See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/5506243

Characterization of high intensity focused ultrasound transducers using acoustic

streaming

Article  in  The Journal of the Acoustical Society of America · April 2008

DOI: 10.1121/1.2835662 · Source: PubMed

CITATIONS READS
23 267

6 authors, including:

Prasanna Hariharan Matthew Myers

U.S. Food and Drug Administration U.S. Food and Drug Administration
42 PUBLICATIONS   447 CITATIONS    94 PUBLICATIONS   716 CITATIONS   

SEE PROFILE SEE PROFILE

Ronald Robinson Subha Maruvada

U.S. Department of Health and Human Services U.S. Food and Drug Administration
7 PUBLICATIONS   65 CITATIONS    54 PUBLICATIONS   416 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Medical Countermeasures View project

B/A Measurements View project

All content following this page was uploaded by Prasanna Hariharan on 01 June 2014.

The user has requested enhancement of the downloaded file.

Characterization of high intensity focused ultrasound
transducers using acoustic streaming
Prasanna Hariharan
Division of Solid and Fluid Mechanics, Center for Devices and Radiological Health, U. S. Food
and Drug Administration, Silver Spring, Maryland 20993, USA and Mechanical, Industrial, and Nuclear
Engineering Department, University of Cincinnati, Cincinnati, Ohio 45221, USA

Matthew R. Myers,a兲 Ronald A. Robinson, and Subha H. Maruvada

Division of Solid and Fluid Mechanics, Center for Devices and Radiological Health, U. S. Food
and Drug Administration, Silver Spring, Maryland 20993, USA

Jack Sliwa
St. Jude Medical, AF Division, 240 Santa Ana Court, Sunnyvale, California 94085, USA

Rupak K. Banerjee
Mechanical, Industrial and Nuclear Engineering Department and Biomedical Engineering Department,
598 Rhodes Hall, P.O. Box 210072, University of Cincinnati, Cincinnati, Ohio 45221, USA

共Received 3 August 2007; revised 20 December 2007; accepted 26 December 2007兲

A new approach for characterizing high intensity focused ultrasound 共HIFU兲 transducers is
presented. The technique is based upon the acoustic streaming field generated by absorption of the
HIFU beam in a liquid medium. The streaming field is quantified using digital particle image
velocimetry, and a numerical algorithm is employed to compute the acoustic intensity field giving
rise to the observed streaming field. The method as presented here is applicable to moderate
intensity regimes, above the intensities which may be damaging to conventional hydrophones, but
below the levels where nonlinear propagation effects are appreciable. Intensity fields and acoustic
powers predicted using the streaming method were found to agree within 10% with measurements
obtained using hydrophones and radiation force balances. Besides acoustic intensity fields, the
streaming technique may be used to determine other important HIFU parameters, such as beam tilt
angle or absorption of the propagation medium. © 2008 Acoustical Society of America.
关DOI: 10.1121/1.2835662兴
PACS number共s兲: 43.80.Ev, 43.35.Yb, 43.25.Nm 关CCC兴 Pages: 1706–1719

I. INTRODUCTION inaccuracies due to strong focusing; and 共iii兲 inaccurate fre-

Most therapeutic ultrasound procedures, including tumor
ablation, hemostasis, and gene activation, rely on the ability
of high intensity focused ultrasound 共HIFU兲 to rapidly el-
evate tissue temperatures. In order to maximize the effective-
ness of the procedure, as well as to avoid collateral tissue
damage, it is desirable to predict the energy distribution of
the HIFU beam within the propagation medium. An impor-
tant first step in this process is the characterization of the
ultrasound beam in a liquid medium. In this characterization,
the acoustic intensity is determined throughout the spatial
volume of interest, for transducer power levels of practical
importance.
Currently, HIFU fields are often characterized using ra-
diation force balance and hydrophone techniques, which
measure the ultrasonic power and intensity distribution, re-
spectively 共Shaw and ter Haar, 2006; Harris 2005兲. Although
these two techniques are well established and widely used,
there are known limitations to both of these methods, includ-
ing: 共i兲 sensor damage due to heating and cavitation; 共ii兲
a兲
Author to whom correspondence should be addressed. Electronic mail:
matthew.myers@fda.hhs.gov
quency response due to generation of higher harmonics
共Shaw and ter Haar, 2006兲. Consequently, these techniques
can accurately characterize HIFU transducers only at low
power. For clinically relevant high powers, there are no al-
ternative measurement standards available to accurately
characterize medical ultrasound fields generated by HIFU
transducers 共Shaw and ter Haar, 2006; Harris 2005兲.
Several new methods for measuring HIFU fields are be-
ing researched, including development of robust sensors and
hydrophones 共Wang et al., 1999; Shaw, 2004; Schafer et al.,
2006; Zanelli and Howard, 2006; Shaw and ter Haar, 2006兲.
An alternative approach to overcome the sensor-induced in-
accuracies is to eliminate the use of sensors, and noninva-
sively measure the pressure field. One such commercially
available noninvasive method is the schlieren imaging tech-
nique 共Harland et al., 2002; Theobald et al., 2004兲, which
utilizes changes in the optical index of refraction to qualita-
tively define the ultrasound field. However, for quantitative
evaluation, the pressure field must be reconstructed tomo-
graphically. Other than schlieren imaging, there are no non-
intrusive techniques reported in recent literature capable of
measuring ultrasound field at high powers.

1706 J. Acoust. Soc. Am. 123 共3兲, March 2008 0001-4966/2008/123共3兲/1706/14/$23.00 © 2008 Acoustical Society of America
 This paper describes a noninvasive method that is ca-
pable of measuring the acoustic intensity in a free field. The
method incorporates acoustic streaming, the steady fluid
movement generated when propagating acoustic waves are
attenuated by viscosity of the fluid medium 共Nyborg, 1965兲.
Acoustic streaming arising from ultrasound absorption was
first discussed by Eckart 共1948兲, who derived an expression
for the streaming velocity by applying the method of succes-
sive approximations to the Navier-Stokes equations. Accord-
ing to Eckart’s theory, the acoustic streaming velocity is di-
rectly proportional to the square of acoustic pressure, and
inversely proportional to the shear viscosity. However, Eck-
art’s expression was derived ignoring the hydrodynamic non-
linearity term in the Navier-Stokes equation. Later, Lighthill
共1978a, b兲 established that this steady streaming motion is
due to the mean momentum flux 共Reynolds stress兲 created by
the viscous dissipation of acoustic energy in the fluid me-
dium. Subsequently, Starritt et al. 共1989, 1991兲, Tjotta and
Tjotta 共1993兲, and Kamakura et al. 共1995兲 investigated, both
experimentally and numerically, the effect of acoustic and
hydrodynamic nonlinearity on the streaming velocity for a
focused ultrasound source. Results obtained from their ex-
periments and computations suggest that both acoustic and
hydrodynamic nonlinearities can play a major role in genera-
tion of acoustic streaming.
Previous studies have exploited the relationship between
the acoustic streaming field and the ultrasound intensity to
various degrees. Nowicki et al. 共1998兲 used both particle
image velocimetry 共PIV兲 and Doppler ultrasound to measure
streaming motion generated by a weakly focused ultrasound
beam in a solution of water and cornstarch. They found that
similar streaming velocities were obtained for both the meth-
ods and the velocity magnitude was directly proportional to
the acoustic power emitted by the transducer. Hartley et al.
共1997兲 and Shi et al. 共2002兲used Doppler ultrasound to quan-
titatively measure acoustic streaming velocity in blood when
exposed to an ultrasound source. They used streaming as a
potential tool for improving hemorrhage diagnosis. Hartley
et al. 共1997兲 found that the streaming velocity increases with
acoustic power and reduces with increase in viscosity during
blood coagulation. Choi et al. 共2004兲 used PIV to character-
ize streaming motion induced by a lithotripter. They ob-
served that the streaming velocity correlated linearly with the
peak negative pressure of the acoustic field measured at the
focus. More recently, Madelin et al. 共2006兲 used MRI to
measure streaming velocity in a Glycerol water mixture.
From the streaming velocity they tried to estimate the fluid
properties such as attenuation and bulk viscosity. They used
the expression derived by Eckart 共1948兲 to calculate the
acoustic field and time-averaged acoustic power of a plane
ultrasound transducer from the streaming velocity data.
While all of these experimental studies have observed a cor-
relation between the acoustic intensity field 共and acoustic
power兲 and the acoustic streaming field, none of them at-
tempted to use this correlation to accurately predict the
acoustic intensity field.
Our characterization technique employs a predictor-
corrector type method that back calculates the total ultra-
sonic power and acoustic intensity field from the streaming
J. Acoust. Soc. Am., Vol. 123, No. 3, March 2008 Hariharan et al.: Transducer characterization using acoustic streaming
velocity field generated by the HIFU transducers. The acous-
tic streaming field set up by the HIFU beam with unknown
energy distribution is measured experimentally using digital
particle image velocimetry 共DPIV兲 共Prasad, 2000兲. Then, an
initial guess for the unknown acoustic intensity field is made.
Based upon the Reynolds stress derived from this intensity,
the streaming velocity is computed and compared with the
experimental values. Using the error between the computed
and observed velocity fields, an optimization routine is em-
ployed to refine the guess for the intensity field and the com-
putation procedure is repeated. When the difference between
the computed and measured velocity fields falls below a
threshold value, the final estimate for the ultrasound intensity
field is obtained. This inverse method can also be used to
estimate physical properties of the fluid medium such as ul-
trasound absorption coefficient, or properties of the beam
such as tilt angle.
The acoustic intensities of interest in this paper are those
that are low enough that nonlinear propagation may be
neglected, yet high enough to produce temperature increases
in tissue of a few degrees per second. We label this range
the “moderate” temperature regime 共Hariharan et al., 2007兲.
An order-of-magnitude estimate of this regime is
100– 1000 W / cm2, with precise values depending upon the
amount of focusing, nonlinear parameter of the tissue, etc.
Despite the assumption of linear propagation, intensities in
the moderate regime may be damaging to conventional hy-
drophones, and hence good candidates for measurement via
the streaming method. In terms of transducer powers, the
values considered were in the range 5 – 30 W.
The streaming technique is described in Sec. II. In Sec.
III, the intensity field and acoustic power calculated using
this approach are compared with measurements made using
standard measurement techniques such as hydrophone scans
and radiation force balance 共RFB兲. The applicability of the
technique is summarized in Sec. IV.
II. MATERIALS AND METHODS
The transducer-characterization method utilizes an opti-
mization algorithm, in which the difference between the ex-
perimental streaming velocity and the computed velocity is
minimized as a parameter 共or parameter set兲 characterizing
the acoustic field is varied. The parameter of interest is often
the total power, though quantities such as the transducer fo-
cal length, transducer effective diameter, or medium attenu-
ation may also be determined. A flow chart of the algorithm
is shown in Fig. 1. The following sections describe the steps
of the flow chart.
A. Experimental measurement of streaming velocity
Figure 2 shows the experimental apparatus. Our experi-
mental setup combines two different systems—共i兲 acoustic
streaming generation system 关Fig. 2共a兲兴 and 共ii兲 DPIV mea-
surement system 共Integrated Design Tools, Tallahassee, FL兲
关Fig. 2共b兲兴.
In the streaming generation system, the HIFU transducer
to be characterized is mounted vertically in a Plexiglas™
tank using a triaxis positioning system 关Fig. 2共a兲兴. Sound
1707

sources considered in this study are single element, spheri-
cally focused, piezoceramic transducers with diameter rang-
ing from 6.4 to 10.0 cm and focal length varying between
6.3 and 15.0 cm 共Table I兲. The central frequency of the trans-
ducers ranges from 1.1 to 1.5 MHz. The driving signal for
the transducer comes from a wave form generator 共Wavetek
81, Fluke Corp., Everett, WA兲 in conjunction with an rf am-
plifier 共Model 150A 100B, Amplifier Research, Souderton,
PA兲. The output signal from the amplifier was monitored
using an oscilloscope 共Model 54615B, Agilent, Santa Clara,
CA兲.
The measurement system employs a cubical Plexiglas
tank 共20⫻ 20⫻ 20 cm兲 filled with water or other appropriate
fluid medium. The fluid medium was seeded with 10 ␮m
hollow glass spheres 共Dantec Dynamics Inc., Ramsey, NJ兲
which act as tracer particles for capturing the flow. The spe-
cific gravity of the spheres was approximately 1.1. Radiation
force acting on the tracer particles was calculated using the
expressions derived by King 共1934兲 and Doinikov 共1994兲.
Using the radiation force data, the terminal velocity of hol-
low glass sphere particles was calculated to be negligible
with respect to the measured streaming velocity 共from DPIV
system兲. This suggests that the tracer particles used in this
study do not undergo motion relative to the streaming flow.
In the experiments, particles were added until 5 to 10 particle
pairs per interrogation area were obtained.
The DPIV system incorporates a dual pulsed
15 mJ/ pulse Q-switched multimode Nd:YAG laser 共␭
= 532 nm兲, with a beam diameter of 2.5 mm, repetition rate
of 15 Hz, and a measured pulse width of 10⫾ 1 ns, as the
TABLE I. Physical characteristics of HIFU transducers used in the experi-
ments.
Parameters Transducers
HIFU-1 HIFU-2 HIFU-3
Transducer radius 5 cm 3.8 3.2
Operating frequency 1.5 MHz 1.107 MHz 1.1 MHz
Focal distance 15 cm 11 cm 6.264 cm
1708 J. Acoust. Soc. Am., Vol. 123, No. 3, March 2008
FIG. 1. 共Color online兲 Flow chart
summarizing the inverse methodology
used to determine acoustic power and
intensity.
illumination source 共New Wave Solo 1, New Wave Corp.,
Freemont, CA兲. A 100 mm focusing lens was used as laser-
to-fiber coupler 关Fig. 2共b兲兴 for focusing the 2.5 mm laser on
to the fiber face. The focal spot diameter at the fiber face was
measured as 178 ␮m ⫾ 1 ␮m.
Instead of solid silica-core fibers, which are not suitable
for higher power delivery, this system incorporated a
700 ␮m core diameter hollow waveguide 共HW兲 delivery fi-
ber 共Robinson and Ilev, 2004兲. This fiber was coated with
cyclic olefin polymer to minimize the HW attenuation losses
at the 532 nm wavelength. The output of the fiber was then
directed through the microscope objective lens 共4 ⫻ 兲. The
laser beam was then passed through the final Powell lens
optics, which produced a thin laser sheet for illumination of
the flow model.
Flow visualization and recording were done using an
8 bit 1K⫻ 1K CCD camera 共double exposure Kodak ES
1.0兲, which has a field of view of 2 cm⫻ 2 cm. The CCD
camera was connected to a computer via frame grabber to
store and process the recorded images. The computer also
synchronizes the camera with the laser pulse source. The
time between consecutive image pairs, which is the inverse
of laser pulse repetition rate, was set at 1 / 15 s. The pulse
delay time or the time between the images was varied be-
tween 1000 and 10 000 ␮s depending on the streaming ve-
locity.
Measurement of the streaming field was initiated with
the positioning of the axis of the HIFU beam within the laser
sheet. To perform this alignment, a hydrophone was placed
in the laser sheet, which was parallel to the beam axis 关z axis
in Fig. 2共a兲兴. The transducer was then moved using a posi-
tioning system, so that the beam axis traversed a path per-
pendicular to the sheet 关x axis in Fig. 2共a兲兴. When the hydro-
phone registered a maximum value, it was assumed that the
beam axis resided in the plane of the laser sheet. As an al-
ternative to hydrophone measurements, the transducer could
be moved in the x direction until the maximum streaming
velocity was observed.
After the beam axis was aligned with the laser sheet, the
ultrasound was turned off and the fluid motion was allowed
Hariharan et al.: Transducer characterization using acoustic streaming

to dissipate. The ultrasound was turned on and the flow field
near the focus was allowed to reach steady state. Our nu-
merical calculations predicted that flow near the focus
reaches steady state in ⬃3 s, while outside the focus it takes
longer depending on the boundary conditions. Thus, after
6.6 s the CCD camera was triggered to capture images of the
flow field. Our PIV data also showed that velocity fields
captured after 5, 10, and 15 s were the same.
Though a single image pair can give the instantaneous
flow field, in order to avoid random errors, 50 image pairs
were captured in a 5 s duration. For the 50 image pairs, the
relative standard deviation in the velocity was less than 2%.
All the image pairs were processed using a standard cross-
correlation algorithm 共IDT Provision, Tallahassee, FL兲 to get
the streaming velocity field. Both the magnitude and direc-
tion of the velocity are provided by the algorithm. Addition-
ally, no out-of-sheet components need to be resolved, since
the flow is axisymmetric 共no flow through the laser sheet兲 for
axisymmetric transducers.
J. Acoust. Soc. Am., Vol. 123, No. 3, March 2008 Hariharan et al.: Transducer characterization using acoustic streaming
FIG. 2. 共a兲 Schematic of the acoustic
streaming generation system. 共b兲
Block diagram of the DPIV measure-
ment system.
B. Numerical computation of streaming velocity
The first step in calculating the streaming velocity was
to compute the momentum transferred to the test fluid by the
acoustic field. Predictions of the acoustic pressure were made
by solving the Khokhlov–Zabolotskaya–Kuznetsov 共KZK兲
nonlinear parabolic wave equation. The KZK equation for an
axisymmetric sound beam propagating in the z direction is
共Hamilton and Morfey, 1998兲
冋 册 冉 冊
⳵ ⳵ p D ⳵2 p ␤ ⳵ p2 c0 ⳵2 p 1 ⳵ p
+ 3 2+ = + . 共1兲
⳵ t⬘ ⳵ z 2c0 ⳵ t⬘ 2␳0c30 ⳵ t⬘ 2 ⳵ r2 r ⳵ r
Here p is the acoustic pressure, t⬘ = tc−z0 is the retarded time, t
is the time, c0 is the speed of sound in the medium, r
= 冑x2 + y 2 is the radial distance from the axis of the beam, ␳0
is the ambient density of medium, D is the sound diffusivity
of fluid medium, and ␤ is the coefficient of nonlinearity de-
fined by ␤ = 1 + B / 2A, where B / A is the nonlinearity param-
eter for the fluid medium. As indicated by the second tem-
1709
poral derivative in the time-domain representation, the model ⳵ uz 1 ⳵ 共rur兲
employs the classic thermoviscous model of absorption, pro- + = 0, 共6兲
⳵z r ⳵r
portional to the square of the frequency in a frequency-

冋 冉 冊册
domain representation 共Hamilton and Morfey, 1998兲. The
numerical solution is implemented using the time-domain ⳵ uz ⳵ uz Fz 1 ⳵ P̄ ⳵ 2u z 1 ⳵ ⳵ u z
ur + uz = − +␯ + r ,
code KZKTEXAS2 developed by Lee 共1993兲. In the execution ⳵r ⳵ z ␳0 ␳0 ⳵ z ⳵ z2 r ⳵ r ⳵ r
of the code the pressure was assumed to be constant across 共7兲
the transducer face, and the transducer characteristics were
taken from Table I.
A steady streaming motion is generated by the absorbed
acoustic energy through the spatial variation in the Reynolds
ur
⳵ ur
⳵r
+ uz
⳵ ur
⳵z
=−
1 ⳵ P̄
␳0 ⳵ r
+␯ 冋 + r 冉 冊册
⳵ 2u r 1 ⳵ ⳵ u r
⳵ z2 r ⳵ r ⳵ r
. 共8兲

stress associated with the acoustic field 共Lighthill, 1978a, b兲.
In terms of the acoustic particle velocity ui, the Reynolds
stress is given by ␳uiu j, where the overbar denotes a time
average. The force 共per unit volume兲 associated with the
Reynolds stress 共Lighthill, 1978a, b兲 is
⳵ 共 ␳ u iu j 兲
Fj = − . 共2兲
⳵ xi
Here repeated indices denote summation. The streaming mo-
tion satisfies the equation of motion expressed as 共Lighthill,
1978a, b兲,
Equations 共6兲–共8兲 were solved using the Galerkin finite ele-
ment method 共Fluent Inc., 2002兲, in a geometry that simu-
lated the tank of Fig. 2共a兲.
C. Iterative characterization
To illustrate the iterative approach, we take the unknown
parameter of interest to be the acoustic power. For the first
iteration, a guess is made for the power, enabling the KZK
equation to be solved for the first iterate of the acoustic field.
From the KZK equation, the axial component of the driving
force Fz is calculated from Eq. 共5兲. Equations 共6兲–共8兲 are

冉 冊
then solved to obtain the first iterate of the velocity field.
⳵ ui ⳵ p̄ Streaming velocity fields obtained from both experiment
␳0 ui = Fj − + ␮ ⵜ 2u j 共3兲
⳵ xi ⳵xj and computation are then input to a Nelder Mead multidi-
mension optimization algorithm 共Lagarias et al., 1995; Math-
along with the conservation of mass equation
works, 2002兲. The objective function in this algorithm is

␳
⳵ ui
⳵ xi
=0 共4兲
Errorrms = 冑兺 n

i=1
共ui,exp − ui,num兲2 , 共9兲

for an incompressible fluid.
Of primary interest in transducer characterization is the
acoustic intensity in the focal region. In the focal region, the
acoustic field may be modeled as a beam of locally planar
waves traveling in the z direction. For this quasiplanar field,
the force associated with the Reynolds stress 关Eq. 共2兲兴 is in
the axial direction and has the form 共Nyborg, 1965兲
2␣ 2 2␣
Fz = p = I, 共5a兲
共 ␳ 0c 0兲 2 ␳ 0c 0
where ␣ is the absorption coefficient of the medium, and
p2
I= 共5b兲
共 ␳ 0c 0兲
is the time averaged acoustic intensity which is calculated by
solving Eq. 共1兲. In principle, a more complex intensity rela-
tion than Eq. 共5b兲 that accounts for the motion of the me-
dium must be used 共Morse and Ingard, 1968兲 in intensity
calculations for a streaming medium. However, the error in-
curred in using Eq. 共5b兲 is on the order of the Mach number
v0 / c0, v0 being a measure of the streaming speed. The Mach
number for the streaming jets reported on in this paper is on
the order of 0.0001 共Figs. 5 and 8兲, and hence the use Eq.
共5b兲 for a stationary medium expression is justified.
For an axisymmetric 共e.g. spherically concave兲 trans-
ducer in an infinite medium, the governing equations for the
streaming field 关Eqs. 共3兲 and 共4兲兴 may be rewritten as
where n is the number of velocity nodes in the camera’s field
of view. Errorrms measures the deviation of numerical veloc-
ity profile from the experimental values. The optimization
routine then attempts to reduce the error by adjusting the
power estimate and recalculating the intensity field. The en-
tire procedure is repeated until the rms error is minimized.
共Further iterations produce no reduction in error.兲 Provided
the numerical velocity field approximates the experimental
one reasonably well, the acoustic power is taken to be the
last iterate of the power, and the intensity field given by the
most recent output of the KZK equation. The total number of
iterations required for convergence of the optimization algo-
rithm is typically 30–40, and the duration for the entire back-
calculation procedure is of the order of 3 h. Computations
were conducted on a 3 GHz processor with 2.0 Gbytes
RAM.
D. Validation experiments
The total acoustic power and intensity field estimated
from the inverse approach were compared with the measure-
ments from radiation force balance 共RFB兲 and hydrophone-
scanning techniques. The RFB used in this study consists of
a highly absorbing target suspended in a water bath 共Maru-
vada et al., 2007兲. The transducer to be calibrated was
mounted directly above the target. When the ultrasound is
turned on, the target experiences a net force arising due to
the momentum associated with the ultrasound wave. The tar-
get was connected to a sensitive balance which measures the

1710 J. Acoust. Soc. Am., Vol. 123, No. 3, March 2008 Hariharan et al.: Transducer characterization using acoustic streaming

FIG. 3. Unprocessed DPIV image and streaming velocity contour, for
HIFU-2 in water. Acoustic power: 3.6 W. Streaming velocity contour is
obtained from the average of 50 image pairs captured in a 5 s duration. The
standard deviation for 50 images at the point of maximum velocity is
⬃1.93%.
force acting on it. From the measured force, acoustic power
was obtained by multiplying it with the sound speed, and the
factor 1 + 共a / 2d兲2 共a⫽transducer radius, d⫽focal length兲 to
account for focusing effects 共IEC 2005兲. Powers determined
from the RFB were estimated to have an uncertainty of less
than 10%.
Pressure measurements were made using 0.6 mm piezo-
electric ceramic hydrophones 共Dapco Industries, Ridgefield,
CT兲. The hydrophones were calibrated using a planar scan-
ning technique 共Herman and Harris, 1982兲. The uncertainty
in the pressure measurements acquired using these hydro-
phones was approximately 20% 共Herman and Harris, 1982兲.
The transducer to be characterized was mounted horizontally
in a Plexiglas tank surrounded with sound absorbing materi-
als, and the hydrophone was moved within the HIFU beam.
The movement of the hydrophone in all three coordinates
was monitored and controlled using computer controlled
stepper motors. A scanning step size of 0.4 and 1 mm was
used along radial and axial directions, respectively. At each
step, measurements of the peak positive pressure were made.
J. Acoust. Soc. Am., Vol. 123, No. 3, March 2008
Since the hydrophone scanning method is accurate only at
low power levels, this validation was done at acoustic pow-
ers less than 5 W.
III. RESULTS
In Sec. III A, computed streaming velocities are com-
pared with previously published results. Computed and mea-
sured streaming fields are also compared using a known
sound source. Additionally, the stability of the streaming jet
is analyzed. In Sec. III B our inverse algorithm is used to
determine acoustic power and intensity in water, at low
power. The inverse method is then used to determine the
absorption of a more viscous medium, at moderate power.
Characterization of a transducer of unknown power is then
performed in the higher-viscosity medium.
A. Forward problem: Experimental and numerical
streaming velocity fields
Figure 3 shows a single raw photographic image ob-
tained from DPIV measurements and the corresponding
streaming velocity contour obtained after postprocessing us-
ing a standard cross-correlation algorithm 共IDT Provision,
Tallahassee, FL兲. This velocity contour is the average of 50
image pairs captured in a 5 s duration. The standard devia-
tion in peak velocity magnitude for 50 images, divided by
the mean velocity of the 50 images, was 1.93%. This flow
field was obtained in degassed water using HIFU-2 as the
sound source for an input acoustic power of 3.6 W, as mea-
sured using a radiation force balance 共Maruvada et al.,
2007兲. The contour origin, which is the center of field of
view of the camera, does not coincide with ultrasound focus.
The velocity field at focus has the cigar shape characteristic
of focused ultrasound beams.
1. Validation of numerical model
Our computational model for calculating streaming ve-
locity was validated with the data of Kamakura et al. 共1995兲.
Figure 4 shows the transient streaming velocity generated by
a Gaussian shaded transducer with diameter 1 cm, focal
FIG. 4. 共Color online兲 Validation of
computational model with streaming-
velocity data from Fig. 2 of Kamakura
et al. 共1995兲. Gaussian shaded trans-
ducer with diameter 1 cm, focal length
5 cm, and central frequency 5 MHz,
driven at a maximum source pressure
of 30 kPa.
Hariharan et al.: Transducer characterization using acoustic streaming 1711

FIG. 5. 共a兲 Radial plot of streaming velocity in water. Transducer is HIFU-2;
power⫽5.4 W. 共b兲 Photograph of the experimental apparatus showing the
thickness of the laser sheet.
length 5 cm, and central frequency 5 MHz, driven at a maxi-
mum source pressure of 30 kPa. The transient speed calcu-
lated using the present model matches with the result pub-
lished by Kamakura et al. 共Fig. 2 of Kamakura et al., 1995兲
to within about 1%.
2. Comparison of experimental and numerical
streaming fields
Figure 5共a兲 compares the radial speed profiles obtained
experimentally using DPIV, and computationally by solving
Eqs. 共6兲–共8兲 关highest and lowest curves in Fig. 5共a兲, middle
curve containing “averaged data” will be discussed subse-
quently兴. The sound source used for this validation experi-
ment is the HIFU-2 transducer 共Table I兲, driven at a known
acoustic power of 5.4 W, as measured using the RFB. The
peak velocity profile calculated from the computational
model is ⬃15% higher than the peak velocity measured us-
ing the DPIV system. A similar trend is observed for other
transducers and other power levels.
The reason for this offset involves the finite thickness of
the laser sheet. As mentioned earlier, the thickness of the
laser sheet 关Fig. 2共a兲兴 when it exits the Powell lens is around
0.5– 1 mm. However, as the laser passes through the flow
medium, the sheet diverges gradually and becomes ⬃2 mm
thick when it reaches the ultrasound beam axis 关Fig. 5共b兲兴.
Therefore, instead of visualizing a single plane 共2 ⫻ 2 cm兲,
the CCD camera captures particle motion in a three-
1712 J. Acoust. Soc. Am., Vol. 123, No. 3, March 2008
dimensional space of size 2 ⫻ 2 ⫻ 0.2 cm. As a result, the
streaming velocity measured at a particular point in the YZ
plane 关Fig. 2共a兲兴 is essentially the spatial average of the ve-
locity occurring along the thickness of 0.2 cm of the sheet in
the x direction.
In order to account for this measurement error in our
numerical algorithm, the velocity field obtained numerically
was averaged over the 0.2 cm thickness of the laser sheet in
the x direction 关Fig. 5共b兲兴:
1 L
ūaveraged =
L 0
冕ū共r,z兲dx, r = 冑x2 + y 2 , 共10兲
where ū is the velocity estimated from the computations and
L is the half thickness of the laser sheet 共⬃1 mm兲. This
averaged velocity, ūaveraged, was then used in the optimization
routine to compute Errorrms in Eq. 共9兲. Figure 5 共middle
curve兲 shows the velocity profile calculated from our com-
putations after doing the spatial averaging. The averaged ve-
locity profile more closely matches the experimental data;
the difference is approximately 6%.
3. Higher-viscosity streaming medium
DPIV experiments performed in water at higher acoustic
powers 共5 – 30 W兲 revealed the onset of instability in the
streaming jet. This instability was manifested in the form of
small puffs occurring at irregular times, just below the focus.
This behavior was observed for all three HIFU transducers.
As a result of the behavior, unrepeatable velocity fields were
obtained.
To overcome the stability problem, the viscosity of the
medium was increased 共Reynolds number decreased兲. An al-
ternative fluid medium composed of degassed water mixed
with Natrosol-L 共Hydroxy ethyl cellulose, Hercules, Wilm-
ington, DE兲 in various concentrations was developed, and
used in experiments where the acoustic power exceeded
5 W. The mixture was degassed using a vacuum pump prior
to sonication. The viscosities of 1.4% 共1.4 g / 100 ml water兲,
2.4% and 3.4% Natrosol-L solutions were measured using a
Cannon type-E viscometer to be ⬃5, ⬃12, and ⬃25 cp, re-
spectively. Speed of sound for the 2.4% Natrosol solution
was measured using an ultrasonic time delayed spectrometry
system 共Harris et al., 2004兲 to be 1480 m / s 共same as for
degassed water兲. Finally, as discussed in Sec. III B, the at-
tenuation coefficient of the 2.4% Natrosol solution was
found using our streaming method to be roughly twice that
of water.
Figure 6 shows radial speed profiles obtained using wa-
ter as well as 1.4%, 2.4%, and 3.4% Natrosol solutions as the
flow medium. The transducer used in this comparison study
is HIFU-1 excited at an output power of 30 W. Broader,
smoother, and more axisymmetric profiles can be seen with
increasing Natrosol content. As the fluid viscosity increases
from 1 共water兲 to 25 cp, the peak velocity reduces from 8.2
to 5.8 cm/ s.
Figure 7 shows the fluid speed obtained in a 2.4% Na-
trosol solution for three different transducers driven at same
voltage 共42 Vrms兲. The plot shows that the velocity distribu-
tion at the focal region depends upon the focusing character-
istics of the transducer. Focusing gain, 3, and 6 dB beam
Hariharan et al.: Transducer characterization using acoustic streaming

dimensions 共measured from hydrophone scans兲 of the trans-
ducers are listed in Table II. The focusing gain 共Lee, 1993兲,
␻a2
given by the expression 2c 共␻ being the central frequency
0d
of the transducer, a the transducer radius, c0 the speed of
sound in the medium, and d the focal length兲, is the highest
for HIFU-1. The corresponding peak streaming velocity is
maximum for this transducer. The velocity contours for
HIFU-1 are relatively narrow, in accordance with the small
6 dB beam width 共Table II兲. In contrast, the velocity con-
tours for HIFU-3 are short in axial direction, consistent with
the smaller 6 dB width in the axial direction 共Table II兲. From
the 6 dB dimensions shown in Table II, it can be seen that
focal region of HIFU-2 is larger than HIFU-1 and HIFU-3.
Consequently, velocity contours obtained using this trans-
ducer are elongated.
Figure 8 compares the streaming velocity profile ob-
tained numerically and experimentally while driving HIFU-3
at an input voltage of 35 Vrms, in a 12 cp medium. The cor-
responding acoustic power measured using the RFB is 19 W.
It can be seen that experimental and numerical velocity pro-
files match closely 共⬃1 % 兲 in both the axial and radial direc-
tions.
FIG. 7. Speed contours obtained in 2.4% Natrosol solution for 共i兲 HIFU-1, 共ii兲 HIFU-2, and 共iii兲 HIFU-3 transducers. Transducer input voltage⫽42 Vrms.
J. Acoust. Soc. Am., Vol. 123, No. 3, March 2008
FIG. 6. 共Color online兲 Velocity magni-
tude vs radial distance for 共i兲 water,
共ii兲 1.4% Natrosol solution, 共iii兲 2.4%
Natrosol solution, and 共iv兲 3.4% Na-
trosol solution.
The width of the velocity distribution in Fig. 8共b兲, based
upon the distances at which the speed has dropped to half its
maximum value 共kinetic energy reduced to 41 the maximum兲,
is around 0.5 cm. By contrast, the intensity width for the
same transducer 共HIFU-3兲 is about 0.2 cm 共Table II兲. In gen-
eral, the width of the velocity distribution is larger than that
of the intensity distribution, and the effect becomes more
pronounced as the viscosity of the medium increases. In Fig.
6, for example, the width of the velocity profile for HIFU-1
in the 2.4% Natrosol solution 共viscosity 12 cp兲 is about
5.4 mm. The width of the velocity profile in water 共viscosity
1 cp兲 is between 2.4 and 3.3 mm, depending upon how the
asymmetry is treated. Within the limits of linear acoustics,
the width of the intensity distribution is independent of vis-
cosity, with the value of about 1.7 mm for HIFU-1.
B. Inverse problem: Total acoustic power and HIFU
beam characteristics
In the following, the optimization algorithm is used to
backcalculate various parameters characterizing the trans-
ducer and the propagation medium.
Hariharan et al.: Transducer characterization using acoustic streaming 1713
TABLE II. Focusing characteristics of HIFU transducers used in the experi-
ments.
Transducer 6 dB dimension 共axial⫻ radial兲 Focusing gain
HIFU-1 2.4 cm⫻ 1.7 mm 53
HIFU-2 2.95 cm⫻ 3 mm 31
HIFU-3 1.35 cm⫻ 2 mm 38
1. Characterizing acoustic field at low power levels
„<5 W…
Since current field hydrophone mapping techniques are
practical only at low power levels, characterization was first
performed at low input voltages with water as the fluid me-
dium. Figure 9 compares the source pressure 共p0
= 冑W ⫻ 2␳0c0 / ␲a2, p0 being the peak source pressure in
N / m2, W being the acoustic power in watts, and a the trans-
ducer radius in m兲 obtained using RFB and streaming back-
calculation methods. Using the procedure outlined in Fig. 1
FIG. 8. 共Color online兲 Velocity profiles obtained from computations and
experiments in 2.4% Natrosol solution; 共a兲 velocity profile along axial di-
rection and 共b兲 velocity profile along radial direction. Transducer is HIFU-3;
input voltage⫽35 Vrms 共19 W兲. For the experimental data, standard devia-
tion error bars for five trials are also shown. The maximum standard devia-
tion for five trials is 0.02 cm/ s.
1714 J. Acoust. Soc. Am., Vol. 123, No. 3, March 2008
and discussed in Sec. II, acoustic power was calculated from
the streaming velocity for driving voltages of 15 and 18 Vrms.
The source pressure of HIFU-2 was 0.048 and 0.058 MPa
for driving voltages of 15 and 18 Vrms, respectively. The
inverse method predicted the source pressure to be 0.046 and
0.056 MPa, which is within 5% of the values obtained from
the RFB measurements. When the averaging prescribed by
Eq. 共10兲 is not done, this deviation was ⬃10%.
Figure 10 compares the axial and radial intensity fields
of HIFU-2, obtained via hydrophone scanning and the in-
verse method for a transducer voltage of 10 Vrms. Peak inten-
sity at the focus measured by the inverse method is
65 W / cm2, approximately 4% below the peak intensity mea-
sured by hydrophone 共67.8 W / cm2兲. The 6 dB length and
width, backcalculated from the streaming velocities, are
2.7 cm⫻ 2.7 mm; the corresponding values obtained from
hydrophone scanning are 2.95 cm⫻ 3 mm.
2. Characterizing acoustic field at moderate power
levels „<30 W…
As noted earlier, a Natrosol-based solution is the pre-
ferred medium for characterization when the acoustic power
exceeds about 5 W. However, unlike water, little is known
regarding the absorption properties of Natrosol solutions.
Equation 共5兲 shows that the Reynolds stress, which accounts
for momentum transfer from the sound field to fluid, is pro-
portional to the absorption coefficient of the medium. Using
a transducer of known intensity, this parameter can be treated
as the unknown in the optimization algorithm, in the follow-
ing manner.
Figure 11共a兲 summarizes the procedure for calculating
the acoustic absorption coefficient of the medium. Using
DPIV, the streaming velocity generated by the HIFU-2 trans-
ducer in the 2.4% Natrosol-L solution was measured. First,
an initial guess for the unknown ␣ was made. In these deter-
minations of absorption coefficient using the inverse algo-
rithm, the attenuation coefficient of water 共2.52
⫻ 10−4 Np cm−1 MHz−2兲 was taken as the first guess for ␣.
Then, the forcing term was calculated using Eq. 共5兲 and the
streaming velocity was calculated numerically by solving
Eqs. 共6兲–共8兲. The Errorrms value was then computed from the
experiment and numerical velocity profiles and input into the
optimization routine. The procedure was repeated until the
relative error was reduced to a value of 10−3.
The absorption coefficient of 2.4% Natrosol measured
using this procedure is shown in Fig. 11共b兲. The ␣ values
measured at the power levels 3.6, 5.8, and 11.8 W were
fairly constant, approximately 5.65⫻ 10−4 Np/ cm for the
1.1 MHz frequency considered. This value of ␣ is used in the
subsequent calculations to characterize transducer output at
moderate and high power/intensity levels.
Figure 12 compares the power measurements made us-
ing the RFB and streaming in the 2.4% Natrosol medium.
Transducers HIFU-1 and HIFU-3 were driven at voltages
ranging from 19 Vrms to 42 Vrms. The inverse method was
able to measure the acoustic power with a maximum error of
about 10%. Computed intensity profiles in the Natrosol me-
dium are shown in Fig. 13, for a power of 30 W. The 6 dB
length and width, backcalculated from the streaming veloci-
Hariharan et al.: Transducer characterization using acoustic streaming

ties, are 1.2 cm⫻ 1.85 mm. The values determined using hy-
drophone scanning at low power, both in water and Natrosol,
are 1.35 cm⫻ 2 mm 共Table II兲.
IV. DISCUSSION
The streaming technique as presented in this paper is
useful for characterizing HIFU transducers in the “moderate”
intensity regime, roughly 100 to 1000 W / cm2. The intensi-
ties are low enough that linear acoustic propagation is a rea-
sonable assumption, yet high enough to possibly damage
conventional hydrophones.
At low powers—around 3 W or 70 W / cm2 focal inten-
sity, close agreement with piezoceramic hydrophones 共Fig.
10兲 was obtained. At focal intensities above about
1000 W / cm2 共transducer powers around 20 W兲, divergence
between the measured and predicted power levels can be
seen 共Fig. 12兲, probably due to nonlinear propagation effects.
While the KZK equation can predict the generation of addi-
tional frequency components by nonlinear mechanisms, the
absorption model used in our iterative method does not ac-
count for the frequency dependence of the absorption. A
more comprehensive model of the absorbed acoustic energy
would involve a sum of the form ⌺␣nIn, where In is the
intensity of the nth acoustic harmonic and ␣n the correspond-
ing absorption. The iterative scheme would then involve op-
timization over an n-dimensional space containing the inten-
sity modes. These nonlinear-propagation effects will be
treated in a future generation of the model.
The inverse method was first used to determine un-
known transducer power. Because we employed the linear-
ized version of the KZK equation, the intensity field is lin-
early proportional to the transducer power, and the KZK
equation actually needed to be solved only once. Subsequent
intensity fields required in the iteration process 共iteration is
still required due to the more complex dependence of veloc-
ity upon intensity in the Navier-Stokes equations兲 could be
obtained by simply multiplying the first intensity field by a
scaling factor dependent upon the current value of the acous-
tic power. This would not be true for other transducer-
J. Acoust. Soc. Am., Vol. 123, No. 3, March 2008 Hariharan et al.: Transducer characterization using acoustic streaming
FIG. 9. Comparison of source pres-
sure for HIFU-2 in water, obtained via
the streaming method and by radiation
force balance. Results based upon the
streaming technique are shown with
and without averaging over the 2 mm
laser sheet thickness. Standard devia-
tion error bars for three trials are also
shown. For RFB measurements, the
maximum standard deviation for three
trials is less than 1% of the mean
value. For the streaming measure-
ments, the maximum standard devia-
tion for three trials is 3% of the mean.
characterization parameters, such as focal length, beam tilt
angle relative to the desired axis, or the absorption of the
medium.
Determining an unknown absorption 共Fig. 11兲 is a char-
acterization problem for which the streaming technique is
well suited. Conventional absorption-measurement tech-
niques such as time delay spectroscopy 共Peters and Petit,
2003兲 measure attenuation over a prescribed propagation dis-
tance. When the attenuation is small over the given distance,
these techniques are susceptible to large error. With the
streaming-based technique, however, even slight attenuations
such as that of water can yield readily measurable fluid ve-
locities. Mathematically, the forcing term in the momentum
equation 关Eq. 共5兲兴 contains a factor of ␣, producing relative
changes in the fluid velocity field that are commensurate
with relative changes in the absorption coefficient. This com-
mensurate change in velocity with variation in ␣ also allows
for stable operation of the optimization algorithm; nonu-
niqueness issues arise when the streaming field is insensitive
to changes in the parameters being determined.
The optimization technique may be used to estimate
multiple parameters simultaneously, though the computation
time increases significantly, and convergence can be more
difficult to achieve. For example, determining the transducer
intensity simultaneously with the absorption coefficient
would be difficult, since the two quantities ␣ and I0 appear
primarily as a product 关Eq. 共5兲兴. The absorption coefficient
also appears in the exponential attenuating factor of the in-
tensity but, as is the case with conventional absorption mea-
surement techniques, this exponential factor is relatively in-
sensitive to changes in ␣.
The streaming algorithm in this paper assumes an axi-
symmetric ultrasound beam. The technique can be extended
to three-dimensional 共x , y , z兲, nonsymmetric beams with no
change in the procedures performed 共Fig. 1兲 in the algorithm.
A three-dimensional propagation code and fluid-flow simula-
tor would be necessary, and an increase in computation time
would result. The three-dimensional version of the technique
would be useful in diagnosing imperfections or problems in
transducer fabrication, such as a misfiring array element or a
propagation axis that was skewed relative to the axis of sym-
1715

metry of the transducer face. The problem would first be
manifested in an asymmetry of the experimental streaming
velocity field. The three-dimensional inverse algorithm could
be used to help predict the transducer characteristics giving
rise to the asymmetric velocity field that was observed. For
example, the location of a nonfunctioning array element
could be left as a free parameter in the optimization proce-
dure, to be determined as part of the iterative solution.
Figure 6 illustrates the complex dependence of velocity
upon fluid viscosity in a streaming medium. An increase in
viscosity increases acoustic absorption, i.e., produces a larger
transfer of momentum from the acoustic field to the hydro-
dynamic field, resulting in a higher peak 共on-axis兲 velocity.
However, the higher viscosity also serves to radially diffuse
momentum, resulting in a lower peak velocity. Thus, the
peak velocity in Fig. 6 is essentially unchanged as the vis-
cosity is increased by a factor of 5 from 1 to 5 cp, but drops
about 25% when the viscosity is doubled from 12 to 25 cp.
1716 J. Acoust. Soc. Am., Vol. 123, No. 3, March 2008
FIG. 10. 共Color online兲 Acoustic in-
tensity profiles as a function of 共a兲
axial and 共b兲 radial distances 共cm兲, ob-
tained from backcalculation and hy-
drophone scanning in water for
HIFU-2 transducer. Transducer input
voltage⫽10 Vrms 共3.2 W兲.
The higher viscosity test fluids are essential at higher
streaming velocities 共higher intensities兲 due to stability re-
quirements for the streaming jet. The higher viscosity test
fluids also yield more symmetric velocity profiles, as can be
seen by comparing the natrosol-based curves in Fig. 6 with
that for water. An upper limit on the viscosity of the test fluid
is imposed by the low resulting velocities resulting at high
viscosities. These velocities can become difficult to quantify
accurately using DPIV systems. For the intensity range con-
sidered in our studies, the agreement between experimental
and computational velocity profiles was best for the 12 cp
fluid 共Fig. 8兲.
The KZK equation was used in our simulations, how-
ever, any beam propagation code can be used in the iterative
algorithm outlined in Fig. 1. As noted by Tjotta and Tjotta
共1993兲, the accuracy of the KZK degrades when the beam
becomes highly focused 共large aperature transducers兲. Tjotta
and Tjotta 共1993兲 noted that the ratio a / d of transducer ra-
Hariharan et al.: Transducer characterization using acoustic streaming
 FIG. 11. 共Color online兲 共a兲 Flow chart
of the inverse problem to estimate
acoustic absorption coefficient, ␣ 共Np/
cm兲, of the 2.4% Natrosol fluid me-
dium. 共b兲 Absorption coefficient of
2.4% Natrosol fluid obtained using the
inverse method, for different powers
using HIFU-2 transducer. Standard de-
viation error bars for three trials are
also shown. The maximum standard
deviation for three trials is less than
1% of the mean.

dius to focal length should be small in order to ensure accu- 0.51, respectively. While the a / d value for HIFU-3 was rela-
racy of the KZK approach. For our transducers HIFU-1, tively large, the agreement with experimental measurements
HIFU-2, and HIFU-3, the a / d values were 0.33, 0.35, and of streaming velocity 共Fig. 8兲 and total power 共Fig. 12兲 indi-

FIG. 12. 共Color online兲 Comparison

of acoustic power obtained from back-
calculation and RFB in 2.4% Natrosol
medium, for transducers HIFU-1 and
HIFU-3.

J. Acoust. Soc. Am., Vol. 123, No. 3, March 2008 Hariharan et al.: Transducer characterization using acoustic streaming 1717
FIG. 13. 共Color online兲 Acoustic intensity as a function of 共a兲 axial and 共b兲 radial distances obtained using HIFU-3 transducer in 2.4% Natrosol solution.
Acoustic power⫽30 W.

cates that the criterion of Tjotta may be conservative. In any V. CONCLUSION

event, when high accuracy is required, a full-wave model
The streaming technique presented in this paper allows
such as the Helmholtz equation in the linear propagation
free-field characterization of HIFU transducers to be per-
regime or the Westervelt equation in the nonlinear regime
formed in an intensity range that may be hazardous to con-
should be employed in the iterative algorithm. The tradeoff is
ventional hydrophones. The method is noninvasive and
the increase in computation time associated with using a full-
agrees well with hydrophone and radiation-force balance
wave model compared to a parabolic equation.
techniques in common ranges of applicability. In addition to
A limitation of our inverse methodology is that the re-
predicting acoustic intensity fields, the approach may also be
petitive solution of the acoustic propagation equation and
used to determine other important HIFU parameters, such as
fluid-flow equations can involve significant computation
medium absorption, transducer focal length, or beam tilt
time 共⬃3 h兲. Our technique is also limited by the accuracy of
angle.
the sound propagation code. For example, the KZK code
currently employed by our model degrades in accuracy when
the transducer gain becomes large. The above-mentioned ACKNOWLEDGMENTS
drawbacks can be circumvented by developing an alternate
method which determines the acoustic field directly from the The mention of commercial products, their sources, or
experimental streaming velocity. In subsequent studies, an their use in connection with material reported herein is not to
attempt will be made to estimate the acoustic field directly be construed as either an actual or implied endorsement of
from the experimental streaming velocity data. such products by the U. S. Department of Health and Human

1718 J. Acoust. Soc. Am., Vol. 123, No. 3, March 2008 Hariharan et al.: Transducer characterization using acoustic streaming
Services. Financial support from the FDA Center for Device
and Radiological Health and the National Science Founda-
tion 共Grant No. 0552036兲 is gratefully acknowledged.
Choi, M. J., Doh, D. H., Kang, K. S., Paeng, D. G., Ko, N. H., Kim, K. S.,
Rim, G. H., and Coleman, A. J. 共2004兲. “Visualization of acoustic stream-
ing produced by lithotripsy field using a PIV method,” J. Phys.: Conf. Ser.
1, 217–223.
Doinikov, A. A. 共1994兲. “Acoustic radiation pressure on a rigid sphere in a
viscous fluid,” Proc. R. Soc. London, Ser. A 447, 447–466.
Eckart, C. 共1948兲. “Vortices and streams caused by sound waves,” Phys.
Rev. 73, 68–76.
Fluent Inc. 共2002兲. FIDAP User Manual ver. 8.6.2 共Fluent, Inc., Lebanon,
NH兲.
Hamilton, M. F., and Morfey, C. L. 共1998兲. “Model equations,” in Nonlinear
Acoustics, edited by M. F. Hamilton and D. T. Blackstock 共Academic,
New York兲, Chap. 3.
Hariharan, P., Banerjee, R. K., and Myers, M. R. 共2007兲. “HIFU procedures
at moderate intensities: Effect of large blood vessels,” Phys. Med. Biol.
52, 3493–3513.
Harland, A. R., Petzing, J. N., and Tyrer, J. R. 共2002兲. “Non-invasive mea-
surements of underwater pressure fields using laser Doppler velocimetry,”
J. Sound Vib. 252, 169–177.
Harris, G. R. 共2005兲. “Progress in medical ultrasound exposimetry,” IEEE
Trans. Ultrason. Ferroelectr. Freq. Control 52, 717–736.
Harris, G. R., Gammell, P. M., Lewin, P. A., and Radulescu, E. 共2004兲.
“Interlaboratory evaluation of hydrophone sensitivity calibration from
0.1 MHz to 2 MHz via time delay spectrometry,” Ultrasonics 42, 349–
353.
Hartley, C. J. 共1997兲. “Characteristics of acoustic streaming created and
measured by pulsed Doppler ultrasound,” IEEE Trans. Ultrason. Ferro-
electr. Freq. Control 44, 1278–1285.
Herman, B., and Harris, G. R. 共1982兲. “Calibration of miniature ultrasonic
transducers using a planar scanning technique,” J. Acoust. Soc. Am. 72,
1357–1362.
International Electrotechnical Commission 共IEC兲. 共2005兲. “Power
measurement—Radiation force balances and performance requirements up
to 1 W in the frequency range 0.5 MHz to 25 MHz and up to 20 W in the
frequency range 0.75 MHz and 5 MHz,” Ultrasonics 87, 1–44.
Kamakura, T., Kazuhisa, M., Kumamoto, Y., and Breazeale, M. A. 共1995兲.
“Acoustic streaming induced in focused Gaussian beams,” J. Acoust. Soc.
Am. 97, 2740–2746.
King, L. V. 共1934兲. “On the acoustic radiation pressure on spheres,” Proc. R.
Soc. London, Ser. A 147, 212–240.
Lagarias, J. C., Reeds, J. A., Wright, M. H., and Wright, P. E. 共1995兲.
“Convergence properties of the Nelder-Mead algorithm in low dimen-
sions,” AT&T Bell Laboratories, Technical Report, Murray Hill, NJ.
Lee, Y. S. 共1993兲. “Numerical solution of the KZK equation for pulsed finite
amplitude sound beams in thermoviscous fluids,” Ph.D. dissertation, Uni-
versity of Texas at Austin, Austin, TX.
Lighthill, J. 共1978a兲. “Acoustic streaming,” J. Sound Vib. 61, 391–418.
Lighthill, J. 共1978b兲. Waves in Fluids 共Cambridge University Press, Cam-
bridge, U.K兲.
J. Acoust. Soc. Am., Vol. 123, No. 3, March 2008 Hariharan et al.: Transducer characterization using acoustic streaming
View publication stats
Madelin, G., Grucker, D., Franconi, J. M., and Thiaudiere, E. 共2006兲. “Mag-
netic resonance imaging of acoustic streaming: Absorption coefficient and
acoustic field shape estimation,” Ultrasonics 44, 272–278.
Maruvada, S., Harris, G. R., Herman, B. A., and King, R. L. 共2007兲.
“Acoustic power calibration of high-intensity focused ultrasound transduc-
ers using a radiation force technique,” J. Acoust. Soc. Am. 121, 1434–
1439.
Mathworks Inc. 共2002兲. Matlab Manual 共Mathworks, Inc., Natick, MA兲, ver.
6.5.
Morse, P. M., and Ingard, K. U. 共1968兲. Theoretical Acoustics 共McGraw-
Hill, New York兲.
Nowicki, A., Kowalewski, T., Secomski, W., and Wojcik, J. 共1998兲. “Esti-
mation of acoustical streaming: Theoretical model, Doppler measurements
and optical visualization,” Eur. J. Ultrasound 7, 73–81.
Nyborg, W. L. 共1965兲. Physical Acoustics 共Academic, New York兲, Vol 2B.
Peters, F., and Petit, L. 共2003兲. “A broad band spectroscopy method for
ultrasound wave velocity and attenuation measurement in dispersive me-
dia,” Ultrasonics 41, 357–363.
Prasad, A. K. 共2000兲. “Particle image velocimetry,” Eur. Trans Telecommun.
Relat. Technol. 79, 51–60.
Robinson, R. A., and Ilev, I. K. 共2004兲. “Design and optimization of a
flexible high-peak-power laser-to-fiber coupled illumination system used
in digital particle image velocimetry,” Rev. Sci. Instrum. 75, 4856–4863.
Schafer, M. E., Gessert, J. E., and Moore, W. 共2006兲. “Development of a
high intensity focused ultrasound 共HIFU兲 hydrophone system,” Proc. Int.
Symp. Therapeutic Ultrasound 829, 609–613.
Shaw, A. 共2004兲. “Delivering the right dose, advanced metrology for ultra-
sound in medicine,” J. Phys.: Conf. Ser. 1, 174–179.
Shaw, A., and ter Haar, G. R. 共2006兲. “Requirements for measurement stan-
dards in high intensity focused 共HIFU兲 ultrasound fields,” National Physi-
cal Laboratory 共NPL兲 Report, ISSN 1744-0599.
Shi, X., Martin, R. W., Vaezy, S., and Crum, L. A. 共2002兲. “Quantitative
estimation of acoustic streaming in blood,” J. Acoust. Soc. Am. 111, 1110–
1121.
Starritt, H. C., Duck, F. A., and Humphrey, V. F. 共1989兲. “An experimental
investigation of streaming in pulsed diagnostic ultrasound beams,” Ultra-
sound Med. Biol. 15, 363–373.
Starritt, H. C., Duck, F. A., and Humphrey, V. F. 共1991兲. “Force acting in the
direction of propagation in pulsed ultrasound fields,” Phys. Med. Biol. 36,
1465–1474.
Theobald, P., Thompson, A., Robinson, S., Preston, R., Lepper, P., Swift, C.,
Yuebing, W., and Tyrer, J. 共2004兲. “Fundamental standards for acoustics
based on optical methods–Phase three report for sound in water,” Report
No. CAIR9 ISSN 1740-0953, National Physical Laboratory 共NPL兲, Ted-
dington, England.
Tjotta, S., and Tjotta, N. J. 共1993兲. “Acoustic streaming in ultrasound
beams,” in Advances in Nonlinear Acoustics, 13th International Sympo-
sium on Nonlinear Acoustics, edited by H. Hobaek 共World Scientific, Sin-
gapore兲, pp. 601–606.
Wang, Z. Q., Lauxmann, P., Wurster, C., Kohler, M., Gompf, B., and Eisen-
menger, W. 共1999兲. “Impulse response of a fiber optic hydrophone deter-
mined with shock waves in water,” J. Appl. Phys. 85, 2514–2516.
Zanelli, C. I., and Howard, S. M. 共2006兲. “A robust hydrophone for HIFU
Metrology,” Proc. Int. Symp. Therapeutic Ultrasound 829, 618–622.
1719