Current Voltage Characteristics through Grains and Grain Boundaries of
Highk Dielectric Thin Films Measured by Tunneling Atomic Force
Microscopy
Katsuhisa Murakami, Mathias Rommel, Vasil Yanev, Anton J. Bauer, and Lothar Frey
Citation: AIP Conf. Proc. 1395, 134 (2011); doi: 10.1063/1.3657879
View online: http://dx.doi.org/10.1063/1.3657879
View Table of Contents: http://proceedings.aip.org/dbt/dbt.jsp?KEY=APCPCS&Volume=1395&Issue=1
Published by the American Institute of Physics.
Related Articles
Intrinsic effect of porosity on mechanical and fracture properties of nanoporous ultralow-k dielectrics
Appl. Phys. Lett. 101, 123109 (2012)
Low-κ organic layer as a top gate dielectric for graphene field effect transistors
Appl. Phys. Lett. 100, 193117 (2012)
Time-dependent dielectric breakdown of plasma-exposed porous organosilicate glass
Appl. Phys. Lett. 100, 112905 (2012)
H2/N2 plasma damage on porous dielectric SiOCH film evaluated by in situ film characterization and plasma
diagnostics
J. Appl. Phys. 109, 084112 (2011)
Effect of bake/cure temperature of an advanced organic ultra low-k material on the interface adhesion strength to
metal barriers
J. Appl. Phys. 109, 074301 (2011)
Additional information on AIP Conf. Proc.
Journal Homepage: http://proceedings.aip.org/
Journal Information: http://proceedings.aip.org/about/about_the_proceedings
Top downloads: http://proceedings.aip.org/dbt/most_downloaded.jsp?KEY=APCPCS
Information for Authors: http://proceedings.aip.org/authors/information_for_authors
Downloaded 06 Nov 2012 to 131.188.201.33. Redistribution subject to AIP license or copyright; see http://proceedings.aip.org/about/rights_permissions
Current Voltage Characteristics through Grains and Grain
Boundaries of High-k Dielectric Thin Films Measured by
Tunneling Atomic Force Microscopy
Katsuhisa Murakami
1
, Mathias Rommel
2
, Vasil Yanev
2
,
Anton J. Bauer
2
, and Lothar Frey
1, 2
1
Chair of Electron Devices, University of Erlangen-Nuremberg, Cauerstrasse 6, 91058 Erlangen, Germany
2
Fraunhofer Institute for Integrated Systems and Device Technology (IISB), Schottkystrasse 10, 91058 Erlangen,
Germany
Abstract. Leakage current distributions of high-k dielectric thin films (8 nm HfSi
x
O
y
and 5 nm ZrO
2
) were measured by
tunneling atomic force microscopy (TUNA). In contrast to the thick HfSi
x
O
y
film, where grains and grain boundaries
can be seen by TUNA topography maps, ZrO
2
films show no topography roughness. But large leakage current
fluctuations can be seen for both dielectrics by TUNA current images. Higher leakage currents were found to flow
through the grain boundaries of both analyzed high-k dielectric films. Furthermore, local current voltage (I-V)
characteristics could successfully be measured precisely localized at grains and at grain boundaries, respectively, of the
ZrO
2
film. The obtained local I-V curves showed significant differences between grains and grain boundaries,
respectively. In the case of the ZrO
2
film, the leakage current through the grain boundaries was up to 8 times larger than
that through the grains at a substrate voltage of -3.5 V.
Keywords: Tunneling AFM (TUNA), high-k, grain and grain boundary.
PACS: 77.55.D-
INTRODUCTION
The strong interest in high-k dielectric films is
triggered by the applications of these materials as new
gate oxides for metal insulator semiconductor (MIS)
devices
1
and as capacitor materials for memory
devices.
2
Required electrical properties of high-k
dielectric films for such applications are low leakage
currents and high dielectric constants. The electrical
properties (i.e., leakage current and dielectric constant)
of conventional dielectric films based on SiO
2
are
usually characterized and evaluated by macroscopic
measurements on MOS capacitors. But, in contrast to
SiO
2
, it is difficult to fully analyze and evaluate the
electrical properties of high-k materials using
conventional macroscopic measurements, since
surface morphology and electrical properties of high-k
materials are usually inhomogeneous on nanoscale.
This is mainly due to their nanocrystalline structure
after typical fabrication procedures. It is, therefore,
important and necessary to investigate the surface
morphology and the leakage current of high-k
dielectric films with very high lateral resolution.
Knowledge obtained from those investigations would
be extremely helpful to fabricate optimized high-k
dielectric films with low leakage current.
The tunneling atomic force microscopy (TUNA)
technique is well suited to this demand.
3
The TUNA
technique allows to simultaneously measure the local
leakage current and the surface morphology of high-k
dielectric films with extremely high current sensitivity
(sensitivity limit is approximately 40 fA) and very
high lateral resolution (< 20 nm), respectively. In our
previous study, the surface morphology and leakage
current distribution of high-k dielectric films (HfSi
x
O
y
and ZrO
2
) has been simultaneously investigated at
nanoscale by TUNA.
4
It was found that the high
leakage current paths through high-k dielectric films
were located at the grain boundaries. Possible reasons
for the larger leakage current through grain boundaries
are different current conduction mechanisms for grains
and grain boundaries or the minor thickness of grain
boundaries compared to that of grains. So far,
however, the mechanisms for the leakage current
conductions of high-k dielectric films through grains
and grain boundaries has not been investigated in
detail and has not been fully understood.
Frontiers of Characterization and Metrology for Nanoelectronics: 2011
AIP Conf. Proc. 1395, 134-138 (2011); doi: 10.1063/1.3657879
© 2011 American Institute of Physics 978-0-7354-0965-1/$30.00
134
Downloaded 06 Nov 2012 to 131.188.201.33. Redistribution subject to AIP license or copyright; see http://proceedings.aip.org/about/rights_permissions
The purpose of this paper is to measure local
current-voltage (I-V) characteristics precisely at grains
and grain boundaries for HfSi
x
O
y
and ZrO
2
high-k
dielectric films by TUNA in order to clarify the
current conduction difference between the grains and
grain boundaries of high-k dielectric films.
EXPERIMENTAL PROCEDURES
Both, HfSi
x
O
y
and ZrO
2
high-k films (with
thicknesses of about 8 and 5 nm, respectively) were
deposited by atomic layer deposition on native oxide
(thickness of 1.1 nm) on p-type silicon wafers.
Subsequently, rapid thermal annealing (RTA) was
performed for HfSi
x
O
y
at a temperature of about
1000
o
C for 60 s to crystallize the films. For ZrO
2
films, RTA was performed in Ar atmosphere at 450
o
C
for 30 s.
Topography maps, leakage current maps, and local
I-V curves were measured with a Bruker Dimension
ICON tool equipped with NanoSope V controller
(closed-loop system), an extended TUNA module
(sensitivity 1pA/V), and Pt/Ir coated silicon tips (the
nominal tip radius of Pt/Ir coated probes is less than 25
nm). Within this work, substrate voltages from -10 V
to +10 V can be applied to the sample. In all TUNA
measurements, the substrate was negatively biased to
avoid anodic oxidation of the sample surface in air.
For negative substrate bias, the MIS structure formed
by the conductive tip and the high-k dielectrics on p-Si
substrate is in inversion, resulting in electrons being
injected from the Si substrate. For all local I-V
measurements in this study, the substrate bias was
ramped from 0V to -10V. For all measurement (i.e.,
topography map, TUNA current map, and local I-V
measurement) in this study the probe is in contact with
the surface of the high-k layer in air. The TUNA
experimental parameters for HfSi
x
O
y
and ZrO
2
are
summarized in Table 1. Deflection setpoint is the
parameter related to the contact force between the
probe and sample during TUNA measurements (both,
current maps and local I-V measurements). The
contact force increases with increasing the positive
value of the deflection set point.
Table 1. Parameters of TUNA current maps and local I-V
measurements for HfSi
x
O
y
and ZrO
2
film.
HfSi O
x y
ZrO
2
Substrate voltage for TUNA current map (V)
-9.1
-2.7
Tip velocity for TUNA current map ( m/s)m
1.50
1.50
Ramp rate for local measurement (V/s)IV
0.50
0.25
Deflection setpoint (V)
-0.90
2.00
Additionally, topography map measurements and
roughness analyses were carried out by tapping mode
atomic force microscopy (AFM). High-resolution
transmission electron microscopy (HR-TEM) images
were obtained by a FEI CM300 to analyze the film
structure and morphology of the ZrO
2
films.
RESULTS AND DISCUSSION
Figure 1 shows a typical topography and
corresponding current map of the 8.0 nm thick
HfSi
x
O
y
film measured by TUNA. Crystalline grains
can clearly be observed in the topography map. The
height of the film at the grain boundaries is lower than
that of the grains. The edges of the grains seem to be
protuberant. Therefore, grains also exhibit local height
variations. In the TUNA current map, the conductive
structures (dark color), corresponding to the grain
boundaries in the topography map, are clearly visible
which indicate larger currents. In addition, two types
of the grain boundaries are found to exist in this
sample. One is a grain boundary with high leakage
current (grain boundary A). Another is a grain
boundary with low leakage current (grain boundary B)
2 nm
0 pA
-1.2 pA
Grain boundary A
Grain boundary B
Figure 1. (a) TUNA topography map and (b) corresponding
TUNA current map of the HfSi
x
O
y
film. Scan size: 2x0.5
µm
2
.
Figure 2 shows corresponding local I-V curves for
the grains and grain boundaries of the HfSi
x
O
y
film.
The local positions for the I-V curves through grains
and grain boundaries were deduced from the TUNA
topography map only. A difference between I-V curves
through grains and grain boundaries can be observed.
135
Downloaded 06 Nov 2012 to 131.188.201.33. Redistribution subject to AIP license or copyright; see http://proceedings.aip.org/about/rights_permissions
The distribution of the local I-V curves through grains
has a wide spread, while that through grain boundaries
shows two tight spreads (i.e., spread A and B). The
reason for the wide spread in the local I-V curves
through grains is not apparent. Further measurements
(e.g., HR-TEM analysis of the cross sectioned sample)
are needed for clarification. The reasons for the two
spreads in the local I-V curves through grain
boundaries are also not fully understood yet but might
be due to different current conduction mechanisms for
two types of grain boundaries (i.e., grain boundary A
and grain boundary B). Another reason might be a
slight error in the measurement positions, since – in
contrast to the measurements on ZrO
2
shown
afterwards – the location of the measurement positions
of the local I-V curves were not confirmed by
subsequent TUNA current maps. In any case it should
be emphasized that it can be observed that different
types of grain boundaries as determined from the
topography map show also very different current
levels in the TUNA current map. Most important,
leakage currents through grain boundaries are found to
start at lower voltages for spread A. However, in the
case of HfSi
x
O
y
films, the effect of the thickness
variations on the I-V characteristics difference between
grains and grain boundaries cannot be neglected.
Namely, the reason for the larger leakage current at
grain boundaries of spread A might be just due to a
thinner thickness of the dielectric layer at the grain
boundaries compared to that of the grains. It is,
therefore, difficult to clarify the origin of the
difference for the I-V characteristics between grains
and grain boundaries for the given HfSi
x
O
y
films.
-4 -5 -6 -7 -8 -9
1E-13
1E-12
1E-11
Grain
Grain boundary
Curren t (A)
Substrate voltage (V)
A
B
sensor limit (12.3 pA)
Figure 2. Local I-V curves through grains and grain
boundaries of the HfSi
x
O
y
film measured by TUNA.
In contrast to the HfSi
x
O
y
thin film, it was found
that no crystal grains could be observed for the given
ZrO
2
film by tapping mode AFM (see Fig. 3). The
surface of the ZrO
2
film was rather smooth. The
average roughness and root mean square (RMS)
roughness of the ZrO
2
surface was 0.12 nm and 0.18
nm, respectively. In order to confirm the morphology
of the ZrO
2
film, HR-TEM analysis of the cross
sectioned sample was performed in a previous study
4
,
as shown in Fig. 4. The HR-TEM image of the ZrO
2
clearly showed the nanocrystalline structure. It is also
visible that the ZrO
2
surface and the underlying silicon
oxide surface of the samples are very smooth, which
confirm the results from the tapping mode AFM
topography. Therefore, TUNA current maps will not
be influenced by any artifacts arising potentially from
topography (e.g., increased contact area at the grain
boundaries or thickness variations).
0 nm
Figure 3. Topography map of the ZrO
2
film measured by
tapping mode AFM. Scan size: 500 x 500 nm
2
.
crystalline
Figure 4. HR-TEM image of ZrO
2
film annealed at 450 °C
for 30 s (This image was reported in Ref. 4).
Figure 5 shows TUNA current maps of the ZrO
2
thin film before and after local I-V measurements at a
grain and a grain boundary, respectively. Figure 5 (a)
shows the initial TUNA current map of the ZrO
2
film
before any local I-V measurement. A structure of areas
with different current densities which will be attributed
to grains and grain boundaries of the ZrO
2
film can
clearly be observed in the TUNA current maps. The
larger leakage current paths are located at the grain
boundaries of the ZrO
2
film, although the surface of
the ZrO
2
is rather smooth compared to the HfSi
x
O
y
film. Those findings in our previous study
4
could be
confirmed using the closed-loop system. This result
indicates that the main reason for the larger leakage
current at grain boundaries for this film is not due to a
thickness variation of the layer.
Figure 5 (b) shows the TUNA current map after the
first local I-V measurement at a grain boundary. The
measured spot (spot 1) showed very high leakage
current with an oscillating noise along the scan
direction of the cantilever due to the breakdown of the
136
Downloaded 06 Nov 2012 to 131.188.201.33. Redistribution subject to AIP license or copyright; see http://proceedings.aip.org/about/rights_permissions
measured spot. Interestingly, another weak spot at a
different grain boundary (spot 2) was observed in the
second TUNA current map which already showed
rather high leakage current in the initial TUNA current
map (Fig. 1(a)). Probably the electrical stress applied
during the initial mapping led to a dielectric
breakdown of this spot.
Figure 5 (c) shows the TUNA current map after the
second local I-V measurement through a grain. The
measured spot at the center of the grain (spot 3) also
showed very high leakage current similar to the
measured spot at the grain boundary (spot 1). The
measured points were exactly at the grain and grain
boundary of the ZrO
2
films, respectively (compare the
TUNA current maps before and after the local I-V
measurements).
Spot 2 (weak spot)
Spot 3 (Grain)
0 pA
-0.5 pA
Figure 5. TUNA current maps of ZrO
2
films (a) before any
local I-V measurement, (b) after the first local I-V
measurement at a grain boundary, (c) after the second local
I-V measurement at a grain. Scan size: 1.33x 0.5 µm
2
.
For our setup, the measurement positions for the
local I-V curves could be very accurately set into the
grains or the grain boundaries without any image drift
by the closed-loop system. An additional remark
which should be made here is that the resolution of the
TUNA current map is not deteriorated after the local I-
V measurements. This result indicates that the shape of
the tip apex for the conductive tip is practically
unchanged by the local I-V measurements. Therefore,
the I-V characteristics at grains and grain boundaries
can be compared neglecting the tip shape degradation
during the measurements. This measurement cycle
(i.e., TUNA current map → local I-V at grain
boundary → TUNA current map → local I-V at grain)
was repeated 10 times using the identical experimental
parameters (i.e., the same deflection set point, scan
rate, substrate voltage, and voltage ramp rate for local
I-V measurements) summarized in Table 1.
Figure 6 shows the local I-V curves for the ZrO
2
film at grains and grain boundaries, and the TUNA
current map after the fifth cycle of the measurement,
respectively. The local I-V curves show significant
differences between curves at grains and grain
boundaries, respectively. The leakage current through
the grain boundaries exceeds the noise level and
reaches the sensor limit for the TUNA module at lower
substrate voltages, than the current through the grains.
The leakage current through the grain boundaries is up
to 8 times larger than that through the grains at a
substrate voltage of -3.5 V. In addition, the TUNA
current map in Fig. 6 (b) shows that the measured
spots are accurately located at grains and grain
boundaries. This result ensures the reproducibility and
reliability of the local I-V curves. Namely, the local I-
V curves could be exactly measured at grains and grain
boundaries. Future work will focus on the detailed
evaluation of the current conduction mechanisms
through the grains and grain boundaries.
0 -1 -2 -3 -4 -5
1E-13
1E-12
1E-11
Grain
Grain boundary
Curren t (A)
Substrate voltage (V)
sensor limit (12.3 pA)
(a)
forward voltage ram p
reverse voltage ram p
(i.e., after breakdown)
0 pA
Figure 6. (a) Local I-V curves for the ZrO
2
film at grains and
grain boundaries. (b) TUNA current map after several local
I-V measurements at grains and grain boundaries. Scan size:
2 x 0.5 µm
2
.
137
Downloaded 06 Nov 2012 to 131.188.201.33. Redistribution subject to AIP license or copyright; see http://proceedings.aip.org/about/rights_permissions
