Effect of cantilever–sample interaction on piezoelectric force microscopy Seungbum Hong

APPLIED PHYSICS LETTERS
VOLUME 80, NUMBER 8
25 FEBRUARY 2002
Effect of cantilever–sample interaction on piezoelectric force microscopy
Seungbum Honga) and Hyunjung Shin
Storage Lab, Samsung Advanced Institute of Technology, Suwon 440-600, Korea
Jungwon Woo and Kwangsoo No
Department of Materials Science and Engineering, Korea Advanced Institute of Science and Technology,
Daejon 305-701, Korea
共Received 9 October 2001; accepted for publication 17 December 2001兲
We report on the evidence for the cantilever–sample 共CS兲 capacitive force contribution to the
piezoelectric force microscopy 共PFM兲. In addition, we present that positioning of the tip near the
edge of the sample surface can significantly reduce this spurious contribution for any combinations
of tip cantilever and film. As proof of both the existence of CS interaction and its reduction, the
domains formed by the application of voltage pulses through the tip are observed by PFM at two
different positions, i.e., sample center and edge. In accordance with the model that a piezoresponse
consists of a piezoelectric vibration of the film and an electrostatic force induced vibration of
cantilever, the domain contrasts are characterized by dot structure in the amplitude and negligible
contrast in the phase images when the tip is placed in the center of the sample surface. However,
reducing the CS interaction by placing the tip near the sample edge yields domain contrasts showing
ring structure in the amplitude and a clear 180° phase shift in the phase images. Accompanying
resolution enhancement in phase images results in smaller size of domains 共bits兲 produced by
identical voltage pulses as is evidenced from bit size estimation. Additional evidence for reduction
of CS interaction is obtained from piezoresponse hysteresis measurement. © 2002 American
Institute of Physics. 关DOI: 10.1063/1.1454219兴
Ferroelectric domain imaging using contact mode atomic
force microscopy, often called piezoelectric force microscopy 共PFM兲, is widely used as a nondestructive probe to
observe the domain configuration and dynamics.1– 4 However, the probing mechanism and resulting domain contrasts
were not clearly understood due to the competitive contributions, to tip vibration signal, from electrostatic and piezoelectric forces between tip and sample, and capacitive force
between cantilever and sample.
Regarding the tip–sample interaction, many researchers
believe that piezoelectric vibration under the tip yields the
domain contrast.1– 4 However, it is clear that electrostatic
force effect exists and can be the dominant factor when converse piezoelectric effect is small.5 Regardless of the origin
of whether the domain contrast comes from either piezoelectricity or electrostatic force effect or their combination, there
is a consensus of the fact that the contrast represents the
ferroelectric domain, thereby enabling us to measure domain
size and determine projected polarization direction.
Many researchers calculated the cantilever–sample 共CS兲
capacitive force and its contribution to electrostatic force
microscopy,4,6,7 and concluded that it does not have a significant effect on the obtained domain contrast. This is true for
noncontact imaging when the tip–sample distance is small,
and for contact imaging when the tip height is large or the
spring constant of the cantilever is high. However, in case
one uses a low-aspect ratio tip with a soft cantilever, CS
interaction strongly interferes with the domain signal. Specific quantitative description of the aforementioned condition
can be obtained when the information on the sample piezoa兲
Electronic mail: [email protected]
electric coefficient, applied field, tip height, tip radius, spring
constant of the cantilever, and Young’s modulus of the
sample surface are available. For the combination of tips
共UL06A or B, Park Scientific Instruments兲 and films used in
our laboratory, it was clear that the CS interaction was nonnegligible, and often contributed to the blurred phase
images.8 Using a stiff cantilever can certainly reduce this
artifact, but there are some drawbacks. The tip can damage
the surface of the films by scratching. Also, there are some
materials that need a soft cantilever such as polymer ferroelectric films.
We previously explored the tip–sample and cantilever–
sample interactions to clarify each contribution and their
origins.8 In this letter, we present further evidence of CS
interaction affecting the PFM images and piezoresponse hysteresis measurement. Then, we present a method that can
effectively reduce the spurious effect from the cantilever and
can be applied to virtually any combinations of tip–
cantilever system and films.
We used a commercial atomic force microscope 共Autoprobe M5, Park Scientific Instruments兲 and a lock-in amplifier 共SR830, Standford Research Systems兲 for topography
and domain image acquisition. Detailed descriptions of the
equipment can be found in a previous publication.9 The tip
共UL06B兲 was positioned at the center 共‘‘C’’ position兲 and
near the edge 共E position兲 共⬃40 ␮m distance from the
sample edge兲 of the sample. 17 kHz ac voltage of 1 Vpp
共peak to peak兲 was applied to the tip while the bottom electrode was grounded. The scan frequency was 0.5 Hz, and the
scan size was 2.5 ␮m⫻2.5 ␮m.
Initially, the domains were oriented from bottom to top
by applying ⫹20 V to the bottom electrode over the scan
0003-6951/2002/80(8)/1453/3/$19.00
1453
© 2002 American Institute of Physics
Downloaded 12 Jan 2004 to 143.248.115.81. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/aplo/aplcr.jsp
1454
Hong et al.
Appl. Phys. Lett., Vol. 80, No. 8, 25 February 2002
FIG. 1. Topography 关共a兲 and 共d兲兴, amplitude 关共b兲 and 共e兲兴, and phase 关共c兲 and
共f兲兴 images of reversed domains 共bits兲 for C 关共a兲–共c兲兴 and E 关共d兲–共f兲兴 positions. The bits were formed by applying ⫺18 V pulses of 33, 8.25, 1.65. and
0.66 ms to bottom electrode. The white bars in 共a兲 and 共d兲 represent 500 nm.
area of 5⫻5 ␮m2 共which is referred to as back poling兲. After
the back poling process, series of ⫺18 V pulses with pulse
widths of 33, 8.25, 1.65, and 0.66 ms were applied to the
bottom electrode over the scan area of 2.5⫻2.5 ␮ m2 inside
the back poled area 共Fig. 1兲. Round-shaped bright dots 共top
to bottom oriented domains兲 were formed and observed in
amplitude 共A兲 for C position 关Fig. 1共b兲兴 and in phase 共␾兲
images for E position 关Fig. 1共f兲兴. Dark ring structure or dots
were found in an amplitude image for the E position 关Fig.
1共e兲兴. No distinct features were found in phase image for the
C position 关Fig. 1共c兲兴.
Our observation in Fig. 1 supports the CS capacitive
force contribution in C position to the tip vibration signal,
because it matches the situation where CS interaction acts as
an additional offset in piezoresponse. The following equation
describes the piezoresponse, A cos ⌽, in terms of tip–sample
and cantilever–sample interactions.
A cos ⌽
⫽
再
冋
冋
d 33⫺
册
2K⌫ lever
共 V dc⫹V c 兲 V ac
k lever
⫺d 33⫺
for
↓domain
2K⌫ lever
共 V dc⫹V c 兲 V ac for
k lever
↑domain,
册
共1兲
where all of the parameters are defined in Ref. 8. The first
term of each of the right-hand sides of Eq. 共1兲 is the piezoelectric contribution of the film, whereas the second term is
the capacitive contribution of the cantilever.
For domain imaging V dc equals 0, and the second term in
Eq. 共1兲 becomes positive in our case because ⌫ lever⬍0 and
V c ⫽0.3– 0.7 V for p-type Si tips and Pt bottom electrode.
FIG. 2. Expected piezoresponse 关共a兲 and 共d兲兴, amplitude 关共b兲 and 共e兲兴 and
phase 关共c兲 and 共f兲兴 signals and accompanying images of a reversed domain
for C 关共a兲–共c兲兴 and E 关共d兲–共f兲兴 positions.
Since the curves obtained by letting A cos ⌽⫽0 in Eq. 共1兲
serve as asymptotes for the hysteresis loop,8 it is clear that
CS interaction also leads to a horizontal shift of the loop.
This reasoning explains both the positive vertical shifts of
the piezoresponse in Fig. 2共a兲 and the constant phase value in
Fig. 2共c兲.
On the other hand, when CS interaction does not exist,
one can put ⌫ lever⫽0 so that the second term in the bracket
vanishes. Then, the piezoresponse represents the piezoelectric coefficient multiplied by V ac . This can explain the drop
in the amplitude signal at the domain boundary by competitive contribution of positive 共⫺d 33兲 and negative 共⫹d 33兲 domains in Fig. 2共e兲.
Figures 2共a兲–2共c兲 illustrate the resulting piezoresponse,
amplitude, phase signals and accompanying images when CS
capacitive force is present. Its contribution is marked as an
arrow in Fig. 2共a兲. Dot structure10 in amplitude 关cf. Figs. 1共b兲
and 2共b兲兴, and no significant features in phase 关cf. Figs. 1共c兲
and 2共c兲兴 images can be explained by Eq. 共1兲.
Positioning the tip near the edge of the sample 共E position兲 can significantly reduce CS interaction due to the small
capacitive coupling between the cantilever and bottom electrode. The expected piezoresponse, amplitude and phase signals, and accompanying images in the absence of CS interaction are depicted in Figs. 2共d兲–2共f兲. Ring structure in
amplitude 关Fig. 2共e兲兴 and dot structure in phase 关Fig. 2共f兲兴
images are expected for fully penetrating domains of opposite polarity to that of the matrix. This is in good agreement
with the results shown in Figs. 1共e兲 共white dotted area兲 and
Downloaded 12 Jan 2004 to 143.248.115.81. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/aplo/aplcr.jsp
Hong et al.
Appl. Phys. Lett., Vol. 80, No. 8, 25 February 2002
1455
FIG. 3. Dependence of bit size 共reversed domain size兲 on voltage pulse
width for C and E positions.
1共f兲. Detailed analysis of the evolution from ring to dot structure in amplitude image 关Fig. 1共e兲兴 is published in our previous publication,9 which is related to a different domain
boundary effect for fully and nonfully penetrating domains in
thickness direction.
The presence of CS interaction broadens the boundary
width between the domains of opposite polarity. It is clear
that the broadening effect is more prominent if the domain
size reduces. The broadening is expected to be proportional
to the areal ratio of the bit to the cantilever. Figure 3 was
obtained by measuring the diameter of dot-shaped domains
in Figs. 1共b兲 and 1共f兲. The diameter was measured from the
line profile of each opposite domain 共defined by ‘‘bit’’兲
where full width half maximum 共FWHM兲 was the criterion
for the size measurement. As can be seen in Fig. 3, the bit
size varied almost linearly with logarithmic value of pulse
width for both C and E positions, of which the origin can be
understood by the coercive field contour and its dependence
on pulse width.9 The bit size measured from Fig. 1共b兲
showed a larger value than that from Fig. 1共f兲. Also, the
difference between the bit sizes measured at different positions came closer to each other as the pulse width increased.
This phenomenon is likely to originate from the broadening
effect of CS interaction on the bit boundaries that has a decreasing tendency with bit size 共or pulse width兲. Also, it
strongly supports the argument that removing spurious CS
interaction can further enhance the resolution of domain images.
Furthermore, the piezoresponse hysteresis curves were
collected from position C and E as shown in Fig. 4. The
vertical shift of the loop, which is defined by
冋
册
p r,max⫹p r,min
2
⫻100共 % 兲 ,
共 p r,max⫺p r,min兲
共2兲
共where p r represents remnant piezoresponse兲 changed from
⫹55% to ⫺10% as the tip moved from C to E position. The
latter value is closer to the vertical shift 共⫺1.4%兲 of polarization electric field 共PE兲 hysteresis curve measured by
RT66A on Pt top electrode region.11 The horizontal shift 共defined in the same way as for vertical shift兲 changed from
FIG. 4. Piezoresponse hysteresis curve measurement for 共a兲 C and 共b兲 E
positions.
⫺57% to ⫹16%, which is in good agreement with⫹12% of
PE curve. The coercive voltages from PE measurement were
⫺1.5 and 2.4 V, which are in better agreement with those
from E position 共⫺1.2 and 2.3 V兲 than C position 共⫺1.6 and
⫺0.1 V兲. The slope of the curve 共defined as (p s,max
⫺ p s,min)/20, where p s is spontaneous piezoresponse兲, which
is proportional to ⌫ lever in Eq. 共1兲, reduced from 60 to 7. This
suggests that significant portion of CS interaction was removed by placing the tip near the edge of the sample.
Dr. C. J. Kim is gratefully acknowledged for supplying
the piezoelectric transducer sample.
A. Gruverman and M. Tanaka, J. Appl. Phys. 89, 1836 共2001兲.
E. Z. Luo, Z. Xie, J. B. Xu, I. H. Wilson, and L. H. Zhao, Phys. Rev. B 61,
203 共2000兲.
3
L. M. Eng, M. Abplanalop, and P. Guenter, Appl. Phys. A: Mater. Sci.
Process. 66, S679 共1998兲.
4
S. V. Kalinin, and D. A. Bonnell, Phys. Rev. B 63, 125411 共2001兲.
5
J. W. Hong, G. H. Noh, S. -I. Park, S. I. Kwun, and Z. G. Kim, Phys. Rev.
B 58, 5078 共1998兲.
6
T. Hochwitz, A. K. Henning, C. Levey, C. Daghlian, and J. Slinkman, J.
Vac. Sci. Technol. B 14, 457 共1996兲.
7
S. Belaidi, P. Girard, and G. Leveque, J. Appl. Phys. 81, 1023 共1997兲.
8
S. Hong, J. Woo, H. Shin, J. U. Jeon, Y. E. Pak, E. L. Colla, N. Setter, E.
Kim, and K. No, J. Appl. Phys. 89, 1377 共2001兲.
9
J. Woo, S. Hong, N. Setter, H. Shin, J.-U. Jeon, Y. E. Pak, and K. No, J.
Vac. Sci. Technol. B 19, 818 共2001兲.
10
P. Paruch, T. Tybell, and J. -M. Triscone, Appl. Phys. Lett. 79, 530 共2001兲.
11
However, the reason for decrease in absolute value of p r in the E position
down to 1/4 of that in the C position is not clearly understood. It is under
further investigation using high aspect ratio tip and will be presented in a
future publication.
1
2
Downloaded 12 Jan 2004 to 143.248.115.81. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/aplo/aplcr.jsp