Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

S4-3 (L. N.) Ext. Abs.

the 13th International Workshop on Junction Technology 2013

Thermal Behavior of Residual Damage in Low-Dose Implanted Silicon after


High-Temperature Rapid Thermal Annealing
Akihiko Sagara and Satoshi Shibata
Panasonic Corporation
3-1-1, Yagumo-nakamachi, Moriguchi, Osaka 570-8501, Japan
Phone: +81-6-6906-4916 Fax: +81-6-6906-3407 E-mail: sagara.akihiko@jp.panasonic.com

1. Introduction 2. Experimental

Along with the development of the Si semiconductor Figure 1 shows the process of making of the samples
industry, numerous studies have been carried out on the used in this investigation. The base material was a p-type
defects that remain after ion-implantation processes [1]. For Cz-Si wafer and a thermal oxide layer was grown to a
15 –2 +
example, in the case of high-dose (~10 cm ) implantation, thickness of about 40 nm. As ions were implanted through
13 –2 +
dislocation loops can be created even after annealing. These the oxide layer at 150 keV to a dose of 1.0×10 cm . B
12
defects are typically evaluated by transmission electron ions were also implanted at 250 keV to a dose of 1.0×10
–2 12 –2
microscopy (TEM) and have been confirmed as a reason for cm and 400 keV to a dose of 1.0×10 cm . These samples
13 –2
junction leakage [2][3]. Even in low-dose (< 10 cm ) were annealed at the top temperature of 1100 ˚C for 30
implantation, some intrinsic point defects remain at seconds using a hot-wall type RTA system. The heating-up
relatively low annealing temperatures (< 700 ˚C). These rate and cooling-down rate were set at 40 ˚C/sec and 80
defects have been conventionally analyzed and investigated ˚C/sec respectively. Additional FA was applied at 300 - 900
by optical and electrical characterization techniques, such ˚C in steps of 100 ˚C. The top temperature was held for 60
as photoluminescence (PL) and deep transient level minutes. The heating and cooling rates were slower than
spectroscopy (DLTS) [4]-[6]. In contrast, residual damage with the RTA system: 8 ˚C/min and 4 ˚C/min respectively.
in low-dose implanted and high-temperature annealed Si An oxide-layer-grown sample was prepared as reference.
has not been detected and reported. Therefore, it is believed We also prepared samples without RTA treatment, which
that there is no damage remains in this condition, and, if were FA-applied at 800 - 900 ˚C immediately after
exists, it has no influence on device performance. Little implantation, for comparison. CL spectra were analyzed
attention has been paid to the defects that remain after using a scanning electron microscope (SEM) with a single
low-dose implantation processes. monochromator and InGaAs multi-channel detector. All CL
In our previous studies, we have shown the presence of measurements were performed at 15 K. For PAS analysis,
residual damage when using a rapid thermal process as heat the oxide layers were removed using hydrofluoric (HF)
treatment. Cathodoluminescence (CL) analysis clearly acid. The Doppler broadening spectra of the positron
13
revealed a non-radiative decay center in low-dose (10 annihilation radiation were measured as a function of the
–2
cm ) arsenic (As) implanted Si after high-temperature incident positron energy. The low-momentum part of the
(1100 ˚C) rapid thermal annealing (RTA) [7]. This defect spectrum was characterized by the S parameter, which was
has been identified by positron annihilation spectroscopy defined as the number of annihilation events over the
analysis as a vacancy-type defect (PAS). Moreover, it has energy range of 511 keV ± ΔE (where ΔE = 0.76 keV).
been shown that this defect is transformed into an This S parameter is an index of
oxygen-related defect by a non-equilibrium reaction with
oxygen (O2) in the annealing chamber during the extremely
rapid cooling step in the RTA sequence [8]. SiO2 40 nm
SiO2-grown
In this study, we analyze the thermal behavior of the p-type Cz-Si
defects remaining after RTA. We prepared low-dose
implanted and high-temperature RTA-applied Si wafers. Impurity
As 150keV 1× 1013 cm-2
Implantation B 250keV 1× 1012 cm-2
We also applied additional furnace annealing (FA) as
B 400keV 1× 1012 cm-2
equilibrium heat treatment at various temperatures. The
defects were characterized using a carrier lifetime Rapid Thermal
Annealing (RTA) 1100 ˚C, 30 sec
measurement system in addition to the CL and PAS ↑40 ˚C/sec, ↓80 ˚C/sec
methods. The details of the defects and their evolution
during annealing will be discussed by comparison with Furnace
carrier properties obtained by sheet resistance Annealing (FA) 300~900 ˚C, 60 min
↑8 ˚C/min, ↓4 ˚C/min
measurements.
Figure 1. Fabrication of the samples used in this investigation.

978-1-4799-0580-5/13/$31.00 ©2013 IEEE 45

Authorized licensed use limited to: Cardiff University. Downloaded on May 02,2023 at 16:20:09 UTC from IEEE Xplore. Restrictions apply.
S4-3 (L. N.) Ext. Abs. the 13th International Workshop on Junction Technology 2013

the number and size of vacancy-type defects [9]. Effective 1.2

Normalized TO-line Intensity (a.u.)


minority carrier lifetime was measured using a 9.35-GHz (a) RTA+FA reference (b) FA only
microwave transmittance measurement system and 1.0
analyzed on the basis of carrier diffusion and annihilation Annealing conditions
0.8 RTA: 1100˚C, 30sec.
theories [10][11]. A 635-nm CW laser diode light, with its FA: 300-900˚C, 60min.
2
intensity set at 1.5 mW/cm , was used to generate carriers.
The microwaves transmitted through the samples were 0.6
rectified using a high-speed diode and integrated. The
detection accuracy of the transmissivity was ± 0.1% in the 0.4
present system [12]. The sheet resistance was measured FA:
0.2 800-900˚C
using Frontier Semiconductor RsL300 non-contact 60min.
photovoltage-based equipment. [13] The wavelength and
0
Rs frequency of the modulated LED were 700 nm and 100 0 100 200 300 400 500 600 700 800 900 800 900
kHz respectively. The 1000 iterated measurements of each FA temperature (˚C)
measurement point improved the repeatability by up to Figure 3. Comparisons of TO-line intensity of the samples after
0.08%. annealing. (a) RTA + FA. (b) FA. All intensities are normalized
by those of the reference sample.

3. Results and discussion


peaks can be seen other than these TO and TO+OΓ-lines, it
Figure 2 shows the CL spectra of reference, can be concluded that the defect density is low just after the
as-implanted, and the annealed samples subjected to FA in oxide layer has grown. In contrast, the intensity of TO-line
addition to RTA. Two emission lines, labeled TO and is reduced after implantation and a W-line is newly
TO+OΓ, can be observed in the spectrum of the reference detected in the spectrum. W-line emission is considered to
sample. These lines are typical of band-to-band transition be originated from Si self-interstitial clusters [14]-[17].
followed by the emission of transverse optical (TO) This defect-related line reflects the development of implant
phonons and optical phonons at k = 0 (OΓ) [14]. Since no damage caused by the impact of accelerated ions. When
this implanted sample was annealed using the RTA system
at a high temperature of 1100 ˚C, the W-line disappeared
T=15K and the TO-line reappeared. This indicates that the implant
Annealing conditions
damage had been annealed out and the crystallinity had
TO RTA: 1100˚C, 30sec. been regained to some extent. However, the intensity of the
FA: 300-900˚C, 60min. TO-line gives new information on the residual defects.
TO+OΓ Figure 3 shows a comparison of the TO-line intensity of the
RTA+FA900˚C
samples after annealing. Note that the repeatability of
TO-line intensity, obtained by taking the average of 5
RTA+FA800˚C
measurements, was realized to within about 6% by
optimizing the optical systems and controlling the
RTA+FA700˚C
CL intensity (a.u.)

fluctuation of the sample temperature. Its intensity was


H normalized by the value of the reference sample, for
RTA+FA600˚C
comparison. The TO-line intensity of the only-RTA-applied
sample (i.e., additional FA temperature was 0 ˚C) is clearly
RTA+FA500˚C much lower than that of the reference sample. This suggests
that the TO-line emission had been suppressed by
RTA+FA400˚C non-radiative decay centers, i.e., some kind of defect and/or
crystal disorder. This provides strong evidence of the
RTA+FA300˚C presence of residual damage, as we had discovered in
previous studies [7]. In contrast, the TO-line intensity of the
RTA samples with additional FA at 300 ˚C and 400 ˚C gradually
W increased with increased annealing temperature. This
as-impla. indicates that the vacancy-type defects mentioned above
tended to be removed and the total number of defects was
reference suppressed after FA. When the FA temperature was set at
1100 1200 1300 1400 1500 500 - 600 ˚C, an H-line was observed, as shown in Figure 2.
Wavelength (nm) Although the origin of this H-line is not entirely clear, it
Figure 2. CL spectra of reference, as-implanted and the annealed appears to be due to some kind of point defect related to
samples subjected to FA in addition to RTA. interstitial carbon or interstitial oxygen [14][18]. This

978-1-4799-0580-5/13/$31.00 ©2013 IEEE 46

Authorized licensed use limited to: Cardiff University. Downloaded on May 02,2023 at 16:20:09 UTC from IEEE Xplore. Restrictions apply.
S4-3 (L. N.) Ext. Abs. the 13th International Workshop on Junction Technology 2013

defect line disappeared after FA at 700 - 900 ˚C. The


103
TO-line intensity was drastically increased and saturated at reference

carrier lifetime (usec)


the value of the reference sample. This demonstrates that
crystallinity is likely to be regained to almost the same
level as the reference sample. For comparison, the samples 102

Effective minority
to which FA had been applied instead of RTA were
prepared and measured by CL. Only TO and TO+OΓ-lines
were detected in the spectra (not shown). The TO-line
101
intensities of these samples were slightly lower than those
Annealing conditions
of the samples subjected to RTA before FA at 700 - 900 ˚C,
RTA: 1100˚C, 30sec.
as shown in Figure 3. This result indicates that some kind FA: 700-900˚C, 60min.
of damage is present in the only-FA-applied sample (i.e.,
100


without RTA). The difference between these samples was 0 700 800 900
the presence or absence of RTA at 1100 ˚C, the maximum FA temperature (˚C)
temperature of the annealing process. We therefore Figure 5. Effective minority carrier lifetime of some of the
presume that a maximum temperature of annealing of 800 - annealed samples subjected to FA in addition to RTA
900 ˚C was insufficient to remove the implant damage
and/or activate impurities. Therefore, high-temperature
RTA appears to be necessary before FA to suppress the reference sample’s profiles. This result clearly indicates
creation of defects. that vacancy-type defects were abolished after FA at 700 ˚C.
Several samples were evaluated by PAS, and the S This conclusion matches the CL analysis.
parameters as a function of incident positron energy E are Figure 5 shows the effective minority carrier lifetime of
shown in Figure 4. The mean implantation depth of several samples. Carrier lifetime is very sensitive to the
positrons in Si crystal is calculated and described at the total number of defects and crystal quality. Compared to
upper x-axis in the same Figure [9]. The S value of the the reference sample, the lifetime of the only-RTA-applied
only-RTA-applied sample was higher than that of the sample was one order lower. This is the influence of the
reference sample, and it was distributed near the surface residual damage detected by CL and PAS measurements. In
region up to a few hundreds of nm. This provides further contrast, the lifetime of the samples to which additional FA
evidence of the presence of damage as suggested by the CL had been applied at 700 ˚C was one order higher than the
results; and a part of this damage was identified as only-RTA-applied sample, and was saturated with
vacancy-type defects. When additional FA was applied increasing annealing temperature from 700 ˚C to 900 ˚C.
after RTA, the S value gradually decreased with increasing This suggests that almost all the defects that act as lifetime
FA temperature, and finally its distribution in the killers in the lattice were removed by additional FA at 700
FA-applied sample at 700 ˚C coincided exactly with the ˚C or higher. This result also corresponds to the
expectations arrived at after CL measurements.
The sheet resistance variation of all annealed samples is
Mean Implantation depth shown in Figure 6. Unlike the only-RTA-applied samples,
of positrons (nm) the sheet resistance of the samples applied with additional
10 50 100 300 500
0.560 FA fell with increasing FA temperature. Since it was
□ reference
confirmed by PAS analysis that the vacancy-type defects
0.555 ● RTA gradually disappeared in the temperature range of 300 - 400
◆ RTA+FA300˚C ˚C, this reduction of sheet resistance might be due to a
▲ RTA+FA700˚C
reduction in carrier-trapping sites. In contrast, a significant
S parameter

0.550
Annealing conditions change of the sheet resistance was observed in the samples
RTA: 1100˚C, 30sec. annealed at 500 - 600 ˚C. In this temperature range, the CL
FA: 300,700˚C, 60min.
0.545 results confirmed that intrinsic point defects were being
created. This means that the variation in sheet resistance is
0.540 due to the creation of defect levels related to point defects
in the band gap that are likely to act as carrier-trapping sites
0.535
which keep the sheet resistance still higher. When the
0 2 4 6 8 10 temperature of additional FA was set at 700 - 900 ˚C, the
Positron Energy (keV) sheet resistance decreased and seemed to level off at a
Figure 4. S parameters as a function of incident positron energy E lower level. This result clearly indicates that most
for RTA and FA-applied samples. A plot of the reference sample carrier-trapping sites, i.e., residual defects, were removed
is also shown for comparison. Mean implantation depth of by additional FA at 700 ˚C or higher, as shown by using
positrons in Si crystal are calculated and recorded on the upper various techniques for measuring its characteristics. Th e
x-axis. sheet resistance under this particular annealing condition

978-1-4799-0580-5/13/$31.00 ©2013 IEEE 47

Authorized licensed use limited to: Cardiff University. Downloaded on May 02,2023 at 16:20:09 UTC from IEEE Xplore. Restrictions apply.
S4-3 (L. N.) Ext. Abs. the 13th International Workshop on Junction Technology 2013

2150 Yoshikawa for analyzing the SIMS profiles. We are also grateful
(a) RTA+FA (b) FA only to K. Ito and E. Igaki for their support.
Sheet resistance (orm/sq.)

2100
Annealing conditions
2050 RTA: 1100˚C, 30sec.
FA: 300-900˚C, 60min. References
2000
[1] See review, K.S. Jones, S. Prussin, and E.R. Weber, Appl.
1950
FA:
Phys. A 45, 1 (1988).
1900 800-900˚C [2] N.E. Cowern, Proc. Mater. Res. Soc. Symp. 1070, 143
60min.
(2008).
1850 [3] S. Prussin, D.I. Margolese, and R.N. Tauber, J. Appl. Phys.
54, 2316 (1983).
1800
0 100 200 300 400 500 600 700 800 900 800 900 [4] K. Terashima, T, Ikarashi, M. Watanabe, and T. Kitano, NEC
FA temperature (˚C) Res. & Develop., 39, 289 (1998).
Figure 6. Sheet resistance of the samples after annealing. (a) RTA [5] O.O. Awadelkarim, A. Henry, B. Monemar, J.L. Lindstrom,
+ FA. (b) FA. Y. Zhang, and J.W. Corbett, Phys. Rev. B 42, 5635 (1990).
[6] S. Libertino, S. Coffa, and J.L. Benton, Phys. Rev. B 63,
195206 (2001).
might thus correspond to the expected value under [7] A. Sagara, M. Hiraiwa, S. Shibata, R. Sugie, and K. Yamada,
conditions where the implant impurities are fully activated. AIP Proc. 1321, 225 (2008).
Figure 3 shows that the sheet resistance of the [8] A. Sagara, M. Hiraiwa, A. Uedono, and S. Shibata,
th
only-FA-applied sample was slightly lower than that of the Proceedings of the 12 International Workshop on Junction
samples to which RTA had been applied before FA at 700 - Technology, 81 (2012).
900 ˚C. We surmise this to be due to insufficient activation [9] A. Uedono, K. Tsutsui, S. Ishibashi, H. Watanabe, S. Kubota,
of B at the FA temperature of 800 - 900 ˚C. This conclusion Y. Nakagawa, B. Mizuno, T. Hattori and H. Iwai, Jpn. J.
is also consistent with the trend of the TO-line intensity as Appl. Phys. 49, 051301 (2010).
evaluated by CL analysis. [10] T. Sameshima, H. Hayasaka, and T. Haba, Jpn. J. Appl. Phys.
48 021204 (2009).
[11] T. Sameshima, T. Nagao, S. Yoshidomi, K. Kogure, and M.
4. Conclusions Hasumi, Jpn. J. Appl. Phys. 50 03CA02 (2011).
[12] T. Sameshima, K. Betsuin, T. Mizuno, and N. Sano, Jpn. J.
We analyzed the thermal behavior of residual damage Appl. Phys. 51 03CA04 (2012).
13 –2
in low-dose (< 10 cm ) As and B implanted Si after [13] V.N. Faifer, M.I. Current, T.M.H. Wong and V.V. Souchkov,
high-temperature (1100 ˚C) RTA after applying additional J. Vac. Sci. Technol. B 24(1) 414-420 (2006).
FA at 300 - 900 ˚C. The characterization techniques we [14] G. Davies, Phys. Rep. 176, 83 (1989).
selected were CL, PAS, lifetime, and sheet resistance [15] G. Davies, E.C. Lightowlers and Z.E. Ciechanowska, J.
measurement. All the results revealed the following new Phys. C 20, 191 (1987).
information about the defects. Vacancy-type defects were [16] P.K. Giri, S. Coffa and E. Rimini, Appl. Phys. Lett. 78, 291
gradually removed in the temperature range of 300 - 400 ˚C. (2001).
In contrast, several kinds of point defects were created at [17] P.K. Giri, Semicond. Sci. Technol. 20, 638 (2005).
500 - 600 ˚C. These defects were confirmed to be [18] C.E. Jones, E. S. Johnson, W. D. Compton, J.R. Noonan,
eliminated at 700 ˚C and crystal quality was significantly and B.G. Streetman, J. Appl. Phys. 44, 5402 (1973).
improved, exceeding that of the only-FA-applied samples
without RTA. The reason is presumed to be that the
maximum temperature of FA (800 - 900 ˚C) is insufficient
to activate and/or remove the implant damage.
High-temperature RTA is thus necessary before FA to be
able to activate the impurities and reduce residual damage
after the implantation process.

Acknowledgements

We would like to express our gratitude to Prof. A. Uedono for


performing the PAS measurements and to Prof. T. Sameshima for
FC measurements and their analysis. We would like to thank F.
Kawase and T. Ishisaki for preparing the samples and S.

978-1-4799-0580-5/13/$31.00 ©2013 IEEE 48

Authorized licensed use limited to: Cardiff University. Downloaded on May 02,2023 at 16:20:09 UTC from IEEE Xplore. Restrictions apply.

You might also like