INSTITUTE OF PHYSICS PUBLISHING PHYSICS IN MEDICINE AND BIOLOGY

Phys. Med. Biol. 51 (2006) 1463–1477 doi:10.1088/0031-9155/51/6/007

The effect of increase in dielectric values on specific

absorption rate (SAR) in eye and head tissues
following 900, 1800 and 2450 MHz radio frequency
(RF) exposure
Jafar Keshvari1, Rahim Keshvari2 and Sakari Lang1
1 Technology Platforms, Nokia Corporation, PO Box 301, FIN-00045 Nokia Group,

Linnoitustie 6, 02600 Espoo, Finland

2 Department of Ophthalmology, Savonlinna Central Hospital, Savonlinna, Finland

E-mail: Jafar.keshvari@nokia.com and sakari.lang@nokia.com

Received 16 August 2005, in final form 16 January 2006

Published 1 March 2006
Online at stacks.iop.org/PMB/51/1463

Abstract
Numerous studies have attempted to address the question of the RF energy
absorption difference between children and adults using computational
methods. They have assumed the same dielectric parameters for child and adult
head models in SAR calculations. This has been criticized by many researchers
who have stated that child organs are not fully developed, their anatomy is
different and also their tissue composition is slightly different with higher
water content. Higher water content would affect dielectric values, which in
turn would have an effect on RF energy absorption. The objective of this study
was to investigate possible variation in specific absorption rate (SAR) in the
head region of children and adults by applying the finite-difference time-domain
(FDTD) method and using anatomically correct child and adult head models. In
the calculations, the conductivity and permittivity of all tissues were increased
from 5 to 20% but using otherwise the same exposure conditions. A half-wave
dipole antenna was used as an exposure source to minimize the uncertainties
of the positioning of a real mobile device and making the simulations
easily replicable. Common mobile telephony frequencies of 900, 1800 and
2450 MHz were used in this study. The exposures of ear and eye regions
were investigated. The SARs of models with increased dielectric values were
compared to the SARs of the models where dielectric values were unchanged.
The analyses suggest that increasing the value of dielectric parameters does
not necessarily mean that volume-averaged SAR would increase. Under many
exposure conditions, specifically at higher frequencies in eye exposure, volume-
averaged SAR decreases. An increase of up to 20% in dielectric conductivity
or both conductivity and permittivity always caused a SAR variation of less
than 20%, usually about 5%, when it was averaged over 1, 5 or 10 g of cubic
mass for all models. The thickness and composition of different tissue layers
0031-9155/06/061463+15$30.00 © 2006 IOP Publishing Ltd Printed in the UK 1463
1464 J Keshvari et al

in the exposed regions within the human head play a more significant role
in SAR variation compared to the variations (5–20%) of the tissue dielectric
parameters.

1. Introduction

The increasing use of wireless communication devices has led to concerns that exposure to
electromagnetic (EM) waves emitted by these devices may cause adverse health effects. There
have also been some discussions that children’s heads would absorb more RF energy under
the same exposure conditions compared to adults’ heads. However, this is not necessarily the
case and not supported by many studies (Kuster and Balzano 1992, Schoenborn et al 1998,
Keshvari and Lang 2005, Christ and Kuster 2005).
The basic safety limits for RF exposure are defined in terms of absorbed power per unit
mass, which is expressed by SAR in W kg–1. Depending on the exposure condition, SAR is
expressed either as a localized SAR value or averaged over the whole body. It describes the
amount of energy W absorbed in a dielectric material in time (dt) and mass unit (dm).


d dW
SAR = . (1)
dt dm
It can be related to the electric field or temperature rise (dT) at a point by
σeff |E|2rms |E|2 dT
SAR = = (σ + ω ε0 ε

) =c (2)
ρ ρ dt
where E is the root-mean-square (rms) electric field; σeff is the effective conductivity (S m–1);
ρ is the mass density (kg m–3); ω = 2πf , with f the frequency; ε0 is the permittivity of free
space; ε

is the loss factor and c is the specific heat capacity (J kg–1 ◦ C–1) of the material.
Relation (2) to temperature is limited to ‘ideal’ non-thermodynamic circumstances with
no heat loss by thermal diffusion, heat radiation or thermoregulation. Equation (2) shows that
the SAR varies directly with both σ and ε

, or in other words, σ and ε

indicate how much
energy will be absorbed by tissues for a given electric field E. Theoretically the greater the σ
or ε

, the greater the loss in material, and more power is absorbed for the same electric field
E. Generally, tissues with larger water content such as muscle absorb more energy than tissues
with lower water content, such as bone and fat. Both conductivity and permittivity of tissues
are frequency dependent.
There are a number of studies (Gandhi et al 1996, Gandhi and Gang 2002, Burkhardt
and Kuster 1999, Dimbylow 1993, 1994, Keshvari and Lang 2005, Martinez-Burdalo et al
2004, Schoenborn et al 1998, Martens et al 1995) which have investigated the SAR difference
between children and adults. Most of them have used computational techniques. In previous
numerical SAR studies, the researchers have assumed the same dielectric parameters for both
child and adult head models.
Peyman et al (2001) reported measured changes in the dielectric properties of various
tissues in the rat. This triggered the discussion about the validity of the studies, which have
used the same dielectric parameters both in child and adult head models. They reported that
conductivities of both cerebral tissue and the cranium decrease in rats from birth up to an age
of 70 days (figure 1(a)). This raised questions that how much would the SAR increase if larger
dielectric values were applied for child models.
Dielectric variation effect on SAR 1465

(a)

(b)

Figure 1. (a) Left: conductivity of rat tissues in the head at 900 MHz, adapted from Peyman et al
(2001). Right: the variation in total body water (TBW) content of rats with age expressed in ml of
water per kg of body weight, adapted from Altman (1974). (b) The variation in total body water
(TBW) content of humans with age expressed as ml of water per kg of body weight, adapted from
the published data compiled by Altman (1974). The horizontal error bars indicate the age range
for the data collected at each point and the vertical error bars indicate the range of TBW data for
each data point.

The main objective of this study was to respond to this question by investigating the effect
of increased dielectric values on mass-averaged SAR. The distributions in human ear and eye
regions were studied using anatomically correct MRI-based head models as well as a layered
flat phantom model. The ear and eye regions were exposed to a half-wave dipole antenna at
900, 1800 and 2450 MHz frequencies.

2. Dielectric characteristics of biological tissues and their age dependence

Information about the dielectric properties of biological tissues is essential to both RF and
thermal dosimetry. Parameters such as the conductivity, σ , and the dielectric permittivity,
1466 J Keshvari et al

ε, of human tissues are fundamental parameters in RF dosimetry because of their role in the
field distribution inside the body. In general, the parameters of interest are the real part of the
dielectric constant and the conductivity.
The dielectric properties of tissues above 100 MHz are determined by the intra-cellular
electrolytes, principally water (0.9% saline solution). These properties are consistent with
the suspensions of low conductivity and low permittivity particles such as cells in an aqueous
electrolyte. At the microwave frequencies, tissue properties can be attributed to their ‘free’
water content and the dispersion of normal bulk water.
Brain growth is the result of an increase in both the number of brain cells and their weight
during the first year of life (Fein 1978). After the first year, only the weight of the cells
increases. Myelination of the brain mainly occurs in the first 2 years of life (Holland et al
1986, van der Knaap and Valk 1990). Unlike intercellular fluid, myelin does not contain free
ions. This means that if the amount of myelin in the brain increases, it will result in a reduction
of the ion concentration and consequently to a reduction of the overall electrical conductivity
of brain tissue. This occurs during early development, but there is very little change thereafter.
It is known that total body water content (TBW) between different individuals may differ
and tissues with larger TBW are more conductive. Newborn infants have larger water content
than older infants and children. This is partially because the newborn has less fat and a
greater proportion of body mass composed of visceral organs. The extra cellular fluid (ECF)
compartment comprises over half of the total body water at birth and contains a greater relative
content of extra cellular sodium and chloride. Most of this neo-natal ‘excess’ ECF is lost
during the first 10 days of life through insensible perspiration, which can amount from 5% to
10% of the infant’s birth weight. Until about 2 years of age, the infant maintains a larger ECF
than the adult in terms of overall percentage of body water.
There is a very good correlation between the age changes in TBW and the conductivity
in physiological organs. Like tissue conductivity, the TBW is highest at birth and decreases
steadily to about the age of 20–25 days, and then flattens off (figure 1(a)). Thus, it appears
reasonable to make the general assumption that at the age at which the TBW of an animal
ceases to decline, tissue conductivity also becomes constant. However, this conclusion is based
on animal data. By using human TBW data, tissue conductivity changes can be calculated
for children as well. In figure 1(b), a compilation of published TBW data for children of
varying ages is shown (Altman 1974). The data show an initial decline after birth and then a
relatively constant TBW after the age of 1 year. Consequently, it may be concluded that tissue
conductivity in children also ceases to decline after infancy. Hence, using increased dielectric
values in the 3 year old and 7 year old child head models does not reflect the realistic exposure
conditions but they are rather used in this study to investigate SAR variations by changing the
dielectric parameters.
One would always anticipate higher RF energy absorption in the tissues with larger TBW
such as infants and very young children compared to adults; however, the results of this study
will show that it is not always the case.

3. Materials and methods

Material and methods used in this study are explained in detail in our previous paper (Keshvari
and Lang 2005). Only a short description of the head models, exposure conditions and
computational models is given here.
Finite-difference time domain was applied as a numerical method to carry out the SAR
calculations. The calculations were carried out using a commercially available EM solver,
SEMCAD 1.8.9b. The grading mesh technique was used to reduce the number of voxels
Dielectric variation effect on SAR 1467

Figure 2. High-resolution anatomical head models of the 3 year old (left) and 7 year old child
(right).

Figure 3. Specifications of the layered flat phantom model used in this study.

and computational time. In particular, in the regions where high fields were expected, the
maximum voxel size was limited to 1 mm in all directions, whereas outside that region the
grid size was increased to 6 mm.
Two anatomically correct MRI-based child head models, 3 year old and 7 year old
(figure 2), were used. In addition, two adult head models, one male and one female, were
used. In order to have a better understanding of SAR variations in complex biological tissue
models, even a simple flat phantom model consisting of skin, fat and muscle tissues was used.
The structure of the three-dimensional flat phantom model and antenna position is shown in
figure 3. The thickness value for the muscle layer was chosen large enough to ensure enough
penetration depth for the RF energy. The phantom had the same dielectric parameters for skin,
fat and muscle layer as in the MRI models.
A half-wave dipole antenna was used as an exposure source. The exposure conditions
were as follows: first, the antenna was placed vertically 20 mm away from the reference points
of the right ear (figure 4), oriented parallel to the body axis. In order to study the SAR in the
eye region, the antenna was placed horizontally 40 mm away from the left eye. In the case of
the layered flat phantom model, the antenna was placed 20 mm away from the phantom.
The dielectric parameter values were increased from 5 to 20% in three steps, +5%, +10%
and +20%, by first increasing the conductivity alone and then increasing both the conductivity
and permittivity simultaneously. Since the TBW of children differs very little compared to
1468 J Keshvari et al

Figure 4. The antenna position: left—antenna vertically placed in front of the ear at a distance of
20 mm; right—antenna horizontally placed in front of the eye at a distance of 40 mm.

that of adults, increasing dielectric values up to 20% is sufficient to study SAR variation in
tissue.
SAR values averaged over 1 g and 10 g tissue masses were calculated for the ear exposure
region, whereas SAR values averaged over 1 g and 5 g were calculated for the eye exposure
region. The average SAR is calculated for a cube, which includes the voxel with the largest
SAR value, and the volume of the cube is increased until the desired mass is achieved
(IEEE-1529 2000).

4. Results

The calculated SAR values were first normalized to 1 W net antenna input power. Then they
were compared to the SAR of the model with unchanged dielectric values.
The calculated SAR values are shown as bar graphs. Each bar represents the SAR
corresponding to the increased dielectric parameters. All values presented in the graphs are
percentage values. They represent the SAR variations in the models when dielectric values
were increased. Each colour represents the amount of increase (5, 10 and 20%) in dielectric
values giving a SAR variation in relation to conductivity, or both conductivity and permittivity.
The percentage variation of SAR has been calculated for all three models at 900, 1800 and
2450 MHz frequency. The reference dielectric values were selected from the literature (Gabriel
and Gabriel 1996).
In the ear exposure region, the SAR values were averaged over 1 g and 10 g tissue masses.
Pinna was not included in the SAR calculations. Because in some of the models the total eye
tissue mass was less than 10 g, the SAR values were averaged over 1 g and 5 g masses.

4.1. SAR variation in the ear exposure region

4.1.1. 900 MHz. Figures 5(a) and (b) represent the variation of SAR for 1 g and 10 g
averaging masses when the dielectric values were increased from 5 to 20% at 900 MHz.
At this frequency, the maximum variation occurs in the adult head model. In most of the
cases, all models show an increase in SAR when the conductivity, or both conductivity and
permittivity, is increased. The SAR variation is less for the10 g averaging mass compared to
the 1 g averaging mass. Interestingly, there is a large decrease in SAR for the 10 g averaging
volume compared to the 1 g averaging volume, especially in the male head model. When
Dielectric variation effect on SAR 1469

900MHz_1g average

(a) 16
14.82
14
13.27

12

10
8.61
8 7.64
7.02 7.08 sigma_+5%
6.67
6.23 6.23 sigma_+5%_epsilon_+5%
% SAR variation 6
sigma_+10%
4.70 4.46 4.66
4.13 sigma_+10%_epsilon_+10%
4
sigma_+20%
2.92
2.80
2.39 2.42
2.26 sigma_+20%_epsilon_+20%
2 1.89

0
(0.66)
(1.14)
(1.78)
-2 (2.38) (2.10)

-4
Layered flat Adult male 7-years old child 3-years old child
phantom

900MHz_10g average

(b) 8
6.96

6 5.80
4.94 4.67
4.28
4 3.69
2.69 2.99

2 1.89
1.49
0.93 0.71
0.50
0 (0.20)
sigma_+5%
% SAR variation (1.27) (1.26)
(1.42) (1.37)
(1.82) sigma_+5%_epsilon_+5%
-2
sigma_+10%
sigma_+10%_epsilon_+10%
-4
sigma_+20%
(5.15)
sigma_+20%_epsilon_+20%
-6 (6.45)

-8 (8.32)
(9.13)
(9.76)
-10
Layered flat phantom Adult male 7-years old child 3-years old child

Figure 5. (a) Relative 1 g SAR variation in the ear region of the MRI models and layered flat
phantom model due to an increase of 5–20% of dielectric values at 900 MHz. (b) Relative 10 g
SAR variation in the ear region of the MRI models and layered flat phantom model due to an
increase of 5–20% of dielectric values at 900 MHz.

both conductivity and permittivity values are increased simultaneously, the SAR values also
increase in all three models. The SAR increase in the child head models is less than that in
the adult head model, but the values are closer to the layered phantom values.
A possible explanation for the decrease of SAR when dielectric values were increased
is the field distribution for the specific averaging volume. Unlike in child head models
and a layered flat phantom model, the increase in dielectric values at this frequency causes
a concentrated electric field distribution that leads to larger SAR values in smaller tissue
volumes. As shown in figure 8, for the 1 g averaging volume the field is more concentrated
compared to the case where the dielectric value is increased. However, for the 10 g averaging
volume, the distribution of the field is more spread when the conductivity is increased.
This, in turn, causes a smaller SAR when averaged over 10 g compared to 1 g tissue mass
(figure 8).

4.1.2. 1800 MHz. Figure 6(a) shows a 1 g averaged SAR variation for all models at
1800 MHz exposure in the ear region when the dielectric values were increased between
1470 J Keshvari et al

1800MHz_1g average

8
(a)
7.03
6.34
6
5.57

4.65
4.36
4 3.83

2.71
2.50
2.24
% SAR variation 2 2.00 sigma_+5%
1.35 1.44 1.46
1.27
1.00 1.19 sigma_+5%_epsilon_+5%
0.92
sigma_+10%
0.35 0.33 0.30
0 sigma_+10%_epsilon_+10%
(0.61) sigma_+20%
(0.97) (0.93)
(1.33) (1.38) sigma_+20%_epsilon_+20%

-2 (2.11)

(3.07)
(3.48) (3.57)
(3.76)
-4
Layered flat Adult male 7-years old 3-years old Adult female
phantom child child

1800MHz_10g average
4
(b) 3.35
3 2.69

2 1.97

1.07
1 0.69
0.49
0.26 0.19 0.16
0 (0.03) (0.22) (0.08)
(0.33) (0.28)
(0.57)
(0.70)(0.60)
-1 (0.96) (0.90)
(1.12) (1.20)
% SAR variation (1.39) sigma_+5%
(1.60)
-2 sigma_+5%_epsilon_+5%
(2.61) sigma_+10%
-3 (2.91)
(3.32) (3.15) sigma_+10%_epsilon_+10%
(3.40)
(3.70) sigma_+20%
-4
sigma_+20%_epsilon_+20%
-5

-6 (6.23)

-7
Layered flat Adult male 7-years old 3-years old Adult female
phantom child child

Figure 6. (a) Relative 1 g SAR variation in the ear region of the MRI models and layered flat
phantom model due to an increase of 5–20% of dielectric values at 1800 MHz. (b) Relative 10 g
SAR variation in ear region of the MRI models and layered flat phantom model due to an increase
of 5–20% of dielectric values at 1800 MHz.

5 and 20%. For 1800 MHz (and 2450 MHz), the SAR variation in the adult female head
model is also presented. At this frequency, the increase in dielectric values caused a smaller
SAR variation. Both adult models showed an increase in SAR when the conductivity alone
or both conductivity and permittivity were increased. In the 1 g averaging volume, both
child models showed a decrease in SAR when only the conductivity was increased. The
largest variation was an increase of 7% for the adult model when the conductivity was
increased by 20%, and 5.6% for the 7 year old child model when both conductivity and
permittivity were increased by 20% (figure 6(a)). Child models follow a similar variation
but with different values, i.e. increasing the conductivity alone causes a decrease in SAR
but increasing both conductivity and permittivity causes an increase in SAR. Averaging
over 10 g mass yields in most of the cases a decrease in SAR. For example, a decrease
of 6.2% for the 3 year old child head model occurs when the conductivity was increased
by 20% (figure 6(b)). For the 7 year old child model, there is a very small change in
SAR when it is averaged over 10 g mass. At 1800 MHz for the adult male model, the
SAR increase in the 10 g average mass is about half of the value in the 1 g averaged
mass.
Dielectric variation effect on SAR 1471

2450MHz_1g average

(a) 12
10.54
10 9.62 9.27

6 5.85
5.10 4.97
4 3.77
3.00
2.63 2.57
2.15
% SAR variation 2 sigma_+5%
1.16 0.81 1.00 sigma_+5%_epsilon_+5%
0.29 0.39 0.06 0.38 0.31
0 (0.04)
(0.21) (0.35) (0.28) sigma_+10%
(0.82)
(1.32) sigma_+10%_epsilon_+10%
-2 sigma_+20%
(2.81)
sigma_+20%_epsilon_+20%
(3.76)
-4
(5.25)
-6 (6.15) (6.48)

-8
Layered flat Adult male 7-years old 3-years old Adult female
phantom child child

2450MHz_10g average
6
(b)
4.22
4
2.39 2.44
2 1.48
1.27
0.83 0.71 1.16
0.32
0
(0.82) (0.86)
(1.65) (1.40)(1.49) (1.71)
-2 (2.05)
(2.45) (2.71)
(3.43) (3.26)
(3.54) (3.62)
% SAR variation -4 (3.89) sigma_+5%
(4.74)
(5.48) sigma_+5%_epsilon_+5%
-6 (5.79)
sigma_+10%
(6.79)
sigma_+10%_epsilon_+10%
-8 sigma_+20%
(9.42) sigma_+20%_epsilon_+20%
-10
(11.07)
-12
(13.30)
-14
Layered flat Adult male 7-years old 3-years old Adult female
phantom child child

Figure 7. (a) Relative 1 g SAR variation in the ear region of the MRI models and layered flat
phantom model due to an increase of 5–20% of dielectric values at 2450 MHz. (b) Relative
10 g SAR variation in the ear region of the MRI models and layered flat phantom model due to an
increase of 5–20% of dielectric values at 2450 MHz.

Figure 8. SAR distribution in the ear region for 5% conductivity increase. Two pictures on left:
1 g averaging volume. Two pictures on the right: 10 g averaging volume. The green line is the line
passing through the voxel, where the SAR has its maximum value and the red-coloured squares
are SAR averaging volumes.

4.1.3. 2450 MHz. At this frequency the largest SAR variation (10.5%) occurs in the adult
male model when the averaging mass is 1 g. There are also decreases in SAR, especially for
the 3 year old and adult female models (figure 7(a)). For the 3 year old child model, the 1 g
1472 J Keshvari et al

900MHz_1g_Eye

(a) 8
7.55

sigma_+5%
4.00
4 sigma_+5%_epsilon_+5%
3.47
% SAR variation sigma_+10%
3 2.89 sigma_+10%_epsilon_+10%
sigma_+20%
2.05 "sigma_+20%_epsilon_+20%
2 1.85 1.79
1.35
1.04 1.16
1 0.85 0.89 0.910.96
0.53 0.66 0.57

(0.89)
-1
Adult male 7-years old child 3-years old child

900MHz_5g_Eye

(b) 8 7.82

6 sigma_+5%

sigma_+5%_epsilon_+5%
5 4.72
sigma_+10%
4.05 sigma_+10%_epsilon_+10%
4
% SAR variation sigma_+20%

3 2.84 "sigma_+20%_epsilon_+20%
2.41
2.06
2 1.76
1.51
1.10 1.20
0.94
1
0.36 0.50
0.23
0.07
0 (0.04)

(0.81) (0.74)
-1
Adult male 7-years old child 3-years old child

Figure 9. (a) Relative 1 g SAR variation in the eye region due to an increase of 5–20% of dielectric
values at 900 MHz RF exposure for the child and adult models. (b) Relative 5 g SAR variation in
the eye region due to an increase of 5–20% of dielectric values at 900 MHz RF exposure for the
child and adult models.

SAR value decreases about 6.2% when the conductivity increases by 20%. In the 7 year old
child model, the increased conductivity causes a very small change in SAR, but increasing
both dielectric parameters yields a 3.8% increase in SAR when averaged over 1 g. Unlike in
child models, the 1 g SAR at 2450 MHz increases almost linearly up to 10.5% in the male adult
model when either the conductivity or both conductivity and permittivity values are increased.
When averaged over 10 g mass, the SAR decreases for all MRI models, except for the
adult male model, where there is a 2.4% increase in SAR if the conductivity alone is increased
by 20% (figure 7(b)). In the layered phantom model, there is a very small change in SAR if
both dielectric parameters are increased, but rather a linear increase of up to 9.6% when the
conductivity alone was increased. At this frequency the effect of averaging volume is more
dominant than that at other frequencies. For example, the SAR of the adult model goes down
to +2.4% when averaged over 10 g compared with the value of 9.3% in 1 g mass. In both
child head models, the SAR decreases significantly when increasing both the conductivity and
permittivity. In the adult female head model, the decrease is about 12.3% (figure 7(b)).
Dielectric variation effect on SAR 1473

1800MHz_1g_Eye

(a) 6
4.65
4.36
4
2.71
2.50
2 1.46 sigma_+5%
1.27
0.51 0.33 sigma_+5%_epsilon_+5%
0 (0.37)
(1.08) sigma_+10%
(1.54) (1.32) (1.26)
(1.81)
-2 sigma_+10%_epsilon_+10%
(2.67)
% SAR variation (3.06)
(3.70) sigma_+20%
(3.83)
-4
(4.51)
"sigma_+20%_epsilon_+20%
(5.78)
(6.02) (5.85)
-6

(7.65)
-8

-10
(10.69)

-12
Adult male 7-years old child 3-years old child Adult female

1800MHz_5g_Eye

(b) 0
(0.57)
(0.89) (0.70)(0.60)
(0.90)
(1.20)

-2 (2.15)
(2.16)
(2.38) (2.44)
(2.79) (2.61)
(2.96) sigma_+5%
(3.29)
sigma_+5%_epsilon_+5%
-4
(4.43)
(4.71) (4.87) sigma_+10%

sigma_+10%_epsilon_+10%
% SAR variation -6 (6.18)
(6.84) sigma_+20%

"sigma_+20%_epsilon_+20%
-8

(9.09) (9.26)
(9.39)
(9.66)
-10

(11.35)
-12
Adult male 7-years old child 3-years old child Adult female

Figure 10. (a) Relative 1 g SAR variation in the eye region due to an increase of 5–20% of
dielectric values at 1800 MHz RF exposure for the child and adult models. (b) Relative 5 g SAR
variation in the eye region due to an increase of 5–20% of dielectric values at 1800 MHz RF
exposure for the child and adult models.

4.2. Eye region exposure and SAR variation in eye tissues

Because of the small volume of the eyes in child models, instead of 10 g, the SAR was
averaged over 5 g tissue mass. As in the ear region exposure, even in the eye region, SAR
variation is frequency dependent. The anatomy of the eye region in different models is more
similar to each other compared to that of the ear region where more significant differences
can be observed. This is why the difference of the SAR variation between different models is
smaller in the eye exposure compared to the ear exposure.
At 900 MHz, an increase in the dielectric values yields a maximum SAR increase of 7.6%
in the adult head model, a 3.5% increase in the 3 year old child head model and a 1.2% in the
7 year old child model (figure 9(a)). When the field penetrates well into the whole eye region,
increasing the average volume to 5 g does not make much difference in the SAR variation.
At 1800 and 2450 MHz when the dielectric values were increased from 5 to 20% the
1 g and 5 g averaged SAR values decrease almost in all cases (figures 10(a), (b) and 11(a),
(b)). Increasing the conductivity by 20%, at 1800 MHz, the adult female model is the only
1474 J Keshvari et al

2450MHz_1g_Eye

(a) 2

0.38 0.31
0 (0.28)
(1.24)
(1.48) (1.23)
(1.58) sigma_+5%
-2 (1.93)
(2.70) (2.58) sigma_+5%_epsilon_+5%
(2.96)
(3.06) (3.04) (3.13) (3.25)
sigma_+10%
-4
sigma_+10%_epsilon_+10%
(5.24) (5.25)
% SAR variation -6 sigma_+20%
(6.46) (6.27) (6.48)
"sigma_+20%_epsilon_+20%
(7.48)
-8
(8.47)
(9.29)
-10

-12
(12.56)

-14
Adult male 7-years old child 3-years old child Adult female

2450MHz_5g_Eye

(b) 0 (0.14)
(0.99) (0.86)
(1.09)

-2 (1.92) (2.05)
(2.64) (2.83) (2.73)
(3.18) sigma_+5%
(3.51) (3.62)
(3.85)
-4
sigma_+5%_epsilon_+5%
(4.74)
(5.53) (5.64) sigma_+10%
-6 (6.30)
(6.79) sigma_+10%_epsilon_+10%
% SAR variation
(7.47)
(7.74) sigma_+20%
-8
"sigma_+20%_epsilon_+20%

-10

(11.15)

-12 (12.11)
(12.32)

(13.30)
-14
Adult male 7-years old child 3-years old child Adult female

Figure 11. (a) Relative 1 g SAR variation in the eye region due to an increase of 5–20% of
dielectric values at 2450 MHz RF exposure for the child and adult models. (b) Relative 5 g SAR
variation in the eye region due to an increase of 5–20% of dielectric values at 2450 MHz RF
exposure for the child and adult models.

model which shows a SAR increase of up to 4.7% when the averaging volume is 1 g. By
increasing the averaging volume to 5 g, the SAR decreases in all models with 11.4% decrease
the largest decrease in the 7 year old child head model. When the dielectric values were
increased from 5 to 20% at 2450 MHz, there were substantial decreases in the SAR values for
all models for both 1 g and 5 g averaging volumes.

4.3. Single tissue phantom

The field behaviour is better understood if a phantom of cubical shape of a single tissue layer
is considered. In order to see how increasing the dielectric values would affect the field
behaviour, we ran a simulation at 900 MHz for a cube of the same dimension as the layered
phantom, but with only a single tissue type. A high water content tissue, muscle, was chosen
for this purpose. In figure 12, it is clearly seen that increasing the dielectric value by 20%
causes the field to converge more and hence to give different SARs at different averaging
volumes.
Dielectric variation effect on SAR 1475

Figure 12. Peak SAR distribution in a single tissue cubical shaped phantom with unchanged
dielectric values (top) and 20% increased conductivity (bottom). The values are normalized to the
peak SAR of the top figure, i.e., to the SAR of the unchanged dielectric muscle tissue.

5. Conclusions and discussions

Interpretation of equation (2) may give an impression that by increasing the conductivity
(sigma) and permittivity (epsilon) values, the SAR would increase. This could mostly be the
case if only peak SAR values are considered, i.e., if only the maximum peak SAR value in a
single voxel in the whole computational field is considered. In general, when the frequency
increases, the penetration depth of the RF energy decreases, resulting in a more superficial RF
energy absorption and a larger effect of averaging mass on SAR variation.
The analysis of this study reveals that SAR values may even decrease although conduc-
tivity or conductivity and permittivity values are increased to even unrealistically high levels
(compared with the values given in the literature) which are not plausible in human tissues.
For example, a decrease in the eye SAR values is significant both at 1800 and 2450 MHz
although the dielectric parameters are increased. The SAR variation in the eye region shows
that a curvilinear shape of the exposed organ together with its relative size to the wavelength
has a direct effect on the SAR variation. The bending effect is more significant for curvi-
linear shapes as shown in figure 12. However, for the eye tissues, the SAR variation for
all models is about the same. The difference in the SAR variation between 1 g and 5 g
masses is less in the eye region exposure compared to the ear region exposure (1 g and 10 g
tissue masses) because of a more uniform SAR distribution in eye tissues.
In the ear region exposure, a decrease in SAR values can also be observed when both
frequency and averaging volume are increased. Because the thickness of the tissue layers in
the ear region is relatively similar between the child head models, the SAR variations in the
child head models are also more consistent with each other compared to the variation between
the adult head models. For example, at 1800 and 2450 MHz the SAR decreases almost for all
cases when the dielectric values increase. This is because of the lower depth of penetration
of EM fields in a medium with higher conductivity resulting in a decrease in SAR in 1 g and
10 g averaging masses.
1476 J Keshvari et al

Examining the SAR variation in a layered flat phantom model, reveals that the prediction
of SAR variation, when the dielectric values are increased, is much easier in a well-defined
geometrical shape than in a complex anatomical tissue structure such as the human head. Since
the head models are anatomically different in shape with different tissue layer compositions, the
field penetration depth and field distribution also differ when the dielectric values are increased.
The SAR variation in the eye region is also a good example of the fact that the prediction
of SAR variation when the dielectric values are increased is easier for geometrically more
uniform structures. For a single tissue phantom, the analysis of RF field behaviour becomes
even more predictable.
The change in dielectric values in geometrical or anatomical lossy material means that
impedance matching of the antenna will change accordingly. The SAR variation depends not
only on the increase in the dielectric values but also on the variation of impedance matching
of the antenna, external anatomical shape of the models and the ratio of the dielectric values
in different tissue layers.
This study also shows that the same increase in the dielectric values in a layered flat
phantom model, and adult and child head models results in different SAR variations. This
indicates that anatomy of the head plays a key role in SAR variation when the dielectric values
are increased. The maximum 1 g averaged SAR increase of 15% was observed at 900 MHz
in the adult head model when both the conductivity and permittivity values were increased by
20%.
The averaging mass is also an important factor when assessing RF energy absorption. If
the peak SAR in a single voxel is examined, there is usually an increase in the SAR value.
However if the SAR is calculated for specific averaging masses such as 1 g, 5 g or 10 g, the
SAR may increase or decrease depending on the anatomy of the exposed region.
SAR variation is frequency dependent showing that by increasing the dielectric values,
SAR value changes are not the same between the selected frequencies. This is due to a lower
depth of penetration of the E field and a higher surface SAR when the frequency is higher.
In summary, this study shows that although dielectric parameters are increased, the SAR
value does not necessarily increase in human head exposure to RF energy; it may even decrease
under certain exposure conditions, such as most of the cases in eye exposure. The RF energy
absorption is, as expected, frequency dependent and the choice of averaging volume is an
important variable when analysing SAR with computational methods. The thickness and
composition of different tissue layers in the exposed regions within the human head play
a more significant role in SAR variation compared to the variation of the tissue dielectric
parameters. Although the findings of this study are based on computational dosimetry, it can
be concluded that interaction of RF energy with biological tissue, such as the human head,
is an extremely complex biophysical phenomenon. Even significant theoretical changes in
the fundamental physical parameters, tissue conductivity and permittivity, in the commonly
accepted ‘SAR equation’, do not allow one to make far-reaching conclusions on RF energy
absorption in the human body without in-depth analyses.

References

Altman P L 1974 Biology Data Book: Blood And Other Body Fluids ed D S Dittmer (Washington, DC: Federation of
American Societies for Experimental Biology) pp 350–3
Burkhardt M and Kuster N 1999 Review of exposure assessment for handheld mobile communications devices and
antenna studies for optimized performance Review of Radio Science 1996–1999 ed W R Stone (Oxford, UK:
Oxford University Press) chapter 34
Christ A and Kuster N 2005 Differences in RF energy absorption in the heads of adults and children
Bioelectromagnetics 26 S45–50
Dielectric variation effect on SAR 1477

Dimbylow P J 1993 FDTD calculations of the SAR for a dipole closely coupled to the head at 900 MHz and 1.9 GHz
Phys. Med. Biol 38 361–8
Dimbylow P J C 1994 Calculations in an anatomically realistic model of the head for mobile communication
transceivers at 900 MHz and 1.8 GHz Phys. Med. Biol. 39 1537–53
Fein G G 1978 Child development Englewood Cliffs (NJ: Prentice-Hall)
Gabriel C and Gabriel S 1996 Compilation of the dielectric properties of body tissues at RF and microwave frequencies
Technical Report AL/OE-TR-1996-0037 (London: Physics Department, King’s College)
Gandhi O P, Lazzi G and Furse C M 1996 Electromagnetic absorption in the human head and neck at 835 and
1900 MHz IEEE Trans. Microw. Theory Tech. 44 1884–7
Gandhi O P and Gang Kang 2002 Some present problems and a proposed experimental phantom for SAR compliance
testing of cellular telephones at 835 and 1900 MHz Phys. Med. Biol. 47 1501–18
Holland B A et al 1986 MRI of normal brain maturation Am. J. Neuroradiol. 7 201–208
IEEE-1529 2000 Recommended Practice for Determining the Spatial-Peak Specific Absorption Rate (SAR) Associated
with the Use of Wireless Handsets—Computational Techniques (New York: Institute of Electrical and Electronics
Engineers) (draft standard)
Keshvari J and Lang S 2005 Comparison of radio frequency energy absorption in ear and eye region of children and
adults at 900, 1800 and 2450 MHz Phys. Med. Biol. 50 4355–69
Kuster N and Balzano Q 1992 Energy absorption mechanism by biological bodies in the near field of dipole antennas
above 300 MHz IEEE Trans. VT 41 17–23
Martens L, De Moerloose J and De Zutter D 1995 Calculation of the electromagnetic fields induced in the head of an
operator of a cordless telephone Radio Sci. 30 283–90
Martinez-Burdalo M et al 2004 Comparison of FDTD-calculated specific absorption rate in adults and children when
using a mobile phone at 900 and 1800 MHz Phys. Med. Biol. 49 345–54
Peyman A, Rezazadeh A A and Gabriel C 2001 Changes in the dielectric properties of rat tissue as a function of age
at microwave frequencies Phys. Med. Biol. 46 1617–29
Schoenborn F, Burkhardt M and Kuster N 1998 Differences in energy absorption between heads of adults and children
in the near field of sources, Health Phys. 74 160–68
van der Knaap M S and Valk J 1990 MR imaging of the various stages of normal myelination during the first year of
life Neuroradiology 31 459–70