close

Вход

Забыли?

вход по аккаунту

?

08002.0141ecst

код для вставкиСкачать
ECS Transactions, 80 (2) 141-146 (2017)
10.1149/08002.0141ecst ©The Electrochemical Society
Study of SiGe Surface Cleaning
K. Komoria,b, K. Wostynb, D. Rondasb, J.L. Pradob, T. Conardb, R. Loob, L.-Å.
Ragnarssonb, N. Horiguchib and F. Holsteynsb
a
SCREEN Semiconductor Solutions Co., Ltd., 480-1, Takamiya-cho, Hikone, Shiga
522-0292, Japan
b
imec vzw, Kapeldreef 75, 3001 Leuven, Belgium
SiGe is a promising candidate to replace Si in Fin-shaped and
GAA (gate all around) FETs (field-effect transistors) due to its
higher carrier mobility. However, the presence of Ge oxide in the
interfacial layer (IL) between the SiGe channel and HfO2 causes an
increase in interface trap density (DIT) and becomes in considerate
as a defect of the device. In this study, the IL formation by a
combined wet cleaning process with a subsequent annealing step is
investigated by X-ray Photo-electron Spectroscopy (XPS). Finally
we will present a process sequence leading up to a Ge-oxide free
interlayer.
Introduction
Recently novel approaches for the microfabrication of semiconductor devices are being
studied and reported. These approaches include the integration schemes such as FinFETs
(field-effect transistors) and GAA (gate all around) structures, and novel materials such
as SiGe (Silicon Germanium) and Ge (Germanium). The introduction of SiGe for PMOS
logic devices for <7nm technology node device is under development due to its higher
electron and hole mobility, but presents new challenges [1]. One possible integration
scheme for a SiGe PMOS device uses a Si cap on the IL of a SiGe based channel;
however, the device characteristics are sensitive to the cap film thickness [2]. Depositing
a uniform Si cap on SiGe FinFETs and GAA structures becomes difficult, because it is
necessary to control the cap film thickness on all the channel surfaces exposed, i.e. (100)
and (110). Therefore, a Si-cap-free integration scheme for SiGe CMOS devices is
preferred, but this approach also comes with some challenges. Due to the presence of
Ge-oxide in the IL of SiGe the interface trap density (DIT) increases compared to Si
devices. Recently a reduction in Ge concentration in the SiGe oxide was demonstrated
using an unspecified surface preparation method [3-5]. The study reports a clear
correlation between the total number of interface traps (NIT) and the Ge-oxide
concentration in the IL.This study focused on the reduction of Ge oxide concentration in
SiGe IL oxide. A wet-only approach and a combination of wet and dry approaches were
performed and the resulting Ge oxide concentration in the SiGe IL oxide was measured.
141
Downloaded on 2017-10-25 to IP 80.82.77.83 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
ECS Transactions, 80 (2) 141-146 (2017)
Experimental
Blanket Si0.75Ge0.25 wafers of 50-nm thickness were prepared by epitaxial growth
(ASM Epsilon). Ge atomic arrangement on the wafer surface was measured by Angle
Resolved X-ray photoelectron spectroscopy (AR-XPS, Theta300, Thermo Scientific).
The atomic arrangement on the solid surface was examined in detail by rotating the
sample at 22-78 degrees in the direction of the tilt angle (take-off angle) and measuring
the angle resolved spectrum of the photoelectron intensity. 4 elements (O1s, C1s, Si2p
and Ge3d) are taken into account for the calculation of atomic concentrations. The
angle-resolved Ge-oxide fraction was calculated from Si-oxide and Ge-oxide. Figure 1
summarizes the different methods used for the preparation of a set of different
SiGe-oxides.
Figure 1. Schematic description of the different steps used to prepare the different oxides
formed on Si0.75Ge0.25 evaluated in this study.
Characteristic of Native Oxide and Chemical Oxide
Native and chemical oxide were used to understand the Si0.75Ge0.25 aqueous
oxidation process. A native oxide surface was prepared by exposing the wafer to
cleanroom air for approximately one week. A chemical oxide was made on Si0.75Ge0.25
wafer by a sequence of an HF/HCl mixture and an O3-based aqueous oxidation (10ppm,
25˚C).
Effect of O3 and O3/HCl Process
O3 (aq) and O3/HCl (aq) (O3 water and HCl (0.1, 1, 10M)) was used to investigate
the impact of HCl on the Ge oxide concentration in the Si0.75Ge0.25 IL oxide.
Wet and Anneal Combination
Combinations of wet and anneal treatments were performed to investigate their
impact on the Ge oxide in Si0.75Ge0.25 IL oxide as indicated in Figure 1. The wet treatment
consisted of either a water or an HCl-spiked water rinse (0.01M).
All wet processing was performed using a single wafer wet cleaning tool (SU3000
or SU-3200, SCREEN Semiconductor Solutions) or batch wet tool (FC-821L, SCREEN
Semiconductor Solutions). The anneal process was performed on an epi tool (Epsilon,
ASM).
142
Downloaded on 2017-10-25 to IP 80.82.77.83 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
ECS Transactions, 80 (2) 141-146 (2017)
Results and Discussion
Characteristic of Native Oxide and Chemical Oxide
Figure 2 shows the angle-resolved Ge-oxide fraction of the native and chemical
oxide on Si0.75Ge0.25. The Ge-oxide concentration in the native oxide is, within the
experimental variation, identical to the Ge fraction in the strained Si0.75Ge0.25 layer. The
Ge oxide fraction of the chemical oxide is approximately half the Ge oxide fraction of the
native oxide. A reduction of GeO2 on Ge during a rinse process has been reported and
was attributed to the solubility of GeO2 [6]. Therefore, this reduction in Ge-oxide fraction
is attributed to the high water solubility of GeO2. To confirm the nature of Ge oxide
components, the GeO2 and GeO surface concentrations are compared. Figure 3 shows the
atomic concentrations of GeO2 and GeO in Ge oxide. GeO2 was reduced by
approximately 50% when comparing the aqueous O3-based chemical and native oxide.
The Ge-oxide content of the native and chemical oxide as determined by AR-XPS is
independent from the take-off angle.
Figure 2.
Ge oxide concentration in Si0.75Ge0.25 oxide
(a) Native oxide
(b) Chemical oxide
Figure 3. Atomic concentration of GeO2 and GeO on Si0.75Ge0.25 wafer (a) Native oxide,
(b) Chemical oxide by O3
143
Downloaded on 2017-10-25 to IP 80.82.77.83 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
ECS Transactions, 80 (2) 141-146 (2017)
Effect of HCl Process
Figure 4 shows that the Ge oxide concentration in Si0.75Ge0.25 oxide was slightly
reduced by the addition of HCl to the O3 solution. However, the concentration of HCl in
the mixture had no major effect in the range studied between 0.1 M and 10 M to further
reduce the Ge-oxide concentration.
Figure 4.
Ge oxide concentration in Si0.75Ge0.25 oxide
Wet and Anneal Combination
Since a wet-only process was insufficient to remove Ge oxide in SiGe oxide,
combinations of wet and anneal process steps were investigated. Figure 5 shows the
effect of varying process sequences on the Ge oxide concentration on Si1-xGex oxide. As
previously shown, a first wet process step reduced the Ge oxide by 50%, a subsequent
annealing step reduced the oxide an additional 50%, and a final HCl or rinse step reduced
the Ge-oxide fraction even lower. Figure 6 details the GeO2 and GeO atomic
concentrations after each process with XPS angle of 78˚. The anneal process reduced
both the fraction of GeO2 and GeO. The HCl and rinse processes after anneal removed
GeO2 completely. From the results in Figure 6, we assume that only GeO2 dissolves
easily during a wet process and the main fraction of GeO is being removed during the
anneal.
144
Downloaded on 2017-10-25 to IP 80.82.77.83 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
ECS Transactions, 80 (2) 141-146 (2017)
Figure 5. Ge oxide concentration in Si0.75Ge0.25 oxide with wet and anneal combination
process by XPS angle of 22˚and 78˚
Figure 6. Atomic concentration of GeO2 and GeO in Si0.75Ge0.25 oxide with XPS angle of
78˚
Conclusion
In order to obtain Ge-oxide-free oxide on Si0.75Ge0.25 a sequence of steps has been
developed. An aqueous O3/HCl process reduced the Ge-oxide fraction in the chemical
oxide by approximately 50% compared to its native oxide. An anneal process after
chemical oxide formation reduced the Ge-oxide content a further 50%. A wet water or
HCl-spiked water rinse after anneal resulted in an approximately Ge-oxide-free
145
Downloaded on 2017-10-25 to IP 80.82.77.83 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
ECS Transactions, 80 (2) 141-146 (2017)
Si0.75Ge0.25 oxide. Only a small fraction of GeO remains, a fraction that is close to the
noise level of the AR-XPS setup used. From these results, the best process sequence is
(1) reduction of Ge oxide by chemical oxidation, (2) anneal step to reduce both GeO2 and
GeO and (3) an water or HCl-spiked rinse step to make the layer GeO2 free.
Acknowledgments
We wish to thank Y. Yoshida, A. Iwasaki, H. Takahashi and J. Snow for many
kind supports and advices on experiment.
References
1.
2.
3.
4.
D. Guo, G. Karve, G. Tsutsui, K-Y Lim, R. Robison, et al.,VLSI Symp.Tech.Dig.,(2016)
J. Franco, B.Kaczer,et al., IEDM ,18.5.1-4 (2011)
C.H. Lee, H. Kim, P. Jamison et al.,VLSI Symp.Tech.Dig.,(2016)
S.L. Heslop, P. Engesser, H. Okorn-Schmidt and A.J. Muscat, ECS Transactions,69 (8)
287-293 (2015)
5. S.W. Park, T.K. Osborn, H. Kim, S. Siddiqui, B. Sahu, N. Yoshida, A. Brandt and A.C.
Kummel, J.Vac.Sci.Technol.A,Vol.33,No.4,Jul/Aug 2015
6. Y. Yoshida, H. Takahashi et al.,UCPSS.1012-3094,Vol.255,22-26(2016)
146
Downloaded on 2017-10-25 to IP 80.82.77.83 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
Документ
Категория
Без категории
Просмотров
0
Размер файла
207 Кб
Теги
0141ecst, 08002
1/--страниц
Пожаловаться на содержимое документа