Nanotechnology

PAPER Related content

- Thermal management of thermoacoustic
Thermoacoustic sound projector: exceeding the sound projectors using a free-standing
carbon nanotube aerogel sheet as a heat
fundamental efficiency of carbon nanotubes source
Ali E Aliev, Nathanael K Mayo, Ray H
Baughman et al.
To cite this article: Ali E Aliev et al 2018 Nanotechnology 29 325704 - Increasing the efficiency of thermoacoustic
carbon nanotube sound projectors
Ali E Aliev, Yuri N Gartstein and Ray H
Baughman

- Thermoacoustic excitation of sonar

View the article online for updates and enhancements. projector plates by free-standing carbon
nanotube sheets
Ali E Aliev, Nathanael K Mayo, Ray H
Baughman et al.

This content was downloaded from IP address 86.27.85.58 on 18/10/2020 at 23:42

 Nanotechnology

Nanotechnology 29 (2018) 325704 (17pp) https://doi.org/10.1088/1361-6528/aac509

Thermoacoustic sound projector: exceeding

the fundamental efficiency of carbon
nanotubes
Ali E Aliev1,7 , Daniel Codoluto1, Ray H Baughman1,
Raquel Ovalle-Robles2, Kanzan Inoue2, Stepan A Romanov3,
Albert G Nasibulin3,4, Prashant Kumar5, Shashank Priya5,
Nathanael K Mayo6 and John B Blottman6
1
Alan G. MacDiarmid NanoTech Institute, University of Texas at Dallas, Richardson, TX 75083, United
States of America
2
Nano-Science & Technology Center, LINTEC OF AMERICA, INC. Richardson, TX 75083, United
States of America
3
Skolkovo Institute of Science and Technology, Skolkovo, Moscow Region, 143025, Russia
4
Department of Applied Physics, Aalto University, Puumiehenkuja 2, FI-00076 AALTO, Finland
5
Center for Energy Harvesting Materials and Systems, Virginia Tech, Blacksburg, VA 24061, United
States of America
6
Sensor and Sonar Systems, Naval Undersea Warfare Center, B1170/R109, Newport, RI, United States of America

E-mail: Ali.Aliev@utdallas.edu

Received 2 March 2018, revised 1 May 2018

Accepted for publication 15 May 2018
Published 5 June 2018

Abstract
The combination of smooth, continuous sound spectra produced by a sound source having no
vibrating parts, a nanoscale thickness of a flexible active layer and the feasibility of creating large,
conformal projectors provoke interest in thermoacoustic phenomena. However, at low frequencies,
the sound pressure level (SPL) and the sound generation efficiency of an open carbon nanotube
sheet (CNTS) is low. In addition, the nanoscale thickness of fragile heating elements, their high
sensitivity to the environment and the high surface temperatures practical for thermoacoustic sound
generation necessitate protective encapsulation of a freestanding CNTS in inert gases.
Encapsulation provides the desired increase of sound pressure towards low frequencies. However,
the protective enclosure restricts heat dissipation from the resistively heated CNTS and the interior
of the encapsulated device. Here, the heat dissipation issue is addressed by short pulse excitations
of the CNTS. An overall increase of energy conversion efficiency by more than four orders (from
10−5 to 0.1) and the SPL of 120 dB re 20 μPa @ 1 m in air and 170 dB re 1 μPa @ 1 m in water
were demonstrated. The short pulse excitation provides a stable linear increase of output sound
pressure with substantially increased input power density (>2.5 W cm−2). We provide an extensive
experimental study of pulse excitations in different thermodynamic regimes for freestanding
CNTSs with varying thermal inertias (single-walled and multiwalled with varying diameters and
numbers of superimposed sheet layers) in vacuum and in air. The acoustical and geometrical
parameters providing further enhancement of energy conversion efficiency are discussed.
Supplementary material for this article is available online

Keywords: thermoacoustics, carbon nanotubes, pulse excitation, heat transfer, sound, sonar projectors
(Some figures may appear in colour only in the online journal)

7
Author to whom any correspondence should be addressed.

0957-4484/18/325704+17$33.00 1 © 2018 IOP Publishing Ltd Printed in the UK

Nanotechnology 29 (2018) 325704
1. Introduction
Thermoacoustic (TA) sound transduction shows promise to
fill the low frequency niche for light-weight, thin and flexible
loudspeakers and underwater sonar systems. Recent studies
on TA sound generation using freestanding carbon nanotubes
(CNTs) [1–8], alternative nanostructures [9–13], and micro-
fabricated, suspended aluminum nanowires [14, 15] have
found that the sound pressure in open systems is proportional
to the applied power and frequency. The ultra-small heat
capacity per unit area of a thin sheet CNT heater, Ch*, and the
efficient thermal coupling between CNTs and the surrounding
gas provide a wide frequency response (1 Hz–1 MHz) and
relatively high sound pressure level (SPL) at high frequencies.
However, in the low frequency region, where the demand for
large area, flexible sound projectors is high, sound generation
efficiency is disappointingly low [3, 6, 16]: ηmax„10−5 for
f„10 kHz. Other challenges are relate to the air-like density
of CNTS, resulting in low mechanical strength and high
sensitivity to the environment at the high surface temperatures
necessary for high-power TA sound generation.
TA sound generation combines the principles of a heat
engine, Carnot’s theorem, and fluid dynamics at the nanos-
cale. Expansion and contraction of a gas in vicinity of resis-
tively heated CNTS induces sound wave propagation in the
surrounding medium. Increasing the thermal gradient created
between the surface of the CNTS and the surrounding med-
ium results in increased sound pressure intensity. Therefore,
high electrical power is required to achieve high sound
intensity and high energy conversion efficiency. The extre-
mely low heat capacity of the aerogel density CNTS and the
small, but finite heat capacity of air, are accompanied by low
thermal conductivities to external heat sinks. At high applied
AC power, this leads to an accumulation of heat in the vici-
nity of the heater, i.e., the CNTS. The elevated background
temperature decreases the efficiency of the TA projector. The
heat dissipation issue is even more severe for encapsulated
devices [17, 18]. The heat dissipation becomes the main
obstacle limiting the acoustic power density of the TA
projector.
For the purpose of benchmarking advances of nanos-
tructured TA materials, like CNTs, wherein the heat capacity
of the projector material is much less than the surrounding
gas, we define the fundamental efficiency of sound genera-
tion. By fundamental efficiency, we mean the ratio of far field
acoustic power output to applied electrical power when the
electrical input is a sine wave signal. This fundamental effi-
ciency will depend upon whether the system is open or
closed. Hence, we have exceeded the fundamental efficiency
for either an open system or a closed system when the ratio of
output acoustic power at a given frequency to the input
electrical power exceeds that for the case where the electrical
input is a single sinusoidal signal.
The heat dissipation issue and the creation of an
enhanced thermal gradient in encapsulated TA projectors is
addressed here by signal processing of the applied AC
power, namely, by short pulse excitation of a nanostructured
heater. The advantages of short pulse modulation are
2
A E Aliev et al
multiple: (i) a small duty cycle provides sufficient time for
cooling the active CNT heat source and surrounding gas
towards ambient temperature within each pulse period;
(ii) the temperature increase in a short pulse and the energy
transfer to the gas medium are nonequilibrium processes,
resulting in sharper spatial thermal gradient and an enhanced
energy conversion efficiency; (iii) the use of a high fre-
quency pulsed carrier signal modulated with a low frequency
sound envelope enables sound generation at the frequency of
the applied current, without the need for additional energy
consuming DC biasing (valid for both open and closed
systems); (iv) the electronics for powering the TA device
with high-power short pulses are more efficient, lighter and
more compact (like electronic converters) than sinusoidal
signal amplifiers.
Two pulse excitation regimes have been examined: a
single pulse (SP) per vibrational displacement and a high
frequency alternating carrier-pulse sequence modulated by an
acoustical (sinusoidal or arbitrary form) envelope. For both,
an extended linear power dependence of the generated sound
pressure and an increased energy conversion efficiency are
observed.
We provide an extensive experimental study of pulse
excitations for different thermodynamic regimes and explain
our observations within basic theoretical models. The
acoustical and geometrical parameters providing further
increases in efficiency and transduction performance for
resonant and non-resonant systems are discussed.
2. What governs the efficiency of a TA projector:
open and closed systems
TA sound generation is directly related to the ideal gas law:
PV=nRT, i.e., the volume, V, occupied by n moles of any
gas has a pressure, P, at temperature, T, in Kelvin, where R is
known as the gas constant. For open systems, the boundary
conditions follow dV/V0=dT/T0|P, and the sound pressure
wavelength is much larger than temperature wavelength,
λP?λT, while P remains constant. For closed systems,
dP/P0=dT/T0|V, and the volume remains constant, dV=0
[19]. The general solution to spherical TA radiation from any
solid pulsating sphere into gas suggests that the heat capacity
per unit surface area of the heater, C*h=hρhCh, should be
lower than the heat capacitance of the ambient gas involved in
the TA interaction, C*g=lρgCg [6, 20]:
- iro w 2a2 Ph
prms = ⋅
2 k1To r - i4pa2 (wCh* + k1k)
f 1 1
= ⋅ ⋅ ⋅ Ph. (1 )
2 2 Cg T0 2c + 2c + 1 r
2
The root-mean-square (rms) value of sound pressure,
prms, produced by an applied AC power, Ph=I2oR, is
increased with frequency, f, and the slope of this dependence
is a function of material parameters and ambient conditions.
Here, T0 is the temperature of the surrounding gas, ω=2πf is
Nanotechnology 29 (2018) 325704
the angular output frequency, |k1|=1/l=(ω/2α)1/2 is the
real part of the thermal wavenumber, α and κ are the thermal
diffusivity and the thermal conductivity of the surrounding
gas, respectively, a is the radius of the TA pulsating spherical
heater, and r is the distance to the microphone. The dimen-
sionless coefficient χ=(hρhCh)/(lρgCg) is the ratio of the
heat capacity per unit area, C*h=hρhCh, to the heat capacity
of the gas layer involved in the TA interaction. Here, C is the
gravimetric specific heat capacity of the heater (Ch) or gas
(Cg), ρ is the corresponding density, h is the thickness of the
heater and l is the thermal diffusion length in the gas, deter-
mining the thickness of air layer involved in heating process.
The derivation of equation (1) is given in S1 of the supporting
information (SI) available online at stacks.iop.org/NANO/
29/325704/mmedia. The following conclusions are espe-
cially important for identifying new TA heaters: first, the low
density, ρh, and small thickness, h, of the TA sheet are
essential for having both a lower thermal inertia and a higher
heat exchange with the surrounding gas; second, for an
enhanced energy conversion efficiency, the TA material
should withstand a high applied power, i.e., high temperatures
are necessary for a high Carnot’s efficiency, η=1−To/Th;
third, the heat capacity per unit area of the heater source, C*h,
needs only be small relative to C*g; finally, the effect of the
thermal inertia of the heater is diminished at low frequencies
due to an increasing thermal diffusion length in air,
l=(α/πf )1/2.
At a low applied power (linear regime) and when
Ch=C*g, i.e., χ → 0, equation (1) reduces to the well-known
* substantially affects the output sound. For a heater encapsu-
dependence [1],
f 1
prms = ⋅ ⋅ Ph
2 2 T0 Cg r
and the energy conversion efficiency for a flat TA source
radiating acoustic power Pac in both direction (2S), or for an
omnidirectional source, i.e., for k2a„1 [3], is
2 achieved. Encapsulation using resonantly vibrating rigid plates
Pac 2SI 2Sprms p ⋅ f2
h= = = = Ph.
Ph Ph rg ug Ph rg ug Cg2 T02
In the far field, k2a…1 (when the acoustical wave-
number k2=2π/λa becomes comparable, or larger than the
source size a), the directionality of the source and the inter-
ference forms a radiation pattern:
p ⋅ f2 1 2p 3r2p h p2 hg2 f 2 u f
h= 2 2 h
rg ug Cg T0
P ´
2p
∬ D2 sin q dq df ,
where D(θ, j) is the directivity function. For k2a>2 the
directivity integral gives I ∝ f −2, which is exactly compen-
sated by the TA coupling proportionality, f 2 in equation (3)
[3]. Therefore, the maximum efficiency in the far field is
ug
h= Ph.
2Srg Cg2 T02
A E Aliev et al
Note, an increase of applied AC power (i.e., an increase
of modulated temperature), increases the efficiency, similar to
that of a heat engine. If the input power is totally transferred
to the surrounding gas, Ph=2SβTa, the efficiency of an open
TA projector in the far field is depends on the intrinsic
properties of the ambient gas, υg, ρg, Cg, the mean temper-
ature of the thin TA heater above ambient, Ta, and heat
exchange coefficient of the TA heater, β:
ug ug b Ta
h= ⋅ Ph = ⋅ . (5 )
2Srg Cg2 T02 rg Cg2 T02
Thus, the fundamental efficiency of an open system is
much below the thermodynamic limit of the Carnot’s cycle,
due to the low coupling coefficient. There are a few ways to
improve the energy conversion efficiency of a TA projector:
(i) utilize a gas with a low Cg near the TA heater; (ii) increase
the heat exchange coefficient of the TA heater, β; (iii) increase
the temperature modulation amplitude of the TA heater, Ta.
Among the proper gases, the monoatomic, high mole-
cular weight inert gases show the best performance. For
example, the thermoacoustically induced sound pressure in
xenon is ∼7 times higher than in air [6]. Moreover, nanos-
tructured TA heaters, like freestanding CNTS or thin gra-
phene sponges, can withstand very high temperatures (>2000
K) and enormous applied power densities (>28 W cm−2) [8]
in inert gas environments. Thus, an encapsulation in inert
gases provides protection against oxygen in air and severe
environmental conditions. At the same time, encapsulation
lated in a small rigid enclosure (closed system), the pressure
variation is less sensitive to the gas parameters (except the
heat capacity ratio, γ=Cp/Cv) and has an opposite fre-
(2 ) quency dependence compared to the open system [6]
(g - 1 )
prms = Ph. (6 )
2 pfV
Here, V is the enclosed volume of gas. The smaller the
heated gas volume, the higher the sound pressure that can be
(3 ) of density ρp and thickness of hp provides a Q times higher
external sound pressure and magnifies efficiency by a factor of
Q2, where Q is the mechanical quality factor at a resonance [17]
rg (a ⋅ b) 1
p (r ) = ⋅Q⋅ ⋅ prms , (7 )
rp 2ph p r
(g - 1)2 rg Q 2
h= ⋅ Ph. (8 )
Despite the above-mentioned benefits from encapsulation
in an inert gas, the limited heat dissipation from the chamber
interior, which increases the ambient temperature, T0, restricts
the applied power density and Carnot’s efficiency. The use of
heat sinks, thermoelectric or water cooling only partially solves
the problem. This paper describes the merit of pulse excitation
that drives the encapsulated TA heater for a short portion of
(4 ) the sound period, thereby allowing sufficient time for heat
dissipation between pulses. Magnification of η by over a factor
3
Nanotechnology 29 (2018) 325704
of 104 is demonstrated by using different short pulse excitation
regimes for thin, flat TA heaters encapsulated in argon gas.
Three orders of this magnification, compared to the open sys-
tem, mostly comes from Q2 and the enhanced heat dissipation,
which allows a higher Ph and a higher temperature gradient,
Ta/T0. The further enhancement of energy conversion effi-
ciency (>10) comes from the reduced filling factor of the input
electrical signal and the shorter time for accelerating fluid
particles, which does not provide enough time for equilibrium
gas heating. The nonequilibrium process reduces the increase
of gas entropy and enhances the one-dimensional expansion of
enclosed gas [21].
3. Experimental methods
3.1. Temperature and blackbody radiation in vacuum and air
The precise measurement of temperature on the surface of
nanostructured materials is challenging. The temperature on
the surface of freestanding CNTS is controlled using an
infrared thermal imaging camera FLIR T650-sc (FLIR Sys-
tems, Inc.) with a preinstalled emissivity, ε. This temperature
from the thermal camera is compared with the temperature
obtained from the measurement of resistance change (R/R0
versus T) in the supported CNTS [8, 12]. The emissivity of
the CNTS was obtained from the optical transmission, τ,
using Kirchhoff’s law for thermal radiation (ε=α=1−τ).
Here, the dimensionless coefficient of absorption (or the
absorptivity), α, is the fraction of incident light (power) that is
absorbed by the body when it is radiating and absorbing in
thermodynamic equilibrium.
The thermal infrared camera is designed for precision
measurements of steady or slowly changing temperatures
with a frame rate of 30 Hz. The response frequency of the
microbolometer matrix in the T650-sc is 125 Hz (time
constant<8 ms). For periodic temperature waves with a
frequency higher than 30 Hz, the thermal camera measures
the mean value of temperature, which for pure sinusoidal
waves, corresponds to half of the peak value (the same as the
power, Ph=UpIp/2). For example, by applying 30 V (the
peak value is 42.4 V) of sinusoidal rms voltage to a free-
standing CNTS (S=1.25×2.3 cm2, R0=841 Ω) in air at
f>60 Hz, one can obtain Ta=T0−Tb=100 °C above the
ambient (background) temperature, Tb. But the peak value of
Ta measured at a frequency of 1 mHz is 186 °C, which is 7%
lower than the expected 200 °C, because of increasing power
consumption by blackbody radiation at peak temperatures. By
applying square wave signals of 30 V (Upeak=Urms= 30 V),
which creates constant, non-alternating temperature, the
thermal camera shows a temperature of 100 °C, the same as
for 30 V DC.
To control the shape of the thermal signal, the instant
amplitude of the modulated part of the temperature and the
evolution of the background temperature, we used a high-
speed infrared photo-detector (InAsSb amplified detector
PDA10PT from ThorLabs Inc., which has a wavelength range
from 1.0 to 5.8 μm, and a maximum sensitivity at 4.9 μm).
4
A E Aliev et al
Figure 1. Schematic diagram of the experimental setup. The
loudspeaker testing chamber is acoustically insulated from the
laboratory environment and ‘noisy’ high-power amplifiers using
absorptive fabric and anechoic foam wedges.
The temperature, T, and the intensity of the blackbody
radiation, UIR, in air and in a high vacuum (P<0.1 mTorr)
were measured in optical cryostat (VPF-475 from Janis
Research Company, LLC). One glass window of the cryostat
was substituted with an infrared-transparent germanium
window (2 mm thick and 50 mm in diameter) coated with an
antireflection film (Edmunds Optics Inc.) which has a wave-
length range of 3–12 μm. An absorption factor of 1.0424 was
used to correct the radiation intensity transmitted through the
Ge window.
3.2. Acoustic setup
The schematic diagram of the setup is shown in figure 1. A
sine or pulsed signal of frequency f (0–80 MHz) generated by
a Digital Function/Arbitrary Waveform Generator (Agilent
33250A), G, is amplified by a high-power amplifier, A, and
applied to the TA projector, TA, which is grounded through a
reference resistor (Rref=1 Ω) to control the applied current.
IntuiLink software was used for bipolar short pulses and
complex waveforms created and edited on a personal comp-
uter, PC. Three high-power amplifiers were used in this work
for different experimental conditions: (1) a wide frequency
range (DC to 250 kHz) high-voltage (±350 V) piezo-driver
amplifier PZD350-M/S (Trek Inc.) was used for high-impe-
dance projectors (R0=100 Ω–100 kΩ, output current of
±0.4 A); (2) a AE-Techron 8102 (AE-Techron Inc.) power
supply amplifier with a frequency bandwidth from DC to
5 kHz and a maximum power output of about 1100 W was
used for low-impedance projectors (R0<100 Ω); and (3) a
AG1012 (T&C Power Conversion, Inc.) 10 kHz to 10 MHz
RF generator and amplifier with an output RF power up to
1000 W was used for ultra-short pulse sequences (Δτ<
100 ns). The thermoacoustically generated sound pressure
was measured using precision microphones, Mic., with
Nanotechnology 29 (2018) 325704 A E Aliev et al

different apertures, A, and sensitivities, S:A=12.5 mm for an Table 1. Physical parameters of the studied CNTS.
ACO Pacific Model 7046 (2 Hz–20 kHz, S=54.3 mV Pa−1);
SWNT-2 MWNT-10 MWNT-50
A=6.25 mm for an ACO Pacific Model 7016 (5 Hz–
120 kHz, S=5.43 mV Pa−1); and A=3.1 mm for a B&K
25 22
Diameter of individual 2.2 10 50
pressure-field microphone 4138-A-015 (6 Hz–140 kHz, CNTs, nm
S=0.543 mV Pa−1). The SPL in air was measured using an Length of individual 1.5–3 200 600
CNTs, μm
impulse SPL-meter (Quest Technologies, Model 2700) pro-
Number of shells 1 7–8 40–50
vided with a QC-10 acoustic calibrator.
Sheet thickness, μm 0.05825 2022 ∼30
The charge conditioning preamplifier, CA (B&K Type Areal density, μg cm−2 6.025 1.522 9
2670 or ‘Sound Connect’ Listen Inc.), amplifies the 2f signal Optical transparency, % 85 82 44
and sends it to another high-gain, low-noise amplifier marked Sheet resistance, Ω/square 260 1000 150
LNA, which increases the detected signal further and sends it Thermal coefficient of +1.0 −0.7 −0.4
to the SR830 lock-in amplifier, LA (Stanford Research resistivity, ×10−3 K−1
Instruments, Inc.). The rms amplitude and phase delay of the Thermal conductivity, 500 50 28
second harmonic signal are read from the lock-in amplifier W m−1 K−1
displays. Simultaneously, the signal goes to a four-channel Rate of temperature rise, 53 63 8
digital oscilloscope DPO4104B-L, Osc. (Tektronix, Inc.), for K ms−1 (figure 5)
visualization and recording the shape of the received signal.
The detection of the second harmonic signal using the phase-
sensitive lock-in amplifier substantially enhances the signal/ DC, AC, and pulsed electrical power were applied to the
noise ratio. In addition to the reference resistor, Rref=1 Ω, freestanding CNTS and the dependence of temperature on
the shape of the pulse current was controlled by 120 MHz area-normalized input power was recorded in vacuum and in
air. Figure 2 shows that in air (a) and in vacuum (a) and (b)
bandwidth AC/DC current probe TCP0030A (Tek-
the temperature on the surface of resistively heated free-
tronix, Inc).
standing CNTS is a nonlinear function of applied power. In
vacuum, the main heat dissipation channel at T>100 °C is
blackbody radiation. In air, this nonlinearity is caused mainly
by the temperature dependent thermal conductivity of air,
4. Results and discussion
convection, and by blackbody radiation at elevated tempera-
tures. For aerogel-type transparent sheets at low frequencies
4.1. Thermal inertia of a freestanding TA heater
( f<1 kHz), we can neglect the heat capacitance of the
To provide results that enable optimization of signal proces- heater, Ch=0. Thus, in equilibrium, for a large heater
sing conditions (signal shape, duty cycle, number of pulses k1a?1 [6], the main dissipative mechanism of the applied
per period), we first studied the dynamics of temperature power, Ph, in air is the thermal conduction to air, κτ:
development on the surface of freestanding CNTS having Ph Ph
different average CNT diameters, sheet densities and number Ta = » . (9 )
- i4pa2 (wCh* + k1k) pf
of superimposed layers: (i) a single layer of a highly aligned 2Sk
multiwalled CNTS (MWNT-10) drawn from a chemical a
vapor deposition (CVD) grown forest (the height of the In the linearized energy balance equations [6, 8] we
aligned forest is ∼200 μm, the average diameter of the omitted the temperature dependence of the thermal conductivity
nanotubes is ∼10 nm, the optical transparency of the sheet (thermal diffusivity) of the gas. However, for large temperature
between 0.5 and 20 μm is ∼82%, and the thermal radiation variations, the temperature dependence of the thermal con-
emissivity is 0.18) [22, 23]; (ii) a single layer of an aligned ductivity, κτ ∝ T½ [28], and the thermal diffusivity, α=
multiwalled CNTS (MWNT-50) drawn from a forest grown κτ/ρgCg ∝ T3/2, should be restored. The substitution for an
by the floating catalyst based CVD method (the forest height ideal gas (ρ ∝ T−1 and Cg=(7/2)R, where R is the ideal gas
is 600 μm, the average diameter of the nanotubes is ∼50 nm, constant) gives Ta ∝ Ph/(ρgCgκτ)1/2 ∝ PhT1/4. However, for
the optical transparency of the sheet is ∼44%, and the air, which contains mainly the molecular gases N2 and O2, the
emissivity is 0.56) [24]; (iii) six layers of MWNT-10 sheets thermal conductivity has a much sharper temperature depend-
stacked parallel to each other, (MWNT-6L) (wherein the ence above 200 K (κτ =0.026 24·(T/300)0.8646 W m−1 K−1)
optical transparency of the sheet assembly is ∼44%, and the and the heat capacity gradually increases (Cg=1002.6+
emissivity is 0.56) [8]; and (iv) a single layer of a randomly 275×10−6·(T – 200)2 J kg−1 K−1) [29]. As a result, the
deposited freestanding single-walled CNTS (SWNT-2) grown strong increase in the thermal conductivity enhanced by an
by the floating catalyst based CVD method (the average increase in the heat capacity is almost totally balanced by the
diameter of the nanotubes is ∼2.2 nm, the optical transpar- decrease of air density, ρ ∝ 1/T. Thus, the power dependence
ency of the studied sheet is 85%, and the emissivity is ∼0.15) of Ta should deviate little from the linear dependence, Ta ∝
[25–27]. Some useful parameters of these CNTS are shown in Ph·T−1/20. In fact, we observe a much stronger deviation, Ta ∝
table 1 and details are described in S2 of SI. PhT−1/4. This behavior is shown in the inset of figure 2(a) as

5
Nanotechnology 29 (2018) 325704 A E Aliev et al

Figure 2. (a) Temperature at the middle of MWNT-10 surface in air (open blue circles) and in vacuum (open green circles for AC excitation
with f=100 Hz and open red squares for the DC excitation) versus applied power (normalized to the surface area of the sheet, S=1.25×2
cm2). The solid blue circles show the dependence of temperature on the power consumed only by thermal conduction to air, Ph−Pε (Pε is
the radiated power). The inset shows the linear dependence of T5/4 on the power consumed by air transport. (b) The temperature on the
surface of four freestanding CNTS in vacuum: SWNT-2; MWNT-10; MWNT-6L; and MWNT-50, as a function of the area-normalized input
power.

T5/4
a versus (Ph−Pε)/S. The quantity Ph−Pε is the power
consumed only by the thermal conduction to air, Pκ. To plot
Pκ(T), the power consumed by blackbody radiation for each
temperature point measured in vacuum (open green circles) was
subtracted from the applied power, Ph, for the corresponding
temperature measured in air (open blue circles). Apparently, the
small decline of the resulting curve from the linear dependence
(<5% for solid blue circles in the main panel of figure 2(a)) is
caused by a convective heat loss (air motion in a gravitational
field) from the vertically (or horizontally) oriented sheet. Note,
that convection is a secondary process producing a redistribu-
tion of the accumulated part of the heat. The natural convection
can be substantially reduced by partitioning the space near the
sheet using short fins (i.e., reducing the Rayleigh number [30])
or by placing the heater in a closed chamber, which is the case
for encapsulated TA projectors. The natural convection for the
vertically aligned MWNT-10 sheet (1.25×2 cm2) creates a
temperature gradient between the bottom and top edges of
∼25 °C for Ph=1 W and Ta=130 °C. Enclosing the sheet in
an aluminum foil chamber of 5×5×2 cm3 partially reduces
the temperature gradient to ∼15 K. At the same time, the par-
tition increases the absolute temperature in the center of the
sheet by ∼3%.
In general, the total heat loss from a TA heater includes
three components: the heat loss through thermal conduction
along the TA heater to the electrodes, Q0=−Aκl (∂T/∂x),
where κl is the thermal conductivity along the sheet having
cross-section A; blackbody radiation 2SεσT4, where ε is the
thermal radiation emissivity of the surface and σ=5.6×
10−8 W m−2 K−4 is the Stefan–Boltzmann constant; and the
heat loss through thermal conduction (κτ) to the surrounding
fluid, 2Sκτ [(Ta−T0)/l]=2SβT. Thus
dT dT
Ph = I 2R = rVCh - Akl + 2SesT 4 + 2SbT . (10)
dt dx
The 1 W of applied power heats up the 1.25×2 cm2
freestanding MWNT-10 sheet in air to ∼130 °C. The heat loss
by thermal conductivity along the sheet to the current elec-
trodes is the smallest term in equation (10), Pκl =
−2Aκl(∂Ta/∂x)≈0.0009 W, where κl =50 W m−1 K−1,
A=8×10−10 m2 is the cross-section area of the sheet with
thickness h=20 μm (h=40 nm for the densified structure),
and x=L/2=6.25×10−3 m is the distance from the sheet
center to the electrodes. The heat loss through radiation from
both sides of the sheet can be estimated as Pσ =2Sεσ
(Ta 4 - T0 4 )=0.096 W, where ε=0.18, S=2.5 cm2 and
T0=295 K. Assuming that the contribution of the convective
part is lower than 10% (the power that is needed to linearize
the solid blue dotted curve in figure 2(a)) and the contribution
of the thermal conduction along the nanotubes together with a
radiation heat loss at Ta =130 °C do not exceed 9.7%, one
can conclude that more than 80% of an applied AC power in
equilibrium is consumed by heat conduction to the sur-
rounding gas. In the case of a negligible heat capacity of the
heater, Ch=0, the heat exchange coefficient, β, can be
directly estimated from the slope of T(Ph) at low tempera-
tures. For the MWNT-10 sheet (see figure 2(a)), β=(Ph−
Pκl−Pσ−Pconv)/2S(T−T0)≈18 W m−2 K−1.
Despite the higher surface temperatures found for
SWNT-2 sheet at high input power (figure 2(b)), the slope at
low input power is essentially the same as for the MWNT-10
sheet suggesting very close thermal inertia. However, the heat
exchange coefficient, β=0.16 W m−2 K−1, for the SWNT-2
sheet (with 85% transparency) is little lower than for the
MWNT-10 sheet, mostly due to the higher density of the
SWNT-2 sheet, ρ=6 μg cm−2 [25], and the extremely high
thermal conductivity along the sheet, κl ≈500 W m−1 K−1,
which provides a one order higher heat dissipation to the
electrodes, Pκl =0.012 W (ΔT=100 °C). Note, the thinner
SWNT-2-90% and SWNT-2-95% sheets, having the same

6
Nanotechnology 29 (2018) 325704
volumetric density as SWNT-2-85% but much better access
to the fluid, show superior TA performance (will be published
elsewhere).
In vacuum, there are only two heat dissipation channels:
thermal conduction along the nanotube sheet to the electrodes
and blackbody radiation. At a low applied power
(Ph/S<0.1 W cm−2), the TA heater temperature, that is
mainly determined by the heat capacitance of the sheet, ρhCh,
increases linearly. We will use this fact for a direct mea-
surement of ρhCh for a freestanding sheet. At higher tem-
peratures, blackbody radiation becomes the main contributor
to the T(Ph) decline.
To avoid the development of a temperature background
in the above experiments, we intentionally used a low fre-
quency excitations (an applied AC voltage of frequency
100 Hz corresponds to a temperature modulation frequency of
200 Hz). At the same time, to obtain a stable reading (to avoid
the flickering of the thermal camera image), the temperature
modulation frequency was chosen much above the frame rate
of the thermal camera ( ffr=30 Hz) and the response fre-
quency of the microbolometer matrix of T650-sc, f∼125 Hz.
The measured temperature versus applied AC power, both in
air and in vacuum, is consistent with the temperature com-
puted from the resistance change technique that correlates a
decrease in resistance (for MWNT-10) to increasing
temperature.
4.2. Shape of the thermal pulse
To improve heat dissipation from the interior of the nanos-
tructured heater, the pulse duration should be short enough to
give sufficient time for heat relaxation to return to the back-
ground level. Following the sampling theorem of digital
signal processing, we introduce here a thermal sampling cri-
teria for short pulse excitation of an encapsulated TA pro-
jector. The sampling interval, Δt, depends on the acoustical
signal bandwidth, Δfa, the thermal inertia of the filled gas,
ρgCg, and the heat exchange coefficient, β:
1 Hrg Cg Hrg Cg
Dt = + » . (11)
2Dfa 2b 2b
The new term, HρgCg/2β, derived from the heat balance
equation (10) in the equilibrium state
dT
VrCh = 2 S bT , (12)
dt
introduces the time that is necessary for the temperature to
return to the background level after each pulse excitation.
Since Cg?Ch, here we replace Ch with Cg. The temperature
response for a short electrical pulse is limited by the heat
capacitance of the gas layer within the thermal diffusion
length and the heat exchange coefficient.
The development of the thermal pulse measured for three
types of freestanding CNTS in high vacuum and air shows that
the heat capacitance of the gas governs the rising and falling
edges of the heat pulse. The heat capacitance of a single
freestanding MWNT-10 sheet (S=1.25×2.0 cm2), estimated
in vacuum from the slope of the T(Ph) plot in figure 2(a), and
7
A E Aliev et al
Figure 3. The shape of the blackbody radiation signal (open blue
circles) and the temperature rise (open red squares) versus pulse duty
cycle for a MWNT-10 TA heater. The temperature was recovered
from blackbody radiation intensities using the mechanically chopped
radiation of a DC excited TA heater. The black dashed line shows
the shape of the applied voltage for a duty cycle of 10%, f=10 Hz,
and Ph=0.35 W. T∞=220 °C for a duty cycle of 100%.
using the reduced equation (9), to be ρhCh=Ph/SωTa, is
∼0.01 J m−2 K−1. Hence, the specific heat capacity of the
MWNT-10 sheet for the previously measured density per unit
area, ρ=0.015 g m−2, is Ch=670±30 J kg−1 K−1. The
heat capacitance of the air layer involved in the heating process
depends upon the pulse duration, Cg·H, where Cg=
1006.3 J kg−1 K−1=1191.5 J m−3 K−1 is the specific heat
capacity of air at T=295 K and H=h+2l is the thickness
of the heater, h=∼20 μm, and l = 2a/w = 2at is the
thermal diffusion length. For a pulse width of 1 ms (1% duty
cycle at pulse repetition rate of 10 Hz), l=0.21 mm and
Cg·H=0.52 J m−2 K−1. Hence, the heat capacitance of air is
∼50 times greater than the heat capacitance of the TA heater.
In the case of a pulse width for a high frequency alternating
carrier-pulse sequence of 1% duty cycle at f=35 kHz, that
corresponds to a 286 ns pulse, in which the pulse amplitude is
modulated by an acoustic frequency envelope, l=3.5 μm.
Now the heat capacitance of the air layer, H=27 μm, is much
lower,∼0.032 J m−2 K−1, and comparable with the heat
capacitance of the CNT heater.
Figure 3 shows the infrared signal, UIR, radiated from the
surface of the MWNT-10 TA heater and the temperature, T,
recovered from peak amplitudes of the radiated signal by
comparing it with mechanically chopped radiation from a DC
excited heater, over a range of pulse duty cycles in vacuum.
The linear increase of temperature (highlighted by green
dashed line) and the quadratic rise of blackbody radiation
signal (for a given spectrum of infrared detector) are typical
for low applied power.
For TA sheets in vacuum, figure 4(a) shows the rise of an
infrared signal’s pulse front from the surface of SWNT-2,
MWNT-10, MWNT-6L, and MWNT-50 sheets (S=2.5 cm2)
as a function of pulse width (duty cycle in %). Since the
intensity of the infrared signal for a narrow bandwidth receiver
(1–5.8 μm) is proportional to the square of the applied power
(which was verified for all studied CNTS), the normalization to
Ph2 allows comparison between different TA heaters. The
Nanotechnology 29 (2018) 325704 A E Aliev et al

Figure 4. (a) The shape of the radiated signal at an increasing pulse width for four CNTS placed in vacuum. The frequency of the pulse
sequence is 10 Hz. The applied pulse power of 0.61 W, 0.48 W, 0.85 W, and 0.8 W are chosen to reach the maximum temperature of 220 °C
at a 100% duty cycle. (b) The comparison of the pulse shape for MWNT-10 placed in vacuum and air.

Figure 5. The shape of the received blackbody radiated signal recorded for increasing duty cycles: (a) MWNT-10 sheet excited by 20 V
pulses; (b) SWNT-2 sheet excited by 13 V pulses; and (c) MWNT-50 sheet excited by 10 V pulses. The duty cycle in % for a pulse sequence
of 10 Hz corresponds to the pulse width in ms, i.e., a 1% duty cycle corresponds to a pulse width of 1 ms.

obtained pulse development curves show the highest rate of
temperature increase for MWNT-10 with a lower density of
sheet than for SWNT-2 with smaller diameter of nanotubes, but
much higher density of the sheet. Along with small diameter,
the spacing between adjacent nanotubes is a crucial property
for TA heaters.
Figure 4(b) shows the development of the infrared signal
as a function of pulse width for MWNT-10 in vacuum
(UDC=16.24 V, Ph=0.376 W, T∞=220 °C) and air
(UDC=41.2 V, Ph=2.2 W, T∞=220 °C). For clarity, the
signal intensity in air is magnified by a factor of 100.
Figures 5(a)–(c) show the shape of the front and rear edge of
the infrared signal for MWNT-10, SWNT-2, and MWNT-50
in vacuum for an increased pulse duty cycle recorded directly
from the digital oscilloscope. For details of light-to-voltage
conversion, see S8 in the SI.
For low temperatures on the surface of a TA source
(Δt<2 ms in figures 3–5), when the blackbody radiation can
be neglected and two other heat dissipation channels in the
vacuum are canceled, one can again estimate the heat capa-
citance of the freestanding sheet from the slope of the
temperature rise versus time using equation (10):
Ph/S=ρhCh(dT/dt). The temperature slope in figure 3 at
input power of 0.35 W gives ρhCh=0.01 J m−2 K−1 for the
above MWNT-10 sheet. Control measurement of pulse slope
in air gives the precise specific heat capacity of air,
1010 J kg−1 K−1 (ρhCh=150 J m−2 K−1, where the thick-
ness of the air layer involved in the heating process is
l=(2αΔt)1/2=0.8 mm, and Ta=100 °C). Thus, monitor-
ing the blackbody radiation from the freestanding TA heater
in air and vacuum is a very sensitive and powerful tool for
characterization of nanostructured materials and the thermal
management of TA devices.
4.3. Pulse excitations of encapsulated TA projectors
Two pulse excitation regimes have been examined and
compared with a continuous sinusoidal excitation: (1) SP per
vibration displacement with a constant amplitude; (2) high
frequency alternating carrier-pulse sequence in which the
pulse amplitude is modulated by an acoustic envelope. In both
regimes, an extended linear dependence of the acoustic

8
Nanotechnology 29 (2018) 325704 A E Aliev et al

Figure 6. Excitation of a TA transducer by short pulses (blue lines show the pulse sequences and black dashed lines show the displacement of
the vibrating plates): (a) constant amplitude and width bipolar pulses (SP) synchronized with the resonant vibration of the plate; (b) single
(bipolar or unipolar) constant amplitude and width pulses per several periods of mechanical vibration; (c) pulse width modulation (PWM)
using constant pulse amplitude; (d) high frequency alternating carrier-pulse sequence in which the pulse amplitude is modulated by an
acoustical envelope (PAM).

pressure with respect to the applied power and an increased

energy conversion efficiency were observed.

4.3.1. Single pulse (SP) per vibration displacement. SP

excitation with constant width and amplitude is the simplest
excitation suitable for encapsulated TA projectors to generate
monochromatic sound. Several regimes are available,
depending on the quality factor, Q, of the vibrating plates:
(i) single bipolar pulses per period for low Q (figure 6(a)); and
(ii) SP per several periods for high Q (figure 6(b)). Bipolar
excitation is preferable at high frequencies to avoid the
development of a mean voltage (current). Despite significant
distortions, this regime is extremely efficient in air and Figure 7. The amplitude-modulated carrier sinusoidal signal has a
underwater. 1.4 times lower filling factor than the directly applied sound
envelope.
4.3.2. Short pulse sequence modulated by an acoustical
signal. In a previous paper [6] we have demonstrated that applied power. The enhancement depends on the quality
the sound pressure generated by an encapsulated TA projector factor of the vibrating plates averaging the internal pressure
is enhanced nearly two-fold by using a high frequency oscillations. This mode of excitation enables generation of
sinusoidal carrier signal ( fc∼50 kHz), whose amplitude is any arbitrary acoustic signal defined by the modulated signal.
modulated by an acoustic signal envelope (AM) (figure 7). If the sinusoidal carrier filler is substituted with short,
The enhancement results from the reduced filling factor of the uni- or bipolar pulses, then the fill-factor can be substantially
envelope, which is (2)1/2 for voltage and a factor of 2 for reduced. For the optimized carrier signal frequency, the pulse

9
Nanotechnology 29 (2018) 325704 A E Aliev et al

Figure 8. (a) The spectra of the input power-normalized sound pressure for four types of excitations of a glass plate TA projector which has a
mechanical resonance of the vibrating plates at ∼750 Hz: (1) the blue line is for the simple sinusoidal excitation at an applied frequency of
375 Hz; (2) the red line is for a 35 kHz sinusoidal carrier signal modulated by a 750 Hz sinusoidal sound envelope (AM); (3) the green line is
for single bipolar pulse excitation at 750 Hz, with a pulse duty cycle of 1% (SP); (4) the brown line is for a 35 kHz pulse sequence with a duty
cycle of 1% AM modulated by a 750 Hz sinusoidal signal (PAM). The distance between the vibrating plate and the microphone (model 7046
with type 2618 preamplifier) was 3 cm. (b) The vertically placed two-glass-plate TA projector comprises two 12.7×17.8 cm2 glass plates
with a thickness of 1.85 mm and four layers of freestanding MWNT-10 sheets (R=370 Ω) suspended 1 mm apart between the two-glass
plates.

duty cycle can be adjusted so that after each short pulse, the
system has ample time to cool down to near room temperature
(see equation (11)). Hence, the large temperature gradient and
energy conversion efficiency is preserved.
There are three types of pulse modulation to consider:
(i) amplitude modulation of a short pulse sequence (PAM,
see figure 6(d); (ii) pulse width modulation (PWM, see
figure 6(c), and (iii) the pulse density modulation (PDM),
where the frequency of constant width short pulses is
modulated by a sound wave. While PWM and PDM permit
very simple, efficient electronic circuitry, and commonly used
for excitations of flat, coil loudspeakers (like found in
postcards), here we focus attention on PAM for potentially a
more efficient excitation providing much lower distortion.
In what follows, we examine short pulse excitation by
spectra, acoustical sensitivity, and efficiency for MWNT-10
based encapsulated TA projectors with vibrating plates of
various size, stiffness and mass.
4.3.3. 12.7×17.8 cm2 two-glass-plate TA projector. The
enhanced output sound pressure and efficiency of
encapsulated TA projectors results from an optimal balance
between elastic modulus and mass of the vibrating plates,
providing a high quality factor, Q, and an acoustical impedance
mismatch (∝ ρf υf/ρpυp), according to equations (7) and (8).
Figure 8(a) demonstrates the increases of sound pressure at
resonance ( fr∼750 Hz) for four different types of excitations.
Sound pressure for the TA projector shown in figure 8(b) was
measured when the device was horizontally suspended on a
marble table covered with a sound absorber (1 cm separation
between the bottom plate and the absorber). Large area,
1.8 mm thick glass plates sealed at the edges by highly elastic
silicone strips (see for details S2 in SI) provided a high quality
factor, Q≈36. PAM is found to be the most efficient
excitation for the given combination of Q and impedance
mismatch for this device.
The sound pressure for a simple sinusoidal excitation,
which should linearly increase with applied power for low
power levels (equations (6) and (7)), starts to degrade at a
relatively low power (open blue circles in figure 9(a)). In
contrast, SP excitation maintains the expected linear depend-
ence up to the highest power (∼65 W) available for this TA
projector, which has a high input resistance (two electrode
devices). The acoustic sensitivity (prms/Ph) increased by a factor
of 1.8 times for the SP excitations (figure 9(b)) compared to a
sinusoidal excitation. The sensitivity of the device increases for
shorter pulses and saturates at duty cycles below 10% (see
figure S2.3(a) in SI). SP excitation at sub-harmonic frequencies
(see S3.1 in SI for 375, 250, 187. 5, 150 and 125 Hz) does not
change sensitivity. Due to the high mass of the vibrating plates,
the highest achieved sensitivity, prms/Ph=0.73 Pa W−1 (r=
3 cm)=60.8 dB W−1 (r= 1 m), and efficiency, η=1.4% (for
Ph=65 W), are relatively low.
The theoretical value for sound pressure can be estimated
from equations (6) and (7): the internal pressure generated by
the encapsulated TA heater at an applied power of 1 W at
f=750 Hz is ∼2.65 Pa, whereas the external sound pressure
at the centers of two resonantly vibrating plates calculated at a
distance of 3 cm is ∼3.3 Pa W−1 (see calculations on page 12
in S3 of SI). The measured value for a horizontally suspended
TA projector driven by the most efficient PAM excitation was
∼1.56 Pa W−1 (see figure 8(a)). Apparently, the damping of

10
Nanotechnology 29 (2018) 325704 A E Aliev et al

Figure 9. (a) The dependence of sound pressure on input power that is provided by a two-glass-plate TA projector using SP excitation at
f=750 Hz for an increasing duty cycle: 1%; 2%; 4%; and 10%. The sound pressure generated by continuous sinusoidal excitation (open
blue circles) is shown for comparison. (b) The acoustic sensitivity versus applied power for the TA projector vertically suspended in an
anechoic room. The distance to microphone is 3 cm.

Figure 10. (a) The sound spectrum of a single plate TA projector comprising the 15×15 cm2 carbon-fiber composite plate shows broad
resonance peaks at 430 Hz and 910 Hz. The inset shows an extended view at the main resonance peak. (b) A picture of the TA projector with
the microphone (model 7046, ACO Pacific, Inc.) placed 3 cm above the projector. The inset shows a thermal image of the projector at
Ph=20 W.

plate vibrations by the silicone sealant and the entrained
gas substantially reduces the external sound pressure, causing
this discrepancy. Note the contribution of the entrained gas
increases for thinner enclosures (see equation (S3.3) in [6]).
4.3.4. 15×15 cm2 carbon-fiber composite plate on a
heatsink. The relatively lighter 15×15 cm2 carbon-fiber
composite plate has a much lower quality factor (Q=18.2)
and a much broader resonance peak (figure 10(a)). This TA
projector comprised five superimposed layers of freestanding
MWNT-10 sheets attached to an array of interdigitated
copper-foil electrodes. The MWNT sheets were encapsulated
in argon gas between a carbon-fiber composite plate (15×
15× 0.08 cm3, Dragon Plate Inc.) placed on the top of the
device and a heatsink (15×15 cm2, Aavid thermalloy LLC)
on the bottom (figure 10 (b)). The input resistance of the
device was 18.4 Ω (the sheet resistance of MWNT-10 is 2000
Ω/,) and fr=910 Hz (see details in S4 of SI).
The sound pressure and efficiency (figure 11(a)) linearly
increase with increasing power up to the highest applied power
available for short, SPs (164 W) and for a modulated sequence
of pulses (156 W). Despite the lower sound pressure achieved
for sinusoidal excitation, the prms(Ph) curve departs from a
linear dependence at a relatively high applied power of ∼20 W.
This is attributed to the enhanced heat dissipation through the
sink. Note, the higher carrier frequency for AM excitation (open
green triangles for fc=5 kHz and solid green triangles for
fc=50 kHz in figure 11(b)) results in a higher sound pressure.

11
Nanotechnology 29 (2018) 325704 A E Aliev et al

Figure 11. (a) The sound pressure and efficiency of the carbon-fiber-plate TA projector versus applied power for three different types of
excitations: continuous sine wave (blue circles, open for sound pressure and solid for efficiency, respectively); SP with duty cycles of 10%
(red circles); and PAM with fc=50 kHz and duty cycles of 10% (green squares). (b) The sensitivity of a TA projector versus applied power
for Sine, AM, SP and PAM excitations. For these measurements, fr=910 Hz and r=3 cm.

Figure 12. (a) The sound spectra of the single plate TA projector #3, comprising a 7.6×6.6 cm thin (250 μm) AlN ceramic plate on top
2

and an aluminum heatsink on the bottom. The fundamental resonance peak is at 1366 Hz. (b) Top view of the ceramic plate TA projector.

Perhaps at higher carrier frequencies, when the acoustical
wavelength of carrier excitation becomes comparable to the
heater-plate distance, both regimes of open and closed systems
described by equations (2) and (6) are combined.
The efficiency of a carbon-fiber-plate TA projector in air
reaches 0.24% at the highest available applied power of
164 W (solid green squares in figure 11(a)). This efficiency
was calculated using a direct method for near field sound
pressure (see figure S4.2(b) in SI). The sensitivity of the
device is twice as higher for a SP and PAM excitations
(0.47 and 0.48 Pa W−1 in figure 11(b)) than for sinusoidal
excitations. However, the lower density of the vibrating plate
(∼1.33 g cm−3) is totally canceled by two-times lower quality
factor of plates in this device, and results in a sensitivity close
to the glass plate TA projector. Nevertheless, at a maximum
applied power of 164 W driven with the SP-10% excitation, a
SPL reached 120.5 dB re 20 μPa @ 1 m distance.
4.3.5. 6.6×7.6 cm2 AlN ceramic plate on heatsink. The
argon-filled aluminum nitride (AlN) ceramic plate TA projector
comprises four superimposed layers of freestanding MWNT-10
sheets attached to an array of interdigitated copper-foil
electrodes. The MWNT sheets were encapsulated between an
AlN ceramic plate on top (7.6×6.6×0.025 cm3) and an
aluminum heatsink on the bottom. Six TA projectors with an
input resistance between 40 and 60 Ω were assembled with
different attachments of the ceramic plate to the supporting
sealing silicone strips (see details in S5 of SI). The low mass
and thickness of the plate and the high bending modulus of the
used AlN ceramic results in a high resonance frequency,
fr=1366 Hz, and a high quality factor for mechanical
vibrations, Q=43 (figure 12(a)).
The encapsulated AlN ceramic TA projector generates a
sound pressure that is 33 times higher than for an equivalent
open system (green dashed line in figure 13(b)) for a sinusoidal

12
Nanotechnology 29 (2018) 325704 A E Aliev et al

Figure 13. The sound pressure (a), sensitivity (b) and efficiency (c) of the encapsulated AlN ceramic plate TA projector (Pyramid #3) versus
applied AC power for four types of excitations measured in air: a continuous sine wave (open blue circles); AM excitation with sinusoidal
carrier signal ( fc=35 kHz) modulated with a sinusoidal envelope of fr=1366 Hz (open red squares); SP excitation using three different
duty cycles: 0.1%; 1%; and 10% (green triangles); and PAM excitation with a duty cycle of the short pulse sequence of 10% at fc=50 kHz.
For these measurements fr=1366 Hz and r=3 cm. (d) The sound pressure in water versus applied AC power for sinusoidal and SP types of
excitations with fr=315 Hz, where the sound pressure was measured at a distance of r=3 cm (Pyramid #4).

drive, mostly due to a high quality factor of the vibrating plate,
Q=43. Further enhancement of sound transduction sensitivity
(figure 13(b)) and energy conversion efficiency (figure 13(c))
were achieved by lowering the filling factor of the excitation
signals, which provides another factor of ∼2.9 increase in
sensitivity and more than one order in efficiency. A deviation
of pulse shape from rectangular at a duty cycle <1% results in
a decline of the sound pressure (top curve, open dark-green
triangles) at a high applied power. The PAM excitation with a
pulse duty cycle of 10% at fc=50 kHz produced a sound
pressure of ∼140 Pa (r=3 cm) and a SPL of ∼120 dB re
20 μPa @ 1 m. The energy conversion efficiency calculated by
the direct method using equation (3), with the sound pressure
measured at the surface of the vibrating plate (r=1 mm and
S=50 cm2) and an applied power of 42 W (open brown
squares in figures 13(a)–(c)) reaches 8.7%. Note, an efficiency
of 15% obtained from the sound pressure measured in the far
field at the front of the TA projector (20 Pa at r=1 m) is an
overestimated value due to the radiation pattern of the flat
projector [31].
The double-plate radiating TA projector composed the
same AlN ceramic plates placed in water shows broad-band
spectra (Q<26) with a resonance frequency reduced down to
∼328 Hz (see details in S6 of SI). Apparently, much higher
sound pressure (prms>10.6 kPa at r=3 cm), sound pressure
level (SPL=170 dB re 1 μPa @ 1 m) and sensitivity
(prms/Ph>250 Pa W−1) results in water from the improved
impedance matching and cooling conditions. At a sound
pressure of ∼10 kPa ( f=315 Hz) we observed bubble
nucliation (cavitation) at the center of the vibrating plate,
which spreads across the whole plate at prms∼10.5 kPa (see
figure S5.4 (d) in SI). Above this sound pressure the cavitation
cloud penetrates deeper into the water medium (up to 10 cm)
and limits the radiated acoustic power. Rough estimation of

13
Nanotechnology 29 (2018) 325704
Figure 14. (a) The sound spectra of a two-plate TA projector comprising a single layer of a freestanding MWNT-10 sheet encapsulated
between two thin (96 μm) mica plates (5×5 cm2). The insets show the expanded spectrum of the main resonance peak at 6110 Hz. (b) The
first three plate vibration modes. The insets show the modal patterns whose relative frequencies are shown on the bottom of the squares.
cavitation limited sound intensity in the near field for a rigid
clamped plate [32] confirms that the acoustic pressure
amplitude exceeds atmospheric pressure (101 kPa).
A larger size TA projector (11.43×11.43 cm2), with
0.5 mm thick AlN ceramic plate on a heatsink driven with SP-
10% excitation provides superior sound power at lower
frequencies (890 Hz in air and 240 Hz in water), with sound
pressures above 12 kPa in water limited by intense cavitation
(see details in S7 of SI). In this device we modified the
attachment of MWNT sheet to the interdigitated copper
electrodes by densification. At high applied AC current or
voltage the abundance of free dangling CNTs and nanotube
bundles in a freestanding aerogel sheet creates uncontrollable
nanoscale shortages at terminals. Tiny sparks eventually
develop into larger openings, which propagate along the
electrodes. To avoid the dangling CNTs we densified the
freestanding sheets by using methanol vapor. The densifica-
tion at electrodes (see figure S7.1(c)) substantially enhanced
the robustness of the TA heater’s structure and significantly
increased the power limit.
The sound pressure, sensitivity and efficiency enhance-
ments in water by using rigid vibrating plates and short pulse
excitations is the same as in air. This might be attributed to
the high density and thicknes of studied vibrating plates (see
equation (7)). Below, we evaluate the short pulse excitations
in encapsulated TA projectors composed light-weight mica
membranes.
4.3.6. Light-weight mica membranes. Highly transparent
(T=90%) plates of mica (H20-5691BS from Fire
Mountain Gems and Beads Inc.) with a thickness of
80–160 μm, show excellent mechanical and thermal
properties. The low density (ρ=1.6–3 g cm−3), excellent
flexural strength (>180 MPa) and high thermal resistance
(>750 °C) of muscovite and phlogopite mica sheets make
14
A E Aliev et al
them a suitable for vibrating membranes of flexible and
transparent loudspeakers and underwater TA projectors.
The resonance frequency of a square plate with rigid
clamped edges for the first fundamental (0, 0) mode is
fr=B·υl·h/L2=457 Hz, where h=96 μm is the plate
thickness, Lx=Ly=5 cm are the plate dimensions, υl =
7200 m s−1 is the longitudinal sound velocity, and B=1.654
is the coefficient of the Bessel function, determined by the
boundary conditions [33]. The longitudinal sound velocity
was chosen from a list of 36 different Muscovites [34] to fit a
measured 457 Hz resonance frequency for the first funda-
mental mode. The layered structure of the mica plates
provides a low bending modulus. The sound pressure of the
first three resonances, corresponding to the (0, 0), (1, 0), and
(1, 1) modes, is low (see figure 14(b)).
The main resonance peak of the device at 6110 Hz has
relatively high sound pressure per input power and a quality
factor Q=18 (see details in S8 of SI). The short pulse
excitation of these light and thin mica membranes
(h=96 μm, S=5×5 cm2) significantly increases the
sensitivity of the encapsulated device (open squares in
figure 15). The increase of energy conversion efficiency to
0.45% at an applied power of 35.7 W is due to the mechanical
response of the light membrane to the pressure variation
created by short thermal pulses. With a total weight of 2.5 g,
the 5×5 cm2 TA projector radiates in air the SPL of 100 dB
re 20 μPa at 1 m and 170 dB re 1 μPa at 1 m underwater.
For all above examples of flat encapsulated TA
projectors excited with single short pulses (SP excitation) or
sequence of short pulses modulated with a sound envelope
(PAM excitation), we observed enhancement of sound
pressure and energy conversion efficiency compared to
sinusoidal excitation. For the case of a thin, light AlN
ceramic plate (see Pyramid #3 in figure 13), this enhance-
ment reaches 2.7 and 7.2 fold, respectively. For sinusoidal
excitation, the slow motion of the flat piston provides enough
time to thermalize the energy of excited gas particles across
Nanotechnology 29 (2018) 325704 A E Aliev et al

Figure 15. The power dependence of sound pressure (a) and sensitivity (b) for an argon-filled two-plate TA projector comprising a single
freestanding MWNT-10 sheet (R=1680 Ω) encapsulated between two 96 μm thick mica membranes (S=5×5 cm2) for two types of
excitations: a continuous sine wave (open blue circles) and pulsed sine wave (tone burst) with a pulse frequency of 10 Hz and duty cycle of
7% (PS-7%); and SP excitation using two different duty cycles: 2% and 10% (open green and red squares, respectively). The sound pressure
and sensitivity measured in air at a resonance frequency of 6110 Hz and at a distance of 3 cm. The inset to (a) shows the shape of the sound
pressure signal for the sine tone burst excitation PS-7%.

the enclosure. In this case, the gas particles can be treated as a
three-dimensional thermodynamic system, reaching equili-
brium during the sound period. Equation (6) with γ=5/3
(for monoatomic gas as argon) well describes this case. For
short pulse excitations, the motion of accelerated particles can
be treated as effectively a one-dimensional gas in none-
quilibrium state, since insufficient time is provided for
thermalization across the thin, flat enclosure. Accelerated
particles do not lose energy by thermalizing the system in the
horizontal direction and behave as an adiabatically expanding
gas with γ=3 [21]. The rate of entropy increase is also
reduced. The sound pressure and efficiency for this none-
quilibrium one-dimensional thermodynamic system, as
roughly estimated by equation (6), are increased by factor 3
and 9, respectively, which are very close to the observed
results.
5. Conclusions
Small diameter CNTs in a low density, porous, freestanding
sheet are preferable for TA sound generation. Therefore,
aerogel structures enabling enhanced fluid accessibility to the
inner parts of the freestanding sheet provide the highest TA
performance. To minimize internal thermal interaction, the
distance between separate nanostructures in an aerogel TA
heater should exceed the thermal diffusion length of the
surrounding fluid. The comparison of three aerogel CNTS
containing differing diameter nanotubes shows that the
aerogel structure of MWNT-10 sheets (D=10 nm with 7–8
shells) which has an average distance between individual
nanotubes and bundles in the freestanding sheet greater than
20 μm, provides superior TA performance to SWNT-2
(D=2.2 nm) and MWNT-50 (D=50 nm) sheets. Despite
similar to MWNT-10 optical transparency, ∼85%, SWNT-2
has much denser sheet structure (two orders thinner) and one
order higher thermal conductivity along the sheet. However,
the thinner SWNT-2%-90% and SWNT-2%-95% sheets,
having the same volumetric density as SWNT-2%-85% but
much better access to the fluid, show enhanced TA perfor-
mance. Moreover, the metallic temperature dependence of
resistivity, the one order lower sheet resistance, and the
unique mechanical properties of the studied SWNT films
make them very attractive for high frequency TA transduction
(>100 kHz) and underwater applications (higher hydro-
phobicity). On the other hand, the ability of large diameter
nanotubes, like MWNT-50 (D≈50 nm), to avoid bundling
and the extended length of individual nanotubes (∼0.6 mm)
make them very attractive for low frequency TA projectors
( f<1 kHz) working in harsh conditions of high static
potentials and strong vibrations.
It was demonstrated that blackbody radiation from the
resistively heated CNTS can be used as a very sensitive tool
for examining the modulated and accumulated parts of the
heat on freestanding nanostructured TA heaters and for
extracting the thermal inertia and heat capacity of the heater
material. The precise control of heat dissipation channels
(internal heating of the sheet, thermal conduction to the
electrodes along the heater and to the fluid, and blackbody
radiation) allows direct estimation of the TA (heat-to-sound)
coupling coefficient.
Encapsulation of a nanostructured TA projector heater
working in harsh conditions in air and underwater is neces-
sary to achieve desirable robustness and efficiency. Employ-
ing rigid, vibrating plates softly clamped on the edges, an
approximately 50-fold improvement of device sensitivity
(prms/Ph) and a more than four order enhancement in energy
conversion efficiency, compared to the open system, were

15
Nanotechnology 29 (2018) 325704
demonstrated. These enhancements were achieved by using
short pulse excitation that mostly improves the heat dissipa-
tion conditions from the interior of the encapsulated device
and allows higher input power densities.
There are a number of geometric and thermodynamic
parameters of the projector heater, chamber and the vibrating
plate to consider in the design of high-performance TA pro-
jector. However, the crucial step to overcoming the steady-
state fundamental efficiency limit of open TA system is in
encapsulation and using short pulses having a low duty cycle
(<1%). The rapid acceleration of fluid particles within the
front of short thermal pulses creates nonequilibrium one-
dimensional expansion of the filled gas. The accelerated
particles do not have enough time to create an equilibrium
state. Hence, the entropy of the system is lower than for the
equilibrated system. These particles create stronger impact on
a vibrating plate placed within the thermal diffusion length.
The energy conversion efficiency of this non-ergodic engine
is still below the thermodynamic limit of Carnot’s cycle, but it
is much higher than the fundamental efficiency of either open
or closed systems.
Short temperature pulses at an applied power of 10 kW
per pulse can permit over 10% energy conversion efficiency
in air and over 60% efficiency underwater. Since the SPL for
an encapsulated device is independent of the properties of
enclosed inert gas, other than the heat capacity ratio
γ=Cp/Cv, the use of helium should improve heat dissipa-
tion [35] for extreme conditions of high-power >10 kW per
pulse.
Using a pulse excitation (PAM, PWM, PDM), one can
generate an acoustic output having the same shape as the
input signal, and use linear signal processing techniques to
avoid the distortions specific to second harmonic signals and
the need for biased excitation current.
The obtained SPL at relatively low volume and mass of
encapsulated TA projectors by using short pulse excitations
show promising opportunities for noise cancelation, fire
extinction and low frequency sonars.
Acknowledgments
We thank JNC Corp. for providing spinnable MWNT-50,
Juan Pablo Oviedo Robles (Material Science Engineering, UT
Dallas) for the TEM images of MWNT-50. SAR and AGN
acknowledge the Russian Science Foundation for the support
by project # 17-19-01787. This research work was supported
by the Office of Naval Research grants N00014-14-1-0152,
N00014-17-1-2521, Army Research Office STTR Contract
#W911NF-15-P-0023, and the Robert A Welch Foundation
Grant AT-0029.
Author contributions
The manuscript was written with contributions from all
authors. All authors have given approval to the final version
of the manuscript.
16
A E Aliev et al
Figure A1. The time dependence of the normalized temperature
modulation of a CNT sheet when driven by an AC current,
I0·sin(ωt), is shown for different values of ωτ, where τ is the
characteristic heat dissipation time constant. The fluctuation
amplitude reaches a maximum of 2 at the limit ωτ→0, and shrinks
to a narrow line on the background of [T(x, t)−T0]/Δ0=1
as ωτ?1.
Notes
The authors declare no competing financial interest.
Appendix
The applied current Io·sin(ωt+jo) generates heat at an
angular frequency of 2ω. The power applied to the sheet,
Ph=I2·R·sin2(ωt)=I2·R·(1−cos(2ωt))/2, results in
a temperature increase, ρh·Ch(dT/dt), if there are no other
heat dissipation channels, where ρh is the density and Ch is the
heat capacity of a resistively heated TA heater. The full heat
balance equation in vacuum is:
¶ ¶2
I 2R sin2 wt = r h Vh Ch T (x , t ) - k 2 T (x , t )
¶t ¶x
+ 2AesT 4 (x , t ) , (A1)
where κ is the thermal conductivity along the sheet, ε is the
emissivity, and σ is the Stefan–Boltzmann constant.
For small temperature fluctuations, the temperature
change on the surface of CNT sheet is [36]:
⎡ ⎤
sin (2wt + j) ⎥
T (x , t ) - T0 = D 0 ⎢1 - , (A2)
⎢⎣ 1 + (2wt )2 ⎥⎦
where Δ0 is the maximum temperature change. At low tem-
peratures in vacuum, Δ0 depends only upon the thermal
conduction along the TA heater, Δ0=2I20R/(πκS/L), and τ
is the characteristic heat dissipation time constant, determined
exclusively by the thermal diffusivity, α, along the length of
the nanotubes and nanotube bundles, namely, τα =L2/π2α.
However, for temperatures above 400 K and the sample
length above 2 mm, the contribution of blackbody radiation
should also be considered.
The temperature, normalized to Δ0, alternates between 0
and 2 as ωτ→0, i.e., when the heat accumulation during a
temperature oscillation period is negligible. However, this
Nanotechnology 29 (2018) 325704
normalized temperature modulation approaches zero along
the line of the mean temperature accumulation, Δ0, when
ωτ?1. Figure A1 schematically illustrates the dependence
of the amplitude of temperature fluctuations on the double-
frequency of the electrical current. Note that the thermal
camera measures Δ0, which is the mean value of the modu-
lated and accumulated temperature.
ORCID iDs
Ali E Aliev https://orcid.org/0000-0003-0034-9707
References
[1] Xiao L et al 2008 Flexible, stretchable, transparent carbon
nanotube thin film loudspeakers Nano Lett. 8 4539–45
[2] Aliev A E, Lima M D, Fang S and Baughman R H 2010
Underwater sound generation using carbon nanotube
projectors Nano Lett. 10 2374–80
[3] Vesterinen V, Niskanen A O, Hassel J and Helistö P 2010
Fundamental efficiency of nanothermo-phones: modeling
and experiments Nano Lett. 10 5020–4
[4] Xiao L, Liu P, Liu L, Li Q, Feng Z, Fan S and Jiang K 2011
High frequency response of carbon nanotube thin film
speaker in gases J. Appl. Phys. 110 084311–5
[5] Nasibulin A G et al 2011 Multifunctional freestanding single-
walled carbon nanotube films ACS Nano 5 3214–21
[6] Aliev A E, Gartstein Y N and Baughman R H 2013 Increasing
the efficiency of thermoacoustic carbon nanotube sound
projectors Nanotechnology 24 235501
[7] Barnard A R, Brungart T A, McDevitt T E, Aliev A E,
Jenkins D M, Kline B L and Baughman R H 2014
Advancements toward a high-power, carbon nanotube, thin-
film loudspeaker Noise Control Eng. J. 62 360–7
[8] Aliev A E, Mayo N K, Baughman R H, Avirovik D, Priya S,
Zarnetske M R and Blottman J B 2014 Thermal
management of thermoacoustic sound projectors using a
free-standing carbon nanotube aerogel sheet as heat source
Nanotechnology 25 405704
[9] Shinoda H, Nakajima T, Ueno K and Koshida N 1999
Thermally induced ultrasonic emission from porous silicon
Nature 400 853–5
[10] Tian H et al 2011 Graphene-on-paper sound source devices
ACS Nano 5 4878–85
[11] Suk J W, Kirk K, Hao Y, Hall N A and Ruoff R S 2012
Thermoacoustic sound generation from monolayer graphene
for transparent and flexible sound sources Adv. Mater. 24
6342–7
[12] Aliev A E, Mayo N K, Jung de Andrade M, Robles R O,
Fang S, Baughman R H, Zhang M, Chen Y, Lee J A and
Kim S J 2015 Alternative nanostructures for thermophones
ACS Nano 9 4743–56
[13] Aliev A E, Perananthan S and Ferraris J P 2016 Carbonized
electrospun nanofiber sheets for thermophones ACS Appl.
Mater. Interfaces 8 31192–201
[14] Niskanen A O, Hasse J, Tikander M, Maijala P, Grönberg L and
Helistö P 2009 Suspended metal wire array as a
thermoacoustic sound source Appl. Phys. Lett. 95 163102
17
A E Aliev et al
[15] Brown J J, Moore N C, Supekar O D, Gertsch J C and
Bright V M 2016 Ultrathin thermoacoustic nanobridge
loudspeakers from ALD on polyimide Nanotechnology 27
475504
[16] Bouman T M, Barnard A R and Asgarisabet M 2016
Experimental quantification of the true efficiency of carbon
nanotube thin-film thermophones J. Acoust. Soc. Am. 139
1353–63
[17] Aliev A E, Mayo N K, Baughman R H, Avirovik D, Priya S,
Zarnetske M R and Blottman J B 2014 Thermoacoustic
excitation of sonar projector plates by free-standing carbon
nanotube sheets J. Phys. D: Appl. Phys. 47 355302
[18] Tong L H, Lim C W and Li Y C 2013 Gas-filled encapsulated
thermal-acoustictransducer J. Vib. Acoust. 135 051033
[19] Arnold H D and Crandall I B 1917 The thermophone as a
precision source of sound Phys. Rev. 10 22–38
[20] Mayo N K 2015 Improving efficiency and performance of
carbon nanotube based thermophones PhD Thesis The
University of Texas at Dallas, Dallas, TX, USA
[21] Mehta P and Polkovnikov A 2013 Efficiency bounds for
nonequilibrium heat engines Ann. Phys. 332 110–26
[22] Zhang M, Fang S L, Zakhidov A A, Lee S B, Aliev A E,
Williams C D, Atkinson K R and Baughman R H 2005
Strong, transparent, multifunctional, carbon nanotube sheets
Science 19 1215–9
[23] The spinnable forest for MWNT-10 sheets was provided by
Lintec of America Inc., R&D division of Lintec Corp., Japan
[24] The spinnable forest for MWNT-50 sheets was provided by
JNC Corp., Japan
[25] Moisala A, Nasibulin A G, Brown D P, Jiang H,
Khriachtchev L and Kauppinen E I 2006 Single-walled
carbon nanotube synthesis using ferrocene and iron
pentacarbonyl in a laminar flow reactor Chem. Eng. Sci. 61
4393–402
[26] Kanninen P, Luong N D, Sinh L H, Anoshlin I V, Tsapenko A,
Seppala J, Nasibulin A G and Kallio T 2016 Transparent
and flexible high-performance supercapacitors based on
single-walled carbon nanotube films Nanotechnology 27
235403
[27] The sheets of SWNT-2 were provided by Skolkovo Institute of
Science and Technology (A G Nasibulin), Russia
[28] Atkins P W 1991 Physical Chemistry (Oxford: Oxford
Univesity Press)
[29] Dixon J C 2007 The Shock Absorber Handbook 2nd edn (New
York: Wiley)
[30] Landau L D and Lifshitz E M 1987 Fluid Mechanics 2nd edn
(Oxford: Elsevier)
[31] Rossing T D 2014 Handbook of Acoustics 2nd edn (Berlin:
Springer)
[32] Sherman C H and Butler J L 2007 Transducers and Arrays for
Underwater Sound (The Underwater Acoustics Series)
(Berlin: Springer)
[33] Rossing T D and Fletcher N H 2004 Principles of Vibration
and Sound 2nd edn (New York: Springer)
[34] Vaughan M T and Guggenheim S 1986 Elasticity of muscovite
and its relationship to crystal structure J. Geophys. Res. 91
4657–64
[35] Dzikowicz B R, Tressler J F and Baldwin J W 2017 Cylindrical
heat conduction and structural acoustic models for enclosed
fiber array thermophones J. Acoust. Soc. Am. 142 3187
[36] Lu L, Yi W and Zhang D L 2001 3ω method for specific heat
and thermal conductivity measurements Rev. Sci. Instrum.
72 2996–3003