CSIRO - June 1994




Basic Concepts

Time-varying electric and magnetic fields induce electric fields and electric currents in conducting materials, including biological tissues. In general, biological effects depend on the strength of induced current and fields, although effects have been observed that appear to depend on other parameters. These effects are still not well understood and if they are confirmed, appropriate dosimetry may need to be developed.

Given the complex shape and non-homogeneous character of biological systems, it is exceedingly difficult to characterize completely the propagation of electromagnetic fields in the human body. The strength of the field reaching subcutaneous tissue is partly determined by absorption in the outer layers and by the reflection at the interface between different media. The amplitude of the electric and magnetic fields decreases exponentially with distance. The penetration depth of the field (ò) is defined as the distance at which the field is attenuated by a factor of e-1 = 0.368. It decreases with increasing frequency according to the equation:

ò = 1/(sq root(pi fso)

where f is the frequency, o is the permeability of the free space (1.26 x 10-6 H/m) and sis the conductivity of the medium. The last quantity, in turn, depends on the frequency.

The penetration depths of high-water-content biological medium at various frequencies are given below:

   f (MHz)    s (S/m)   ò(mm)
     10     0.625      200
    100     0.889       53
    300      1.37        25
    915        1.6       13
   2450      2.21        6.8
   3000       2.26        6.1
   5000       3.92        3.6
  10000      10.3        1.6

In the radiofrequency band, two quantities are commonly used to quantify the interaction between the field and the biologic medium, ie the current density (used predominantly at frequencies below 100 kHz) and the specific absorption rate (SAR). The latter describes the rate at which radiant energy is deposited into a unit mass of tissue and is usually expressed in units of W/kg. The current density can be usefully compared to those known to produce physiological responses (eg muscle stimulation) or to endogenous body currents.

The interaction between radiant energy and an absorbing medium is particularly efficient when the dimension of the medium is equal, or approximately equal, to a multiple of 1/2 the wavelength (resonance condition). Therefore, the peak absorption for an adult man exposed to a wave with the electric field parallel to the length of the body, occurs at about 75 MHz. The head resonates at much higher frequencies (approx 1GHz).

At higher frequencies, the basic interaction mechanism is the rotation of molecules in which positive and negative charges are separated in space (polar molecules). The most common such molecule in biological matter is water. Polar molecules tend to align themselves with the electric field and, as this oscillates, they tend to follow these oscillations. In this process energy is dissipated in the form of heat. The resulting increase in temperature (delta T) measured in °K can be expressed as delta T = (SAR -HLR)xt/C, where HLR is the rate of heat loss per unit mass, due to thermal conduction and convection, and t is the time in seconds.

The SAR is not readily measurable in practice, therefore in order to prevent overexposure, it is necessary to resort to published data that relate the SAR to the electric and magnetic field strengths or the power density of the incidence radiation.

The SAR can be determined empirically or theoretically. Both of these methods have limitations and rely on each other for validation and complementation. Computational methods indicate that the SAR is a function of frequency, of the wave polarization and that it peaks at resonant frequencies.

With respect to the absorption characteristics of the human body, the RF range can be divided into four regions:

  • up to 30 MHz (sub-resonance range), radiation incident on the trunk is predominantly absorbed at the surface, whereas that incident on the legs and neck may result in significant energy absorption. In this range, absorption increases rapidly with frequency.
  • between 30 MHz and 300 MHz (resonance range), the SAR per unit incident power density reaches a peak, as resonance conditions are attained for the whole body and body parts.
  • between about 400 MHz and 3 GHz (hot spot range), significant heating may occur in particular sites in the body. The size of these 'hot spots' decreases from several cm to about 1 cm as the frequency increases.
  • for frequencies above about 3 GHz (surface absorption range), radiant energy is absorbed at, and heating is limited to, the surface of the body.


Fig. 11.1 Variation of normalised SAR with frequency and related absorption characteristics in living organisms. (from WHO, 1993)

It must be stressed that the above considerations depend on the body dimensions, therefore adequate allowance must be made when extrapolating results obtained by animal studies. For example, at 2450 MHz, the SAR resulting from exposure to 10 W/m2, with the E vector parallel to the long axis of the body is about 70 times higher for a mouse (whose length is comparable to the radiation's wavelengths) than for a man (Durney et al 1980).

SARs have also been empirically measured on human volunteers or on human models. For frequencies below and close to the resonant frequencies, experimental values exceeded the calculated values by factors of 3-4 (Hill 1984 a,b,c; Guy 1987).

A good agreement between calculated values and measurements in models were found for frequencies at and above the resonant frequency for irradiation in free space and with the electric field parallel to the long axis of the body.

The SAR is also affected by several practical exposure conditions (position of the body and of the limbs, distance from the ground, footwear etc). Using the results of these computations and measurements and including appropriate safety margins, limits to the maximum acceptable SAR or, as appropriate, maximum induced current density, are translated into maximum permissible exposure levels.

Within the context of the current concerns about brain tumors allegedly caused by exposure to RF radiation from cellular phones, it is interesting to examine the absorption properties of the brain itself and of the DNA molecule.

The brain is a tissue rich in water, but with a substantial proportion of fatty tissue. It has a conductivity at about 1 GHz of approximately 1.1 S/m and a penetration depth of approximately 1.5 cm. While the brain has a metabolic rate about 16 times higher than that of muscle tissue (and therefore generates much more heat that muscle tissue), this is more than compensated by 20-fold higher rate of blood flow and a somewhat higher thermal conductivity (Guy 1974). Therefore, brain tissue is no more prone to RF heating that muscle tissue.

There have been conflicting results on the question of resonant absorption of microwave radiation in DNA. Resonance absorption peaks were reported by Swicord et al (1983) and Edwards et al (1984, 1985). Zhang (1989) calculated that resonance absorption of microwave energy in DNA is possible in the GHz and sub-THz frequency ranges. The most recent reports, however (Foster et al 1987, Gabriel et al 1987 Maleev et 1987, Rhee et al 1988, Garner et al (1990) , tend to argue against the presence of resonant absorption.

Current Issues

A new challenge has arisen recently, due to the relatively new situation of radiating antennae being routinely placed very close to the human body. Data calculated or measured in this condition are extremely dependent on the geometry of the model used.

Gandhi and co-workers used both computational and experimental techniques to obtain SARs in the human head for ten cellular phone from four different manufacturers. For their computations, they used a high resolution model consisting of very small cells (2x2x3 mm) each having appropriately defined properties reflecting their anatomical equivalent. The computation results were verified using an head-shaped model made of tissue equivalent materials. By contrast, Balzano (1994) and Kuster (1993) used a much cruder approach, relying on a head-shaped mannequin filled with a solution that simulates brain tissue and, approximately, bone and fatty tissue.

Gandhi's results were summarized as follows:

  • peak SAR over any 1 g of tissue         0.09 - 0.29 W/kg
  • peak SAR over any 1 g of brain tissue     0.04 - 0.17 W/kg
  • whole body SAR                      0.5 - 1.1 mW/kg

The highest SAR were found to occur in tissue in the upper ear.

These figures contrast sharply with those obtained by Balzano et al (1994) and Meier and Kuster (1993), although the phones output power was the same in all cases (0.6 W).

Balzano et al (1994) measured the SAR induced in human-equivalent phantoms by two types of Motorola cellular phones. They found SARs as high as 1.4 W/kg for "Flip" phones (ie phones with a very thin radio case and a collapsible antenna; when the antenna is extended, the radiation emitted is slightly further away from the head and this results in lower SARs).


"Classic" phones Peak SAR 0.2 - 0.4 W/kg
"Flip" phones Antenna collapsed 0.8 - 1.4 W/kg
Antenna extended 0.6 - 1.0 W/kg

SARs of up to 1.7 W/kg were measured by Kuster et al (1993) under "standard" conditions and SARs as high as 5.3 W/kg under "worst case" conditions, with the antenna actually touching the head.

Mokhtech et al (1994, BEMS Newsletter March/Apr p8), also reported that local peak SARs exceeding the ANSI safety limits may be encountered.

Meier and Kuster (1993) argue that "anatomically correct shell phantoms [such as that used by Gandhi] have been proven poorly suitable to achieve good reproducibility because the position of the radio with respect to the phantom is difficult to define". Their and Balzano's approach is of determining the SAR under 'worst case conditions'.

Which of the two approaches is more appropriate is as much a political as a scientific decision. From a technical view point, the main concern is the view expressed by Balzano (quoted in Microwave News, Jan/Feb 1994, p 13) that Gandhi's results were not obtained by pressing the phone against the ear. At the time of writing, details of Gandhi's measurements have not yet appeared in the scientific literature. If Balzano's claim is correct, it casts some doubt on the usefulness of Gandhi's results. Kuster also suggests that "for some cellular phones, it is not unusual for the antenna to touch the skull" (Microwave News, Jan/Feb 1994, p 14). Nevertheless, even Kuster's results indicate that the majority of the phones tested complied with the ANSI/IEEE C95.1 Standard (and consequently, with the Australian/New Zealand Standard), when the phones are tested under standard conditions. When tested under worst-case conditions (ie with the antenna touching the head), they all exceeded the limit, with 2 of the 6 models tested being about 50% above the recommended limit.

In order to ensure compliance with safety limits, it is necessary that testing conditions be reproducible. The requirement that compliance be demonstrated under "worst case conditions" overcomes some of the standardization problems, such as the position of the hand holding the phone. In "worst case" tests, the telephone is supported against the head by a non-conducting prop. In practice, the presence of the hand reduces the SAR measured in the head.

The question of low-level dosimetry has been given very little consideration to date. The evidence of non-thermal effects is very complex to interpret and generalize. WHO (1993) suggest that the SAR may also provide a valid measure of all intensity dependent interaction mechanisms, although some additional information may be required, such as modulation characteristics and amplitude "windows" that are biologically active. However, at this stage, it is difficult to see how these additional characteristics could be defined and even whether the observed non-thermal effects are intensity dependent.

For the purpose of further research on human subjects, "dose" or "exposure" need to be defined not necessarily in rigorous, biologically validated terms, but at least in terms that will allow to identify reliably 'cohorts' of subjects whose exposure conditions are clearly different from that of the general population and, preferably, to establish an exposure gradient. The most obvious, but not necessarily accurate, tentative definition of exposure is the time-weighted average of the field power density.

For each of these tentative definitions, the following questions need to be addressed and answered:

  • how is 'exposure' distributed among the community?
  • how is 'exposure' distributed among specified occupational groups?
  • what attributes can be used as 'proxy' for exposure? (eg occupational title, proximity to specific sources etc)
  • how strong is the association between a proxy and the 'exposure'?
  • if proxies are used, what is the likely effect of inaccurate exposure assessment on the results of the study?
  • are any of these proxies associated with other possible risk factors (eg chemical carcinogens, other EMR frequencies) that may confound the results of new studies?


[BACK]                                         [NEXT]


Current Research Abstracts Index

Please e-mail comments, information and updates to DON MAISCH: