
z
E-mail: nipoti@bo.imm.cnr.it
material cannot exceed the value corresponding to the solid solubility
of the implanted species in the semiconductor lattice at the temperature
of post implantation annealing. Generally, solid solubility increases
with the increase of the processing temperature. This explains the
improving of the electrical activation of a given implanted dopant
species in SiC with the increase of the post implantation annealing
temperature. Nevertheless, to the best of these authors knowledge, the
concentration of the implanted dopant species in 4H-SiC has often
overcome their solid solubility values at the used annealing tempera-
ture; examples can be found in Refs. 9 and 13. This is an experimental
configuration that should be avoided because it favors the formations
of both dopant precipitates and structural extended defects
14–16
that
may affect the carrier transport in the implanted layer.
16
The solid
solubility limits of the p-type doping species B and Al in 4H-SiC have
been measured in the temperature range 1700–2000
◦
C and they are
published in Ref. 17 and in Ref. 18, respectively.
Al is generally preferred to B when elevated p-type doping levels
are desired. In fact, Al has an higher solid solubility in 4H-SiC at
the temperatures necessary to obtain a significant electrical activa-
tion. For example, the Al solubility in 4H-SiC at 1900
◦
Cis2× 10
20
cm
−318
while that of B is 5 × 10
19
cm
−3
.
17
Al has also a lower thermal
ionization energy in 4H-SiC than B, about 200 meV
19
against about
600 meV.
20
Moreover, the Al ionization energy shows an important
dependence on the Al acceptor concentration.
21
In the case of identi-
cal B and Al acceptor concentrations, such differences favor a higher
p-type conductivity in the Al case because of its lower partial ion-
ization at any sample temperature. Last but not least, a reason for
preferring Al to B, is that Al is almost immobile during post implan-
tation annealing while B not at all.
22
This works is a study on the relevance of the annealing time during
the electrical activation of Al implanted in 4H-SiC at 1950
◦
C anneal-
ing temperature. In this study the implanted Al concentration has been
kept almost equal to 1 × 10
20
cm
−3
that is lower than the Al solubil-
ity in 4H-SiC at the annealing temperature of 1950
◦
C.
18
Moreover,
to avoid the activation of phenomena favored by temperatures lower
than 1950
◦
C, a fast heating rate has been used. At the same time,
for quenching the annealing process at the end of the time spent at
1950
◦
C, a fast cooling rate has been used too.
Experimental
A High Purity Semi-Insulating (HPSI) 8
◦
off-axis <0001> 4H-
SiC wafer was Al
+
implanted at 400
◦
C with various doses and energy
to obtain an almost box shaped Al depth profile, 1 × 10
20
cm
−3
high
ECS Journal of Solid State Science and Technology, 5 (9) P534-P539 (2016) P535
and about 1.3 μm thick next to the wafer surface. Ion implantation was
performed by using a Tandentron 1.7 MV accelerator (High Voltage
Engineering Europa B.V.), equipped with a 4 inches wafer holder.
AthickSiO
2
film was deposited on the wafer before implantation.
This film has the function to decrease the ion kinetic energy so that a
flat Al depth profile next to the wafer surface can be obtained. After
the implantation, the wafer was diced in 5 mm × 5 mm pieces. The
SiO
2
film was accurately removed in a hydrofluoric bath. Immediately
after, a resist film was spun on the implanted surface of each sample.
This film was transformed in a carbon layer (C-cap) by a pyrolysis at
900
◦
C for 2 min in forming gas.
23
Post implantation annealing processes were performed in high
purity Ar ambient. The samples were inside an inductively heated
graphite box (crucible). In such a system, crucible heating and cool-
ing are controlled by regulating the power of the radio frequency
generator. The maximum heating rate of our setup is 40
◦
C/s, while
the faster cooling transient is determined by the thermal inertia of
the crucible once the generator is switched off. The temperature of
the crucible is measured by an optical pyrometer. All the samples of
this study were treated with annealing temperature, heating rate and
cooling cycle constant and equal to 1950
◦
C, 38
◦
C/s, exponential with
65 s time constant, respectively. The latter corresponds to a time t
dependence of the cooling temperature T given by the equation T =
1950 exp[(t
0
-t)/65] with t
0
(time at which T starts to decrease) and
t expressed by “second” unit. All that corresponds to about 1 min
(0.1 min spent above 1600
◦
C) to reach the annealing temperature, and
to about 5 min (0.2 min spent above 1600
◦
C) to cool down to RT
after annealing. To the best of these authors experience, 1600
◦
Cis
the minimum annealing temperature for measuring a significant elec-
trical activation of implanted Al in 4H-SiC. Moreover, temperature
overshoot and time to reach temperature stationarity were 15
◦
Cand
20 s, respectively. Four samples with different annealing times of
5–10–25–40 min were processed. After annealing, the C-cap was
removedbya850
◦
C/15 min dry oxidation process.
van der Pauw (vdP) devices were obtained by forming triangular
Ti(80 nm)/Al(350 nm) ohmic contacts on the four corners of the
annealed specimens. These contacts were alloyed 1000
◦
C/2 min in
vacuum.
The root mean square (rms) surface roughness, not shown in this
work, was measured on virgin and annealed materials, the latter after
C-cap removal. This parameter was found 0.2 nm and ≤ 5 nm for the
virgin sample and the sample of longer annealing time, respectively.
This is an expected trend.
The Al depth profiles after annealing were measured by Secondary
Ion Mass Spectrometry (SIMS) by using a Cameca IMS-4f spectrom-
eter with an 8 keV O
2
+
primary ion beam.
Four point sheet resistance and Hall effect measurements on the van
der Pauw devices were performed in vacuum at different temperatures
in the range 30–680 K. Two different set-ups, equipped with magnets
of different intensity, were used for measurements above (1 T magnet)
and below (0.8 T magnet) room temperature (RT). Measurements
of the same sample were taken during heating and during cooling,
moreover they were repeated at distance of time. The spread of all
the data at a given temperature was less than one percent. Due to the
non-negligible dimensions of the contacts with respect to the device
size, proper correction factors were applied to the results of both sheet
resistance and Hall coefficient measurements.
24
To convert the Hall
data in drift data the Hall factor r
H
of Ref. 25 for holes in 4H-SiC has
been used. It is worthwhile to remember that this factor is temperature
dependent r
H
(T).InRef.26 it has been shown that this r
H
25
is valid
for Al acceptor densities in 4H-SiC going from 1.8 × 10
15
cm
−3
to
1.1 × 10
20
cm
−3
.
Results
Fig. 1a shows the SIMS Al depth profiles of the 5 min and
40 min annealed specimens. The differences between these two Al
depth profiles are consistent both with the uncertainty of the SIMS
measurements and with differences in the thickness of the SiO
2
film
Figure 1. (a) SIMS Al depth profiles of the 5 min and 40 min annealed 4H-SiC
samples. (b) Schematic representation of the Al depth profiles of (a).
used as stopping power of the ion energy during the implantation
process. Every other sample of this study had an Al depth profile
overlapping with those of Fig. 1a. Fig. 1b is a schematic drawing
of the Al depth profiles of Fig. 1a. The Al depth profiles show the
scheduled 1.1 × 10
20
cm
−3
plateau about 1250 nm thick but with a
significant depletion around 700 nm. Such a depletion is the conse-
quence of an error in the performed implantation process. Fortunately,
such an error does not hamper the possibility to perform the desired
study (see the Discussion section).
Fig. 2a shows the Arrhenius plot of the temperature dependence
of the sheet resistance of all the samples of this study. Fig. 2b is
the enlarged view of the high temperature region of Fig. 2a.This
figure shows that at high T the sheet resistance values of the Al
implanted layer of this study decreases with the increasing of the
annealing time and attains saturation for annealing times longer than
25 min. No dependence of the sheet resistance curves on the annealing
time is evident at low T. Each curve of Fig. 2a features two almost
exponential trends, at low and at high temperatures, as for a carrier
transport that is thermally activated over the whole T interval, but
with two different thermal activation energies, and thus two different
mechanisms involved, depending on T. Thermal activation energy
values in the high T region attain hundreds of meV while those in the
low T region, tens of meV. Moreover, at high T, these values decrease
with the increasing of the annealing time, varying from 116 meV for
5 min annealing to 107 meV for 25–40 min annealing. At low T a
single thermal activation energy of about 20 meV accounts for all the
curves.
The RT sheet resistance values of Fig. 2 are plotted versus the
annealing time in Fig. 3. Two straight lines interpolate the 5 min and
10 min data, and the 25 min and 40 min ones. The cross point between
these straight lines falls at ≈ 12 min. This value is an estimation of the
minimum time spent at 1950
◦
C for achieving the lowest p-type sheet
resistance for the Al implanted layer of this study.
Fig. 4a shows the temperature dependence of the ratio r
H
(T)/
[e R
H
(T)], i.e. the ratio between the Hall factor r
H
(T) and the product
P536 ECS Journal of Solid State Science and Technology, 5 (9) P534-P539 (2016)
Figure 2. (a) Temperature dependence of the sheet resistance of the Al implanted HPSI 4H-SiC samples of this study after 1950
◦
C annealing, 5−10−25−40 min
long (see inset). (b) Enlarged view of the high temperature region of (a). The dimension of symbols includes the measurement error bars.
Figure 3. Room temperature values of sheet resistance (●) and acceptor ion-
ization energy () of the samples of this study after post implantation anneal-
ing at 1950
◦
C (see text) and different annealing times. Dashed straight lines
interpolates the decreasing and the saturated data (see text). The cross point
between these two trends is 12 min and 22 min for sheet resistance (●)and
acceptor ionization energy (), respectively.
of the measured Hall coefficient R
H
(T) times the elementary charge
e. This quantity, that has the dimension “cm
−2
”, in the case of homo-
geneously doped layers, is the drift hole area density of the measured
layers. In the Discussion section it will be explained why, in the
case of the Al implanted layers of this study, the quantity plotted in
Fig. 4a can be equal to the effective carrier area density in the layer.
Fig. 4b is an enlarged view of the high temperature region of Fig. 4a.
With the increase of the annealing time, the curves of Fig. 4a attain
higher values, up to reach saturation values for annealing time above
25 min. In the low T region, the temperature dependence of the curves
of the10–25–40 min samples of Fig. 4a show a minimum, while in
the high T region all the curves of Fig. 4a show an exponential trend.
These latter have activation energies that decrease from 99 meV down
to 92 meV with the increasing of the annealing time from 5 min to
25 min (see Fig. 4b). Above 25 min, thermal activation energies equal
that of the 25 min specimen. These values have been plotted in Fig. 3
as a function of the annealing time. The cross-point between the two
straight lines that interpolate the 5–10 min data and the 25–40 min
ones falls at ≈22 min. This time is the minimum time spent at 1950
◦
C
for achieving the lowest activation energy in the high T region of the
temperature dependence of the carrier area density in the samples of
this study.
The drift carrier mobility in the Al implanted layers of the sam-
ples of this study has been obtained as the ratio between the Hall
Figure 4. (a) Temperature dependence of the drift hole area density of the Al implanted layer of the 4H-SiC samples of this study (see text). (b) Enlarged view of
the high temperature region of (a). The dimension of symbols includes the measurement error bars.