ArticlePDF Available

Formation of Ultra-shallow Junction with :*10 nm in Si Combined with Low Temperature and Laser Annealing

Authors:

Abstract and Figures

A combination of Ge pre-amorphization implantation (Ge-PAI), low-energy B implantation and laser annealing is a promising method to form highly-activated, abrupt and ultra-shallow junctions (USJ). In our previous report of IIT 2006, we succeeded in forming pn junctions less than 10nm using non-melt double-pulsed green laser. However, a large leakage current under reverse bias was observed consequently due to residual defects in the implanted layer. In this study, a method to form USJ is proposed; a combination of low-temperature solid phase epitaxy and non-melt laser irradiation for B activation. Ge pre-amorphization implantation was performed at energy of 6keV with a dose of 3x1014/cm2. Then B implantation was performed at energy of 0.2keV with a dose of 1.2x1015/cm2. Samples were annealed at 400C for 10h in nitrogen atmosphere. Subsequently, non-melt laser irradiation was performed at energy of 690mJ/cm2 and pulse duration of 100ns with intervals of 300ns. As a result, USJ around 10nm with better crystallinity was successfully formed. And the leakage current of pn diodes was reduced significantly. Moreover, it is proved from secondary ion mass spectroscopy (SIMS) analysis that transient enhanced diffusion (TED) of B is specifically suppressed.
Content may be subject to copyright.
Formation of Ultra-shallow Junction with
10 nm in Si Combined with Low
Temperature and Laser Annealing
Takumi Fukaya
1
, Shuhei Hara
1
, Yuki Tanaka
1
, Satoru Matsumoto
1
, Toshiharu Suzuki
2
, Genshu Fuse
2
, Toshio Kudo
3
and Susumu Sakuragi
3
1
Keio University, 3-14-1 Hiyoshi, Kouhoku-ku, Yokohama, Kanagawa, 223-8522 Japan
2
SEN Corporation, an SHI and Axcelis Company, SBS Tower 9F, 4-10-1 Yoga, Setagaya-ku, Tokyo, 158-0097 Japan
3Sumitomo Heavy Industries Ltd., 19 Natsushima-cho, Yokosuka, Kanagawa, 237-8555 Japan
E-mail: fukaya@a5.keio.jp
, matumoto@elec.keio.ac.jp
Abstract. A combination of Ge pre-amorphization implantation (Ge-PAI), low-energy B
implantation and laser annealing is a promising method to form highly-activated, abrupt and
ultra-shallow junctions (USJ). In our previous report of IIT 2006, we succeeded in forming pn
junctions less than 10nm using non-melt double-pulsed green laser. However, a large leakage
current under reverse bias was observed consequently due to residual defects in the implanted
layer. In this study, a method to form USJ is proposed; a combination of low-temperature
solid phase epitaxy and non-melt laser irradiation for B activation. Ge pre-amorphization
implantation was performed at energy of 6keV with a dose of 3x10
14
/cm
2
. Then B
implantation was performed at energy of 0.2keV with a dose of 1.2x10
15
/cm
2
. Samples were
annealed at 400C for 10h in nitrogen atmosphere. Subsequently, non-melt laser irradiation
was performed at energy of 690mJ/cm
2
and pulse duration of 100ns with intervals of 300ns.
As a result, USJ around 10nm with better crystallinity was successfully formed. And the
leakage current of pn diodes was reduced significantly. Moreover, it is proved from
secondary ion mass spectroscopy (SIMS) analysis that transient enhanced diffusion (TED) of
B is specifically suppressed.
1. INTRODUCTION
Miniaturization of MOSFET’s by scaling-down of device dimensions requires the formation of ultra-shallow
junction with a low sheet resistance [1, 2]. A lot of process techniques have been developed to meet the above
requirements. Among these processes, a combination of Ge pre-amorphization implantation (Ge-PAI), low-energy B
implantation and laser annealing (LA) is a promising method to form highly-activated, abrupt and ultra-shallow
junctions [3, 4]. In the previous report of IIT 2006 [5], we succeeded in fabrication of pn junctions less than 10nm
using non-melt double-pulsed laser. However, a large leakage current under reverse bias of pn diodes was observed
consequently due to residual defects in implanted layers. On the basis of the previous results, in this study, the effect
of low temperature annealing prior to non-melt laser annealing was investigated in order to improve crystallinity of
implanted layers.
2. EXPERIMENTAL
725ȝm thick, n-type Si (100) wafers with more than 10ȍ cm were used as substrates. First, Ge
pre-amorphization implantation to suppress channeling of B ions was performed at energies of 3keV and 6keV with a
dose of 3x10
14
/cm
2
and tilt angle of 7°. Then B low-energy implantation was performed at energy of 0.2keV with a
dose of 1.2x10
15
/cm
2
. Subsequently, samples were annealed in N
2
atmosphere at 400C for 1.5h for Ge-PAI energy of
3keV and 10h for 6keV prior to the laser annealing. Laser annealing was carried out in air using a 527nm
The 5th International Symposium on Advanced Science and Technology of Silicon Materials (JSPS Si Symposium),
Nov. 10-14, 2008, Kona, Hawaii, USA
double-pulsed green laser with a pulse width of 100ns, oscillated at a frequency of 1kHz and delay time between the
first and the second pulse of 300ns. The laser formed homogenized 0.25mm x 2.5mm line beam and samples were
irradiated with overlap ratio of 90%. Laser energy densities were selected between 690-760mJ/cm
2
. Annealed samples
were analyzed with four-point probe method for the sheet resistance, secondary ion mass spectroscopy (SIMS) for the
dopant profiles and atomic force microscopy (AFM) for surface morphology evaluation. For SIMS, O
2+
was used as a
primary ion at energy of 350eV. The junction depth was determined at B concentration of 1x10
18
/cm
3
. The junction
leakage current characteristics of pn diodes were measured. The crystallinity near the junction in implanted layers was
analyzed by transmission electron microscope (TEM).
3. RESULTS AND DISCUSSION
AFM Analysis
Figure 1 shows the surface of samples with Ge-PAI energy of 6keV after laser irradiations with energy densities of
(a) 700mJ/cm
2
and (b) 730mJ/cm
2
respectively.

FIG. 1 AFM images of annealed samples. (a) 700mJ/cm
2
, (b) 730mJ/cm
2
The surface of the sample annealed at 700mJ/cm
2
is smooth compared to the image of the sample annealed at
730mJ/cm
2
. It is considered that there is no melt of surface with the fluence of 700mJ/cm
2
. With energies higher than
710mJ/cm
2
, the surface starts getting rough. Therefore, it is expected that the threshold of laser energy density
between a non-melting and a melting state is in-between 710-720mJ/cm
2
with the Ge-PAI energy of 6keV. Meanwhile,
the threshold energy is in-between 720-730mJ/cm
2
with the Ge-PAI energy of 3keV. This difference indicates that
there would be larger amorphous region after pre-anneal for samples with Ge-PAI energy of 6keV. Since absorption
coefficient is much higher in amorphous silicon, laser energy is reserved more as heat in samples with Ge-PAI energy
of 6keV.
Sheet Resistance Evaluation
Figure 2 shows the sheet resistances as a function of laser energy density after low-temperature pre-anneal and
laser anneal for Ge-PAI energies of 3keV and 6keV. Firstly, the values of sheet resistance decrease greatly after
pre-anneal for both samples. And it decreases gradually with higher energy density. It is thought that values decrease
because of melt of the surface and sequential rapid B diffusion. Samples with Ge-PAI energy of 6keV show lower
sheet resistance than 3keV, due to larger remaining amorphous region in implanted layer after pre-anneal. Also we
can see the typical reverse-annealing phenomenon [6] just before melting energy density for each sample. As a result,
the value of 714ȍ/غ is obtained for the sample with Ge-PAI energy of 6keV at 690mJ/cm
2
.
㪌㪇㪇
㪈㪇㪇㪇
㪈㪌㪇㪇
㪉㪇㪇㪇
㪉㪌㪇㪇
㪊㪇㪇㪇
㪊㪌㪇㪇
㪍㪏㪇 㪍㪐㪇 㪎㪇㪇 㪎㪈㪇 㪎㪉㪇 㪎㪊㪇 㪎㪋㪇 㪎㪌㪇 㪎㪍㪇
㪣㪸㫊㪼㫉㩷㪜㫅㪼㫉㪾㫐㩷㪛㪼㫅㫊㫀㫋㫐㩷䋨㫄㪡㪆㪺㫄
㪪㪿㪼㪼㫋㩷㪩㪼㫊㫀㫊㫋㪸㫅㪺㪼㩷䋨㱅㪆䂔䋩
㪞㪼㩷㪊㫂㪼㪭
㪞㪼㩷㪍㫂㪼㪭
㫄㪼㫃㫋
㪧㪘㩷㫆㫅㫃㫐
㫄㪼㫃㫋㫅㫆㫅㪄㫄㪼㫃㫋
㫅㫆㫅㪄㫄㪼㫃㫋
FIG. 2. Sheet resistance vs. Laser energy density
SIMS Profiles
Figure 3 shows the result of SIMS profiles. Results of this study (with Ge-PAI energy of 6keV), as-implanted, and
our previous data of 2007 [7] are included in the figure. It should be noted that the substrate of 2007 in Fig. 3 had the
same implantation condition except the Ge-PAI energy (3keV). In this case, samples were annealed at 650C for 1h
and were irradiated with 740mJ/cm
2
laser pulses.
We can see noticeable B diffusion over 7nm at the condition of year 2007. It is assumed that transient enhanced
diffusion (TED) of B occurred in the pre-anneal process. On the other hand, the B diffusion is suppressed obviously
with 400C 10h pre-annealing. Annealing with a very low temperature may be capable of removing point defects in the
implanted layer, which are responsible for TED. Though the boron profile shifts about 3nm deeper from that of
as-implanted, the junction around 10nm with an abruptness of ~2nm/dec. is successfully formed with altered
pre-anneal and laser anneal conditions.
㪈㪜㪂㪈㪍
㪈㪜㪂㪈㪎
㪈㪜㪂㪈㪏
㪈㪜㪂㪈㪐
㪈㪜㪂㪉㪇
㪈㪜㪂㪉㪈
㪈㪜㪂㪉㪉
㪈㪇 㪈㪌 㪉㪇 㪉㪌
㪛㪼㫇㫋㪿㩷㩿㫅㫄
㪙㩷㪚㫆㫅㪺㪼㫅㫋㫉㪸㫋㫀㫆㫅㩷㩿㪺㫄
㪄㪊
㪸㫊㪄㫀㫄㫇㫃㪸㫅㫋㪼㪻
㪍㪌㪇㪚㩷㪈㪿䋫㪣㪘㩷㪎㪋㪇㫄㪡㪆㪺㫄㪉㩷䋨㪉㪇㪇㪎䋩
㪋㪇㪇㪚㩷㪈㪇㪿㩷㩿㫋㪿㫀㫊㩷㫊㫋㫌㪻㫐㪀
㪋㪇㪇㪚㩷㪈㪇㪿䋫㪣㪘㩷㪍㪐㪇㫄㪡㪆㪺㫄㪉㩷㩿㫋㪿㫀㫊㩷㫊㫋㫌㪻㫐㪀
FIG. 3. SIMS profiles of this study and the past years
Diode Leakage Current Evaluation
Figure 4 shows the I-V characteristics of pn diodes. The result what we reported at IIT 2006 is also included. There
is a large leakage current under reverse bias for the sample of 2006 due to the residual defects in implanted layer.
Meanwhile, for the sample of this study, a leakage current decreases and current density under forward bias becomes
better at start. Besides following LA with 690mJ/cm
2
makes the reverse current much less (about 5 orders in
magnitude) than that of year 2006. These results indicate that the combination of low temperature pre-anneal and
non-melt LA improve the crystallinity of implanted layers and junction region by removing various defects.
㪈㪅㪇㪜㪄㪈㪇
㪈㪅㪇㪜㪄㪇㪐
㪈㪅㪇㪜㪄㪇㪏
㪈㪅㪇㪜㪄㪇㪎
㪈㪅㪇㪜㪄㪇㪍
㪈㪅㪇㪜㪄㪇㪌
㪈㪅㪇㪜㪄㪇㪋
㪈㪅㪇㪜㪄㪇㪊
㪈㪅㪇㪜㪄㪇㪉
㪈㪅㪇㪜㪄㪇㪈
㪈㪅㪇㪜㪂㪇㪇
㪈㪅㪇㪜㪂㪇㪈
㪈㪅㪇㪜㪂㪇㪉
㪄㪋 㪄㪊 㪄㪉 㪄㪈
㪙㫀㪸㫊㩷㪭㫆㫃㫋㪸㪾㪼㩷㩿㪭㪀
㪚㫌㫉㫉㪼㫅㫋㩷㪛㪼㫅㫊㫀㫋㫐㩷㩿㪘㪆㪺㫄
㪣㪘㩷㪎㪋㪇㫄㪡㪆㪺㫄㪉㩷䋨㪉㪇㪇㪍䋩
㪋㪇㪇㪚㩷㪈㪇㪿㩷㩿㫋㪿㫀㫊㩷㫊㫋㫌㪻㫐㪀
㪋㪇㪇㪚㩷㪈㪇㪿㪂㪣㪘
㪍㪐㪇㫄㪡㪆㪺㫄㪉㩷㩿㫋㪿㫀㫊㩷㫊㫋㫌㪻㫐㪀
FIG. 4. Diode characteristics
TEM Analysis
Finally, Fig. 5 shows the cross sectional TEM (X-TEM) images near the junction in implanted layers of (a) after
LA 690mJ/cm
2
and (b) after Pre-anneal 400C 10h + LA 690mJ/cm
2
. It is obvious that the surface region is destructed
for the sample with only LA. This implies that the crystal regrowth was not enough with only LA and it contains lots
of defects in the implanted layer still. On the other hand, the image of the sample with pre-anneal + LA shows the tidy
lattice of Si suggesting that pre-anneal effectively restored the crystallinity of implanted layers.

FIG. 5.
X-TEM image of implanted layers (a) with only LA 690mJ/cm
2
, (b) with pre-anneal 400C 10h + LA
690mJ/cm
2
4. CONCLUSION
The methods fabricating ultra-shallow junction with the combination of low-temperature anneal and non-melt LA
are shown in this study. It is found that low temperature anneal may be an effective process to reduce defects in
implanted layers without large B diffusion As a result, USJ of ~10nm with 714ȍ/غ and better crystallinity is formed
by using the combination method of Ge-PAI, low-temperature pre-anneal, non-melt LA. Since 10h-annealing is long
for mass production, it is necessary to optimize the pre-anneal condition.
ACKNOWLEDGMENTS
The aouthors would like to thank SEN Corporation, an SHI and Axcelis Company, Sumitomo Heavy Industries
Ltd., MST and Sumika Chemical Analysis Service Ltd. for their technical support.
REFERENCES
1. International Technology Roadmap for Semiconductors (2006)
2. H. W. Kennel, P. H. Keys, M. Kiu, Proc. of Workshop on USJT (2005) 4
3. K. K. Ong, K. L. Pey, P. S. Lee, A. T. S. Wee, Y. F. Chong, K. L. Yeo and X. C. Wang, Material Science and
Engineering B, 114-115 (2004) 25-28
4. M. Hernandez, J. Venturini, D. Zahorski et al., Applied Surface Science 208-209 (2003) 345-351
5. Ryuta Yamada, Singo Seto, Soci Sato, Yuki Tanaka, Satoru Matsumoto, Toshiharu Suzuki, Genshu Fuse, Toshio
Kudo and Susumu Sakuragi, Ion Implantation Technology, American Institute of Physics 978-0-7354-0365-9/06
(2006)
6. Jian-Yue Jin, Jinning Liu, Ukyo Jeong, and Sandeep Metha, J. Vac. Sci. Technol. B 20(1) (2002)
7. Takumi Fukaya, Ryuta Yamada, Yuki Tanaka, Satoru Matsumoto, Toshiharu Suzuki, Genshu Fuse, Toshio Kudo
and Susumu Sakuragi, 15
th
IEEE International Conference on Advanced Thermal Processing of Semiconductors –
RTP2007 (2007) 317-320
... The superior tunable doping depth by SOD also makes it possible to obtain the very thin doped surface region [10][11]. On the other hand, the dopant activation using the LA method sufficiently reduces the dopant diffusion into the deeper region of Ge, compared with the conventional rapid thermal annealing (RTA) for dopant activation, which guarantees the formation of the ultra-shallow Ge S/D junctions [12][13][14][15][16]. Therefore, the SOD followed by LA activation technique could be used for the formation of the high performance Ge S/D junctions for future generations of Ge MOSFETs. ...
... On the first set of samples, with target grounded, the beam with initial energy of 5 keV suffers too much dispersion along the beam line. [16][17][18]The results with deceleration but without the lens show that, as expected, that fluence was lower on the samples located near the target border due to beam divergence (Figure 2). The set, with deceleration and the lens, shows some homogeneity. ...
Article
An ion beam deceleration system was studied for the high current ion implanter at the Laboratório de Aceleradores e Tecnologias de Radiação at the Campus Tecnológico e Nuclear, of Instituto Superior Técnico. The installed system consists of a target plate and one electrostatic focusing lens with one electrode. This article describes the results of the evaluation of the new system. With this upgrade, the ion implanter provides enhanced versatility for decelerating to 5 keV a high current ion beam at the μA level. This implantation provides a wide area and allows for a continuous magnetic beam scanning, extending the energy range to lower values, opening up a wider set of applications.
... On the first set of samples, with target grounded, the beam with initial energy of 5 keV suffers too much dispersion along the beam line. [16][17][18]The results with deceleration but without the lens show that, as expected, that fluence was lower on the samples located near the target border due to beam divergence (Figure 2). The set, with deceleration and the lens, shows some homogeneity. ...
... The third test consists of an implantation with target bias and lens (beam of 15 keV, target bias at 10 kV at lens bias at 15.5 kV). Figure 8 shows the obtained values. The first test shows that it is not possible to have implantation homogeneity without deceleration, a beam with initial energy of 5 keV suffers too much dispersion along the beam line [11]- [13]. Second test, with deceleration but without electrostatic lens, shows that, as predicted by simulation, implantation fluence is higher in the sample located far from the target edge. ...
Article
Full-text available
Ion beam deceleration properties of a newly developed low-energy ion beam implantation system were studied. The objective of this system was to produce general purpose low-energy (5 to 15 keV) implantations with high current beam of hundreds of µA level, providing the most wide im-plantation area possible and allowing continuously magnetic scanning of the beam over the sam-ple(s). This paper describes the developed system installed in the high-current ion implanter at the Laboratory of Accelerators and Radiation Technologies of the Nuclear and Technological Campus , Sacavém, Portugal (CTN).
Genshu Fuse, Toshio Kudo and Susumu Sakuragi, Ion Implantation Technology
  • Ryuta Yamada
  • Singo Seto
  • Soci Sato
  • Yuki Tanaka
  • Satoru Matsumoto
  • Suzuki
Ryuta Yamada, Singo Seto, Soci Sato, Yuki Tanaka, Satoru Matsumoto, Toshiharu Suzuki, Genshu Fuse, Toshio Kudo and Susumu Sakuragi, Ion Implantation Technology, American Institute of Physics 978-0-7354-0365-9/06 (2006)
Genshu Fuse, Toshio Kudo and Susumu Sakuragi
  • Takumi Fukaya
  • Ryuta Yamada
  • Yuki Tanaka
  • Satoru Matsumoto
  • Suzuki
Takumi Fukaya, Ryuta Yamada, Yuki Tanaka, Satoru Matsumoto, Toshiharu Suzuki, Genshu Fuse, Toshio Kudo and Susumu Sakuragi, 15th IEEE International Conference on Advanced Thermal Processing of Semiconductors – RTP2007 (2007) 317-320
  • M Hernandez
  • J Venturini
  • D Zahorski
M. Hernandez, J. Venturini, D. Zahorski et al., Applied Surface Science 208-209 (2003) 345-351
  • K K Ong
  • K L Pey
  • P S Lee
  • A T S Wee
  • Y F Chong
  • K L Yeo
  • X C Wang
K. K. Ong, K. L. Pey, P. S. Lee, A. T. S. Wee, Y. F. Chong, K. L. Yeo and X. C. Wang, Material Science and Engineering B, 114-115 (2004) 25-28
  • Jinning Jian-Yue Jin
  • Ukyo Liu
  • Sandeep Jeong
  • Metha
Jian-Yue Jin, Jinning Liu, Ukyo Jeong, and Sandeep Metha, J. Vac. Sci. Technol. B 20(1) (2002)
Ion Implantation Technology
  • Ryuta Yamada
  • Singo Seto
  • Soci Sato
  • Yuki Tanaka
  • Satoru Matsumoto
  • Toshiharu Suzuki
  • Genshu Fuse
  • Toshio Kudo
  • Susumu Sakuragi
Ryuta Yamada, Singo Seto, Soci Sato, Yuki Tanaka, Satoru Matsumoto, Toshiharu Suzuki, Genshu Fuse, Toshio Kudo and Susumu Sakuragi, Ion Implantation Technology, American Institute of Physics 978-0-7354-0365-9/06 (2006)