1. * Address all correspondence to this author. fshih@seattleu.edu
CHARACTERIZATION OF A FAST RESPONDING COMPOSITE THERMAL
BIMORPH FILM ACTUATOR BASED ON CARBON NANOTUBE SHEETS
Alaina M. Bever, Peter J. Brown, Kerry V. Lane, Benjamin L. Levy-Wendt,
Nathan K. Yasuda, Yen-Lin Han, and Frank J. Shih*
Mechanical Engineering Department
Seattle University
Seattle, Washington 98122, USA
ABSTRACT
As the extraordinary thermal, electrical, and mechanical
properties of carbon nanotubes (CNTs) have become better
understood, they have found their way into a wide range of
engineering applications. Used in conjunction with fiber-
reinforced composite materials, CNTs provide enhanced thermal
conductivity, interlaminar strength, and ballistic resistance of
laminar composite materials. However, the direct application of
the macro form of CNT sheet as a heating element for use in a
thermal actuator has not been reported. In the present study, CNT
sheets are used as a flexible, efficient, and fast-responding
heating element that induces transverse motion in a multilayered
functional polymer composite based on thermal expansion
mismatch between layers. The CNT heating element is designed
to have a specific cross-sectional area to length aspect ratio,
giving it a specific resistance and power consumption
characteristic. The heating element is bonded to a compliant
silicone elastomer substrate and a stiff constraining polyimide
thin film, forming a flap-like actuator. The robust design and
simple operation of the actuator makes it a potential candidate
for control surfaces on micro air vehicles and actuating elements
in microscale fluid pumps. The heating response rate of the
actuator is measured experimentally using an infrared thermal
imager. The temperature change in the thermal actuator is
measured as a function of input voltage. The edge deflection of
the actuator is also measured as function of the applied voltage.
Finally, finite element modeling of the thermal actuator, a
parametric study of material selection, and deflection analysis
are conducted to better understand the result of these
experiments.
NOMENCLATURE
k Thermal conductivity
E Young’s modulus
 Poisson’s ratio
 Plate curvature 
 Coefficient of thermal expansion
MWNT Multi-walled nanotubes
 Vertical tip deflection
INTRODUCTION
Carbon nanotubes (CNTs) have been utilized in a wide range
of applications. These application typically takes advantage of
its superior stiffness and strength properties [1]. One area that
has only been explored recently is in CNTs’ use as a heating
element [2]. CNT sheets are flexible and have a rapid thermal
response rate relative to metallic heating element materials. This
paper explores the use of a CNT sheet as a heating element in a
composite thermal bimorph film actuator, which takes advantage
of both its rapid response and flexibility. While there are recent
advances in CNT based thermal actuators [3] that use CNT in
polymer emulsion as precursor materials, direct applications of
CNT sheets have not been reported.
The pertinent material properties of the thermal actuator
components are listed in Table 1.
Proceedings of the ASME 2015 International Mechanical Engineering Congress and Exposition
IMECE2015
November 13-19, 2015, Houston, Texas
IMECE2015-52576
1 Copyright © 2015 by ASME

2. Table 1: Material Properties
Polyimide
Film
Silicone
Elastomer
Young’s modulus E (GPa) 2.5 0.001
Poisson’s ratio  0.39 0.48
Density (kg/m3) 1420 1250
Coefficient of Thermal Expansion 
(/K)
2 x 10-6 250 x 10-6
Thermal Conductivity k (W/m-K) 0.116 2
EXPERIMENTAL
The CNTs used in this study are in the form of non-woven
sheets supplied by Nanocomp Technologies, Inc. (Merrimack,
NH). These are multi-wall nanotubes (MWNT) consisting of a
few walls bundled together in a quasi-isotropic morphology. The
CNTs have diameter ranges from 10 nm to 25 nm. These CNTs
were vapor formed using a proprietary process and then
deposited onto a moving drum, transforming the CNT fibers into
a non-woven sheet. The CNT sheets were then solvent washed
and pressed. The CNT sheets are 16 grams per square meter and
have a thickness of 23 m (0.0009 in.) for an apparent density of
0.7 g/cm3. A scanning electron microscopy (SEM) image of its
surface morphology is shown in Figure 1.
When a voltage is applied across the CNT sheets, the
temperature of the material can elevate from room temperature
to about 100 oC within a few seconds. For cooling, the
convective cooling of the sheet is achieved by the superb thermal
conductivity of CNT material as well as its thin film surface
geometry, allowing the CNTs to cool down to room temperature
quickly in a convective flow stream. The resistance of the
nanotube sheet was tailor made to 120- through modification
of the width and length of nanotube wire in a serpentine two-
dimensional array. The actuator was driven by a low-voltage DC
source.
2 m
Figure 1. Surface of the non-woven CNT sheets showing bundles of
CNTs in a tangled and fairly isotropic morphology. (photomicrograph
courtesy of Nanocomp Technologies, Inc.)
The layout of the composite thermal bimorph film actuator
is schematically shown in Figure 2. The stamp shaped actuator
measures 25 mm by 54 mm (1 in. x 2.13 in.). The portion of the
actuator that moves is measures 25 mm by 39 mm (1 in. x 1.55
in.). A 0.025 mm (0.001 in.) thick Kapton® polyimide film is the
bottommost layer, serving as a constraining layer. A 0.79 mm
(0.031 in.) thick silicone elastomer sheet with a firm durometer
hardness of 70Ais used as the primary substrate. Lastly, the CNT
sheet is laser-cut to a precision pattern (Universal Laser System,
Scottsdale, AZ) and is adhesive bonded to the top of the silicone
elastomer sheet. The measured thickness of the CNT sheet is
0.023 mm (0.0009 in.). The designed operation temperature is
under 200 oC. The two large CNT sheet patches near one end of
the actuator serve as the electrode for the applied DC voltage.
The laser cut CNT sheet, near final form, is shown in Figure
3. The width of the CNT wire on the 2D array is 1.75 mm (0.069
in.). With a measured resistivity of 3 x 10-5 -m, the resistance
of the heating element array can be tailor-made by controlling
the ratio of wire length to wire width. The resulting electrical
resistance of the CNT sheet array is 120-. Finer thickness wire
can be made to elevate the resistance of the CNT sheet array,
allowing the system to generate similar level of output power
with lower voltage.
Figure 2. Schematic of the thermal actuator with CNT sheet heating
array on top, a silicone rubber sheet in the middle, and a polyimide
film at the bottom
Figure 3. The CNT-sheet 2-D heating element array
2 Copyright © 2015 by ASME

3. Figure 4. Thermal images of the thermal actuator (measured from
the top surface) at room temperature and at the operating temperature
of 100° o
C. The cross-hair denotes the temperature (o
C) at that
location.
A DC power-supply (BK Precision 1667) was used to apply
a DC voltage to the CNT sheet heating element. The voltage was
monitored by a multimeter (Fluke 45 Dual Digital Multimeter)
with 5-digit resolution. The temperature on the actuator was
monitored by a non-contact infrared thermal imager (Fluke Ti-
100) that takes measurements at 9-Hz. A traditional contact
metallic thermocouple draws too much heat away from the CNT
heating element, thus making its use unsuitable in this
experiment. A typical temperature reading of the thermal
actuator is shown in Figure 4, where room temperature (~25 oC)
and operating temperature (~100 oC) measurements are shown.
The vertical deflection at the free cantilever edge was
measured using a digital microscope (Keyence VHX-1000) with
a VH-Z1000R lens set at 100x magnification. The calibrated
travel in z-axis (up and down directions) on the microscope,
focusing on a specific location at the tip of the thermal actuator,
was used to measure vertical deflection while the input voltage
varied. The schematics of the experimental setup for both
temperature and deflection measurements are shown in Figure 5.
Figure 5. Schematic diagram of the experimental setup, showing
temperature measurement (remote infrared imaging) and deflection
measurement (focusing on the point of interest with a digital
microscope that reports its z-travel).
RESULTS
The CNT sheet, when not attached to a thermal substrate,
can change in temperature on the order of 100 oC/sec when
appreciable voltage is applied. For the thermal actuator, the CNT
sheet is attached to a firm silicone elastomer sheet, which acts as
a large heat sink, significantly lowering the heating rate. The
resulting time response of the thermal actuator is shown in Figure
6. Using a fixed input voltage of 40 DC volts that is applied to
the system instantaneously, the temperature rose (from room
temperature) to 100 oC in 8 seconds, and to 72 oC in 4 seconds.
For steady state response, input voltage varying from 0 to
20 volt DC was applied to the thermal actuator. The thermal
imager was used to measure the actuator surface temperature.
The result is plotted in Figure 7. The thermal actuator’s response
to voltage input is relatively quick (~1 sec). The surface
temperature appears proportional to input voltage after voltage
reaches 10 V.
The thermal actuator was set up as a cantilever beam with
the electrode end fixed, allowing 39 mm (1.55 in.) of plate to
curve downward as the top surface heated and expanded. The
vertical displacement at the free edge was measured using a
digital microscope (Keyence VHX-1000) with a VH-Z1000R
lens set at 100x magnification. The calibrated travel in the z-axis
on the microscope, auto focusing on a specific small location at
the tip of the thermal actuator with a piezoelectric motor, was
used to measure vertical deflection. The result is plotted in
Figure 7. The input voltage varied from 0 to 20V while the
deflection varied from 0 to 3.0 mm (0.12 in.)
Figure 6. The thermal response rate of the CNT-based thermal
actuator
20
30
40
50
60
70
80
90
100
110
-2 0 2 4 6 8
Temperature(degC)
Time (sec)
3 Copyright © 2015 by ASME

4. Figure 7. Steady state temperature of the thermal actuator as a
function of applied voltage
Figure 8. Edge deflection as a function of applied voltage.
Figure 9. Thermal actuator temperature versus edge deflection.
DISCUSSION
Combining the results from Figures 7 and 8, we can
determine the relationship between the upper surface
temperature and free edge deflection. The result is shown in
Figure 9. The deflection as a function of temperature does not
appear linear. There appears to be an initial deflection from room
temperature to 40 oC. The deflection temperature curve appears
proportional from 40 oC to 90 oC. The slope nearly doubles after
90 oC, suggesting that there are other physical phenomenon at
this temperature regime in the system that are unaccounted for.
The decision to use silicone elastomer was based on its
relatively high thermal expansion coefficient , and its ability to
operate at a high temperature. Although a relatively firm
variation of the silicone elastomer was used as substrate, it is
several magnitude less in stiffness than polyimide. To determine
the optimal combination of material for use in this application, a
parametric study on material selection was carried out. [4]
The bimaterial cantilever curvature subjected to elevated
temperature described in Chu, et. al. [5] was used as the figure
of merit. The curvature  is written as
  
     
1 2 1 2 1 2 2 1
2 22 2 2 2
1 1 2 2 1 2 1 2 1 1 2 2
6
2 2 3 2
E E t t t t T
E t E t E E t t t t t t
 

  

   
(Eqn 1)
where E1 and E2 are the elastic moduli, t1 and t2 are the respective
thicknesses, α1 and α2 are coefficients of thermal expansion for
materials 1 and 2, and T is the difference between room
temperature and operating temperature of the actuator. The
curvature formulation was simplified, making the assumption
that the two materials have the same width. Keeping material 1
constant as polyimide (E1, t1, 1), while solving for actuator
curvature using all candidate materials as material 2, the optimal
substrate material can be determined.
Every class of materials in the polymers (both
thermoplastics and thermosets) and elastomers material family
were considered as the actuator substrates. The inherent
variability of polymers allows for a range of material properties
[4]. The twenty-three classes of material considered are tabulated
in Table 2.
Keeping the constraining layer of polyimide at the same
thickness 0.025 mm (0.001 inch), the properties of the compliant
substrate materials were entered in a spreadsheet to calculate its
curvature with a T of 200 oC. The thickness of the compliant
substrate is kept at 0.79 mm (0.031 inch). The ranked results,
based on the curvature , along with average thermal
conductivity, maximum operating temperature, and average
elastic moduli are tabulated in Table 3.
0
20
40
60
80
100
120
0 5 10 15 20
Temperature(degC)
Applied Voltage (V)
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
0 5 10 15 20
Deflection(mm)
Applied Voltage (V)
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
20 40 60 80 100 120
Deflection(mm)
Temperature (deg C)
4 Copyright © 2015 by ASME

5. Table 2: Polymers and elastomers considered for this application
[4]
Polymers
Acrylonitrile butadiene
styrene (ABS)
Cellulose polymers (CA)
Ionomers
Epoxies
Phenolic
Polyamides (PA)
Polycarbonate (PC)
Polyesters
Polyetheretherkeytone (PEEK)
Polyethylene (PE)
Polyethylene terephalate (PET)
Polymethylmethacrylate (PMMA)
Polyoxymethylene (POM)
Polypropylene (PP)
Polystyrene (PS)
Polytetrafluoroethylene
(PTFE)
Polyvinylchloride
(PVC)
Elastomers
Butyl rubber
Ethylene-vinyl acetate
(EVA)
Isoprene
Natural rubber
Neoprene
Polyurethane (PU)
Silicone Elastomer
Table 3: Top performers based on the actuator curvature 

(1/m)
kavg
(W/m-K)
Tmax
(o
C)
Eavg
(MPa)
Neoprene* 180 0.11 100 1.4
Natural*
Rubber
96 0.12 70 2.0
Silicones* 92 0.65 260 13
Ionomers 75 0.26 80 310
Butyl*
Rubber
63 0.09 110 1.5
EVA* 57 0.35 70 25
CA 49 0.22 110 1800
PTFE 49 0.25 260 480
PU* 47 0.29 90 2.5
PE 43 0.42 140 760
* Elastomers
With larger values for coefficients of thermal expansion and
lower stiffness, the best performing materials are elastomers. The
silicone elastomer used in this study is one of the better choices
due to its relatively large curvature, high thermal conductivity
(determining how quickly the actuator moves), high operating
temperature (allowing large temperature differences), but low
elastic modulus. Neoprene can achieve the same curvature with
lower temperature, but the actuation is slower by about 1/6 due
to its low thermal conductivity. Cellulose polymer (CA) sheets,
Teflon (PTFE) sheets, and PE are reasonable choices. These
polymers curve less under the same temperature gradient, but
have larger elastic moduli, allowing the actuator to resist greater
force. Increasing the material thickness decreases the curvature,
but adds to the overall structural stiffness (EI) of the system.
Finite Element modeling was conducted using Solidworks
Simulation. Silicone elastomer substrate and polyimide film
were modeled in their approximate dimensions. An elevated
temperature at the top surface of silicone rubber film was
applied. The electrode end was fixed, creating a cantilever. The
resulting deflection, shown in a screenshot for the 80 oC case, is
shown in Figure 10. While the finite element analysis shows that
the deflection of 9 mm, the experiment yielded only 2 mm. The
complete temperature versus deflection plot is shown in Figure
11. The source of the discrepancy is unknown.
It was speculated that the discrepancy may be due to the fact
that the CNT sheet, even though sectioned in a serpentine pattern
that allows for greatest flexibility along the length direction,
contributes to the stiffness to the top layer. As CNT’s stiffness is
several magnitudes higher than that of the silicone elastomer,
some constraining effect may be present, and may be significant
enough to influence the deflection value. It is speculated that the
CNT element may act like a mechanical spring on the top surface
plane. When the top surface expands, the zigzag CNT wires offer
similar resistance as a paper clip being opened up.
We sandwich bonded silicone elastomer strips with CNT
sheet in order to measure the effects of the CNTs. The elastomer
specimen was subjected to a tensile load in a MTS machine to
ascertain the effective modulus of the CNT reinforced composite
laminate. Compared with the stress strain behavior of silicone
elastomer without reinforcing CNTs, the CNT sheets had a
calculated elastic modulus of 700 MPa. The result is lower than
the 1.1 GPa reported in literature [6], but is within the same
range.
Exact solution exists for deflection [5], where the deflection
 is given as
2
2
L
  , (Eqn 2)
where L is the length from the fixed end to the tip of the
cantilever. The stiffness contributed by the CNT heaters
increases E2, and decreases the overall curvature . However, the
analytical solution showed that this effect only decreased the
deflection by about 5%. The modeling of the behavior of the
thermal actuator remains an on-going effort. The direction of the
future work includes: (1) instead of relying on published material
property values, measure the key properties values in-house. (2)
Double check the temperature of the silicone elastomer substrate,
as the top surface temperature may be misleadingly high, (3)
explore two-dimensional plate deformation, where expansion
also occurs in the width direction, causing a saddle-like
geometry in the actuator, contributing to total deflection, and (4)
using constitutive relationship for large deformation mechanics
that better accounts for deformation in elastomers.
5 Copyright © 2015 by ASME

6. Figure 10. Free-edge displacement of the thermal actuator where 80
oC was applied at the top surface of silicone rubber film.
Figure 11. Free-edge deflection of the thermal actuator simulated using
a Finite Element software.
CONCLUSION
A thermal actuator prototype using polyimide, silicone
rubber, and CNT non-woven sheet as heating element was
experimentally characterized. The actuator demonstrated a quick
response when voltage is applied. It demonstrated a transverse
motion of 3 mm in about 8 seconds.
ACKNOWLEGMENT
The authors wish to thank Clare Boothe Luce Foundation
for undergraduate research funding for both project supplies and
stipend support for Clare Boothe Luce Scholar A. M. Bever.
REFERENCES
[1] Shan, F. L., Gu, Y. Z., Li, M., Liu, Y. N. and Zhang, Z. G.,
2013, “Effect of deposited carbon nanotubes on
interlaminar properties of carbon fiber-reinforced epoxy
composites using a developed spraying processing,”
Polymer Composites, 34(1), pp. 41–50.
[2] Jung, D., Han, M., Lee, G.S., 2014, “Flexible transparent
conductive heater using multiwalled carbon nanotube
sheets,” J. Vac. Sci. Technol. B., 32(4), pp. 04E105-1-6
[3] Zeng, Z., Jin, H., Zhang, L., Zhang, H., Chen, Z., Gao, F.,
Zhang, Z., 2015, “Low-voltage and high-performance
electrothermal actuator based on multi-walled carbon
nanotube/polymer composites carbon nanotube sheets,”
Carbon, 84(4), pp. 327-334
[4] Ashby, M. F., 2005, Materials Selection in Mechanical
Design, Butterworth-Heinemann, Oxford, UK
[5] Chu, W.-H., Mehregany, M., Mullen, R. L., 1993,
“Analysis of tip deflection and force of a bimetallic
cantilever microactuator,” J. Micromech. Microeng, 3, pp.
4-7
[6] Cheng, Q., Bao, J., Liang, A., Zhang, C., Wang, B., 2009,
“High mechanical performance composite conductor:
multi-walled carbon nanotube sheet/bismaleimide
nanocomposites,” Adv. Funct. Mater. 19, pp. 3219–3225
2
4
6
8
10
12
14
40 60 80 100
Deflection(mm)
Temperature (deg C)
6 Copyright © 2015 by ASME