2, MAY 2006
εr = εαrw · ε1−α
rt (2)
TABLE I
RAT’S DIELECTRIC PEOPERTIES DERIVED FROM AN EMPIRICIAL FORMULA AND THIER DIFFERENCES [%] IN PARENTHESIS FROM MEASURED DATA
noted that the 1- and 10-g averaged spatial peak SARs were cal-
culated according to the standard averaging procedure defined
in IEEE Std C95.3-2002 [15]. That is to say simply, the spatial
average SAR was evaluated in cubical volumes that contain a
mass within 5% of 1 or 10 g, and the cubical volume centered
at each voxel was expanded in all directions until the desired
mass reached with no surface boundary of the averaging volume
extending beyond the most exterior surface of the head. Since
the pinna should be considered as an extremity, we excluded it
during the averaging process. As a result, we found that both the
1- and 10-g averaged spatial peak SARs had almost the same
values as those with the adult dielectric properties. In com-
parison with the adult dielectric properties, the child dielectric
properties resulted only in a difference of 1.3% at maximum for
was based on the TBW database for humans as a function of the 1- and 10-g averaged spatial peak SAR. This implies that
age, which is shown in Appendix (Fig. A1). Table II gives the the age effect on the spatial peak SAR of dielectric properties is
dielectric properties for both adult and children. The latter was negligible.
derived with the above-mentioned empirical formula. Fig. 4 also shows the resistive components of antenna input
It should be noted that the TBW is not guaranteed to be impedance Re[Zin ] with bars. First, Re[Zin ] for child head mod-
the same in various human tissues because the developmental els exhibited smaller values than that for the adult head model,
change of tissues may be quite different. However, in view of and consequently, the spatial peak SARs in child head models
the acceptable accuracy for dielectric properties of rat tissues were somewhat higher. This phenomenon has been discussed
derived from the TBW, applying the data in Table II for the SAR in detail in [7], and whether higher peak SAR occurs in the
calculation of children should be a reasonable approximation. child heads actually depends on the antenna input impedance.
Second, for the same child head models, the child dielectric
IV. RESULTS AND DISCUSSION properties did not result in a significant change in Re[Zin ], and
consequently no significant change was observed in the 1- or
A. Spatial Peak SAR
10-g averaged spatial peak SAR.
Fig. 4 shows the FDTD-calculated 1-and 10-g averaged spa- However, as shown in Table I, the derived dielectric prop-
tial peak SARs for an antenna output of 0.27 W. It should be erties for rats exhibit a difference of ±20% at maximum from
WANG et al.: APPROXIMATION OF AGING EFFECT ON DIELECTRIC TISSUE PROPERTIES FOR 411
B. Penetration Depth
To clarify whether there is a deeper penetration in the child depth in children heads is actually somewhat shallower than
heads, we attempted the calculation of penetration depth of elec- that in the adult head because the electromagnetic field attenu-
tromagnetic fields for mobile telephones, which is considered ation in higher dielectric property tissues or children tissues is
as the distance where the SAR decreases to e−2 of the value larger.
at head surface. However, there is not a standard method for
the penetration depth calculation. One method often used is
to apply the exponential regression to approximate the SAR C. Physical Consideration
profile along a horizontal line starting from the mobile tele- The above results are easy to understand when we consider
phone side to another side of the head, and then derive the a plane wave incident to a semi-infinite multislab of tissues.
penetration depth where the SAR decreases to e−2 of the sur- Different from the near-field exposure for anatomical head
face value. This method, however, does not give a unique so- models, an analytical expression is now available which
lution because there are so many choices for the horizontal provides a physical insight into the results. As a primary step
lines. for the physical insight, we considered a plane wave incident
We therefore proposed a new method for the penetration depth normally to the multislab.
calculation. Although the method was described for a mobile Fig. 6(a) shows the semi-infinite multislab structure consist-
telephone aligning with the head, it should be applicable to the ing of skin, fat, muscle, skull, and brain. Referring to the data of
more popular usage of mobile telephones such as the cheek human head tissue thickness given in [16], we determined two
and tilted positions defined in compliance testing. As shown models, Models A and B, which have the typical tissue thick-
in Fig. 5, we first defined the SAR value at the head tissue ness for adults and children, respectively, as given in Table IV.
voxel nearest the antenna feeding point as the SAR at the head For the multislab structure, we replaced it by a transmission line
surface. We then searched all the voxels where the SAR values model as given in Fig. 6(b). At depth x from the surface x1 (= 0),
were within (e−2 ± 1%) times the SAR at the head surface. we have (4), shown at the bottom of the next page, based on the
From the average distance between these voxels to the defined transmission line theory, where ρm , σm , γm , and z0m are the
head surface voxel, we derived an averaged penetration depth, mass density, conductivity, propagation constant, and charac-
which is a unique and reasonable solution. teristic impedance, respectively, and zim is the input impedance
Table III gives the penetration depths for the adult, 7-year- looking toward the right side for the mth layer. Denoted by εrm ,
old, and 3-year-old head models. Also shown in the table are the relative permittivity for the mth layer, we have
the standard deviations and the voxel numbers that met with
(e−2 ± 1%) times the surface SAR value. As can be seen in
γm = jω µ0 ε0 (εrm − jσm /ωε0 ) (5)
the table, the electromagnetic penetration for the adult and
children had almost the same depth, both around 50 mm at µ0
Z0m = (6)
900 MHz. Moreover, it is interesting to note that the penetration ε0 (εrm − jσm /ωε0 )
412 IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, VOL. 48, NO. 2, MAY 2006
Fig. 7. Normalized SAR profiles for the two five-layer slab models with dif-
Fig. 6. (a) Five-layer slab model with a plane wave incident normally to it. (b) ferent dielectric properties.
Equivalent transmission line model.
TABLE IV V. CONCLUSION
TISSUE THICKNESS OF MULTISLAB MODEL
Because of the higher conductivity reported for young rats,
a question has risen for the employment of adult’s dielectric
properties in the peak SAR evaluation of child heads. To answer
this question, we have derived an empirical formula according
to Lichtenecker’s exponential law for the dielectric properties
of various tissues as a function of the hydrated rate or the TBW,
and confirmed its validity using the rats’ data measured by Pey-
Zi(m +1) + Z0m tanh[γm (xm +1 − xm )] man et al. Applying the formula via relating the TBW to the
Zim = Z0m (7)
Z0m + Zi(m +1) tanh[γm (xm +1 − xm )] ages, we have derived the dielectric properties in 7-year-old and
Zi5 = Z05 . (8) 3-year-old child head models. We then have calculated the spa-
tial peak SAR for a 900-MHz mobile telephone with respect to
the ages. As a result, we have found that the dielectric properties
Fig. 7 shows normalized SAR profiles for the two multislab for the child head models do not affect significantly the 1- or
models with different dielectric properties, which demonstrates 10-g averaged spatial peak SAR. Even in an extreme case, the
the exponential SAR decrease with depth, although there were increased 1- or 10-g averaged spatial peak SAR is still within
two steep falls in the fat and skull layers. The penetration depths 10%. This finding could be explained as a cancellation of the
were around 43 mm in Model A and were somewhat smaller increased conductivity and decreased electric field penetrating
than that derived from the anatomical human head models. It into tissue because of the same degree of increase between the
should be noted that for Model A with the adult, 7-year-old, conductivity and permittivity in children compared to adults.
and 3-year-old dielectric properties, we found that the higher Moreover, the calculation of the penetration depth in child head
dielectric properties give a higher SAR level at the surface of the models has exhibited an almost same value as or somewhat
brain layer, but a more rapid SAR decrease with deepening into smaller value than that in the adult model.
the brain tissue. Moreover, Model B with 3-year-old dielectric Of course, the validity of the proposed approach to derive the
properties and thinner (child) tissue thickness exhibited a more dielectric properties for aged tissues was confirmed only in rats
rapid SAR decrease and shallower penetration depth than Model and not in humans. Its validity to humans needs to be proven via
A with thicker (adult) tissue thickness. These phenomena are high-precision dielectric property measurement, which should
identical to the FDTD-calculated results for the anatomical head be a subject for further studies.
models. From the definition of SAR, a higher conductivity acts
toward increasing the SAR. On the other hand, the permittivity
also increases in child tissue, which acts toward decreasing APPENDIX
the electric field penetrating into the tissue. The two effects The hydrated rate α is related to the total body water (TBW)
cancel out because of their same degree of increase between by α = ρ · TBW where ρ is the tissue density. Fig. A1 shows the
the conductivity and permittivity so that the spatial averaged TBW as a function of age for humans, which is based on the data
SAR does not change significantly between the adult and child in [12]. As can be seen, the TBW varies to a great extent under
head models. 3-year-old, but becomes insignificant to ages over 3-year-old.
2
cosh{γ m (x−x m )}−(Z 0m /Z i m ) sinh{γ m (x−x m )}
SAR(x) x ≥ xm (m ≥ 2) : ρ1σm
× Πm −1 cosh{γ −x )}+(Z /Z −x
= ρm σ1 k =1
(x
k k +1 k 0k i k +1) sinh{γ (x
k k +1 k )} (4)
SAR(x1 ) x < x2 : | cosh{γ1 (x − x1 )} − (Z01 /Zi1 ) sinh{γ1 (x − x1 )}|2
WANG et al.: APPROXIMATION OF AGING EFFECT ON DIELECTRIC TISSUE PROPERTIES FOR 413