Sensors and Actuators A 104 (2003) 162–170

Fabrication of a micro-scale, indium-tin-oxide thin film strain-sensor

by pulsed laser deposition and focused ion beam machining
Timothy M. Millera, Hui Fangb, Robert H. Magruder IIIc,d, Robert A. Wellerb,d,*
a
Department of Mechanical Engineering, Vanderbilt University, Nashville, TN 37235, USA
b
Department of Physics and Astronomy, Vanderbilt University, Nashville, TN 37235, USA
c
Department of Physics, Belmont University, Nashville, TN 37212, USA
d
Department of Electrical Engineering and Computer Science, Vanderbilt University, P.O. Box 1824 Station B, Nashville, TN 37235, USA
Received 1 October 2002; received in revised form 15 January 2003; accepted 19 January 2003

Abstract

A thin film strain-sensor having an active sensing area of 20 mm
100 mm has been developed using a combination of pulsed laser
deposition (PLD) and focused ion beam (FIB) machining to net shape. Sensors were made of indium-tin-oxide (ITO) thin films in the range
100–246 nm thick. Strain sensitivities of large-scale (smallest dimension ¼ 0:8 mm), small-scale (smallest dimension ¼ 0:25 mm) and FIB
machined gages (smallest dimension ¼ 20 mm) are reported. Large-scale devices were deposited in a background oxygen pressure in the
range 28–50 mTorr, while small-scale and FIB machined gages were deposited in 50 mTorr of oxygen, the largest achievable pressure in our
system, which also yielded the largest strain sensitivities. Active strain-sensors were produced with gage factors ranging from 0.3 to 8.7
and showed a linear room temperature response with minimal hysteresis. The effects of imaging and machining with the FIB as well as the
effects of an SiO2 encapsulation layer on the electrical properties of the gages are reported.
# 2003 Elsevier Science B.V. All rights reserved.

Keywords: Thin film; Strain-sensor; Indium-tin-oxide; Pulsed laser deposition; Focused ion beam; MEMS

1. Introduction microsystem components. This paper addresses the electri-

cal strain response of directly deposited thin film sensing
Goldfarb and Celanovic have described a flexure-based layers, the effects of encapsulation layers, and the effects
gripper for micromanipulation [1]. For a tool such as this, resulting from ion irradiation.
designed to handle objects in the size range from 5 to Directly deposited thin film strain gages offer important
100 mm, monitoring the gripping force is a necessity. A advantages over conventional gages, including unobtrusive
technology that adds sensing capabilities to micro-scale operation, negligible mass and profile, low current opera-
systems will have to surpass the limitations of conventional tion, and the potential for production in complex configura-
strain gages. One possibility uses direct deposition of a tions with little added effort. In addition, the full range of
sensing material and has the advantage of avoiding adhe- processes and techniques of thin film science can be brought
sives and manual manipulation. Subsequent micro-machin- to bear for both manufacture and characterization of the
ing of gage structures has the advantage of tailoring a sensor resulting devices. Thus, structures can be tailored for spe-
for a specific application. The addition of tactile feedback by cific applications, including, for example, layers of electrical
the integration of micro-sensors is crucial to realizing the or thermal insulation, while avoiding complex assembly
full potential of microsystem components. steps involving delicate parts placement and adhesive appli-
A functional sensor suitable for placement on metallic cation.
components must include an electrical insulation layer, a While direct deposition of metallic strain gages is possi-
sensing layer, metallic interconnects, and a protective encap- ble, semiconductor materials have greater strain sensitivities
sulating layer. Thin film technology is the most promising while also being easily deposited in thin films. Although
means by which the necessary layers can be added to silicon is the most commonly used semiconductor material
for strain sensing applications, processing issues make it
*
Corresponding author. Tel.: þ1-615-343-6027; fax: þ1-615-343-6614. undesirable for deposited gage applications. An alternative
E-mail address: robert.a.weller@vanderbilt.edu (R.A. Weller). semiconductor material, indium-tin-oxide (ITO), has been

0924-4247/03/$ – see front matter # 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0924-4247(03)00051-7
 T.M. Miller et al. / Sensors and Actuators A 104 (2003) 162–170 163

reported to exhibit large strain sensitivities [2,3] while also are typically in the range 5 to 175 [13], while those of metal
being readily prepared as a thin film by several processing foil sensors are positive and typically in the range from about
techniques [4]. 2 up to approximately 6 [14]. The origin of piezoresistivity
In this paper, we describe a novel two-step approach, in crystalline semiconductors such as Ge and Si has been
combining pulsed laser deposition of ITO thin films and discussed by various authors [15]. However, in ITO the
subsequent focused ion beam machining, to develop micro- mechanism is unknown. Recently ITO has also been demon-
scale sensors for micro-systems applications. strated to function as a sensor at high temperatures [16]. In
[16], Gregory and Luo report the behavior of ITO strain gages
prepared by magnetron sputtering and used together with
2. Background temperature compensating platinum resistors to achieve a
near zero temperature coefficient of resistance (TCR) over
The responsiveness of a strain-sensor is parameterized by the range 25–1200 8C and operation up to 1450 8C.
a quantity called the gage factor, g, defined in Eq. (1) as the Pulsed laser deposition (PLD) of ITO has several advan-
fractional resistance change per unit strain. In this equation, tages over other thin film deposition techniques including
R is the initial resistance of the gage structure and DR is the the ability to control film stoichiometry through a combina-
change in resistance accompanying a strain e. tion of laser target stoichiometry and background pressure,
and the ability to grow films at room temperature [4,10]. In
DR 1
g (1) addition, ITO is relatively impervious to deleterious effects
R e
of exposure to an oxygen environment, even at elevated
For semiconductor gages, it is useful to elaborate upon temperatures. ITO solid solutions are stable in pure oxygen
Eq. (1) by separating the dependence of R on the geometry of up to 1500 8C and do not undergo any disadvantageous
the gage and the intrinsic resistivity r of the material used to phase transformations up to their decomposition tempera-
make it. The result of this separation is Eq. (2), where ture [16]. In this publication we confine the discussion to
Poisson’s ratio n quantifies the contribution of geometric gage structures made by pulsed laser deposition and focus on
deformation to resistance change, and the effect the variation issues related to making and characterizing gages as small as
in intrinsic resistivity Dr/r is expressed explicitly. 20 mm
100 mm.
Dr 1
g ¼ 1 þ 2n þ (2)
r e
3. Experiment details
The extrinsic geometric effects parameterized by n dominate
in metal foil strain gages, and since most metals have n equal ITO thin films were deposited by pulsed laser ablation.
to 0.2–0.5, the gage factors for most metals are between 1.4 The laser target consisted of a pressed powder alloy of
and 2. In semiconductor gages, geometric effects are mini- indium oxide (90% In2O3) and tin oxide (10% SnO2), having
mal and often ignored. The variability of intrinsic resistivity a diameter of 7.6 cm at a distance of 4 cm from the substrate.
with mechanical stress in an elastically strained material is A schematic of the apparatus is shown in Fig. 1. A krypton–
called piezoresistivity [5]. fluoride excimer laser (Lambda Physik Compex 205) pro-
Indium-tin-oxide (ITO) is known primarily as a transpar- ducing 248 nm pulses at 10 Hz with a 25 ns pulse duration
ent conductor and is the most widely used material with this and energy of 380 mJ/pulse yielded a fluence of 0.13 J/
unusual property [6]. ITO is almost always used commer- (cm2 pulse). Prior to all depositions the chamber was evac-
cially as a thin film and has been prepared in this form by uated to a pressure of approximately 6
106 Torr and the
evaporation, spray pyrolysis, magnetron sputtering, chemi- target was cleaned with 5000 laser pulses.
cal vapor deposition, and pulsed laser deposition [4]. It is an Depositions for large-scale devices were performed at
n-type, degenerately doped, wide band gap semiconductor oxygen partial pressures ranging from 28 to 50 mTorr using
whose plasma frequency is just below the visible range of the 99.999% oxygen delivered by a mass flow controller. Small-
spectrum and whose inter-band absorption edge is just above scale devices were deposited at 50 mTorr, the highest
it. Charge carriers are thought to be provided by tin atoms in achievable level in our system. The number of laser pulses
indium sites acting as donors and also by free electrons from was adjusted to produce film thicknesses in the range 100–
oxygen vacancies. 300 nm.
ITO is used as a transparent electrode in such devices as Substrates were Corning glass 2935 with dimensions of
organic light-emitting diodes [7], touch screen displays [8], 50 mm
24 mm
0:16 mm. Prior to deposition, substrates
flat panel displays [9], and lamps [4]. While ITO has been were cleaned in a series of sonic baths of acetone and
extensively studied to optimize electrical conductivity and methanol, rinsed in distilled water, and finally blown dry
transparency [10–12], it has only recently been reported to with compressed dry N2. Film thicknesses were measured
exhibit strain sensitivities characterized by gage factors with a stylus profilometer (Dektak II).
reported in the range from 2 to 77 [2,3]. By comparison, The substrate containing the active gage element was
the strain sensitivities of semiconductor silicon strain gages strained in a four-point-bending configuration from 0 to 200
164 T.M. Miller et al. / Sensors and Actuators A 104 (2003) 162–170

Fig. 1. Experimental set-up for pulsed laser deposition (PLD).

Fig. 2. Four-point bending apparatus used to induce a constant strain field in glass substrates between the inner pins. The device is actuated by a voice coil,
which is driven by a precision digital power supply.

micro-strains (Fig. 2). This strain range is lower than that of The 50–70 nm chromium adhesion layer deposited by ther-
some commercial gauges, but encompasses the full range mal evaporation was used to improve electrical stability and
anticipated to be important for the microsystems applica- contact resistance. The entire fixture was placed in a Faraday
tions that originally motivated this work. A voice coil, cage during I–V characterization, which was performed using
powered by a digitally controlled constant current source a Keithley 2001 7 1/2 digit multimeter for voltage measure-
(HP-E3631A) induced strain in the substrate. The strain ment and a Keithley 236 constant current source.
may be calculated from simple geometry, e ¼ 2dt/(d2 þ d2 ), The smallest gage structures were machined from larger
where t is the half-thickness of the substrate, d the deflection ones using a model 200
P focused ion beam from FEI, Co.
and 2d is the inner pin spacing. The apparatus was calibrated Initially, gages were cut to 100 mm
780 mm, and then scaled
using a commercial metal foil gage mounted conventionally down to 100 mm
400 mm, and finally 20 mm
100 mm.
with adhesives. Imaging and cutting operations with the focused ion beam
The gage patterns were designed for experimental deter- (FIB) used 30 keV Gaþ ions at selectable current increments
mination of strain sensitivity by means of a four-wire resis- from 1 pA to 11.5 nA. Beam time and current were con-
tance measurement (Fig. 3). They were formed using a trolled during experiments to determine the radiation effects
0.13 mm thick stainless steel shadow mask, shown in of imaging the ITO with the FIB. With the exception of
Fig. 4. Corresponding masks defined the locations for chro- focusing, imaging the ITO films in the FIB was typically
mium metallization of the contact pads. Gage structures were done using a single raster scan of the area of interest at a beam
designed to have an active area of 790 mm
4:8 mm for large- current of 11, 70 or 150 pA. Cutting operations used currents
scale devices and 250 mm
1:5 mm for small-scale devices. of either 11.5 nA or 150 pA. Sections of ITO were electri-
Large-scale devices had 2:38 mm
2:78 mm contact pads cally isolated by making open box cuts that overlapped the
while small-scale devices had 760 mm
890 mm contact edges of the film. This method reduced etching time sig-
pads. Electrical contact was made using 25.4 mm thick nificantly as large areas of the active material did not have to
gold wire and H20E conductive epoxy from Epotek, Inc. be removed by sputtering, but were simply isolated from the
 T.M. Miller et al. / Sensors and Actuators A 104 (2003) 162–170 165

Fig. 3. Schematic of the test pattern used in electrical measurement showing the nominal dimensions for large- and small-scale gages. The central strip is
aligned with the direction of axial strain in the four-point bend.

circuit electrically. Subsequent I–V characterization and gage estimated to be 26 nm with range straggling of 6 nm. (Simi-
factor measurements were performed as outlined above. larly, the range and straggling of 30 keV Gaþ in ITO are about
Prior to FIB machining, an SiO2 encapsulation layer was 13  6 nm.) The thickness of the SiO2 layer was verified by
deposited by PLD onto several ITO gages. The SiO2 layer was ellipsometry using a Si substrate. After the ITO was encap-
deposited using an SiO target and a background pressure sulated, the resistance and gage factor were re-measured.
of 40 mTorr of oxygen. This layer was deposited with a target-
to-substrate distance of 6.5 cm, a height that required 1500
laser pulses to produce an SiO2 film 100 nm thick, or about 4. Results and discussion
4 times the mean range of 30 keV Gaþ ions in SiO2, a value
Typically, the active gaging area for the films had a
resistance of several thousand ohms. Large-scale gages
had a range of resistance between 160 O and 600 kO, while
small-scale gages had an average resistance of 55 kO.
Commercial semiconductor strain gages typically have
resistances of approximately 1 kO, while metal foil gages
are either 120 or 350 O. Thus, ITO gages are potentially very
attractive for low current, low power applications.
Fig. 5 shows gage factors and resistivities as a function
of oxygen pressure during deposition for the large-scale
devices, which were used in a series of experiments designed
to optimize the strain sensitivity. Large-scale gages had a
thickness ranging from 100 to 246 nm with an average
standard deviation of 30 nm for three independent films
deposited at a given pressure. Though not shown graphically,
Fig. 4. A stainless steel shadow mask containing several small-scale gage
the thickness of these films did not show any apparent
patterns used to deposit thin film gage patterns. The four-point pattern correlation with oxygen pressure. All gage factors were
center left is the gage structure used predominantly in this work. negative, though absolute values are plotted for convenience
166 T.M. Miller et al. / Sensors and Actuators A 104 (2003) 162–170

factor ranging between 11.4 and 6.5 for oxygen partial

pressures in the range 1.2–4 mTorr. However, their data did
not show a trend with pressure. In fact, it would be mis-
leading to directly compare our data with theirs, because the
residual atmosphere in their sputter system also contained
Ar gas ranging from approximately 50–85%. The presence
of Ar, which is used to improve sputtering yield, also
substantially alters the energy distribution of the sputtered
particles, which typically peaks at an energy of a few
electronvolts. This, in turn, can have a decisive effect on
the properties of a deposited thin film. In our system, the
residual gas was composed entirely of oxygen. This gas both
modified the velocity of the laser ablated material and
reacted with it at the surface. Additional experiments are
needed to separate the effects of these distinctly different
Fig. 5. Gage factor data (open circles) and resistivity data (open squares) physical processes.
plotted vs. oxygen partial pressure during deposition. As mentioned above, the sensitivity of the electrical
properties of ITO to oxygen background pressure during
of representation. Each data point in the figure represents an deposition is well documented. However, the laser fluence
average for three films made at the given pressure, with the was also found to have a significant effect on the electrical
error bars indicating one standard deviation. properties of the ITO films. In Fig. 6 the resistivity and gage
The increase in resistivity with increasing oxygen pres- factor of ITO films are plotted versus oxygen pressure for
sure is consistent with previously published data and has two different laser fluence levels. A clear shift can be seen in
been shown to be a critical parameter in determining the the curves for both resistivity and gage factor between data
resistivity of ITO films [3,4,10]. Oxygen vacancies contri- published previously (open circles) [3], and new data pre-
bute charge carriers and as a result, varying the oxygen sented for the first time in this paper (open squares). The
stoichiometry through oxygen background pressure during laser fluence previously was 0.18 J/cm2, while for the cur-
deposition is in effect a means for controlling carrier density. rent data it was 0.13 J/cm2.
Fewer oxygen vacancies lead to fewer carriers and conse- All small-scale gages were deposited exclusively in
quently higher resistivity, as shown by the data of Kim et al. 50 mTorr of oxygen background pressure, and had an average
[4]. Thus, the effect of the oxygen pressure on the strain gage factor of 4:0  0:8. The average thickness of nine
sensitivity is of significant interest, although the precise films deposited with identical parameters was 172  20 nm.
mechanism for piezoresistivity has not been identified. In Fig. 7 shows a typical data set for a small-scale gage, in which
semiconductor Si gages and other materials with degenerate the fractional change in resistance is shown as a function of
hole bands, strain breaks the degeneracy and leads to applied strain. The sample shown has a resistance of 91 kO
changes in the proportion of heavy and light holes as well
as in the effective masses themselves. This is the reason that
Si strain gages are consistently made of p-type material [15].
A dependence of gage factor on oxygen pressure during
deposition similar to that of the resistivity was observed and
is also shown in Fig. 5. It is important to emphasize that the
well established oxygen pressure during deposition, which
leads to maximum conductivity in ITO, approximately
10 mTorr, does not similarly maximize gage factor. In fact,
for oxygen pressure values that maximize conductivity,
the amplitude of the gage factor is at a minimum, with an
average value of 0.2. When the oxygen pressure is
increased to it’s maximum attainable level in our system,
both the resistivity and the gage factor are maximized, with
the gage factor having an average value of 7.2. Clearly, the
oxygen pressure maintained during ITO deposition is a
means of controlling both the conductivity and the strain
sensitivity, although not independently. Fig. 6. This graph compares the curves of resistivity (dashed line), as well
as gage factor (solid line), vs. oxygen pressure during deposition for two
The only other data available with which to compare the different laser fluence levels indicated in the legend. Previously published
behavior of gage factor with oxygen partial pressure during data from Fang et al. [3] (open circles) are shown together with new data
deposition are those of Dyer et al. [2]. They found a gage (open squares).
 T.M. Miller et al. / Sensors and Actuators A 104 (2003) 162–170 167

Table 1
A summary of the data from FIB imaging experiments

Sample

1 2 3

Pre-FIB
Resistance (kO) 2.3 280 3.8
G.F. 2.5 3.4 2.4
Post-FIB
Resistance (kO) 2.3 30 3.7
G.F. 2.5 2.4 2.3
Processing
nC/mm2 0.25 10.1 10.1
Beam current (pA) 11 150 150
Parameters
Imaging Minimal Normal Normal
Fig. 7. The fractional change in resistance is shown plotted vs. strain for a
Encapsulation None None 100 nm SiO2
small-scale gage (open circles and squares) and the subsequently FIB
machined gage (open triangles). Results obtained during increasing strain
(open circles) and decreasing strain (open squares) are shown separately to
verify the absence of hysteresis. onto the ITO layer by laser ablation of an SiO target in an
oxygen ambient and experimentally verified to have no
measurable effect on the electrical properties of the ITO.
and a gage factor of 4.8, as indicated by the slope of the Both the resistance and the gage factor of several films
curve. Error bars are not shown explicitly in the figure remained the same as determined by I–V measurements.
because the instrumental error for each point is less than Table 1 summarizes a series of experiments to determine
1%. The graph contains points measured during both increas- the effects of the radiation damage from FIB imaging on
ing strain (circles) and decreasing strain (squares) for two the electrical properties of ITO gages. Three samples were
distinct measurements and indicates a linear piezoresistive imaged with the FIB, two un-encapsulated samples at
response as well as an absence of hysteresis that was typical current and total dose levels differing by more than an
of the films measured in this work. order of magnitude, and the third, an encapsulated sample,
Also shown in Fig. 7 is the piezoresistive response of the subjected to the more intense exposure. The first sample was
same sample after FIB machining to a size of 100 mm
imaged several times using single scans at various scan rates
780 mm. The gage factor of the newly trimmed gage is and with a beam current of 11 pA. The total integrated
smaller in magnitude than before, having a value of 2.9. current deposited on this sample, 329 pC in an area of
However, the response is still linear and without hysteresis, as approximately 1.3 mm2, did not produce any observable
the data once again represent both increasing and decreasing changes in the electrical properties of the ITO gage. The
strain cycles. The decrease in gage factor indicates collateral second sample was also imaged many times using single
radiation damage during ion milling. The amount of FIB scans and different scan rates, but at a higher beam current of
imaging of the full surface of this sample was minimal. 150 pA. The total integrated beam current for this sample
However, the cuts were made using the largest available was approximately 13 nC in 1.3 mm2, and it produced a
beam current, 11.5 nA, which also produces the largest and change in the resistance from 280 to 30 kO, and a reduction
least well defined beam spot. What is more telling is that the in the magnitude of the gage factor from 3.4 to 2.4. The third
resistance of this newly trimmed gage decreased by 86% to sample was encapsulated with 100 nm of SiO2 and was
13 kO. When altering the physical dimensions of a resistor, imaged exactly as the second sample. The observed change,
one expects a change in the resistance, but a decrease in less than 3% in resistance and gage factor, proved that the
resistance is unexpected, since the FIB cuts decreased the SiO2 effectively protects the ITO from the radiation damage
cross-sectional area. Evidently, electrical leakage paths were accompanying imaging, at least for total fluences needed to
established either by imaging the gage area or by the beam set up, conduct, and verify ion machining operations.
halo in the regions adjacent to the FIB cuts. Experiments were also performed to determine the effect
In order to establish the location of the radiation damage that FIB milling has on the electrical properties of ITO
affecting electrical properties, experiments were carried out gages. First, a cut was made in a sample parallel to the
with the goal of differentiating damage from imaging and direction of current flow and in the center of the active gage
that which resulted as a side effect of trimming the gage to element. The cut was made using the largest available beam
net shape. An encapsulation layer of sufficient thickness current, 11.5 nA. The 5 mm width of this cut was insignif-
should, in principle, protect a gage from surface damage, as icant compared with the 300 mm width of the full current
well as minimize the effects of humidity and other environ- path, and should not, therefore, have significantly affected
mental hazards. The SiO2 encapsulation layer was deposited the total resistance of the gage. Nevertheless, the operation
168 T.M. Miller et al. / Sensors and Actuators A 104 (2003) 162–170

resulted in a 17% decrease in the gage resistance. To confirm Table 2

the FIB’s ability to cut through the ITO and effectively A summary of the data from FIB machining experiments
isolate parts of the current path, a second cut was made Sample
perpendicular to the direction of current flow and across the
1 2 3
entire width of the gage. The result was as expected, a
completely open circuit, effectively isolating the two halves Pre-FIB
Size (mm
mm) 1000
250 1000
250 1000
250
of the gage.
Resistance (kO) 91 89 36
Table 2 summarizes the results from three samples that G.F. 4.8 4.7 4.4
were trimmed to different sizes using different beam cur-
Post-FIB
rents. As the table indicates, directly machining un-encap- Size (mm
mm) 780
100 400
80 100
20
sulated ITO reduced both the resistance and the absolute Resistance (kO) 13 40 48
gage factors, even at a beam current of only 150 pA, where G.F. 2.9 2.8 3.6
the size of the beam halo was substantially reduced. The Processing parameters
situation with the encapsulated gage was more complex. Cutting beam (pA) 11500 150 150
The encapsulated sample included 100 nm of SiO2, a Encapsulation None None 100 nm SiO2
layer sufficiently robust to protect the sensing layer from
damage caused by imaging. This sample was trimmed to a
size of 100 mm
20 mm using a beam current of 150 pA. Collateral radiation damage near FIB cuts even at mini-
Prior to trimming, it had a resistance of 36 kO. After mal beam currents and with encapsulated films is still
trimming to a smaller cross-sectional area, the resistance sufficient to produce significant electrical effects. A FIB
rose to 48 kO, a value that nevertheless fell short of the micrograph of this gage structure is shown in Fig. 8. Elec-
expectation based on geometrical considerations. However, trical isolation of the ITO above and below the narrow
the magnitude of the gage factor of this sample still central active strip is seen to be total. These regions are
decreased from a pre-irradiation value of 4.4 to a post- dark as a result of a build-up of positive charge from the Gaþ
irradiation value of 3.6, indicating that despite the pre- ion beam. This positive charge produces a voltage that
sence of the encapsulation layer, measurable radiation suppresses secondary electrons, and results in the ITO in
damage still occurred. these regions appearing dark in the image. This is significant

Fig. 8. An ion-induced secondary electron micrograph made in the FIB showing the 20 mm
100 mm active area trimmed from the original deposited area,
which was approximately 250 mm
1000 mm. The dark regions above and below the strip still contain ITO but it is electrically separated from the sensor by a
FIB cut, which extends into the substrate.
 T.M. Miller et al. / Sensors and Actuators A 104 (2003) 162–170 169

By applying similar reasoning, one can infer the gage

factor of the trimmed element that would result in an overall
change in the gage factor from 4.4 to 3.6. Once again,
one has to assume that each of the 10 hypothetical series
segments has the same response to strain, and that only
the trimmed segment is affected by the FIB. To carry out
the analysis, one must account for the fact that when
strained each uncut segment will undergo identical resis-
tance changes DRi, while the machined segment will experi-
ence a unique resistance change DRG. By substituting
(9DRi þ DRG ) for DR in Eq. (1) and using the measured
Fig. 9. A schematic that models the gage structure shown in Fig. 8. gage factor and resistance of the gage after the cut, DRG can
The leakage paths generated by radiation damage during trimming are be determined as a function of strain. Since the resistance of
modeled as resistors RL. The undamaged gage material is assumed to have the trimmed segment is known to be 15.6 kO, the gage
resistance RG.
factor of the 20 mm
100 mm gage is estimated to be 2.0.
Thus, as in other cases, radiation damage by the FIB
because it visually demonstrates that by selective narrow decreases the gage factor of the ITO. However, it is impor-
cuts the FIB has successfully trimmed the gage to size, tant to note that the trimmed segment still has a gage factor
effectively removing two large regions of ITO electrically comparable to that of metal foil gages.
without removing them physically.
To explore this effect quantitatively, a resistor model, shown
in Fig. 9, was developed to estimate the resistance of the 5. Conclusion
radiation damaged regions at the edges of the active area of
the gage. This model, based on the results presented above, Piezoresistive strain-sensors have been fabricated by
assumes that any change in the film’s electrical properties combining pulsed laser deposition through masks onto room
results from trimming and not from imaging. Since the beam temperature substrates with focused ion beam machining
diameter is much less than a mm, one can assume that only a to net shape. The active material, indium-tin-oxide, was
narrow region of ITO experiences a change in resistance as a fashioned reproducibly into sensors with high impedance,
result of alteration by the ion beam. Nevertheless, these cur- large gage factors, minimal hysteresis, and excellent line-
rent paths modeled as resistors RL in Fig. 9 are effectively in arity up to the 200 micro-strain limit of these measure-
parallel with the gage RG, as shown schematically in the figure. ments. For material deposited in a residual atmosphere of
If one conceptually divides the original 1000 mm gage 28–50 mTorr of oxygen, both resistivity and gage factor
length into 10 segments of equal length, and assumes each were found to increase in magnitude with increasing
segment is identical, then initially each segment has a oxygen partial pressure. In addition, it was observed that
resistance Ri of 3.6 kO. Since the resistance of the entire laser fluence is a critical parameter in determining the
gage after trimming was 48 kO, the resistance of the piezoresistive properties of the films. However, the details
trimmed segment with final width of 20 mm and length of of the mechanism for piezoresistivity in ITO have not been
100 mm must be 15.6 kO. With plausible assumptions about established.
the film thickness profile, one can estimate the change in The effect of radiation damage by the focused ion beam
cross-sectional area resulting from the FIB cuts, and from on the resistivity and gage factor of ITO was investigated.
this infer the resistance of the trimmed element, assuming no The generation of ion-induced secondary electron images
radiation damage. Using the three-resistors model in Fig. 9 resulted in minimal damage if done at low beam currents
one can then estimate the resistance of the areas damaged by (11 pA) while larger beam currents (150 pA–11.5 nA) pro-
the ion beam. Assuming a circular profile for the thickness of duced extensive collateral radiation damage when used for
the original ITO gage, the resistance of the trimmed segment imaging or trimming a structure to net shape. Encapsulation
is estimated to increase from 3.6 to 45 kO. This implies that of strain gages with a 100 nm SiO2 layer produced no
the parallel resistors RL modeling the leakage paths along changes in the resistivities and gage factors, protected the
the edges must have resistances of 46 kO each in order to ITO from ion beam damage during FIB imaging, and
achieve the observed resistance of 15.6 kO. If the thickness significantly reduced the radiation damage caused by trim-
profile is instead assumed to be a rectangle bracketed ming the gages to net shape.
by triangles, the resistance of the trimmed segment is
estimated to increase to 24 kO, implying that resistors RL
must have resistances of 86 kO. Thus, we see that FIB Acknowledgements
machining produces leakage paths with resistance equiva-
lent to 50–90 kO, with the exact value being dependent upon This work was supported by the US Army Research
the thickness profile of the trimmed segment. Office under Grant DAAD 19-99-1-0283 and by NASA,
170 T.M. Miller et al. / Sensors and Actuators A 104 (2003) 162–170

Marshall Space Flight Center, under Graduate Student [12] M.A. Morales Paliza, ITO, Indium Tin Oxide Films by Pulsed-Laser
Researchers Program Grant NGT 8-52905. Ablation at Room Temperature, Ph.D. Dissertation, Vanderbilt
University, 2001, and references therein.
[13] L. Fang, W.L. Wang, P.D. Ding, K.J. Liao, J. Wang, Study on the
References piezoresistive effect of crystalline and polycrystalline diamond under
uniaxial strains, J. Appl. Phys. 86 (1999) 5185–5193.
[14] W.M. Murray, W.R. Miller, The Bonded Electrical Resistance Strain
[1] M. Goldfarb, N. Celanovic, A flexure-based gripper for small scale Gage, Oxford University Press, New York, 1992, pp. 13, 39.
manipulation, Robotica 17 (1999) 181–187. [15] P. Kleimann, B. Semmache, M. Le Berre, D. Barbier, Stress-
[2] S.E. Dyer, O.J. Gregory, P.S. Amons, A. Bruins Slot, Preparation and dependent hole effective masses and piezoresistive properties of
piezoresistive properties of reactively sputtered indium tin oxide thin p-type monocrystalline and polycrystalline silicon, Phys. Rev. B 57
films, Thin Solid Films 288 (1996) 279–286. (1998) 8966–8971.
[3] H. Fang, T.M. Miller, R.H. Magruder II, R.A. Weller, The effect of [16] O.J. Gregory, Q. Luo, A self-compensated ceramic strain gage
strain on the resistivity of indium tin oxide films prepared by pulsed for use at elevated temperatures, Sens. Actuators A 88 (2001)
laser deposition, J. Appl. Phys. 91 (2002) 6194. 234–240.
[4] H. Kim, C.M. Gilmore, A. Piqué, J.S. Horwitz, H. Mattoussi, H.
Murata, Z.H. Kafafi, D.B. Chrisey, Electrical, optical, and structural
properties of indium-tin-oxide thin films for organic light-emitting
devices, J. Appl. Phys. 86 (1999) 6451, and references therein. Biographies
[5] C.S. Smith, Piezoresistive effect in germanium and silicon, Phys.
Rev. 94 (1954) 42. Timothy M. Miller received an MS degree in materials science from
[6] B.G. Lewis, D.C. Paine, Applications and processing of transparent Vanderbilt University in 2000. His thesis topic focused on phase
conducting oxides, MRS Bull. 25 (2000) 22–27. transformations in titanium aluminides with ternary additions. He is
[7] H. Kim, A. Piqué, J.S. Horwitz, H. Mattoussi, H. Murata, Z.H. currently completing an MS degree in mechanical engineering as a part of
Kafafi, D.B. Chrisey, Indium tin oxide thin films for organic light- an interdisciplinary team working to develop thin film sensors for MEMS
emitting devices, Appl. Phys. Lett. 74 (1999) 3444. components.
[8] M.J. Staines, Robust analog resistive touch-screen for computers,
Elektronik Praxis no. 13 (2000) 44–47. Hui Fang is a PhD candidate in the Department of Physics and Astronomy
[9] W.J. Lee, Y.K. Fang, J.J. Ho, C.Y. Chen, R.Y. Tsai, D. Huang, F.C. at Vanderbilt University and is investigating the phenomenon of
Ho, H.W. Chou, C.C. Chen, Pulsed-magnetron-sputtered low- piezoresistivity in ITO thin films.
temperature indium tin oxide films for flat-panel display applications,
J. Electron. Mater. 31 (2002) 129. Robert H. Magruder III is a professor of physics at Belmont University
[10] Y. Wu, C.H.M. Marée, R.F. Haglund Jr, J.D. Hamilton, M.A. Morales with interests in the ion beam and pulsed laser deposition fabrication of
Paliza, R.A. Weller, Resistivity and oxygen content of indium tin thin films and optical characterization of thin films.
oxide films deposited at room temperature by pulsed-laser ablation, J.
Appl. Phys. 86 (1999) 991. Robert A. Weller is an associate professor of electrical engineering at
[11] H. Kim, J.S. Horwitz, G. Kushto, A. Piqué, Z.H. Kafafi, C.M. Vanderbilt, with interests in radiation effects in solids, ion beam
Gilmore, Effect of film thickness on the properties of indium tin fabrication and analysis of materials, electronic materials, and computer
oxide thin films, J. Appl. Phys. 88 (2000) 6021. simulation of radiation effects in materials and devices.