Systematic study of the effect of La
2
O
3
incorporation on the flatband
voltage and Si band bending in the TiN/HfO
2
/SiO
2
/p-Si stack
Ming Di,
1
Eric Bersch,
1,a
Robert D. Clark,
2
Steven Consiglio,
2
Gert J. Leusink,
2
and
Alain C. Diebold
1
1
College of Nanoscale Science and Engineering (CNSE), University at Albany, 255 Fuller Road, Albany,
New York 12203, USA
2
TEL Technology Center, America, LLC, 255 Fuller Road Albany, New York 12203, USA
Received 11 August 2010; accepted 15 October 2010; published online 9 December 2010
Recent studies have shown that La
2
O
3
films can be used to adjust the threshold voltage V
t
of
NMOS Hf-based high-k/metal gate devices to desirable values, and a dipole at the high-k / SiO
2
interface has been proposed to explain the V
t
shifts. In order to investigate the mechanism of the V
t
shift further, we have measured the flatband voltage V
fb
and Si band bending of technologically
relevant TiN / HfO
2
/ La
2
O
3
/ SiO
2
/ p-Si stacks where the thickness and position of the La
2
O
3
layer
have been systematically varied. We observed systematic changes in V
fb
, Si band bending and the
HfO
2
-Si valence band offset as a function of La
2
O
3
layer thickness and position. These changes can
be explained by a band alignment model that includes a dipole at the high-k / SiO
2
interface, thus
supporting the work of previous authors. In addition, we have derived the theoretical relationship
between V
fb
and Si band bending, which agrees well with our experimental measurements. © 2010
American Institute of Physics. doi:10.1063/1.3516483
I. INTRODUCTION
In complementary metal oxide semiconductor CMOS
technology, traditional SiO
2
-based gate dielectrics are being
replaced by high-k dielectrics.
1
An outstanding issue for both
PMOS and NMOS stacks with Hf-based gate dielectrics in-
tegrated in a gate first approach, however, is high threshold
voltage V
t
.
2
This problem has been successfully addressed
by deposition of a so-called V
t
shift layer within the high-k/
metal gate stack. A dipole at the high-k / SiO
2
interface has
been shown to account for the change in V
t
.
3–8
La
2
O
3
is a leading material as a V
t
shift layer for high-
k/metal gate NMOS devices.
5,8
Using capacitance-voltage
C-V measurements, Kakushima et al., demonstrated that
increasing the thickness of La
2
O
3
in a
W / HfO
2
/ La
2
O
3
/ SiO
2
/ Si stack made the flatband voltage
V
fb
more negative, which corresponds to a decrease in V
t
for a NMOS device. In the same study, x-ray photoelectron
spectroscopy XPS measurements of the Si 1s core level in
W / HfO
2
/ La
2
O
3
/ SiO
2
/ Si stacks were reported. These mea-
surements showed that downward Si band bending increased
as La
2
O
3
thickness increased, in agreement with the V
fb
measurements.
8
In the work presented herein, we performed a similar
study to that of Kakushima et al., but with technologically
relevant TiN metal gates and chemically grown 8 Å SiO
2
layers. Moreover, the HfO
2
and La
2
O
3
ALD grown layers
were of device appropriate thicknesses 25–30 Å and 4–9 Å,
respectively and the position of the La
2
O
3
layer as well as
its thickness was varied. We observed clear trends in the V
fb
,
Si band bending and the HfO
2
-Si valence band offset with
changes in the thickness and position of the La
2
O
3
layer that
we account for using a band alignment model that includes
an interface dipole. In addition, we performed a comparison
between the experimentally determined V
fb
and Si band
bending relationship and the theoretically derived relation-
ship between these quantities. This comparison demonstrates
that Si band bending derived from XPS measurements is
predictive of V
fb
in high-k/metal gate capacitors.
II. EXPERIMENTAL
In this study, the film stacks consisted of chemically
grown SiO
2
layers, atomic layer deposition ALD grown
HfO
2
and La
2
O
3
layers, and chemical vapor deposition
CVD grown TiN layers on HF cleaned 300 mm p-Si
10
16
cm
-3
wafers as previously described.
9
To form the
chemical oxide the wafer was first cleaned with dilute HF
followed by treatment with sulfuric peroxide mix, then SC1,
then ozonized deionized water and finally a dilute HCl rinse.
The HfO
2
films were grown using tetrakis ethylmethylamido
hafnium and water at wafer temperature of 305 ° C.
10
The
La
2
O
3
films were grown using precursors of lanthanum
triN,N'-diisopropylformamidinate and oxygen at a wafer
temperature of 305 ° C.
11
The C-V measurements were per-
formed on MOS capacitor MOSCAP stacks with 500 Å
thick TiN layers, and XPS measurements were performed on
identically grown film stacks with 30 Å TiN layers. The
highest temperature the stacks were exposed to was 500 ° C
during TiN deposition. The MOSCAP stacks were also sub-
jected to a 30 min forming gas anneal 4% H
2
at 400 °C.
The XPS measured samples were not annealed after TiN
deposition.
The C-V measurements were performed using an Agilent
Technologies 4284A LCR meter on 40 40 m
2
capacitors.
The V
fb
values were extracted using the method of Hauser
and Ahmed.
12
XPS measurements were performed on blan-
ket films using a Thermo Fisher lab tool, with a monochro-
a
Electronic mail: ebersch@uamail.albany.edu.
JOURNAL OF APPLIED PHYSICS 108, 114107 2010
0021-8979/2010/10811/114107/7/$30.00 © 2010 American Institute of Physics 108, 114107-1