United States
Environmental Protection
Agency
Healt Effects Research
Laboratory
Research Triangle Park NC 27711
Research and Development
EPA/600/S1-87/003 Aug. 1987
&EPA Project Summary
Specific Absorption Rate
Distributions in a Heterogeneous
Model of the Human Body at Radio
Frequencies
Stanislaw S. Stuchly
The electric field distribution or the
rate of energy absorption referred to as
the specific absorption rate (SAR) in a
biological body is a complex function of
several exposure parameters such as
frequency, intensity of the incident field,
polarization, source to object configura-
tion (near-field vs far-field) and the body
characteristics such as size, shape and
dielectric properties. An experimental
approach was employed to determine
SAR patterns in a full-scale heteroge-
neous model of man exposed to radio-
frequency fields at 160, 350 and 915
MHz in the far and near fields for two
polarizations. The model had an ana-
tomically correct shape and contained
a skull, spinal cord, rib cage and all
major bones (except those in the feet
and hands), brains, lungs and muscle
tissue. The square of the electric field
inside the model was measured by a
small diameter electric field probe. Data
acquisition, exposure conditions and
data processing were under computer
control. Special circuitry including an
optical link was used to interface the
electric field probe with the computer.
Extensive data were obtained, analyzed
and compared with the data for a
homogeneous model.
This Project Summary was developed
by EPA's Health Effect* Research Labo-
ratory, Research Triangle Park, NC, to
announce key finding* of the research
project that I* fully documented In a
separate report of the tame title (see
Project Report ordering Information at
back).
Introduction
The electric field distribution or the
specific absorption rate (SAR) distribution
in a biological body is a complex function
of several exposure parameters such as
frequency, intensity of the incident field,
polarization, source to object configura-
tion, presence of other objects nearby
(ground plane, reflectors), and the body
parameters such as size, shape and the
dielectric properties. To quantify the inter-
actions between RF fields and biological
systems both theoretical and experimental
dosimetry methods have been developed.
A few numerical techniques have been
used such as the method of moments,
finite-difference method, fast-Fourier
transform method and time-domain finite
difference method to determine SAR dis-
tributions in simplified models of the
human body. The main limitation of all
the methods is the computer memory
and time required, if a reasonable spatial
resolution and accuracy are to be
achieved.
Experimental dosimetry has been
developing in parallel and frequently in
conjunction with theoretical dosimetry
for the following specific applications:
verification of theoretical predictions,
determination of the total dose rate (total
SAR) in experimental animals, determina-
tion of the average SAR and its distribu-
tion in models more closely resembling
real biological bodies than those treatable
theoretically, determination of the SAR in
models and animals exposed in the near
field under the conditions for which a
theoretical analysis is prohibitively dif-
-------
ficult. Experimental dosimetry has been
performed on full size models and animals,
and on scaled down models (frequency
and dielectric properties).
A computer-controlled system recently
developed at the University of Ottawa
facilitates the effective use of implantable
electric field probes to obtain spatial dis-
tribution of the electric field and the SAR
in full-scale models of humans.
The overall performance of the system
was evaluated by measuring the SAR
distributions in homogeneous lossy
spheres and cylinders at various fre-
quencies of interest using three types of
the electric field probes and by comparing
the results with the theoretical values.
The system was used to determine
spatial distributions of the SAR in a full-
scale homogeneous model of man ex-
posed in the far-field at 160 MHz, 350
MHz and 915 MHz and in the near-field
of a resonant dipole, a resonant dipole
with reflector and a resonant slot.
The objective of the project was to
extend the previously developed experi-
mental dosimetry method to a more
realistic, heterogeneous model of man,
and to determine the spatial distribution
of the electric field and the SAR in such a
model at selected frequencies.
The objective of the project was met
through the following developments:
— development of tissue-equivalent
materials to simulate electrical pro-
perties of bones, lungs, and brain,
— development and manufacture of a
heterogeneous model of man,
— development of a method of probing
the electric field in the hetero-
geneous model (mechanical and
electromagnetic aspects),
— evaluation of electric field probe
responses in materials having in-
homogeneous dielectric properties
(the permittivity of bone is an order
of magnitude different from that of
muscle), in particular close to the
boundary between air and high
water content tissues,
— acquisition and analysis of the dis-
tributions of the SAR at selected
frequencies.
Procedure
The experimental system developed in
our laboratory consists of a scanning
system with an electric field probe, a
radiating antenna and a model of a human
body in which electric fields strengths
are measured. This part of the system is
placed in an anechoic chamber to avoid
reflections of the radiated RF energy.
The mechanical structure used for
supporting the positioning of the electric
field probe consists of three custom-made
independent guiding slides forming an
XYZ-coordinate system. All slides use a
lead-screw arrangement.
The probe can be placed at any location
within a regular solid of 1.9 x 0.5 x 0.45
m. The dimension of the rectangular
volume can be altered by repositioning
limit switches at the end of each slide.
The system can operate in a manual or
a computer-supervised mode. In the
manual mode the probe motion is con-
trolled by switches on a portable, hand-
operated unit in order to select probe
coordinates that are stored in a file in the
minicomputer memory, thus producing a
map of measurement points.
In the computer-supervised mode the
probe automatically scans through all
points of the previously selected map,
and the corresponding values of the
electric field strength for each position
are recorded. After completing the data
acquisition phase, the computer can dis-
play the results in graphical form on a
CRT screen or they can be printed in
tabular form on a line printer.
The software used in the data collection
and analysis consists of three programs
developed for this application and a soft-
ware package purchased for display of
contour and mesh views of the data. The
first program called MAP is used in
conjunction with the hand-held control
unit driving the scanning system to gen-
erate a map of coordinates. The second
program called SCAN uses this map to
automatically take electric field measure-
ments at all of the stored locations. For
each location the electric field strength is
measured five times and averaged. The
same program also controls a digital
attenuator that is used to adjust the
power to the antenna, in order to keep
the signal from the amplifier in a linear
range. Each x,y,z location and the associ-
ated electric field strength are stored in a
disk file. The third program called CHART
allows one to display or plot a graph of,
for instance, the electric field strength vs.
distance (x,y,z). Software developed by
Data Plotting Services called DPICT*
provides the capability of displaying the
electric field strength or SAR, in the form
of a contour diagram of equipotential
lines or as a 3- dimensional representation.
The electronic circuitry which interfaces
the probe output to the computer consists
Mention of trademarks or commercial products
does not constitute endorsement or recommenda-
tion for use.
of three FET input amplifiers, a summing
junction, an active filter, RMS to DQ
converter, a voltage to frequency con-
verter, a line driver, a LED and an optical-
fiber link to the computer. The optical
fiber link brings the signal from the
amplifier out of the anechoic chamber to
a circuit containing a fiber optic receiver
and a frequency to voltage converter.
This voltage, which represents the electric
field, is then sent to the POP 11/34
where an A/D converter inputs the signal
to the computer.
The overall system capabilities can be
summarized as follows:
(1) the SAR measurements repeat-
ability is better than ± 0.5 dB
(—10%) (tested on 15 runs under
various exposure conditions), and
(2) the SAR measurement uncertainty
is ± 1 dB; the main factor is the
uncertainty in the probe calibration.
The model of the human body was
exposed to RF radiation, from resonant
dipole or resonant dipole antennas with
reflectors, incident from the front for the
far- and the near-field exposures.
For exposures in the far-field, the
human model was placed at a distance
sufficiently large to ensure the far-fielc
conditions. A complete map of the powei
density at the plane of the model of mar
was obtained for a given placement o
the source without the model in place
The correction for the nonplanar wave
front (amplitude only) was incorporatec
into the computer program that normal
ized the measured SAR values to '
mW/cm2 of the incident power uniforrt
at the plane of the model of man.
A nonperturbing, implantable electrii
field probe consisting of a very shor
dipole and a miniature Schottky-barrie
diode connected by high-resistance lead:
with the external circuitry was designed
constructed and evaluated. The dipole
and the high-resistance leads wen
deposited on a dielectric substrate. Thre<
dipoles were arranged in a triangula
configuration to provide an isotropic direc
tional response. The external diameter c
the probe was 1 cm. The output from th
Schottky-barrier diode was connected t
the electronic circuitry by high-resistane
leads with a resistivity of 2 Mft/m, and
length of 0.3 m.
The probe was calibrated at thre
frequencies (160, 350 and 915 MHz
The sensitivity of the probes in tissu
material was determined by measurin
the output voltages of the electric fiel
probe placed at various locations in
sphere made of styrofoam and filled wit
a dielectric medium having the sarr
-------
dielectric properties as tissue. The output
signal was compared (using the least-
square fit method) with theoretical values
of the electric field in the same locations.
The sphere was irradiated by a plane
wave of a known power density.
One of the major research objectives
was to produce a realistic model of a man
which replicated the major anatomical
components of the human body. The
components chosen were muscle, brain,
lung and bone since they have dissimilar
dielectric properties. Muscle simulating
material was used to model all of the
other organs of the body since their
properties are similar to muscle (± 10%).
The shape of the body was formed by a
thin shell into which the components
were placed.
The shell was formed by wrapping an
Alderson Nuclear Medicine model of the
human body in a water-activated casting
tape from 3M company. This model was
anatomically correct and of the dimen-
sions of a male 1.62 m tall. After setting,
the cast was removed from the Alderson
Phantom and coated with fiberglass resin.
This produced a hard, watertight shell in
the shape of a man. The cavity for the
lungs was formed in the same way using
a plastic model of the lungs. The bones
were cast from a material simulating the
properties of live human bone using molds
from the bones of a skeleton. The interior
of the skull was used to form the brain
cavity. The lungs were filled with lung-
equivalent material, skull with brain-
equivalent material and the remaining
part of the shell with muscle-equivalent
materials.
The components of the muscle material
were water, hydroxyethylcellulose (HEC),
a compound to increase the viscosity,
sodium chloride (NaC1), to increase the
conductivity, sucrose to decrease the
dielectric constant and a bactericide to
prevent breakdown of the polymer by
bacterial agents.
The properties of brain tissue were
similar to those of skeletal muscle and
the same components, in different
proportions.
The basis of the lung simulation was
the same as the skeletal muscle but with
an addition of hollow silica microspheres.
The bone simulating material was made
from Devcon Two Ton epoxy to which had
been added a highly conductive potassium
chloride (KCI) solution. The concentration
of the salt solution (electrolyte) was
adjusted to vary the conductivity. The
electrolyte was incorporated into the
epoxy, thus forming ionic conductance
carriers in the bone equivalent material.
The dielectric properties of the tissue
simulating materials were measured
using an open-ended coaxial-line sensor
and a computer-controlled automatic
network analyzer. The uncertainty of the
measurements was determined to be less
than 3 percent for the dielectric constant
and 2 percent for the conductivity for the
muscle, brain and lung materials, and
less than 5 and 10 percent for the di-
electric constant and conductivity, respec-
tively, for bone simulating material.
The exterior body dimensions and the
dimensions and placement of the interior
structures were determined by the use of
two techniques: laser and CT scans.
An experimental system developed by
Hymarc Engineering (Ottawa, Ontario,
Canada) in conjunction with the National
Research Council of Canada was used to
acquire the exterior coordinates of the
model. The system accomplishes this task
by scanning the model surface with a
faster light. This data in the form of x, y, z
coordinates was recorded on a floppy
disk and then transferred to a VAX 750
computer system to produce a 3-D view
of the exterior of the model.
A CT-scanner was used to take scans
at 1 cm intervals of the head, neck and
upper body and at 10 cm intervals of the
legs. The X-ray films were transferred to
paper by making contact prints. These
pictures defined the location and dimen-
sions of the interior structure of the
model.
Results and Discussion
Far-Field Exposure
Thirty-eight locations within one half
of the model were selected as measure-
ment sites.
In the far-field and for the E-polarization,
high local SARs are produced in the neck
at all three frequencies measured (160,
350,195 MHz). The highest SAR value is
at 160 MHz, and a substantially lower
SAR value is measured at 915 MHz. The
SAR patterns observed in the mid-plane
of the model were found to be similar for
all frequencies. Much lower SARs were
produced at the lowest two test frequen-
cies, namely 160 and 350 MHz for the
H-polarization of the incident wave. This
is not surprising in view of the lower total
whole body SAR deposited for the
H-polarization as compared with the
E-polarization. At 915 MHz, for the
H-polarization only, somewhat lower
SARs in the neck were produced as com-
pared with the E-polarization. This again
could be explained by comparable whole-
body SARs at this frequency.
The SAR patterns in the center of the
head were measured for two polarizations
(E and H). At all frequencies different
SAR distributions are produced for the
two polarizations, with the most pro-
nounced difference at 160 MHz. Except
at 915 MHz, higher SARs occur for the
E-polarization than the H-polarization.
Because of the high values observed,
the SAR distribution in the neck area is of
particular interest. Maxima in the neck
center were found at two frequencies.
Much lower SARs were produced in the
H-polarization in the neck area.
For the E-polarization the maximum
SAR is consistently in the neck, with the
highest value at the lowest test frequency
of 160 MHz. This observation is consistent
with the earlier findings for a homo-
geneous model. Slight shifts in the actual
position of the maximum for the hetero-
geneous model as compared with the
homogeneous model are due to slightly
different shapes of the two models.
At all test frequencies the maximum
value of the SAR was greater for the
heterogeneous model than for the
homogeneous model. The difference was
relatively small at 160 MHz and 915
MHz, but at 350 MHz the difference was
a factor of approximately 10. Intuitively,
one may try to explain the increases in
the SAR by the smaller cross-sectional
area of the neck associated with the
heterogeneous model (the neck diameters
were 37 cm and 40 cm for the hetero-
geneous and homogeneous, respectively).
However, this is a greatly simplified ex-
planation because of the very complex
shape and electrical properties of the
models, which in turn make the pattern
of the induced electric fields complex.
A more plausible explanation of greater
local SARs in the heterogeneous model
is the body's smaller size. For the E-
polarization at frequencies above reson-
ance, the whole-body average SARs are
greater for an ectomorphic and short
man than for an average man.
For the H-polarization SAR curves along
the mid-plane lack any specific character
contrary to the E-polarization. There is a
significant increase in the SAR in the
head at 160 and 350 MHz for the hetero-
geneous model as compared with the
homogeneous one. The increase is par-
ticularly pronounced at 350 MHz.
In the head the SARs were consistently
larger for the heterogeneous model. As
the conductivity of the brain tissue is
0.62 S/m while the conductivity of the
average tissue used in the homogeneous
model was 0.93 S/m, that means that
the difference in the electric field was
-------
even greater. The shape of the SAR pat-
terns, however, was very similar. The
same trend was observed at 160 and 915
MHz.
Local SARs in the heterogeneous and
homogeneous models along the direction
perpendicular to the body mid-plane ano
located approximately in the eye socket
were also compared for exposures at 160
and 350 MHz in the E-polarization. For
the heterogeneous model a well defined
"valley" in the SAR patterns can be
observed in the vicinity of the nasal
bones.
Local SAR distributions in the neck
were compared at all test frequencies.
For the heterogeneous model the axis of
probing was shifted off the center by 1.2
cm in order to by-pass the vertebra. For
the homogeneous model the electric field
was measured in the center of the neck.
At 160 MHz and 350 MHz a well-defined
effect of the vertebra can be seen when
the shapes of the SAR patterns for the
heterogeneous and homogeneous models
are compared. At 915 MHz the effect is
practically non-existent and the SAR
decreases nearly exponentially with
distance for both models. Greater values
of the SAR in this case can be explained
on the basis of larger conductivity of the
muscle tissue in this region in the hetero-
geneous model as compared with con-
ductivity of the average tissue in the
homogeneous model.
The SAR patterns at three frequencies
along the axis passing through the lung
were also compared. It is interesting to
note that, except at 160 MHz, the SAR
values in the heterogeneous model are
higher than in the homogeneous model,
even though the conductivities are lower
by a factor of approximately two. At 350
and 915 MHz the decrease of SAR with
distance is nearly exponential but of dif-
ferent slopes reflecting the differences in
the tissue dielectric properties.
Near-Field Exposures
Near-field exposures were performed
using resonant dipoles positioned close
to the model (less than 0.1 of the wave-
length). The local values of the SAR in
the mid-plane for a vertical dipole orienta-
tion (E-polarization) and a horizontal dipole
orientation (H-polarization) were mea-
sured. The two orientations may be con-
sidered as the two extreme positions of
the antennas (i.e., parallel or perpen-
dicular to the length of the body). The
SAR along the center axis is consistently
greater for the antenna parallel to the
body at all test frequencies. The differ-
ences in the SAR values for both polariza-
tions are much less along the side of the
model and absolute SAR values decrease
with increasing frequency.
It was found that most of the energy
from resonant dipoles located close to
the body is deposited within about 20%
of the body volume in the vicinity of the
dipole. The SAR values for distances from
about 10 to 40 cm from the top of the
head are much greater than elsewhere
for antennas located at the shoulder level.
The maximum SAR is produced in the
neck, as with the far-field exposures. The
value of the maximum SAR decreases
with frequency from 1 W/kg at 160 MHz,
to 0.7 W/kg at 350 MHz and to 0.1 W/kg
at 915 MHz for 1 Watt input power to the
antenna.
The SAR distributions along the axis
perpendicular to the main body axis and
located in the chest very close to the
center of the antenna varies as follows.
Within the first few centimeters from the
chest surface upon which the wave is
incident (10-15 cm), the SAR decreases
nearly exponentially. The attenuation
coefficients obtained from fitting a
straight line (least-square fit) to the ex-
perimental data and the attenuation co-
efficients calculated for a plane-wave
incident upon a muscle equivalent semi-
infinite plane are in very close agreement.
The local values of the SAR along the
axis penetrating the lung were also mea-
sured for the two polarizations. As in the
chest, the SAR decreases nearly ex-
ponentially along the axis, however, the
attenuation coefficient is different due to
the different tissue properties and complex
path of wave propagation in the near-
field of the dipole. The differences in the
SAR patterns for the two polarizations
appear to decrease as the exposure fre-
quency increases.
The local SARs in the mid-plane were
compared for the heterogeneous and
homogeneous models at 350 MHz for the
E-polarization. There was a small shift in
the location of the SAR maximum and in
its magnitude. The shift in the location of
the maximum SAR was due to the dif-
ference in the model size (the homo-
geneous model was 175 cm tall). Higher
maximum SAR for the heterogeneous
model is likely partly due to a greater total
SAR resulting from the smaller size of
the model. However, the difference
appears to be too large to be fully
accounted for by the size difference.
Similar to the homogeneous model of
man, the highest SARs and local "hot
spots" were found in the neck of the
heterogeneous model.
There is some difference in the shape
of the SAR patterns between the hetero-
geneous and homogeneous models. This
difference likely reflects the presence of
the spinal cord. The slight shift upwards
of the SAR curve at 350 MHz for the
heterogeneous model with respect to the
homogeneous model reflects the higher
conductivity of the muscle tissue in the
neck of the heterogeneous model (a =
1.03 S/m versus the average tissue con-
ductivity a = 0.93 S/m).
The SAR distribution in the location of
the eye, at all three frequencies, for the
E-polarization shows an increase of the
SAR at locations close to the nasal bones.
Conclusions and
Recommendations
A computer-controlled scanning system
and miniature, nonperturbing electric field
probes were used to acquire maps of
spatial distribution of the SAR in a realistic
model of man. The model was anatomi-
cally correct and reflected electrically
heterogeneous composition of the human
body. The model was comprised of a
simulated skull, spinal cord, rib cage and
other major bones as well as brain, lungs
and muscle tissue. Electrical properties
of the simulated tissues were the same
as the electrical properties of cor-
responding live tissues.
SAR patterns were obtained at three
frequencies: 160, 350 and 195 MHz in
the far-field and the near-field of resonam
dipoles. Resonant dipoles are reasonable
models for simulating exposures to
hand-held transmitters, and at the same
time are amenable for theoretical
analyses.
For both the far-field and the near-
field, highly nonuniform distributions o1
the SAR were observed at all three
frequencies and for both polarization:
investigated (the E and H polarizations)
Considerably lower SARs were producec
for the H polarization than for the E
polarization at 160 and 350 MHz, while
comparable SARs resulted from expo
sures at 915 MHz for both polarizations.
For the far-field exposures, highes
local SARs were produced in the centei
of the neck at 160 and 350 MHz in the
E-polarization, while in all other exposurt
conditions the highest SARs appeared t(
be produced at the model surface.
The SAR distributions in the hetero
geneous model were compared with thos<
in the homogeneous model havinj
average tissue properties. In general, th<
heterogeneity of the electrical structun
was reflected in the SAR patterns at 16(
and 350 MHz, but to a lesser extent a
-------
915 MHz. No profound differences in
SAR patterns for the two models were
noted, although local values differed up
to an order of magnitude. In very many
locations the values of the SAR for the
heterogeneous model were greater than
for the homogeneous model. This can be
only partly explained by greater average
SARs due to a slightly smaller size of the
heterogeneous model (162 cm) as com-
pared to the homogeneous model (175
cm).
In the near-field for all test frequencies
and both polarizations most of the energy
was deposited in about 20% of the total
body volume closest to the antenna feed-
point. Higher SARs were produced in
most locations under all circumstances
tested for the dipole parallel to the long-
body axis. Also for the E-polarization high
SARs were consistently produced in the
neck, both on its surface (at all three test
frequencies) and close to the center
(except at 915 MHz).
When compared with the homogeneous
model, in the near-field as with the far-
field, the SAR patterns are quite similar,
however inhomogeneity of the model is
apparent when it appears close to the
path of the electric field probe. In many
cases higher SARs, usually two to five
times, were measured in the hetero-
geneous model as compared with the
homogeneous model. For many, but not
all cases, this can be attributed to the
higher conductivity of the tissue (muscle)
in the heterogeneous model as compared
with the conductivity of the tissue
(average) in the homogeneous model.
A comparison of the two models, both
in the far- and near-field, leads to a
conclusion that a realistic full-scale model
is suitable for gaining a general idea
about the maximal local and regional
SAR values. The locations of the maxima
are, however, different for the two models
and by inference different from the loca-
tions in a living human being since the
heterogeneous model only approximates
a human being. Nevertheless, the data
obtained are useful in assessment of
approximate locations of high SARs and
their magnitude. The size of the electric
field probe (10 mm dia.) and technical
limitations such as lack of possibility of
measurement of the SAR in bones do not
permit obtaining very fine grade patterns.
The results obtained, together with
other data previously published, should
prove of value in deriving safe exposure
limits, in particular for portable trans-
mitters.
Detailed data obtained on the SAR
distributions in a realistic heterogeneous
model of man at three different fre-
quencies and two polarizations should be
used in evaluating numerical methods
for calculations of SAR-distributions. This
applies to both the far-field and the near-
field. Since knowledge of the SAR dis-
tribution under various exposure condi-
tions is important in safety considera-
tions, it is essential that there is a reliable
and technically convenient method es-
tablished for gaining such information.
The experimental method developed here
is highly reliable and relatively accurate
(10-15%). However, in view of the tech-
nical effort and time required to obtain a
complete set of data for any exposure
situation, it appears that a numerical
method would be preferred, if it can
satisfy the requirements of reliability and
accuracy.
The results obtained should be used in
evaluation of ratios of the local peak to
the whole-body average in establishment
of safety standards.
Stanislaw S. Stuchly is with the University of Ottawa, Ottawa, Ontario KIN
6N5 Canada.
Ronald Spiegel is the EPA Project Officer (see below).
The complete report entitled "Specific Absorption Rate Distributions in a
Heterogeneous Model of the Human Body at Radiofrequencies," (Order No.
PB 87-201 3567AS; Cost: $18.95, subject to change) will be available only
from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Health Environmental Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
-------
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
rj • \
OCT.-3V
.i.r^.s^i. f*.
0.£ Z-j?
r
Official Business
Penalty for Private Use S300
EPA/600/S1 -87/003
0000329 PS
U S ENVXR PROTECTION ASENCY
REGION 5 LIBRARY
230 S DEARBORN STREET
CHICAGO It. 60604
------- |