Physica B 376–377 (2006) 420–423
Optical, electrical, and diffusion properties of hafnium and zirconium
in single-crystal silicon
R. Sachdeva
a,
, A.A. Istratov
a
, T. Radetic
b
, X. Xu
a,b
, P.N.K. Deenapanray
c
, E.R. Weber
a
a
Department of Materials Science and Engineering, University of California, Berkeley, CA 94720, USA
b
National Center for Electron Microscopy, Lawrence Berkeley National Labs, Berkeley, CA 94720, USA
c
Center for Sustainable Energy Systems, The Australian National University, Canberra 0200, Australia
Abstract
A study of optical, electrical, and diffusion properties of Hf and Zr in silicon is presented. Photoluminescence spectra were observed in
Hf-implanted silicon. Isotope substitution confirms that the observed signal is Hf related. Several deep-level defects were found for Hf in
both the upper and lower half of silicon band gap, and their parameters were tabulated. Dominant defect in deep-level spectra was
determined to be a donor. Diffusion study of Zr-implanted samples indicated that Zr tends to diffuse out to the surface. Outdiffusion and
precipitation of Zr which was found to form platelets, as confirmed by transmission electron microscopy, does not allow the traditional
diffusion models to be used; however, an estimate of Zr diffusivity was performed, yielding a value of 2.8
10
15
cm
2
/s at 1100
1
C.
r
2005 Elsevier B.V. All rights reserved.
PACS:
61.72.
y; 68.55.Ln; 71.55.
i
Keywords:
Silicon; Hafnium; Optical; Diffusion
1. Introduction
There has been great current interest in replacing silicon
gate oxide in CMOS transistors with high-k dielectics. Most
promising in terms of meeting the requirements and
improving the performance of Si devices appear to be ZrO
2
and HfO
2
[1,2]
. Introduction of these films into production
makes possible unintentional contamination of underlying
silicon with Zr and Hf. However, there is a lack of systematic
data on optical, electrical and diffusion properties of these
metals. In this paper, we address these properties.
2. Experimental details
CZ silicon wafers with a resistivity of 5–10
O
cm were
implanted with hafnium or zirconium at an energy of 50 or
1.82 MeV and dose of 10
13
cm
2
and annealed for various
times. Samples for optical measurements were annealed in
a horizontal furnace in Ar ambient in three steps; 650
1
C
for 30 min, 1000
1
C for 3 h, and then cooled to room
temperature. Photoluminescence (PL) measurements were
performed in a closed-cycle helium cryostat at 12 K unless
noted otherwise. The PL was excited by a cw argon-ion
laser (20 mW at the sample surface) with a wavelength of
514.5 nm and detected with a Ge detector cooled by liquid
nitrogen. Samples for electrical properties were annealed at
650
1
C for 30 min, then 1000
1
C for 1 h, and then cooled to
room temperature. Schottky diodes were prepared for both
n and p-Si. Deep-level transient spectroscopy (DLTS)
measurements were performed under reverse bias 5 V
reduced to 0 V during 100 ms filling pulses with emission
time constant of 8.6 ms. The sample was placed in a
cryostat sample holder which along with the liquid helium
dewar was used to run a temperature scan from 40 K to RT
and record the DLTS spectrum. Samples for diffusion
properties were annealed at 1100
1
C for 2, 24, and 240 h in
a vacuum furnace.
ARTICLE IN PRESS
www.elsevier.com/locate/physb
0921-4526/$ - see front matter
r
2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.physb.2005.12.108
Corresponding author. Tel.: +1 510 486 5569; fax: +1 510 486 4995.
E-mail address:
ravinder@berkeley.edu (R. Sachdeva).
3. Results and discussion
3.1. Optical properties of hafnium in silicon
shows PL spectra for three samples. Curve 1
(dotted line) represents PL spectrum for a Hf-implanted
sample annealed at 650
1
C for 30 min. Curve 2 (dashed line)
is PL signal for a sample that was annealed at 650
1
C for
30 min, 1000
1
C for 3 h, and rapidly quenched. Curve 3
(solid line) shows a sample which was annealed at 650
1
C
for 30 min, 1000
1
C for 3 h, and cooled slowly to room
temperature. It is clear that the 1000
1
C annealing step is
critical to the activation of the optical centers and that the
quenching process plays a significant role in determining
the PL spectrum. From high to low energy, these peaks are
labeled as
a
Hf
0
,
a
Hf
1
,
a
Hf
2
,
b
Hf, and
c
Hf. In this notation,
the left superscript indicates the line system and the right
superscript denotes the number of phonons involved. The
distances between
a
Hf
0
,
a
Hf
1
, and
a
Hf
2
are similar,
20.9
7
0.6 meV, and matches, within the error margins,
the 21.3
7
0.5 meV transverse acoustic (TA) phonon for
silicon
a
Hf
1
and
a
Hf
2
are likely to
be phonon replicas of
a
Hf
0
. Comparison of PL spectra of
samples implanted with two isotopes of Hf,
178
Hf and
180
Hf was made and an isotope shift of 0.1
7
0.02 meV was
observed, which confirmed the Hf-related origin of the PL
signal. Further details of PL studies of Hf in Si are
presented in Ref.
3.2. Energy levels of Hf in n-type crystals
shows DLTS spectrum of Hf in n-Si. The
parameters of the traps were determined from Arrhenius
plots presented in
be studied in sufficient detail because its overlap with the
peak N70, but the strong dependence of its position on
electric field in the depletion region (much stronger than that
of N70) indicates that it is likely to be a donor. For
comparison, we also plotted in
the data from the
study of Lemke
, performed on float zone crystal
contaminated with Hf during crystal growth. Energy levels
for defect N70, N84, N109, N146, N202, and N233 are
E
C
0.17,
E
C
0.2,
E
C
0.22,
E
C
0.27,
E
C
0.4,
and
E
C
0.44 eV, respectively. Capture cross-sections determined
by Arrhenius plots for these defects are 2.9
10
11
,
1.6
10
11
,
1.1
10
13
,
1.5
10
14
,
3.3
10
14
,
and
2.25
10
14
cm
2
, respectively. The capture cross-sections of
the trap labeled N146 was measured directly by varying the
DLTS filling pulse width and found to be 1.6
10
15
cm
2
.
Our level N146 matches well Lemke’s E2(Hf). N202 appears
close to Lemke’s
E
o
(Hf), N84 lies close to E1(Hf), and N109
lies close to E3(Hf), although the identification of these levels
with Lemke’s data is less certain. We did not observe in our
measurements a level which would lie close to Lemke’s
E
o
(Hf). A possible explanation for this is that, according to
Lemke
,
E
o
(Hf) appeared only when a Pd Schottky
contact was used, but not when Au contact was used, as was
the case in our study.
presents detailed analysis of electric field
dependence of carrier emission rate from the peak N146.
ARTICLE IN PRESS
Energy (eV)
0.70
0.75
0.80
0.85
0.90
0.95
1.00
PL Intensity (arb. units)
0.001
0.01
0.1
1
10
Do'stlaringiz bilan baham: |