United States
Environmental Protection
Agency
Health Effects Research
Laboratory
Research Triangle Park NC 27711
EPA-600/8-83-026A
June 1983
External Review Draft
Research and Development
Biological Effects of
Radiofrequency
Radiation
Part 1 of 3
Review
Draft
(Do Not
Cite or Quote)
-------
EPA-600/8-83-026A
June 1983
External Review Draft
Biological Effects of Radiofrequency Radiation
Edited By
Daniel F. Cahill and Joe A. Elder
Health Effects Research Laboratory
Research Triangle Park, North Carolina 27711
NOTICE
This document is a preliminary draft. It has not been formally released by EPA
and should not at this stage be construed to represent Agency policy. It is
being circulated for comment on its technical accuracy and policy implications.
HEALTH EFFECTS RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
DISCLAIMER
This report is an external draft for review purposes only and does not
constitute Agency policy. Mention of trade names or commercial products does
not constitute endorsement or recommendation for use.
i i
-------
FOREWORD
The many benefits of our modern, developing, industrial society are
accompanied by certain hazards. Careful assessment of the risk of existing
and new man-made environmental hazards is necessary to establish sound regula-
tory policy. Environmental regulations enhance the quality of our environment
in order to promote the public health and welfare and the productive capacity
of our nation's population.
The Health Effects Research Laboratory conducts a coordinated environ-
mental health research program in toxicology and clinical studies. These
studies address problems in air pollution, radiofrequency radiation, environ-
mental carcinogenesis and the toxicology of pesticides as well as other chemical
pollutants. The Laboratory participates in the development and revision of
air quality criteria documents on pollutants for which national ambient air
quality standards exist or are proposed, provides the data for registration of
new pesticides or proposed suspension of those already in use, conducts
research on hazardous and toxic materials, and is primarily responsible for
providing the health basis for radiofrequency radiation guidelines. Direct
support to the regulatory function of the Agency is provided in the form of
expert testimony and preparation of affidavits as well as expert advice to the
Administrator.
The intent of this document is to provide a comprehensive review of the
scientific literature on the biological effects of radiofrequency radiation.
The purpose of this effort is to evaluate critically the current state of
knowledge for its pertinence and applicability in developing radiofrequency-
radiation exposure guidelines for the general public.
F. G. Hueter, Ph.D.
Director
Health Effects Research Laboratory
i i i
-------
ABSTRACT
This document presents a critical and comprehensive review of the avail-
able literature on the biological effects of radiofrequency (RF) radiation
through 1980. The objective is to determine whether the existing data base
can contribute to the formulation of RF-radiation exposure guidance for the
general public.
The frequency range of concern in this document is 0.5 MHz to 100 GHz,
which includes all the significant sources of population exposure to RF radia-
tion. Research reports that are judged to be credible according to a set of
objective criteria are examined for the relation between the RF energy ab-
sorbed and the presence or absence of biological effects. The reported conse-
quences of the interaction between RF radiation and biological systems are
examined from four perspectives by 1) RF-energy-induced core temperature
increases, 2) whole-body-averaged specific absorption rate (SAR), 3) the
exogeneous energy burden as a percentage of resting metabolic rate, and 4)
actual human experiences documented in epidemiological studies.
The existing data base does lead to tentative conclusions about the
relation between RF-radiation exposure and biological effects that may serve
as unadjusted upper limits for population exposure guidance; however, the un-
certainties and unknowns in the current state of knowledge are potentially
significant.
-------
CONTRIBUTORS
Ernest N. Albert*
Joseph S. Ali
John W. All is
Ezra Berman
Carl F. Blackman.
Daniel F. Cahill
Joe A. Elder
Michael I. Gage
Christopher J. Gordon
Doreen Hill
William T. Joines
James B. Kinn K
William P. Kirk9
Charles G. Liddle
James R. Rabinowitz
Ralph J. Smialowicz
Ronald J. Spiegel
Claude M. Weil
U.S. Environmental Protection Agency
Office of Research and Development
Office of Health Research
Health Effects Research Laboratory
Research Triangle Park, NC 27711
^Department of Anatomy
The George Washington University
Medical Center
Washington, DC 20037
^Current Address:
Carolina Power and Light Company
Harris Energy and Environmental Center
Route 1, Box 327
New Hill, NC 27562
^Office of Health Research
U.S. Environmental Protection Agency
Washington, DC 20406
^Current Address:
Three Mile Island Field Station
U.S. Environmental Protection Agency
100 Brown Street
Middletown, PA 17057
v
-------
CONTENTS
Foreword iii
Abstract iv
Figures ix
Tables xii
Acknowledgment xv
1 Introduction 1-1
(Daniel F. Cahill^
1.1 Ground Rules and General Assumptions 1-1
1.2 General Approach 1-2
1.3 Specific Approach Used in This Review 1-3
2 Summary and Conclusions 2-1
(Daniel F. Cahill and Joe A. Elder)
3 Physical Principles of Electromagnetic Field Interactions . . . 3-1
3.1 Electromagnetic Field Theory 3-1
(William T. Joines)
3.1.1 Electromagnetic spectrum - 3-1
3.1.2 Wave propagation 3-5
3.1.3 Wave modulation 3-11
3.2 RF-Field Interactions with Biological Systems 3-15
(Claude M. Weil and James R. Rabinowitz)
3.2.1 Scattering and absorption of electromagnetic
waves 3-15
3.2.2 RF dosimetry definitions 3-31
3.2.3 Analytical and numerical RF electromagnetic
interaction models 3-34
3.2.4 Mechanisms of RF interaction with biological
systems 3-52
3.3 Experimental Methods 3-67
(Claude M. Weil and Joseph S. Ali)
3.3.1 Exposure methods used in biological
experimentation 3-67
3.3.2 Animal holders 3-97
3.3.3 Densitometric instrumentation 3-102
3.4 Dosimetric Methods 3-115
(James B. Kinn)
3.4.1 Whole body dosimetry 3-115
3.4.2 Regional dosimetry 3-119
3.4.3 Unresolved questions 3-122
vi i
-------
Page
4 Effect of RF-Rad1ation Exposure on Body Temperature 4-1
4.1 Thermal Physiology 4-1
(Christopher Gordon)
4.1.1 Temperature regulation 4-1
4.1.2 Ambient temperature vs. RF-radiation exposure . . . 4-3
4.1.3 Mechanisms of heat gain during RF-radiation
exposure 4-5
4.1.4 Local thermal responses: Effect of RF-radiation
exposure on blood flow 4-11
4.1.5 Whole-body thermal response to RF-radiation
exposure and dependence on ambient
conditions 4-12
4.1.6 Heat stress and the general adaptation syndrome . . 4-17
4.1.7 Thermal physiology of humans 4-19
4.1.8 Thermoregulatory state and physiological
responsiveness to RF radiation 4-24
4.1.9 RF-radiation exposure and behavioral temperature
regulation 4-26
4.1.10 Unresolved questions 4-27
4.2 Numerical Modeling of Thermoregulatory Systems in Man and
Animals 4-33
(Ronald Spiegel)
4.2.1 Heat-transfer models 4-33
4.2.2 RF-radiation-heat-transfer models 4-37
4.2.3 Numerical results 4-42
4.2.4 Unresolved questions 4-45
5 Biological Effects of RF Radiation 5-1
5.1 Cellular and Subcellular Effects 5-1
(John W. All is)
5.1.1 Effects on molecular systems 5-3
5.1.2 Effects on subcellular organelles 5-8
5.1.3 Effects on single cells 5-13
5.1.4 Unresolved questions 5-23
5.2 Hematologic and Immunologic Effects
(Ralph J. Smialowicz) 5-31
5.2.1 Hematology 5-33
5.2.2 Immunology 5-46
5.2.3 Unresolved questions 5-71
5.3 Reproductive Effects 5-73
(Ezra Berman)
5.3.1 Teratology 5-73
5.3.2 Reproduction efficiency 5-99
5.3.3 Testes 5-103
5.3.4 Unresolved questions 5-112
viii
-------
Page
5.4 Nervous System 5-117
(Ernest N. Albert)
5.4.1 Morphologic observations 5-119
5.4.2 Blood-brain barrier studies . 5-122
5.4.3 Pharmacological effects 5-126
5.4.4 Effects on neurotransmitters 5-129
5.4.5 Unresolved questions 5-131
5.5 Behavior 5-137
(Michael I. Gage)
5.5.1 Introduction 5-137
5.5.2 Summary 5-139
5.5.3 Naturalistic behavior 5-140
5.5.4 Learned behavior 5-145
5.5.5 Interactions with other environmental stimuli . . . 5-159
5.5.6 Unresolved questions 5-163
5.6 Special Senses 5-171
(Joe A. Elder)
5.6.1 Cataractogenic effects 5-171
5.6.2 Unresolved questions 5-184
5.6.3 Auditory effects 5-186
5.6.4 Unresolved questions 5-199
5.6.5 Human cutaneous perception 5-203
5.6.6 Unresolved questions 5-206
5.7 Other Physiological and Biochemical Effects 5-209
(Charles G. Liddle)
5.7.1 Clinical chemistry and metabolism 5-209
5.7.2 Endocrinology 5-219
5.7.3 Growth and development 5-225
5.7.4 Cardiovascular system 5-228
5.7.5 Calcium ion efflux 5-233
5.7.6 Unresolved questions 5-237
5.8 Genetics and Mutagenesis 5-241
(Carl F. Blackman)
5.8.1 Introduction 5-241
5.8.2 Effects on genetic material of cellular
and subcellular systems 5-243
5.8.3 Effects on genetic material of higher-order
biological systems 5-252
5.8.4 Unresolved questions 5-256
5.9 Life Span and Carcinogenesis 5-267
(William P. Kirk)
5.9.1 Life span 5-267
5.9.2 Carcinogenesis 5-273
5.9.3 Prausnitz and Susskind study 5-276
5.9.4 North Karelia connection 5-279
5.9.5 Moscow Embassy study 5-280
5.9.6 U.S. Navy study 5-281
5.9.7 Unresolved questions 5-282
ix
-------
Page
5.10 Human Studies 5-285
(Doreen Hill)
5.10.1 Occupational surveys 5-285
5.10.2 Mortality studies 5-291
5.10.3 Ocular effects 5-297
5.10.4 Reproductive effects 5-303
5.10.5 Unresolved questions 5-307
6 Assessment 6-1
(Daniel F. Cahill and Joe A. Elder)
6.1 Core Temperature 6-2
6.2 Specific Absorption Rate 6-5
6.3 Resting Metabolic Rate 6-13
6.4 Human Studies 6-18
References R-l
Glossary G-l
x
-------
FIGURES
Number Page
3-1 Far-field electromagnetic wave at a particular instant
in time 3-8
3-2 Power density vs. distance along axis from antenna
aperture 3-12
3-3 Interaction of RF radiation with electrical conductors,
biological tissue, and electrical insulators 3-16
3-4 Energy distribution in proximity to man at 1 GHz at the
chest plane contour presentation 3-19
3-5 Dielectric data for tissues in RF range 0.01 to 10 GHz 3-22
3-6 Illustration of object size vs. wavelength dependence 3-24
3-7 Whole-body average SAR vs. frequency for three polarizations
in a prolate spheroidal model of a human 3-26
3-8 Absorption dependence on various ground and multi-path
factors 3-30
3-9 ARD distribution in core of 6-cm radius multi-layered
sphere at 1650 MHz 3-42
3-10 Curve fitting of SAR data for a prolate spheroidal model
of man for three basic orientations 3-44
3-11 Effect of a capacitive gap on average SAR between the man
model and the ground plane 3-46
3-12 A realistic block model of man 3-48
3-13 Absorption for man block model standing on ground plane .... 3-50
3-14 The real component of the complex permittivity of muscle
as a function of frequency 3-55
3-15 EPA 2450-MHz anechoic chamber facility or 2.45-GHz far-field
facility 3-72
xi
-------
Number Page
3-16 EPA 2450-MHz anechoic chamber facility: diagram of the
microwave exposure facility 3-73
3-17 Diagram of absorber-lined horn 3-74
3-18 Diagram of a point-source compact range 3-75
3-19 Miniature anechoic chamber facility 3-77
3-20 Photograph of tapered exposure chamber at EPA facility .... 3-78
3-21 Facility for simultaneous exposure of 10 animals with
minimal inter-animal interaction 3-79
3-22 Monopole-over-ground plane irradiation facility 3-81
3-23 Coaxial air-line system for high power exposures of cell
cultures 3-86
3-24 Parallel-piate (microstrip) exposure system 3-87
3-25 Block diagram of complete RF near-field synthesizer 3-87
3-26 EPA 100-MHz rectangular strip line or Crawford cell 3-89
3-27 Circularly polarized 915-MHz waveguide facility 3-91
3-28 Photograph of exposure chamber with associated instrumentation
or the 970-MHz circularly polarized waveguide facility at EPA . 3-92
3-29 VHF resonant cavity facility 3-95
3-30 Multimodal cavity facility for primate irradiation 3-96
3-31 Water-supply system for exposure chamber 3-102
3-32 Samples of commercially available survey meters for
measuring RF electric-field strength 3-106
3-33 Microprocessor-controlled twin-well calorimeter 3-118
4-1 Simple neural model of thermoregulation in a mammal 4-4
4-2 Simple model of temperature gradients in a homeotherm 4-6
4-3 Power densities at 2450 MHz necessary to raise rectal
temperature 1 °C in 60 min 4-8
4-4 SAR as a function of body weight 4-10
xii
-------
Number Page
4-5 Idealistic response of a homeotherm's metabolism and body
temperature to changes in ambient temperature 4-13
4-6 SAR at 2450 MHz, ambient temperature effects, and EHL
in mice 4-16
4-7 Effect of an increasing THI on the lethal dose of RF
radiation in mice 4-18
4-8 Effects of increasing skin temperature on sweating in
humans 4-21
4-9 Effect of exercise on heat loss, metabolic rate, and
tympanic temperature of humans 4-22
4-10 Effects of 5-HT injections on mice 4-25
4-11 Block diagram for one segment of the thermal model 4-40
4-12 Incident power density vs. exposure duration to obtain a
hot spot 4-45
5-1 Arrhenius plot of Na+ efflux 5-18
5-2 Summed incidence of abnormal and nonviable chick
embryo eggs exposed to radiation 5-78
5-3 Cross-sectional sketch of the human and the rabbit eye 5-174
5-4 Time and power density threshold for cataractogenesis in
rabbits 5-177
2
5-5 Distribution of energy absorption rate per mW/cm
incident power density in the rabbit's eye and head
exposed to 2450-MHz radiation 5-178
2
5-6 Distribution of energy absorption rate per mW/cm
incident power density in the rabbit's eye and head
exposed to 918-MHz radiation 5-179
6-1 Steady-state temperature vs. SAR and power density 6-6
6-2 Relationship of body mass to the SAR necessary to alter
the activity of various physiologic systems 6-17
xi i i
-------
TABLES
Number Page
3-1 Radiofrequency Bands 3-4
3-2 Proposed System of RF Dosimetric Quantities, Definitions,
and Units 3-32
3-3 Range of Resonant Frequencies of Man and Animals Irradiated
by Plane Waves in Free Space at 1 mW/cm2 With Long Axis
Parallel to Electric Field 3-37
3-4 Dielectric Permittivities for Various Tissues 3-54
3-5 Energy Units for RF Radiation 3-58
4-1 Partitioning of Heat Loss in Humans as a Function of
Ambient Temperature 4-20
4-2 Summary of Studies Concerning RF-Radiation Effects on
Thermoregulation 4-30
4-3 Summary of Studies Concerning RF-Radiation
Effects on Blood Flow 4-32
4-4 Steady-State Temperatures in a Human Body after Exposure to
80- and 200-MHz RF Fields 4-43
5-1 Classification of Cellular and Subcellular Experiments 5-2
5-2 Summary of Studies Concerning RF-Radiation Effects on
Molecular Systems 5-5
5-3 Summary of Studies Concerning RF-Radiation Effects on
Subcellular Systems 5-9
5-4 Summary of Studies Concerning RF-Radiation Effects on
Single Cells 5-14
5-5 Summary of Studies Concerning Hematologic Effects of
RF-Radiation Exposure 5-34
5-6 Summary of Studies Concerning Immunological Effects (In Vivo)
of RF-Radiation Exposure 5-50
xi v
-------
Number Page
5-7 Summary of Studies Concerning Immunologic Effects (In Vitro)
of RF-Radiation Exposure 5-66
5-8 Conversion of J/g to W/kg 5-82
5-9 Summary of Studies Concerning Teratologic Effects of
RF-Radiation Exposure 5-100
5-10 Summary of Studies Concerning Reproductive Effects of
RF-Radiation Exposure 5-103
5-11 Summary of Studies Concerning Reproductive Effects of
RF-Radiation Exposure in the Rat 5-112
5-12 Summary of Studies Concerning RF-Radiation Effects on the
Nervous System 5-133
5-13 Summary of Studies Concerning RF-Radiation Effects on
Behavior 5-164
5-14 Summary of Studies Concerning Ocular Effects of Near-Field
Exposures 5-172
5-15 Summary of Studies Concerning Ocular Effects of Far-Field
Exposures 5-173
5-16 Summary of Studies Concerning Auditory Effects of RF Radiation
in Humans 5-188
5-17 Summary of Studies Concerning Threshold Values for Auditory-
Evoked Potentials in Laboratory Animals 5-201
5-18 Summary of Studies Concerning Human Cutaneous Perception of
RF Radiation 5-204
5-19 Summary of Studies Concerning RF-Radiation Effects on
Clinical Chemistry and Metabolism, Endocrinology, and
Growth and Development 5-210
5-20 Summary of Studies Concerning RF-Radiation Effects on
Various Aspects of Cardiac Physiology 5-234
5-21 Summary of Studies Concerning Genetic and Mutagenic
Effects of RF-Radiation Exposure 5-263
5-22 Summary of Studies Concerning RF-Radiation Exposure
Effects on Life Span/Carcinogenesis 5-268
5-23 Summary of Prausnitz and Susskind Data 5-271
xv
-------
Number Page
5-24 Distribution of Years of Exposure for 226 Radar Workers 5-287
5-25 Age Distribution of 226 Microwave Workers and 88 Controls 5-287
5-26 Microwave Exposure Levels at the U.S. Embassy in Moscow 5-292
5-27 Number of Deaths from Disease and Mortality Ratios by
Hazard Number 5-296
5-28 Classification by Military Occupation of World War II and
Korean War Veterans With and Without Cataracts 5-299
5-29 Estimated Relative Risk of Cataracts Among Army and Air
Force Veterans 5-300
5-30 Paternal Radar Exposure Before Conception of Index Child 5-305
5-31 Summary of Selected Human Studies Concerning Effects of
RF-Radiaton Exposure 5-311
6-1 Studies Reporting "No Effects" at SAR's < 10 W/kg Grouped
by Biological Variable 6-10
6-2 Studies with Reported "Effects" at SAR's < 10 W/kg Grouped
by Biological Variables 6-11
6-3 Equivalent Power Density for Five Ages and Mass Sizes at
Resonant Frequency for SAR = 0.4 W/kg 6-14
6-4 The Increased RMR and Equivalent SAR and Power Density
Predicted to be Associated with the Onset of Human
Thermoregulatory Response 6-18
6-5 Summary of Estimates of Unadjusted Limits for RF-Radiation
Exposure at Resonant Frequencies 6-21
xv i
-------
ACKNOWLEDGMENT
The authors wish to thank Jan Parsons and Carole Moussali, Northrop
Services, Inc., for editorial review of this document, and Linda Jones and
Connie Van Oosten, Northrop Services, Inc., for the word processing.
Wanda Jones, Bonnie Waddell, and Barbara Queen are also to be commended for
their word processing assistance on earlier versions of this document.
xv i i
-------
SECTION 1
INTRODUCTION
1.1 GROUND RULES AND GENERAL ASSUMPTIONS
Daniel F. Cahill
The goal and purpose of this document is to review and evaluate the avail-
able scientific information on the biological effects of radiofrequency (RF)
radiation. To address this broad topic most effectively and to present the
information in the most useful form, several ground rules and simplifying
assumptions were adopted. It is in the selection and employment of these tools
that differences of opinion are most likely to arise. These are as follows:
1. The frequency range of interest is defined as 0.5 MHz to 100 GHz.
This range includes nearly all of the frequencies that serve as
significant sources of population exposure (e.g., AM, FM, land
mobile, and amateur radio; UHF and VHF TV; and air traffic
control radar).
2. The general public is the population of concern. Therefore,
far-field exposures will be the more usual condition, as compared
with the near-field exposures commonly associated with the work-
place. Wherever possible, the impact on all ages is considered,
rather than just healthy adults, who are of principal concern in
occupational exposures.
3. The concept of specific absorption rate (SAR) is used to
normalize the rate of RF energy input into biological systems
across this frequency range. The SAR is the mass-normalized
rate at which the energy of an electromagnetic (EM) field
is coupled into an absorbing body, and is reported as
watts per kilogram (W/kg). Inherent in the use of a single,
1-1
-------
whole-body-averaged SAR value is a disregard of nonuniform RF
energy deposition patterns and values in excess of the average.
However, the data that describe the distribution patterns of RF
energy in biological systems over a frequency band as broad as
0.5 MHz to 100 GHz are far too incomplete at this time to be
directly useful.
4. In reports of pulsed RF-radiation exposure experiments the
time-averaged SAR is considered, although controversy exists as
to whether some biological effects are functions of the temporal
pattern of energy delivery or only the result of total energy
input.
5. We have accepted "no effects" data in credible reports as
highly valuable in our assessment of the state of knowledge in
this area. These studies are considered as well as studies
providing good evidence for "effects".
6. We are assuming that the human is of equal sensitivity to the
experimental mammalian systems employed.
7. This document presents only the biological effects of RF radia-
tion. The benefitis of RF radiation are not considered, and
therefore no benefit/risk analysis is undertaken.
1.2 GENERAL APPROACH
Although a comprehensive literature review is useful, it is even more
desirable if the body of literature is consolidated, analyzed, and synthe-
sized into a statement or statements that contribute to the definition of
risk from RF-radiation exposure to the general population. To this end, our
approach is essentially as follows:
1. The reports are evaluated for their scientific quality and
utility. Acceptable reports contain adequate descriptions of
appropriate physical and biological systems and tests.
2. The credible reports are then examined for the relation between
the RF energy absorbed and the presence or absence of biological
effects in the experimental systems.
1-2
-------
3. These relations are reviewed to establish SAR levels that can
be converted to equivalent power densities for humans.
1.3 SPECIFIC APPROACH USED IN THIS REVIEW
The literature evaluated for this document includes English-language
publications, numerous English translations of Soviet research reports
obtained through the U.S. Joint Publications Research Service, and selected
technical reports translated from Polish, French, Italian, and Russian.
While the extant literature base numbers over 5000 citations,
considerably fewer are valuable in developing exposure guidelines for the
following reasons:
1. A significant fraction of the literature is available only in
Slavic languages.
2. Biological research on RF radiation is in its formative stages
compared with that on ionizing radiation and other toxicological
agents, so the research reports are uneven in quality and useful
ness.
3. In many review articles, equal currency is given to the con-
clusions from properly designed and executed studies and to
those from less stringently conducted research. Because these
uncritical reviews have contributed to the general confusion
over the health risks associated with RF radiation, we have
chosen to be highly selective in our review.
In reviewing the literature we first evaluate descriptions of frequency,
exposure parameters, RF source, experimental species, age, sex, environmental
and biological controls, the statistics employed, and whether actual data are
displayed or merely referred to in the text. Many pre-1970 reports are defi-
cient in one or more of these key areas and are either rejected outright if
1-3
-------
the flaws are judged fatal, or are segregated into the category of reports
with "unresolved questions". Reports that provide adequate descriptions of
these parameters are further scrutinized for the appropriateness of biological
systems, tests, sample sizes, controls, and statistics employed, as well as
substantiation of their conclusions. Those that clear this second hurdle are
credible reports, but are considered usable only if biological results are
linked explicitly or implicitly to SAR data from the description of the expo-
sure parameters. Reports that fail to provide these parameters are also
assigned to the "unresolved questions" category.
The reports selected for this review describe a large variety of bio-
logical effects in laboratory animals and humans exposed to RF radiation, but
no assessment is made of the possible human health significance of the effects
reported in animals. The reports are assessed by association with 1) induced
increases of core temperature, 2) whole-body-averaged SAR, 3) SAR as a per-
centage increase over resting metabolic rate, and 4) the documented human
experiences. Possible SAR values are determined for each of the four asso-
ciations. A comparison of the SAR's is made after converting these values to
equivalent power densities for humans of different ages.
1-4
-------
SECTION 2
SUMMARY AND CONCLUSIONS
Daniel F. Cahill
Joe A. Elder
1. This document presents a critical and comprehensive review of the
available literature on the biological effects of RF radiation through 1980.
The objective is to determine whether the existing data base can define the
relation between health risk and exposure to RF radiation.
2. RF radiation is a form of nonionizing electromagnetic radiation of
very low photon energies and frequencies (0-3000 GHz), as distinguished from
the very high photon energies and frequencies associated with ionizing electro-
magnetic radiation, e.g., x- and gamma rays. Included in the RF-radiation
spectrum are AM and FM radio, UHF and VHF TV, radar, and microwave communica-
tion frequencies. The frequency range of concern in this document is 0.5 MHz
to 100 GHz, which includes all the significant sources of population exposure
except 60 Hz electrical power systems. However, there is little information
on responses to exposure at frequencies above 10 GHz and below 10 MHz; most of
the research is concentrated in the range of 900 MHz to 3 GHz.
3. RF-energy absorption by biological systems is a complex function of
frequency and the dimensions, orientation, and dielectric properties of the
absorber. Resonant frequencies (and their related wavelengths) for an absorber
2-1
-------
are those at which maximum RF energy is coupled into the absorbing system.
Resonance occurs when the long dimension of the absorber is approximately
0.4 times the wavelength of the incident RF radiation. The whole-body resonant
frequency range for humans (from adults to infants) is approximately 30 to
300 MHz. The concept of whole-body average specific absorption rate (SAR) is
used in this document to normalize the rate of energy absorption across the
frequency range, 0.5 MHz to 100 GHz. SAR is the mass-normalized rate at which
the energy of an electromagnetic field is coupled into an absorbing body in
watts per kilogram (W/kg).
4. High level RF radiation is a source of thermal energy (e.g., microwave
ovens) that carries all of the known implications of heating for biological
systems. SAR's > 3 W/kg are usually associated with increases of rectal
temperature of approximately 1 °C in experimental animals. Lower SAR's can
activate the heat-dissipating thermoregulatory mechanisms of the body to
maintain core temperature within the normal range. The efficacy of RF radiation
to activate these mechanisms increases with increasing ambient temperature and
humidity, whereas the capacity to withstand high levels of RF radiation is
reduced with increasing ambient temperature and humidity.
5. No consistent biological effects have yet been found with molecular
and subcellular systems exposed iji vitro to RF radiation other than those
resulting from intensities that cause general temperature increases.
Conclusions regarding effects of iji vitro exposure of higher-ordered
biological systems, such as single cells and brain tissue, are given below.
2-2
-------
6. The electrophysiological properties of single cells, especially the
firing rates of neurons in isolated preparations, may be affected by RF
radiation in a manner different from generalized heating.
7. RF radiation does not appear to cause mutations or genetic changes in
bacterial test systems unless temperatures well above the normal physiological
range are produced. Similarly, no changes in chromosomes, DNA, or repro-
ductive potential of animals have been reported in the absence of substantial
rises of temperature. -
8. There is a lack of convincing evidence for RF-radiation effects on
the hematologic and immunologic systems without some form of thermal involve-
ment. The reported effects of RF radiation on the hematologic and immune
systems are similar to those resulting from a stress response involving the
hypothalamic-hypophyseal-adrenal axis or following administration of gluco-
corticoids.
9. RF radiation is teratogenic at high SAR's (> 15 W/kg) that approach
lethal levels for the pregnant animal. High maternal body temperatures are
known to be associated with birth defects. There appears to be a threshold
for the induction of experimental birth defects when a maternal rectal temper-
ature of 41 to 42 °C is reached. Any agent capable of producing elevated
internal temperatures in this range, including RF radiation, is a potential
teratogen.
10. Studies involving prenatal exposure of laboratory animals to RF
radiation, while few, have not shown effects on post-natal growth and development.
2-3
-------
11. Permanent changes in reproductive efficiency have been directly asso-
ciated with RF-radiation exposures that caused temperatures in animal testes
greater than 45 °C. At temperatures of 37 to 42 °C mature sperm may be killed
with a temporary loss of spermatogenic epithelium.
12. Neurons in the central nervous system (CNS) of experimental animals
can be altered by acute, high-level and by chronic, low-level exposures
(> 0.4 W/kg). The changes, which include swollen axons and decreased dendritic
spines, are qualitatively similar to each other but quantitatively different
and are almost always reversible. In general, pulsed RF fields produce greater
changes than continuous wave sources. In addition, pulsed RF radiation may
have a potentiating effect on drugs that affect nervous-system function. At
present, no definitive conclusions are possible concerning RF radiation effects
on the blood-brain barrier (BBB) at lower power densities (<10 mW/cm2).
13. An increased mobilization of calcium ions occurs in brain tissue exposed
in vitro to RF radiation, amplitude modulated at frequencies recorded in the
electroencephalogram (EEG) of awake animals. The response to various intensi-
o
ties around 0.75 mW/cm is unusual, and appears to be based on the intensity
of the electric field within the tissue, that is, the SAR.
14. Some types of animal behavior are disrupted at SAR's that are approxi-
mately 25 to 50 percent of the resting metabolic rates of many species. The
reported behavioral alterations appear to be reversible with time.
2-4
-------
15. High intensity RF radiation applied to the eyes of experimental
animals is cataractogenic; for single acute exposures, the threshold for
2
cataract production exceeds 100 mW/cm . No cataracts have been reported in
animals after whole-body, far-field RF-radiation exposures, even at near-lethal
levels. The cataractogenic potential of microwaves varies with frequency; the
most effective frequencies appear to be in the 1 to 10 GHz range. Similar
ocular effects are produced by both CW and pulsed RF radiation of the'same
average intensity.
16. Pulsed RF radiation in the range 216 to 6500 MHz hais been heard by
some human beings. The sound associated with the "RF hearing" varies with
pulse width and pulse-repetition rate and is described as a click, buzz, or
chirp. The threshold for human perception of this effect is approximately
2
40 pJ/cm (incident energy density per pulse). The most generally accepted
mechanism for the RF-auditory sensation is that the incident pulse induces a
minuscule but rapid thermoelastic expansion within the skull, resulting in an
acoustic pressure wave that is conducted by the bone to the cochlear region of
the ear.
17. Cutaneous perception of heat and pain is an unreliable sensory
mechanism for protection against potentially harmful RF-radiation exposure
level's. Many frequencies deposit most of their energy at depths below the
cutaneous thermal receptors.
2-5
-------
18. Changes reported in endocrine gland function and blood chemistry are
similar to those observed during heat stress and are generally associated with
SAR's > 1 W/kg. Exposures of sufficient intensity to produce whole body
heating will produce an increase in heart rate similar to that caused by
heating from other sources. Changes in whole body metabolism have been
reported following exposures at thermal levels (~ 10 W/kg), while brain energy
metabolism is altered at levels as low as 0.1 W/kg following irradiation of
the exposed surface of the brain of anesthesized animals.
19. RF-radiation exposures below acute lethal levels do not appear to
have any carcinogenic or life shortening potential in experimental animals.
20. Human data are currently limited and incomplete but do not indicate
any obvious relationship between prolonged low level RF-radiation exposure and
increased mortality or morbidity, including cancer incidence.
21. The development of population exposure guidelines necessitates that
the consequences of the interaction between RF radiation and biological
systems be quantified in some manner. Four possible approaches are considered
here, namely, 1) whole-body-average SAR, 2) the exogenous energy burden as a
percentage of resting metabolic rate (RMR), 3) RF-energy-induced core tempe-
rature increases, and 4) the actual human experiences documented in epidem-
iological studies. No one relationship is entirely adequate for estimating
population exposure guidelines. However, the available data base viewed from
each of these four perspectives can contribute some insights about population
exposure guidelines, as detailed below:
2-6
-------
a. Approximately 75 percent of the studies that met the criteria
for acceptability in this document employed SAR.'s. < 10 W/kg.
Virtually all of the "no effects" data are associated with
SAR's < 6 W/kg. Thus, "effects" are highly likely at > 6 W/kg.
Below 6 W/kg nearly 90 percent of the "effects" reported in-
volve behavior, CNS structure, hematologic/immunologic, or
hormonal end points; of these, behavioral and hematologic/
immunologic changes predominate. Despi.te the recognizably
incomplete state of knowledge in this area, the currently
available information suggests that the biologically effective
SAR for mammalian systems lies in the range 0.4 to 6.0 W/kg.
b. An inverse linear correlation appears to exist between the log
of body mass for several mammalian species and the log of the
SAR required to activate a physiological, thermoregulatory
response. For humans this "activation" point would be approxi-
mately 25 percent of the RMR for all ages, or approximately
0.27 to 0.81 W/kg.
c. In humans, a core temperature rise of 0.5 °C is within the
normal range of daily fluctuations and less than the 1 °C rise
permitted for occupational heat stress. RF-radiation-induced,
transient core temperature increases of < 0.5 °C could be
roximations of population exposure guide-
lines. No experimental data are available to provide the SAR
necessary to induce this temperature rise in humans; however,
mathematical models predict that an SAR of approximately
1.4 W/kg would be required.
d. As stated above, human data are limited and incomplete but do
not indicate any obvious relationship between low level
RF-radiation exposure and increased mortality or morbidity.
22. While existing data lead to tentative conclusions about the relation
between RF-radiation exposure and biological effects, the unknowns are poten-
tially significant in attempting to establish exposure guidance. These
unknowns include a lack of knowledge of 1) effects of most frequencies in the
range 0.5 MHz to 100 GHz; 2) effects of chronic, low-level exposures on human
beings and laboratory animals; 3) the most sensitive segment(s) of the popula-
tion; 4) the influence of ambient environmental conditions and of potential
synergistic interaction with other agents; 5) the implication of nonhomo-
geneous
2-7
-------
RF-energy deposition; 6) the existence and significance of frequency-specific
effects and power density windows; and 7) the physical mechanisms of inter-
action at low exposure levels. Given our present, limited knowledge of the
health effects of RF-radiation exposure, the prospects for revision and
refinement of the summary statements and conclusions stated above are
considerable.
2-8
-------
SECTION 3
PHYSICAL PRINCIPLES OF ELECTROMAGNETIC FIELD INTERACTIONS
3.1 ELECTROMAGNETIC FIELD THEORY
William T. Joines
3.1.1 Electromagnetic Spectrum
3.1.1.1 Frequency, Wavelength, and Velocity-
Oscillating electric charges induce an electromagnetic (EM) field within
the region surrounding the charge source. In turn, this oscillating induction
field—often called the near field—of the source generates an EM wave that
radiates energy from the region surrounding the charges. The radiated wave
consists of coupled electric and magnetic fields that oscillate at the same
frequency as the source, and the wave propagates outward from the source at
the velocity of light in the medium. In a vacuum or free space, this is
~ 3 x 10 m/s, whereas in a medium with low EM energy dissipation this value
is divided by the square root of the material's dielectric constant relative
to that of free space. In a low-loss material such as fatty tissue, with
O
a relative dielectric constant of 4, the velocity of light is 1.5 x 10 m/s.
For an EM wave traveling at velocity v, the wavelength in the medium is the
distance the wave travels in one time period or T = 1/f, where f is the
3-1
-------
oscillation frequency of the wave in cycles per second, or hertz (Hz).
Therefore, the wavelength A is expressed as
\ = v/f
25
The EM spectrum extends from zero to 10 Hz (or cycles/s) and includes
the ranges of visible light and ultraviolet (UV), infrared (IR), x-, and gamma
rays. The region from zero to 3000 GHz (lower IR range) is referred to as the
RF region (Sams & Co. 1981). The region of interest here is the RF band from
0.5 MHz to 100 GHz. The wavelengths in free space for this frequency band
range from 1000 m to 7.5 mm.
3.1.1.2 Ionizing and Nonionizing Electromagnetic Radiation—
EM waves of all frequencies carry energy. According to quantum mechanics,
they can also be thought of as packets of energy called photons. The energy
of a photon is given by
Energy = hf
-34
where h = 6.63 x 10 joule seconds (Planck's constant)
f = frequency
We see from the above equation that the energy of a photon is directly
proportional to the frequency of the radiation. When the frequency approaches
1 *") _ "| O
or exceeds 3 x 10 Hz (UV), the photon energies equal or exceed 2 x 10 J,
3-2
-------
or 12.4 eV, and become comparable to the binding energy of electrons to atoms.
This high-frequency radiation (x-rays, gamma rays, etc.) is referred to as
ionizing radiation. Since even the weakest chemical bonds have energies that
are several orders of magnitude greater than those of RF or microwave photons
-3
(10 eV or less), RF waves are referred to as nonionizing radiation.
3.1.1.3 Designation of Microwave and Radiofrequency Bands—
Bands of radiofrequencies have been assigned designations according to
frequency or wavelength as shown in Table 3-1 (Sams & Co. 1981). Typical uses
of the frequencies within a band are also indicated. Certain radiofrequencies--
notably the industrial, scientific, and medical (ISM) frequencies—have been
assigned by the Federal Communications Commission (Sams & Co. 1981) for specific
applications. The ISM frequencies are
13.56
MHz
+
6.78 kHz
27.12
MHz
±
160 kHz
40.68
MHz
+
20 kHz
915
MHz
±
25 MHz
2450
MHz
+
50 MHz
5800
MHz
±
75 MHz
22,125
MHz
+
125 MHz
3-3
-------
TABLE 3-1. RADIOFREQUENCY BANDS*
Frequency
Wavelength
Band Designation
Typical Uses
300-3000 GHz
1-0.1 mm
Supra EHF
(extremely high
frequency)
Not allocated
30-300 GHz
10-1 mm
Extremely high
frequency (EHF)
Satellite communications,
radar, microwave relay,
radionavigation, amateur radio
3-30 GHz
10-1 cm
Super high
frequency (SHF)
Satellite communications, radar
amateur, microwave relay,
airborne weather radar, ISM
0.3-3 GHz
100-10 cm
Ultra high
frequency (UHF)
Microwave point to point,
amateur, taxi, police, fire
radar, citizens band, radio-
navigation, UHF-TV, microwave
ovens, medical diathermy, ISM
30-300 MHz
10-1 m
Very high
frequency (VHF)
Police, fire, amateur FM,
VHF-TV, industrial RF equipment
diathermy, emergency medical
radio, air traffic control
3-30 MHz
100-10 m
High frequency
(HF)
Citizens band, amateur, medical
diathermy, Voice of America,
broadcast, International
Communications, industrial RF
equipment
0.3-3 MHz
3 2
10-10^ m
Medium frequency
(MF)
Communications, radio-
navigation, marine radiophone,
amateur, industrial RF
equipment, AM broadcast
30-300 kHz
10-1 km
Low frequency
(LF)
Radionavigation, marine
communications, long range
communications
3-30 kHz
100-10 km
Very low
frequency (VLF)
Very long range communications,
audiofrequencies, navigation
0.3-3 kHz
103-102 km
Voice frequency
(VF)
Voice, audiofrequencies
30-300 Hz
104-103 km
Extremely low
frequency (ELF)
Power lines, audiofrequencies,
submarine communications
0-30 Hz
oo-10^ km
Sub-ELF
Direct current power lines
*Sams & Co. 1981.
3-4
-------
3.1.2 Wave Propagation
Power Density, Electric Field, and Magnetic Field—
In free space, EM waves spread uniformly in all directions from a theo-
retical point source. The wavefront, or the surface joining all points of
identical phase, is spherical in this case. As the distance from the point
source increases, the area of the wavefront surface increases as a square of
the distance, thus spreading the source power over a larger area. If power
density is defined as the ratio of the total radiated power to the spherical
surface area enclosing the source, the power density W is inversely pro-
portional to the square of the distance from the source, and can be expressed
as
W = P/4nr2 (3-1)
where P = transmitted power
r = distance from the source
An isotropic source radiates uniformly in all directions in space. In
practice, there is no source having this property, although there are close
approximations. The inverse-square law (Equation 3-1) applies also when the
source is anisotropic (directional), as long as the medium is homogeneous and
isotropic (e.g., free space or a medium in which the velocity of wave propaga
tion does not change with distance).
3-5
-------
An EM wave is also characterized by its electric-field (E) intensity and
its magnetic-field (H) intensity. In a medium not bounded by conductors, the
product of E and H is the power density W, or
W = EH
(3-2)
where E is in volts per meter, H is in amperes per meter, and W is in watts
per square meter. The ratio of E to H is the intrinsic impedance of the
medium r| (in ohms), or
where r) is 120n Q in free space (or vacuum). Using Equations 3-1 through 3-3,
it is possible to express E in free space at a distance r from an iso-
tropic source radiating total power P as
The radiated waves that propagate outward from an RF or microwave source
can be confined to travel along a two-conductor transmission line (coaxial,
stripline, microstrip, twin lead), or inside a hollow metal pipe (waveguide).
The waves can also be propagated outward into the space that surrounds a
transmitting antenna, as in communications broadcasting. In this case, the
RF or microwave source forces electrons to oscillate on the surface of a
metal-transmitting antenna, thereby producing EM waves. This is a reciprocal
process, for when these waves strike a receiving antenna, they force electrons
n = E/H
(3-3)
E = V30~P/r
(3-4)
3-6
-------
to oscillate. This commutative phenomenon forms the basis of EM communication
systems. Information can be placed on an EM carrier in several ways, including
by impression of amplitude and frequency modulation.
3.1.2.1 Sources of Radiofrequency Radiation—
There is a widespread interest in possible new applications of RF energy,
especially at microwave frequencies. The growing number of commercial
applications has led to an increased awareness of potential hazards due to
the energy sources used. Typical sources of RF or microwave energy are
klystrons, magnetrons, planar triodes, backward-wave oscillators, and semi-
conductor devices. In all of these sources, energy is imparted to charged
particles or electrons from a convenient supply, usually a direct voltage or
current. A portion of the energy is then given up by the charged particles
in the form of oscillations in a tuned circuit. Such sources may operate to
produce (1) continuous (CW) radiation, as in the case of some communications
systems; (2) intermittently, as in microwave ovens, induction heating equipment,
and diathermy equipment; or (3) in the pulsed (PW) mode in radar systems.
3.1.2.2 Plane-Wave, Far-Field, and Near-Field Concepts--
Usually the region that is more than a few wavelengths from the trans-
mitting antenna is called the far field. In this region, the spatial relation-
ship between the electric and magnetic fields in the EM wave is that shown in
Figure 3-1. In this diagram, an EM wave is propagating in a direction perpen-
dicular to the motion of the electrons in the transmitting antenna. The
electric field E is always perpendicular to the direction of propagation and
3-7
-------
Y
VELOCITY, v
Z
Figure 3-1. Far-field electromagnetic wave at a particular instant in time.
lies in the plane formed by the line of propagation and the motion of electrons
in the antenna. The direction of the electric vector periodically changes
along the direction of propagation, and its magnitude forms a sine-wave
function. The magnetic field H, which is always perpendicular to both the
electric field E and the line of propagation, traces out a sine-wave function
in the same manner as the electric field E. Figure 3-1 depicts the situation
at a particular instant in time. The entire wave can be pictured as moving
O
along the line of propagation at ~ 3 x 10 m/s.
At distances less than a few wavelengths from a transmitting antenna,
in the near field, the situation is somewhat complicated because the maxima
and minima of E and H do not occur at the same points along the direction of
propagation as they do in the far-field case (Figure 3-1). Near-field ex-
posures become particularly important when one is considering radiation from
3-8
-------
microwave ovens, microwave diathermy equipment, RF sealers, broadcast antennas,
and microwave oscillators under test. For investigations into the biological
effects of RF-radiation exposure, studies of exposures in the far field are
usually preferable; field strengths in the near field are more difficult to
specify because both E and H must be measured and because the field patterns
are more complicated.
Incident power density at a point in front of an antenna may be calcu-
lated from the measured power transmitted by the antenna, the known antenna
gain G, and effective area A of the antenna. Antenna gain is defined as the
power density at a point in front of the antenna divided by the power density
at the same point if the antenna were radiating the same total power as an
isotropic source. For any impedance-matched antenna (i.e., one where all the
energy fed to it is transmitted), the ratio of G to A is
G/A = 4nA2 (3-5)
The far-field power density is calculated from the Friis free-space transmission
formula (Mumford 1961) as
W = GP/4rcr2 = AP/A2r2 (3-6)
where A. = wavelength
P = power output of the antenna
W = power density on a surface at distance r from the antenna
3-9
-------
It is convenient to express the far-field free-space power density in
terms of the power density at the antenna aperture, i.e., Wq = P/A, and
hence P = WqA. Substituting this expression for P in Equation 3-6 and dividing
both sides by Wq yields
W/WQ = (A/Ar)2 (3-7)
Equation 3-7 may be rewritten as
W/WQ = 4(A/2Ar)2 (3-8)
This expression applies in the far-field region, and the following simple
modification makes it applicable to the near-field region as well
W/WQ = 4sin2(A/2\r) (3-9)
This formula applies to uniform irradiation of square, round, or rectangular
apertures or to irradiation that is tapered in amplitude for round apertures.
For other shapes and tapers, a more complicated analysis is necessary
(Mumford 1961).
Since the power density at the antenna aperture (Wq = P/A) is greater
than the output power P divided by the cross-sectional area of the aperture,
the effective area A is less than the actual area. For the waveguide horn
and dish antennas commonly used at microwave frequencies, the effective area
ranges from about 50 to 80 percent of the actual area of the cross-section.
3-10
-------
For a given P, with A determined for the particular antenna used, Wq = P/A
is used in Equation 3-9 to compute W at any distance r from the antenna
(Mumford 1961).
Equation 3-9 is plotted in Figure 3-2 along with lines representing the
maximum values in the near-field and the far-field approximations as given by
Equation 3-7. In this graph, the relative power density is plotted in decibels
[dB = 101og1Q(W/W0)] on the ordinate, and \r/A is plotted logarithmically on
the abscissa. Note the alternate maxima and minima in the near field. The
maxima all are 6 dB above (4 times) the power density at the aperture. To be
conservative in possible near-field human exposure situations, the power
density can be approximated as the maximum value or
W = 4Wq = 4P/A (3-10)
Note, also, from Figure 3-2 and Equation 3-9 that for \r/A = 1, W = Wq, the
power density at the antenna aperture. Hence, a convenient point of demarca-
tion between the near field and far field is the distance from the antenna where
r = A/\ (3-11)
3.1.3 Wave Modulation
3.1.3.1 Amplitude and Frequency Modulation—
With an EM wave of a particular frequency serving as a carrier, infor-
mation signals at lower frequencies are translated onto the carrier through
3-11
-------
LOCUS OF MAXIMA
Hw/WnM||
NEAR FIELD
|§§ FAR FIELDS
W(r)/W0 = (A/Xr)2
0.06 0.08 0.10 0.2 0.4 0.6 0.8 1.0 2.0 4.0 6.0 8.0 10
Xr/A
Figure 3-2. Power density vs. distance along axis from antenna aperture.
the process of modulation. Modulation of the carrier can be accomplished
either by continuously varying one of the parameters of the carrier (amplitude,
frequency, or phase) or by interrupting the carrier by chopping it into segments
(PW modulation) If the impression of the information signal on the carrier
causes its amplitude to vary, the process is called amplitude modulation (AM).
If the information signal causes a frequency variation of the carrier, the
3-12
-------
process is called frequency modulation (FM). If the information signal causes
a time-phase variation of the carrier, the process is called phase modulation
(PM). Phase modulation and frequency modulation are closely related; both
cause a change in the time phase angle of the carrier (Froehlich 1969).
Since the purpose of modulation is to improve the overall efficiency of
transmission, the carrier frequency is chosen for its suitability to the
medium in which it is to be transmitted. Transmitting unmodified audio sig-
nals is theoretically possible; however, the antenna required to transmit and
receive such signals is extremely large and impractical. More importantly,
unmodulated signals are not transmitted because no other method exists to
separate several information signals on the same transmission path to provide
for multi-channel capability. This flexibility is obtained if carriers of
higher frequencies are chosen and then spaced in frequency so as to prevent
overlapping. High-frequency carriers are also more suitable for radio pro-
pagation. It, therefore, becomes necessary to shift the information signal
to a high-frequency range through the process of modulation. Introducing
frequency translation to information signals provides a way to separate
information channels and thereby improve the transmission efficiency.
3.1.3.2 Pulse Modulation-
One of the more familiar applications of PW modulation is in radar.
Radar is possible because EM waves at some frequencies are reflected by cer-
tain materials, including water vapor in clouds and metal surfaces. A radar
antenna can transmit and receive and is directional; that is, it transmits
3-13
-------
and receives signals in a limited direction. For most purposes, the radar
antenna sends out short pulses of EM waves and later receives the reflected
pulses. The radar system determines the time that elapses between the depar-
ture of the outgoing pulse and the arrival of the incoming pulse and cor-
relates this elapsed time with the antenna's direction of propagation. This
information is then used to pinpoint accurately the position of the reflecting
objects, such as airplanes or cloud formations. The magnitude of the incoming
reflected pulse is smaller than the outgoing pulse by a factor of millions and
sometimes billions. To detect an incoming pulse in the presence of atmospheric
noise, the power of the outgoing pulse is often made as large as possible.
The duty factor (DF) of a transmitter is the ratio of the pulse width of
the transmitted RF to the time between the leading edge of consecutive pulses.
Hence, average power (Pgv) is determined from peak power (Ppg^) by
P = (DF)P .
av v ' peak
In general, long-range radars have a larger transmitted pulse power and a
longer waiting time between pulses (lower DF) than do short-range radars.
For this reason, long-range and short-range radars may have exactly the
same average transmitted power.
3-14
-------
3.2 RF-FIELD INTERACTIONS WITH BIOLOGICAL SYSTEMS
Claude M. Weil
James R. Rabinowitz
Consideration of the coupling of RF radiation into different biological
objects (human, animal, bird, etc.) is important to any study of the biologi-
cal effects of RF-radiation exposure. From measurements and predictions of
RF-energy absorption that are based on experimentation with lower animals, one
can extrapolate data for estimating effects of RF radiation on humans.
Furthermore, an adequate understanding of the interaction of RF radiation with
humans has been crucial in determining exposure limits recommended in the
American National Standards Institute (ANSI) C95.1 standard (ANSI 1982). The
1982 ANSI guideline essentially is based on the concept of a defined, maximum
whole-body exogeneous heat load (see § 4). Consequently, relating this quan-
tity to a maximum permissible incident power density requires an in-depth
understanding of the RF energy-human interaction.
3.2.1 Scattering and Absorption of Electromagnetic Waves
The interaction of EM waves with any irradiated object is a complex event,
the study of which forms a large part of electromagnetism. When an EM wave
propagating in air impinges normally on any material having dielectric proper-
ties different from those of air, a reflected wave, created at the air-
dielectric interface, propagates in a direction opposite to that of the inci-
dent wave. This interaction is illustrated in Figure 3-3 for three materials
of differing electrical properties. At the top of the figure, an EM wave is
3-15
-------
Incident
~
Microwave
Power
Incident
»-
Microwave
Power
Incident
Microwave
Power
Figure 3-3. Interaction of RF radiation with electrical conductors, bio-
logical tissue, and electrical insulators (Sher 1970).
shown incident on a metallic conductor. In this case, virtually all of the
incident energy is reflected back to its source by the conductive plate. At
the bottom of the figure, the energy is shown incident on a slab of low-loss,
nonconductive material (dielectric insulator) where almost no energy is dis-
sipated within the material. For the case with nonconductive material, most
incident energy is transmitted through the slab to the other side with a small
amount reflected back at the first air-dielectric interface. Because this
material exhibits almost no dielectric losses, it absorbs very little of the
incident radiation. The third case, shown in the middle of Figure 3-3,
illustrates the interaction of an incident wave with a slab of what is termed
a complex dielectric. All biological tissues fall into this category. The
electrical properties of a complex dielectric would place it roughly between
Conductor
Total reflection
Tissue
Part reflection
Part absorption
Part transmission
Nonconductor
Large transmission
Small reflection
Very small absorption
3-16
-------
the conductor and the nonconductor i,n Figure 3-3; i.e., it exhibits values of
relative permittivity (a measure of how strongly the material reflects energy)
and conductivity (how well the material conducts electrically) that fall
between the very high values for a metallic conductor and the near-zero values
for a dielectric insulator. For this reason, part of the energy incident on
the slab of complex dielectric shown in the center of Figure 3-3 is reflected,
part is absorbed by the tissue medium, and part is transmitted. If the tissue
slab is of sufficient thickness, all incident energy will either be reflected
or absorbed, and there will be no energy transmitted through the slab. The
mechanism by which RF energy is absorbed or dissipated within the tissue slab
is discussed in more detail b.elow.
The reflected and transmitted waves that surround a dielectric object
irradiated by an incident RF wave together comprise what is termed the
scattered field. Scattering by a dielectric object plus absorption within
the object, where applicable, represent the two basic components of the
EM interaction phenomenon. The degree to which any object interacts with an
EM field when irradiated by a uniform plane wave is defined in terms of scat-
tering and absorption cross sections. These represent the cross-sectional
area in square meters of an equivalent flat plate placed normal to the
direction of propagation, which will either totally reflect (in the case of
the scattering cross section) or totally absorb (in the case of the absorption
cross section) the radiant energy incident on the plate.
3-17
-------
In general, the EM cross section is smaller than the geometric or optical
cross section that the object creates when illuminated by a distant light
source. However, under certain conditions—termed resonance—when the wave-
length of the incident radiation is comparable to the physical dimensions of
the object, the scattering and absorption effects become enhanced, such that
both respective cross sections exceed the geometric cross section. Resonance
is a significant phenomenon that will be discussed in more detail in the fol-
lowing sections; it can be explained by the apparent ability of the resonant
object to intercept or "pull in" considerably more of the energy incident upon
it relative to the optical case.
The terms "scattering coefficient" and "absorption coefficient," referring
to measures of the efficiency with which an object scatters and absorbs energy,
are defined as the ratio of the scattering-to-optical cross sections and the
absorption-to-optical cross sections, respectively. Of the two basic measures
of interaction, the absorption coefficient is considerably more important when
discussing the interaction of EM waves with biological bodies. This is
because any potentially adverse biological effects are related to the energy
that the body is absorbing, and not to what is scattered. For this reason,
most of the remaining material in this section deals only with the absorptive
aspect of the interaction phenomenon.
Some measurements have been made of the scattered fields that surround
both human subjects and life-size phantom models exposed to incident microwave
fields (Reno 1974). Although this work illustrates well the complex scattered
fields surrounding the irradiated subject (see Figure 3-4), no attempts have
3-18
-------
4 (fsnitTto*
FIELO AHPllTUDf
OBOl 01 02 02-6.3 03-95 85 07 07-1# 18-13 13-16 182.0
;
Figure 3-4. Energy distribution in proximity to man at 1 GHz at the chest plane
contour presentation: A) vertical, B) horizontal (Reno 1974).
3-19
-------
yet been made to relate the field patterns to what the subject is absorbing.
Although it is theoretically possible to determine the energy being absorbed
and the internal distribution from measurements of the distribution of ex-
ternally scattered fields, in practice this determination remains a difficult
task and has not yet been attempted. Interest in this area is growing, as
evidenced by a recent symposium (IEEE 1980) that addressed methods of obtain-
ing high-resolution imagery of the internal dielectric structure of a bio-
logical target. The methods are based on techniques of probing the scattered
fields created when the target is irradiated by an RF field. The scope for
future work in this area appears large, particularly in diagnostic applications.
3.2.1.1 Factors Affecting the Absorption of RF Energy by an Irradiated Subject—
The greater the degree of interaction of an irradiated biological subject
with an EM field, the greater is the efficiency of EM energy coupling into the
body, and the greater the overall whole-body absorption. The factors in-
fluencing the degree of absorptive coupling that takes place are listed below
and will be discussed in turn:
a. Dielectric composition of subject
b. Object size relative to wavelength of incident field
c. Shape or geometry of subject and its orientation with respect
to polarization of incident field
d. Complexity of incident radiation
a. Dielectric composition of subject—As mentioned previously, absorp-
tive coupling to the irradiated subject can take place only if the subject is
3-20
-------
composed of dissipative dielectrics having a finite conductivity value. The
dielectric properties of all biological tissues fall into this category,
primarily because their major constituent is water (average of 70 to 80 per-
cent by mass), which contains electrically polarizable or dipolar molecules
as well as free ions. The basic mechanism of dissipation of incident RF
energy within these tissues is that 1) the free rotations induced in the
dipolar water molecules by the externally impressed electric field are damped
out by collisions with surrounding molecules, and 2) conduction currents are
induced within the tissue medium because of the presence of free ions. These
phenomena are discussed extensively in § 3.2.4, Mechanisms of RF Interaction
with Biological Systems.
Over the past 40 years, many measurements have been made of the dielectric
properties of biological tissues. These data have been collected in the
Radiofrequency Radiation Dosimetry Handbook (Durney et cfL 1978, Tables 9 and 10)
as well as in a more recent paper by Stuchly and Stuchly (1980). The dielectric
properties of biological tissues depend to a considerable extent on the water
content of the tissue; i.e., tissues of high water content (blood, skin,
muscle, brain, etc.) exhibit much higher permittivity and conductivity values
than do tissues of low water content (fat, bone, etc.). As a result, most
incident RF energy will tend to pass through the fatty surface tissues of the
body and be deposited in the deeper tissues such as muscle and brain. Figure 3-5
illustrates some dielectric data for different types of tissues in the RF
range 0.01 to 10 GHz. Note that the dielectric properties of tissue are not
constant, but change with frequency. This is an example of what is termed a
dispersive dielectric. For the case of biological tissues, the dispersive
3-21
-------
RAIN
MUSCLI
AT
.BONE
10'
FREQUENCY (MHz)
10' -
MUSCLE
s
>-
I-
BRAIi
>
j—
o
D
D
Z
o
o
10" —
FAT
10'
FREQUENCY (MHz)
Figure 3-5. Dielectric data for tissues in RF range 0.01 to
10 GHz. a: permittivity, b: conductivity.
3-22
-------
characteristic above 1 GHz is created by the Debye relaxation phenomenon in
water molecules (discussed further in § 3.2.4, Mechanisms of RF Interaction
with Biological Systems). In part b of Figure 3-5, it is apparent that con-
ductivity increases by almost an order of magnitude for all tissues in the
frequency range from 1 to 10 GHz. Because of this fundamental dielectric
property of tissue, microwaves in the frequency range above about 5 GHz become
strongly attenuated in the increasingly lossy tissue medium. Consequently,
microwave energy of such frequencies is unable to penetrate deeply into body
tissues and is deposited near the body surface. As a result of their poor
penetration capability, microwaves in the centimetric wavelength region
(< 6 cm approximately) are less capable of inflicting potential damage on
internal tissue than are those of longer wavelength (> 10 cm).
b. Object size relative to wavelength of incident field—The second major
factor upon which energy absorption depends is the size of the irradiated
object relative to the wavelength of the incident radiation. Figure 3-6
illustrates that when the wavelength is much greater than the object size
(see case A at top of figure), the absorptive coupling is inefficient and
the object is almost transparent to the radiation; i.e., little energy is
deposited in the object. For this case, values of the absorption coefficient
lie in the range 0 to 0.5, and the case is referred to as the subresonant con-
dition. Case B illustrates the resonant case, which occurs when the wave-
length and object dimensions are of comparable size. For this case, the
absorptive efficiency is markedly improved so that significantly more RF
energy is coupled into the object, creating much greater deposition. Absorp-
tion coefficient values in the range 1.5 to 4 are obtained under resonant
3-23
-------
Incident wavelength > object size.
Object is essentially transparent to
radiation; little power absorption,
uniform internal distribution.
Wavelength and object comparable in magnitude.
Resonant case with significant absorption,
internal distribution very nonuniform with
"EM hot spots" present
Deep penetration
Incident wavelength < object size.
Intermediate absorption with internal
energy deposition largely confined to
object surface.
Poor penetration
Figure 3-6. Illustration of object size vs. wavelength dependence.
(Object is composed of tissue-like dispersive dielectric.)
conditions. Furthermore, the incident energy penetrates deeply into the
object and is deposited internally in a characteristically nonuniform
manner, creating localized regions of enhanced energy deposition ("EM hot
spots") at or near the object's center. These effects are discussed further
in later sections of this document. In case c shown at the bottom of
Figure 3-6, the wavelength is much shorter than the object size. This
represents the quasi-optical case, where the absorptive efficiency is similar
3-24
-------
to that occurring at optical wavelengths. The absorption coefficient
approaches ~ 0.5 under these conditions, which gives intermediate absorption
values that fall between those achieved with the conditions of cases A and B.
Case C is also characterized by the inability of the incident energy to pene-
trate much beyond the surface of the object (please refer to the dielectric
factors discussed earlier); consequently, the energy coupled into the object
under these conditions is confined to the object surface.
c. Geometry and orientation of subject—The third factor upon which
energy absorption depends is the shape of the object and its orientation with
respect to the electric-field vector of the incident field. Most biological
subjects used in studying the biological effects of RF radiation have a charac-
teristic shape that is significantly elongated along one axis (similarly to
humans). Although this observation may appear to be simplistic, it is of con-
siderable significance when discussing the interaction of RF waves with such
objects. When the electric-field vector of the incident radiation is oriented
parallel to the major or long axis of the irradiated subject (e.g., a verti-
cally polarized field that is incident on a standing human subject), then the
subject will absorb as much as ten times more energy at resonance than if the
electric-field vector is oriented parallel to either of the two minor axes.
Figure 3-7 illustrates a prolate spheroidal model of an average human. The
ordinate represents the whole-body-averaged specific absorption rate (SAR),
(see § 3.2.2, RF Dosimetry Definitions, for discussion of units), and the
abscissa represents frequency. From Figure 3-7, it is evident that when the
electric-field vector is oriented parallel to the long axis of the prolate
spheroid (electric-field polarization case), the absorption curve has a sharp
3-25
-------
HORIZONTAL
ci vnt
- 10 -
10'
FREQUENCY (MHz)
Figure 3-7. Whole-body average SAR vs. frequency for three polari-
zations in a prolate spheroidal model of a human;
incident power density = 1 mW/cm2.
peak at the resonant frequency of ~ 80 MHz. For the other orientations (bl-
and k- polarization cases), it is apparent that the absorption peak is much
broader and occurs at somewhat higher frequencies. Furthermore, the resonant
absorption values for the H and k cases are considerably lower than that for
the E polarization case. Note that these basic differences in absorption values
hold true in both the subresonant and resonant regions, but no longer hold in
the above-resonant (or quasi-optical) region, where absorption in the E- and H-
polarization cases is approximately equal.
3-26
-------
d. Complexity of incident radiation--The fourth and final factor upon
which RF-energy absorption depends involves the complexity of the incident
radiation. Up to this point, most of the discussions of RF coupling in bio-
logical tragets have assumed the simplest form of free-field exposure, which
involves a unipath plane wave emanating from a distant source and incident
on a subject suspended in space. Virtually all of the RF-radiation protection
guides and standards in use throughout the world are based on this rather
idealized assumption. Furthermore, most of the biological experimentation
involving free-field exposures in anechoic chambers attempts to simulate this
idealized concept. However, real-world, actual exposure conditions are far
more complex.
One complicating circumstance arises in the near field. It is readily
apparent that the most intense target exposure with the resulting greatest
potential for injury will occur close to the RF-radiation source. However,
in this proximate region (termed the near field), the plane-wave assumptions
that are true for the far field of the source's radiating antenna no longer
hold. As discussed in § 3.1, the near field is characterized by complex EM
field properties: the E and H fields are no longer in space quadrature
(E and H vectors separated by 90°), and the value of the E/H ratio (termed
the wave impedance) differs greatly from the constant value 377 0 which
characterizes the far field. In the near field, therefore, the power
density concept is meaningless in its usual sense. For an object placed in
the near field of an antenna the interaction is exceedingly complex because
every type of antenna possesses near-field characteristics that are unique to
that type of radiator. Consequently, every phenomenon involving near-field
3-27
-------
exposures must be individually analyzed and characterized. Although it is
possible to predict the field distribution that exists spatially in the near
field of an antenna, this distribution cannot be used as a basis for pre-
dicting the absorptive properties of an object placed in this field because
the object will substantially alter that field distribution, as well as alter
the radiating characteristics of the antenna itself. When an object is
placed in the near field of an antenna, it will interfere with the basic purpose
for which most antennas are designed, namely, the efficient transfer of RF
energy from source to far field. This is because an object in the near field
tends to load the antenna capacitively, thereby creating an impedance mismatch
condition. As a result, some of the transmitting source's energy output is
reflected by the antenna back to the source. In general, the closer the object
is to the antenna, the greater is the mismatch condition, and the poorer is the
transfer of radiated energy to the far field. Consequently, it is reasonable to
conclude that whole-body-averaged SAR values in the near field do not greatly
exceed those existing at the beginning of the far field although there may be
regions within the object of high energy deposition. The limited data available
from EM models involving near-field exposure (see § 3.2.3) seem to confirm this
conclusion.
A secondary aspect of near-field exposures involves the unwanted exposure
to leakage fields emanating from RF devices, such as microwave ovens, RF heat
sealers, and diathermy units, which are not intended to radiate energy beyond
their immediate surroundings. Such leakage fields are complex, and exposure
invariably takes place in the near zone of the leakage source. Nevertheless,
the limited data available from near-field exposures of EM models to leakage
3-28
-------
fields seem to show that whole-body-averaged SAR values are lower than might
otherwise be expected because of the leakage-source and near-field interaction
effects.
Another aspect of real-world exposures is the so-called multipath problem.
People usually stand on the ground and are not suspended in space as in the
idealized exposure concept; furthermore, they frequently stand close to build-
ings or to other reflective surfaces. This situation creates the multipath
problem, in which a subject is exposed to scattered as well as to direct energy.
Some of these effects are illustrated in Figure 3-8, which shows that there is
a considerable enhancement in energy absorption occurring for various multipath
conditions. These effects have been illustrated using the prolate spheroid
representation of a 70-kg man. Note that when the model is in contact with a
conducting ground plane, there is a doubling of the resonant whole-body energy
absorption compared to that of free-space (Case II vs. Case I). Furthermore,
if the model is standing at a distance (Case V) in front of the reflecting
plane that corresponds to an eighth of the incident wavelength (0.125 A.),
there is a tenfold increase in absorption relative to that of free space.
Even greater increases are recorded when the model is placed at a critical
point inside a 90° corner reflector (Cases IV and VI); however, these cases
are exposure situations unlikely to occur in actual practice. In addition
to these multipath effects, a further enhancement of energy absorption can
occur because of proximity effects that are created when two or more subjects
are simultaneously irradiated. Gandhi et a_L (1979) have found that when
subjects are spaced by a critical separation of 0.65 X, a 50-percent enhance-
ment in absorption can occur. Coupling of RF energy into biological systems
3-29
-------
RATE OF WHOLE-BODY ABSORPTION (WBA) IN WATTS FOR A
PROLATE SPHEROID MODEL OF MAN (weight = 70kg, height = 1.75 m) AT 10 mW/cm2
WBA
SAR
Nonresonant conditions
38 W
(no reflectance)
19 W
(50% reflectance)
0 54 W/kg
0 27 W/kg
Case I. At resonance for free space.
0
151 W
2 16 W/kg
f = 62-68 MHz
Case II. At resonance for conditions of electrical
contact with the ground plane.
0,
-77777777
2 x 151 = 302 W
4 31 W/kg
f = 31-34 MHz
Case III. At resonance for placement in front
of a flat reflector.
0
d = 0 125 A
4 7 x 151 = 710 W
10 14 W/kg
f = 62-68 MHz
WBA
SAR
Case IV At resonances for placement in a 90
corner reflector.
d = 1 5 X
27 x 151 =4077 W
58 24 W/kg
f = 62-68 MHz
Case V. At resonance in electrical contact with
ground plane, in front of a flat reflector.
a
77777-
d = 0125 X
2 x 710 = 1420 W
20 28 W/kg
f = 31-34 MHz
Case VI. At resonance in electrical contact with
ground plane in a 90°corner reflector.
d = 1 5 X
2 x 4077 = 8154 W
116 48 W/kg
f = 31-34 MHz
Figure 3-8. Absorption dependence on various ground and multipath factors
(Gandhi et aj_. 1977). Note that these data are based on ex-
perimental measurements.
3-30
-------
is discussed further in the RF Radiation Dosimetry Handbook (Durney et al.
1978, pp. 27-40).
3.2.2 RF Dosimetry Definitions
Dosimetry is the measurement or estimation of RF energy or power depo-
sition in an irradiated subject, including the internal distribution of that
deposited energy. In recent years, the terminology describing dosimetric
assessment has been evolving. Prior to the mid-seventies most investigators
used terminology that appears to have been conveniently borrowed from the
ionizing radiation field and that employed the same cumulative dosage concept
(Youmans and Ho 1975; Justesen 1975). Many investigators felt that this
terminology was inappropriate for use with nonionizing or RF radiation because
energy absorption in this case is not considered a cumulative phenomenon in a
way that has been accepted for ionizing radiation (Susskind 1975; Guy 1975).
Consequently, a new terminology was proposed that did not use the word "dose."
The now widely accepted term "specific absorption rate" was first suggested
by the late Curtis Johnson (1975) to replace "dose rate," the term used until
then for the absorption rate per unit mass. The National Council on Radiation
Protection and Measurement (NCRP 1981) has formally recommended adoption of
the term SAR in RF dosimetry NCRP has also endorsed use of the term
"specific absorption" (SA) for the absorbed energy per unit mass (referred to
earlier as "dose").
Table 3-2 lists the various RF dosimetric quantities in general use,
together with their definitions and units. Although the terms (particularly
3-31
-------
TABLE 3-2. PROPOSED SYSTEM OF RF DOSIMETRIC QUANTITIES, DEFINITIONS, AND UNITS*
Quantity and Definitions
SI Unit
(Preferred)
Name
Symbol Other Units Used
Whole-Body Absorption
(formerly integral dose)
The total EM energy absorbed
by the irradiated subject
Whole-Body Absorption Rate
(formerly integral dose rate)
The time rate of total EM energy or
total power absorption by the
irradiated subject
Specific Absorption
(formerly dose)
The mass-normalized EM energy absorbed
by the irradiated subject
Specific Absorption Rate
(formerly dose rate)
The mass-normalized rate of energy
or mass-normalized power absorbed by
the irradiated subject
Absorption Density
(also absorbed energy density
or energy density, dissipated)
The volume-normalized EM energy
absorbed by the irradiated subject
Absorption Rate Density
(also absorbed power density,
density of absorbed power, heating
potential, etc.)
The volume-normalized rate of EM energy
or volume-normalized power absorbed by
the irradiated subject
joule
watts
joules
per
kilogram
watts
per
kilogram
joules
per
cubic
meter
watts
per
cubi c
meter
W
J/kg
W/kg
J/m-
W/nT
mi 11ijouleSo(mJ)
(1 mJ = 10"6 J)
calories
(1 cal = 4.187 J)
milliwatts ^(mW)
(1 mW = lO"^ W)
cal/mi n
mJ/g
cal/g
mW/g
cal/g-mi n
mJ/cm^
mW/cm
(1 mW/cm~ = -
10J W/m )
^Throughout this table, "absorption" refers to RF-energy absorption.
3-32
-------
SA and SAR) for some of these quantities have now gained widespread acceptance,
other terms, particularly those dealing with the volume-normalized power or
energy absorption, have only recently been proposed and have not yet gained
wide acceptance.
The mass-normalized and volume-normalized absorption rates are directly
related through the localized or whole-body-averaged tissue mass density p,
that is, absorption-rate density, ARD = p(SAR), where p is in kilograms per
cubic meter. Most biological tissues are composed largely of water so that
it is reasonable to assume a unity tissue density value. In this case, the
two quantities are numerically equivalent. This approach has been adopted
by many of those engaged in predictive EM modeling (see § 3.2.3, Analytical
and Numerical RF-Electromagnetic Interaction Models) where the model is com-
posed of some homogeneous tissue having average physical and dielectric
characteristics that are representative of the various tissues of the body
being modeled. For example, all of the SAR data given in the familiar
RF Dosimetry Handbook were derived this way; the assumption of a unity average
tissue density is specifically stated (Durney et aK 1978).
It should be noted that, for both the mass-normalized and volume-normalized
quantities listed in Table 3-2, there exists a subset of two additional
values. These are a) the whole-body-averaged value, which represents the over-
all absorption or absorption rate divided by the total mass or volume of the
subject, and b) the localized value, which describes the absorption or
absorption rate in an incrementally small mass or volume at some given point
3-33
-------
within the subject. The localized ARD is mathematically defined by the
2 • •
expression a E , where a is the local tissue conductivity, and E represents
the root mean square value of local internal E-field strength.
So far, there has been no discussion of radiant exposure rate assessment,
i.e., measurements of the incident field strength or power density. Regrettably,
many workers continue to include such assessment under the general heading of
"dosimetry." Such inclusion is technically incorrect, because the exposure
rate is not an assessment of dose even though the two may be related. The
correct terminology for incident power-density assessment is "densitometry,"
which should not be confused with dosimetry. Experimental methods used in
densitometry are discussed in § 3.3.3, Densitometric Instrumentation. The
various experimental techniques which have been developed for whole-body as
well as localized or regional dosimetry are discussed in detail in § 3.4,
Experimental Methods.
3.2.3 Analytical and Numerical RF~E1ectromagnetic Interaction Models
Much of the discussion concerning the qualitative nature of EM inter-
action with biological targets contained in § 3.2.1 is based on data derived
from mathematical EM models in conjunction with limited experimental confir-
mation. Predictive physical modeling, used extensively in this research area
by many workers during the past 25 years, has contributed significantly toward
an improved understanding of the nature of EM field interactions with bio-
logical subjects and their absorption characteristics. In these models an
3-34
-------
object of simplified geometry, composed of dissipative dielectric materials
that closely simulate the spatially averaged electrical properties of bio-
logical tissues is irradiated by some form of noncomplex incident field such
as a unipath plane wave. Solutions to the problem setup in the model can then
be derived using several well established analytical and numerical techniques.
Actual computations are performed using high-speed digital computers, and the
outputs represent whole-body-averaged and localized ARD, absorption cross
sections, and absorption coefficients.
Besides the improved understanding of the EM interaction process already
mentioned, modeling has proved to be important for two more basic reasons.
First, mathematical modeling and experimental measurements on scaled-down doll
and figurine phantoms have so far provided us with the only practical method
of gaining qualitative and quantitative insight into the absorption character-
istics of the human body under various conditions of irradiation. Any attempt
to make absorption measurements on actual human subjects has so far proven
impractical because of technical difficulties, high costs, and ethical con-
siderations. Reasonably accurate quantitative data on human-absorption char-
acteristics are necessary to determine the objective exposure limits for RF
radiation. The reasons are discussed below.
The 1974 ANSI-recommended protection guide (ANSI 1974) is based on an
allowable exogenous heat load due to RF-energy absorption and conversion that
cannot exceed the basic metabolic rate (BMR) in an adult man (~ 70 to 100 W).
For a 50th percentile standard-size man weighing 70 kg, this corresponds to a
3-35
-------
whole-body-averaged SAR of ~ 1 to 1.4 W/kg. To determine the power density
level of incident radiation that will create this exogenous heat load in man,
a detailed knowledge of human RF-absorption characteristics is required. At
the time that the first RF-radiation protection guide was proposed in the
early fifties, virtually no data existed on the RF-absorption characteristics
of humans, and, consequently, the assumption had to be made that the EM absorp-
tion cross section for humans was equivalent to man's geometric cross section,
2
which is ~ 1 m (i.e., a unity absorption coefficient was assumed). From this
2
came the well known exposure limit of 10 mW/cm . During the past 10 years or
so, a considerable accumulation of data based largely on human interaction
modeling has been developed (discussed later in this section). These data
showed that under resonant conditions, the RF-absorption cross section of
humans can be 4 to 8 times greater than the geometric cross section. Thus, at
resonance, the RF-absorption rate for humans can potentially exceed the BMR by
2
this factor when exposure is to incident radiation levels of 10 mW/cm . Such
a situation clearly constitutes an excessive thermal burden for exposures of
indefinite duration. Thus, EM modeling has played a significant role in
demonstrating that, under certain conditions, the originally recommended ex-
posure limit is potentially unsafe, and in providing an objective basis for
the revision of that guideline.
The second reason for the importance of EM modeling is that it provides
us with a method of extrapolating absorption data from animal experiments for
application to humans. Using the prolate spheroid representation for humans
and animals, a considerable amount of data has been compiled on whole-body-
3-36
-------
averaged SAR vs. frequency characteristics for several different species.
These data are given in the Radiation Dosimetry Handbook (Durney et aJL 1978);
highlights of this material are presented in Table 3-3, which shows the resonant
frequency range (long axis oriented parallel to E field) and the whole-body-
2
averaged SAR at resonance, normalized to an incident field level of 1 mW/cm
for five different species, including humans. The data given in Table 3-3
enable a researcher to readily determine for a particular animal species the
resonance frequency range to be used during experimentation. For example, to
simulate human resonance in adult rats, exposure would have to be at a fre-
quency in the range of 600 to 700 MHz. This is an example of "frequency
scaling", or adjusting the frequency at which an experimental animal is exposed
in order to approximate certain given conditions of human exposure. In
TABLE 3-3 RANGE OF RESONANT FREQUENCIES OF MAN AND ANIMALS IRRADIATED BY PLANE
WAVES IN FREE SPACE AT 1 mW/cm2 WITH LONG AXIS PARALLEL TO THE
ELECTRIC FIELD*
Species
Average
Weight
(kg)
Average
Length
(m)
Resonant
Frequency
(MHz)
RMR
(W/kg)
Average
SAR
(W/kg)
Man (av)
70
1.75
70-80
1.26
0.25
Woman (av)
61
1.61
80-90
1.15
0.25
Child (10 years
old)
32.2
1.38
90-100
2.00
0.35
Rhesus monkey
(seated)
3.5
0.04
300-350
2.36
0.30
Dog (beagle)
13.5
0.06
200-250
1.72
0.16
Guinea pig
0.58
0.02
550-600
3.83
0.55
Rat (medium)
0.32
0.02
600-700
4.8
0.75
Mouse
0.02
0.007
2400-2600
6.5
0.72
*Data derived from Durney et aK 1978.
3-37
-------
addition, differences in the whole-body-averaged SAR between the two species
must be considered; from the last column of Table 3-3, it is evident that, at
resonance, the whole-body SAR for the rat is 3 times that of a human. Thus,
2
if a certain SAR value is obtained in a rat when exposed at a 10-mW/cm level,
2
roughly 3 times that level, or 30 mW/cm , is needed to obtain the same SAR
value in humans (resonant condition for both).
The data shown in Table 3-3 can likewise be used to determine whether a
given exposure frequency falls in, above, or below the resonance range for
humans. As was discussed earlier, these three categories lead to fundamental
differences in absorption as well as patterns of internal energy deposition.
There will likely be a corresponding difference in biological effects seen in
the different frequency ranges, with the resonant range representing the
potentially most hazardous condition because of the high absorption efficiency
and highly nonuniform internal deposition of energy. As already emphasized,
proper simulation of a given human exposure scenario using experimental ani-
mals requires the use of 1) frequency scaling to ensure that the type of
interaction taking place is roughly comparable (i.e., subresonant, resonant,
or supraresonant), and 2) appropriate adjustment of the incident exposure
level to obtain a specified whole-body-averaged SAR or, where possible, a
given localized SAR in some specified target organ. To check for the possi-
bility of any effects that might be specific to a particular frequency to
which humans are being exposed, animals must necessarily be exposed to that
same frequency. Given such an experiment, the use of frequency scaling is
obviously inappropriate.
3-38
-------
3.2.3.1 Model Details-
Gandhi (1980) and Durney (1980) published excellent reviews on EM model-
ing, summarizing most of the major research contributions to 1980 and detail-
ing the status of the EM modeling field at that time. Consequently, the
material that follows will not duplicate these reviews, but will instead trace
the historical development of EM models with emphasis on the increasing com-
plexity of such models and their application. Only a few relevant publi-
cations in the field will be cited; the interested reader is referred to
Gandhi's and Durney1s reviews for more complete details as well as to the
RF Radiation Dosimetry Handbook (Durney et aK 1978, pp. 64-71), which con-
tains a comprehensive literature survey of all theoretical and experimental
modeling work performed to 1977. The interested reader is also referred to
the original publications for further details of the mathematical techniques
used to solve the various problems that are discussed below.
Planar and spherical models—The earliest model considered involved the
one-dimensional solution to the planar tissue slab irradiated by a plane wave
(Tell 1972). The slab is infinitely wide and can contain several layers of
different tissues. Although the solution to this problem is straightforward,
the model is restricted in its application because it does not account for the
obvious closed-form shape of all human and infrahuman bodies. The only situ-
ation for which this particular model is valid is the case where the wave-
length is small compared to the system size, as depicted earlier in
Figure 3-6, case C; for human irradiation, such wavelengths correspond to
frequencies above approximately 3 GHz. The concept of "penetration depth"
originated from this one-dimensional model. The penetration depth is the
3-39
-------
depth within the tissue slab where the RF energy has been attenuated to a
value of 1/e, or 36.8 percent of its value at the planar surface. Strictly
speaking, this concept applies only to this particular planar model. It is
unfortunate that the concept has been extensively borrowed and frequently is
inappropriately used in discussions of RF absorption in nonplanar models where
the wavelength is either longer than or comparable to the object size. It
must be emphasized that the penetrationJdepth concept can be accurately
applied only when the wavelength is small compared to the object size.
The next level of complexity involves two-dimensional solutions of the
infinitely long cylinder with circular, elliptical, and arbitrary cross sec-
tions; some of these models are of homogeneous tissue and others are multi-
layered (Ho 1975). These models are primarily intended to simulate RF heating
in human limbs under conditions of plane-wave irradiation and using a rectan-
gular applicator in contact with the cylindrical model; the latter problem is
of particular interest to diathermy practitioners. Ruppin (1979) examined a
lossy-dielectric cylinder model placed in front of a perfectly conducting
plane (i.e., a multipath' problem). His data show both enhancement and reduc-
tion of average SAR relative to the isolated cylinder, depending on the con-
ditions of irradiation.
The first attempts at three-dimensional solutions involved homogeneous
spheres of tissue-equivalent dielectrics. Although only a poor approximation
to the actual shape and heterogeneous dielectric composition of humans and
experimental animals, the sphere model has been used extensively because of
the ready availability of solutions based on the classical Mei theory
3-40
-------
involving spherical wave functions (Stratton 1941). These methods can be
readily programmed on high-speed machines that yield data on the EM absorption
cross section and on the distribution of localized ARD for homogeneous spheres
(Kritikos and Schwan 1975). (Some approximations made by Kritikos and Schwan
[1975] can be used in the subresonance or Raleigh region and the supraresonance
or quasi-optical region, greatly simplifying these computations.) This work
demonstrated that a highly nonuniform absorption distribution is obtained
under resonant conditions, including the formation of relatively intense "EM
hot spots" near the sphere's center (Figure 3-9). Other workers (Weil 1975)
have used a multilayered spherical model that consists of several concentric
layers of different tissues representing an idealized model of an adult,
child, or monkey head exposed to plane-wave radiation. However, this concept
has obvious limitations because the spherical model is spatially isolated,
whereas, in reality, the head is attached by the neck to the body's trunk.
Results for the multilayered sphere were similar to those obtained for the
homogeneous sphere with the exception that, in the supraresonance region,
enhanced absorption was noted in the multilayered model. This enhancement is
caused by the presence of the surrounding layers of skin and fat, which appear
to provide an impedance transformation mechanism by which energy transfer into
the sphere is increased. A recent paper by Barber et aK (1979) has shown
that a planar model can accurately predict this enhancement effect for any
three-dimensional nonplanar shape, and that the layering enhancement factor
can be applied to any nonlayered object, such as a human model, to predict the
absorption characteristics of the layered object.
3-41
-------
Figure 3-9. ARD distribution in core of 6-cm radius multilayered
sphere at 1650 MHz (Weil 1975).
The effects of altering irradiation from plane wave to near field on the
absorption characteristics of multilayered sphere models were investigated by
Hizal and Baykal (1978). They found that EM hot spots continue to occur when
the model is excited near resonance in the near field of a loop antenna, but
not when similarly excited by a dipole antenna.
3-42
-------
Prolate spheroidal and ellipsoidal models--Sinee prolate spheroids and
ellipsoids are much better approximations to actual human and animal shapes,
using these models was a logical next step to solving this problem. While in
principle it is possible to obtain solutions to the prolate spheroid and
ellipsoid problem in spheroidal and ellipsoidal wave functions in the manner
already accomplished for the sphere, in practice this has not been possible
because of many mathematical difficulties. Consequently, workers have
resorted to using several different approximation techniques, further details
of which may be found in Durney's review (1980). Such methods have provided
accurate data on absorption in the subresonant and near-resonant region, as
well as in the supraresonant or quasi-optical region. These data have been
compiled in the RF Dosimetry Handbook (Durney et ah 1978, pp. 75-106) in
whole-body-averaged SAR vs. frequency plots for many different animal species.
However, under some conditions, there is a gap covering the resonant and
immediate post-resonant regions, where these approximations no longer hold.
Using approximate data derived from antenna theory and experimental observa-
tions, it has been possible to bridge this gap with the aid of curve-fitting
techniques. This is illustrated in Figure 3-10, which shows SAR data for a
prolate spheroidal model of man for the three basic orientations of the model's
long axis relative to the E, H, and k vectors. Note the strong orientation
effect discussed earlier. Durney et ah (3.978, 1979) have also developed a
useful empirical formulation that describes the E-field polarization curve
shown in Figure 3-10. It can be programmed on a hand calculator and yields
reasonably accurate SAR data for a prolate spheriodal model of any size of
eccentricity.
3-43
-------
10M
.H//J
'tlfa
5
cc
<
in
uilO"1-
o
<
QC
LLI
>
<
Long wavelength analysis
EBCM
Empirical relation
Estimated values
Cylindrical model
Geometrical optics technique
104
FREQUENCY (MHz)
Figure 3-10. Curve fitting of SAR data for a prolate spheroidal model of
man for the three basic orientations of the model's long axis
relative to the E, H, and k vectors of an incident plane wave
(1 mW/cm2 power density). Taken from Durney et a^. (1978).
As discussed earlier, an isolated model, regardless of shape, that is
suspended in space is an inappropriate representation of the real-life
conditions under which humans are exposed. A prolate spheroidal model in con-
tact with or in close proximity to a conductive ground plane is an obvious
improvement over the isolated model. Iskander et al_. (1979) have analyzed
3t44
-------
such a model by both antenna and circuit theory. In this model, the finite
separation between feet and ground created by the usual presence of footwear
can be accounted for by introducing a lossy capacitor in the equivalent elec-
trical circuit of the prolate spheroid. From Iskander's data (Figure 3-11),
it is apparent that the primary effect of the ground plane is to shift the SAR
curve for the isolated model to the left with only a slight evident increase
in peak SAR. This means that the resonant frequency of the prolate spheroidal
model of man is shifted from ~ 75 MHz to 45 MHz when the model is in perfect
contact with a conducting ground plane; the intermediate case shown, in which
resonance occurs at 55 MHz, represents imperfect contact between model and
ground plane due to a 3-cm lossy gap. The latter case is probably the most
realistic of the three considered. Iskander et aK also concluded that when
the model is separated from the ground plane by a gap of 7.5 cm or more, the
whole-body SAR values are essentially the same as those for the isolated
model.
It is evident from oral presentations at recent (1978-1980) scientific
meetings that much effort is being devoted to analyses of the absorption of
prolate spheroid and other models in the near field of an antenna. Iskander
et al. (1980) have studied such a problem for a prolate spheroid model of man
in the near field of a short electric dipole at 27 MHz. They found that
average SAR values in the near field tend to oscillate about the constant
value predicted for plane-wave irradiation. There are also significant alte-
rations in the internal distribution of deposited energy in the near field.
3-45
-------
10° •
>
LU
I
i-
Z
LU
Q
O
E
o
3
E
cc
O
05
CC
<
C/)
LU
CD
<
CC
LU
10"
10
-2.
10
-3.
/
s
/
~ \
/
/
f
' *
\
/
/
~
/
/
t
/
/
/
/
f
i
/
/
V
/
/
/
9
/
f
/
/
*
/
' /
J
f r
t
/
r
~
r
/
/
/
/
/
/
/
rf/
oy /
y *
' /
/
/
/
t
/ //
/
//
/ /
/
/
V . /
/
/ (3 / i
*
/ e' '
<&/
' ,
-------
As discussed previously, the data developed on prolate spheroidal and
ellipsoidal models have played an important role in extrapolating RF
absorption data from lower animals to man. Nonetheless, it should be empha-
sized that these data have limitations: the solutions are not exact but are
approximations, yielding data that are accurate enough (± 10 percent) for most
practical applications. Furthermore, with the exception of the subresonant
region (long wavelength analysis, Figure 3-10), existing solutions provide
data on only the whole-body absorption characteristics of this model. No data
are yet available on the three-dimensional ARD distribution and EM "hot spot"
formation potential in the resonant and immediate supraresonant region. Such
data would be of potential value to those engaged in experimentation with
rodents because of the close similarity between the rodent shape and the
prolate spheroid shape.
Human block models—Although prolate spheroidal and ellipsoidal models
are reasonably accurate representations of many laboratory animals, parti-
cularly rodents, this is not true of the primates, including man, because the
appendages (head, arms, and legs) are much larger and more significant. The
importance of these appendages is underscored by experimentation using saline-
filled dolls and scaled-down human phantoms (Guy £t aK 1977), which showed
relatively intense localized absorption in the ankles, legs, and neck. Since
it was impossible to confirm the localized effects with the prolate spheroidal
model, attempts were made to develop a more realistic human model in which
numerous cubical cells form a so-called block model of man (Figure 3-12). The
EM solution to this type of sophisticated problem is realized through the
solution of a large system of simultaneous equations using matrix inversion or
3-47
-------
Figure 3-12. A realistic block model of man (Hagmann et aK 1979a).
iterative techniques. Further mathematical details are given in Durney's
review (1980). Such solutions can be obtained only by using dedicated large-
scale computing facilities with expensive memory capability. Furthermore,
because of finite lijits in the average memory capacity of all but the largest
computers available, there is a limit to the number of cubical cells that can
be used to make up the model. The accuracy and resolution of the solutions
are dependent on cell size. In the work of Chen and Guru (1977), a maximum of
108 cells was used, whereas Hagmann et aj. (1979a) used 180 cells. These
limits inevitably reduce the spatial resolution with which the localized
3-48
-------
absorption distribution can be determined in the model, and they place an
upper limit on the frequency range over which solutions are valid. (Note that
this range does include the resonant and immediate supraresonant region.)
Since this technique readily allows for differences in the dielectric pro-
perties of individual cells in the models, it is possible to approximate the
inhomogeneous tissue properties of the human body, but there is a limit to how
finely such inhomogeneities can be modeled.
Figure 3-13 shows some data obtained for the human block model (Figure
3-12) when it stands on a ground plane and is irradiated by a vertically
polarized plane wave. The curve labeled "whole body" gives the whole-body-
averaged SAR throughout the model, which agrees closely with that obtained for
the equivalent prolate spheroidal model in contact with a ground plane; whole-
body resonance also occurs at ~ 45 MHz. Two significant conclusions may be
drawn from these data: (a) At whole-body resonance, the localized SAR in the
legs is about five times that of the whole-body average, indicating the poten-
tial for relatively intense absorption in the legs; and (b) part-body reso-
nances exist for the arms and for the head at frequencies of ~ 150 and 375 MHz,
respectively. Localized SAR values for the neck are seen to be relatively
high throughout the frequency range 100 to 400 MHz. The significance of the
head-resonance effect has been emphasized by Hagmann et afL (1979b), who (
modeled the head and neck with much improved resolution. In this study, the
detailed head is attached to the same torso model used earlier (Figure 3-12),
so that the head is not isolated as in previous studies involving inhomo-
geneous spherical models (Weil 1975) and block models of isolated human and
infrahuman heads (Rukspollmuang and Chen 1979). The Hagmann et al_. data show
3-49
-------
= /
O
INCIDENT POWER DENSITY = 1 MW/cm2
FREQUENCY, MHz
Figure 3-13. Absorption for man block model standing on ground plane
(Gandhi et aK 1979).
strong localized absorption at the front and back of the neck and in the
center-base region of the skull; localized SAR values are two to four times
the whole-body average.
RF absorption in block models when irradiated under near-field conditions
is being studied by some workers. Chatterjee et al. (1980) reported a study
involving the near-field interaction of the human block model (Figure 3-12)
with the measured leakage-field distribution of an industrial RF heat sealer
operating at 27 MHz. Their data show that averaged and localized SAR values
3-50
-------
are much lower for near-field exposure than is the case for plane-wave
exposure at equivalent power density levels.
3.2.3.2 Accuracy—
As discussed already, this document and the 1982 ANSI guideline (ANSI
1982) recommend frequency-dependent exposure limits that have been derived
mostly from the EM modeling data just described. The validity of this
approach, therefore, depends strongly on the accuracy of these data. One may
conclude that the accuracy of the whole-body-averaged absorption data is good
(certainly within ± 10 percent). The good agreement obtained between data for
the human prolate spheroidal and block models appears to support this conten-
tion. However, the accuracy of the localized absorption data is probably
inferior. This is particularly true of the block models where accuracy is
probably no better than ± 50 percent, owing to the limitations on numbers of
blocks available. On the other hand, localized absorption data for the spher-
ical model as well as those available for the prolate spheroid are more
accurate, since they were derived using more rigorous techniques.
3.2.3.3 Unresolved Questions—
The need for additional modeling has been discussed by Durney (1980).
Improvements and refinements in existing block models are needed, especially
in spatial resolution and accuracy; the same is also true in modeling tissue
inhomogeneities. In coming years, a new generation of "superpower" computers
with greatly enhanced computational speeds—an order of magnitude increase
over existing machines—will undoubtedly become available for scientific
studies. Such computers will almost certainly be used to refine existing EM
3-51
-------
absorption models, particularly in spatial resolution problems and in solving
more complex problems that hitherto have not been attempted because of com-
puter limitations.
As Durney (1980) emphasized, there is still a need for more solutions to
near-field interaction problems, since the near field more frequently repre-
sents the real-life situation to which people are exposed. Though models
involving near-field irradiation are considerably more complex to solve than
are those for the far field (plane wave), it should be possible to attempt the
solutions of these based on our existing knowledge of plane-wave solutions.
3.2.4 Mechanisms of RF Interaction with Biological Systems
Biophysical models that relate molecular structure to the interaction of
RF radiation with biological systems provide a base for understanding the
potential hazards of this type of RF radiation and for designing relevant
experiments. The purpose of this section is to review these interactions and,
where possible, to discuss the changes in structure and biological function
that may result.
3.2.4.1 Complex Relative Permittivity—
The molecular-level interactions may be averaged over a macroscopic bio-
logical sample and characterized by an average interaction parameter called
the complex relative permittivity (e*) of the sample. This interaction param-
eter facilitates discussion of the effects of a biological sample on thermo-
dynamic properties. It is a measure of the capacity of the charge distri-
3-52
-------
bution within the sample to adjust to a change in an applied electric field.
This measure is composed of two parts, the real component of the complex per-
mittivity (e1) and the imaginary component of the complex permittivity (e").
The real component is a measure of the energy stored in the charge
distribution of the sample by the interaction with the field. The imaginary
component is related to the energy dissipated by the field in the sample.
Initially, the dissipated energy may be restricted to just a few molecular
modes within the sample. Eventually, the dissipated energy will be therma-
lized, but temperature gradients may persist within the sample if the rate of
energy dissipation is a function of position, due to the geometry of the
sample or to field gradients or differences in e" within the system. Finally,
after the RF-radiation exposure has ceased, the dissipated energy will be
equally distributed throughout the sample and will result in either a change
in sample temperature or a transfer of thermal energy from the sample as well
as possible chemical and structural changes.
The e* of many biological tissues has been measured by several investi-
gators. Some excellent reviews are available (Schwan 1957; Johnson and Guy
1972; Durney et aK 1978; Stuchly and Stuchly 1980; and Schwan and Foster
1980). Complex relative permittivity is a function of frequency over our
region of interest.
Table 3-4 shows the values of e' and e" (abstracted from Durney et aK
1978; Schwan and Foster 1980; Stuchly and Stuchly 1980). The dielectric
properties of muscle tissue have been used to characterize the dielectric
properties of all biological tissues with a high water content (Schwan 1957;
3-53
-------
TABLE 3-4. DIELECTRIC PERMITTIVITIES FOR VARIOUS TISSUES*
Frequency
(MHz)
Tissue Dielectric
Type Value 0.1
10
50
100 io-
4 4
10 1.7 x 10H
Muscle
£ 1
3 x
104
2 x
103
220
90
75
54
40
34
e"
105
104
1.2 x 103
400
80
27
21
21
Brain
e1
240
110
80
40
35
e"
2.6
x 103
500
180
95
16.5
15
Liver
e1
104
1.5
x 103
240
90
78
46
36
e"
5 x
104
5 x
103
800
195
110
18
11
Bone
e1
8.0
7.2
5.8
4.9
e"
0.5
10.8
1.3
0.7
Fat
e'
12
10
6.0
4.0
e"
17
12
2.0
0.7
*These values are abstracted from reviews by Schwan and Foster (1980) and
Stuchly and Stuchly (1980). The original sources are quoted therein. The
values have been chosen in the middle of the range for experimental values.
Reported values may differ from one another by as much as a factor of two.
Schwan and Foster 1980). This characterization also serves as an illustration
of some of the underlying phenomena responsible for e*. Figure 3-14 shows the
e1 of muscle as a function of frequency.
3-54
-------
t—i—i—i—i—i—i—i—i—r
J I I I I I I I I L
0 123456789 10
LOG f (CPS)
Figure 3-14. The real component of the complex permittivity
of muscle (s1) as a function of frequency (f)
from the values for muscle (Schwan 1957).
It can be seen in Table 3-4 that in the region between 0.5 MHz (the lower
end of the range of interest) and ~ 50 MHz, the complex permittivity of muscle
decreases by two orders of magnitude. However, between 100 MHz and 10 GHz it
decreases by < 50 percent; and again, at frequencies above ~ 10 GHz, but still
within the range of interest, the real component of complex permittivity
decreases rapidly.
The imaginary component of the complex permittivity (e") has two com-
ponents—one due to the movement of free charges, and the other to the orien-
tation of either permanent dipoles or dipoles resulting from the interaction
of the field with polarizable microscopic structures. The former component
3-55
-------
will be proportional to the inverse of the applied RF-radiation frequency;
energy absorbed this way acts directly to heat the sample. The modes of
absorption of the latter component relate to the specific structure of bio-
logical molecules and molecular complexes, and this component has a maximum
where the real component of the complex permittivity e1 is changing most
rapidly.
Frequency regions with rapidly changing dielectric parameters indicate
that a particular component of the system, or a particular type of response
to the field, is reaching the limit of its capacity to respond (i.e., at
higher frequencies the field is changing too rapidly for a particular com-
ponent to respond). It has been suggested that the rapid change in the
region below 50 MHz results from polarization effects in which cellular
membranes are charged by the surrounding electrolytes (Schwan and Foster
1980). Another response that is limited to this low frequency range is the
rotation of large dipolar macromolecules or molecular complexes orienting with
the changing field (Takashima and Minikata 1975). The dielectric response below
50 MHz may result from more than one mechanism. The response above ~ 5 GHz
results from the rotation of water molecules in the biological systems, and it
is similar to the response seen in pure water. It has been postulated that
a small decrease in e" observed near 1 GHz results from rotations of water
molecules constrained by their interaction with large molecules and molecular
complexes (Schwan 1965; Grant et ah 1968) and, therefore, these water
molecules are not able to respond to the field at higher frequencies.
3-56
-------
The general features of e* of biological tissues can be explained in
terms of the major molecular and structural components of the tissues. However,
biological systems are inhomogeneous and inherently complex at the molecular
level, and interaction with a minor molecular component may have negligible
influence on the dielectric parameters of tissue (an average property of a
macroscopic volume) but may interfere with the capacity of that minor com-
ponent to perform its biological function.
Therefore, it is useful to discuss the specific mechanisms for the
absorption of RF radiation at the molecular level. Furthermore, since the
mechanisms are similar for the major components, this discussion will also
provide a more detailed explanation of the interactions responsible for the
complex relative permittivity e*.
3.2.4.2 Mechanisms for RF-Radiation Absorption at the Molecular Level-
Table 3-5 presents information on the energy of an RF beam relevant to
molecular level interactions. It should be noted that the binding energy of
a single electron in a molecule is typically on the order of tens of electron
volts, whereas excitations of electrons within molecules are on the order of
electron volts, and the energy of a hydrogen bond is > 0.1 eV. Accordingly,
direct linear mechanisms for the absorption of a photon or a few photons by
biological molecules from an RF field cannot involve changes in electronic or
covalent molecular structure, nor can they involve the breaking of hydrogen
bonds. This does not imply that single photons cannot cause changes in the
structure of biological molecules or.molecular complexes. They contain suf-
ficient energy to change the three-dimensional structure of biological
3-57
-------
TABLE 3-5. ENERGY UNITS FOR RF RADIATION
K
f
Energy/photon
(m)
(MHz)
(eV)
10"2
3 x 104
1.2 x 10"4
1.24 x 10_1
2.45 x 103
1 x 10"5
3 x 10_1
103
4.1 x 10"6
10
3 x 101
1.2 x 10"7
3 x 102
1
4.1 x 10"9
molecules in a manner that leaves the covalent and hydrogen-bonded structure
intact. The following paragraphs describe some of the excitations of bio-
logical molecules that can be induced by an RF field and the changes in
three-dimensional molecular structure that may result.
A change in the rotational energy of dipolar molecules or molecular seg-
ments is a ubiquitous mode for energy absorption from an RF field by molecules
in a biological system. The absorbing molecular structure increases its
angular momentum by quantized units. The size of these units and the
characteristic frequencies for absorption depend on the distribution of mass
within that structure. In general, the more massive the rotating structure,
the lower the characteristic frequencies.
The rotational mode of interaction is relevant through the entire range of
the frequencies of interest for the purpose of this document. Free rotation of
3-58
-------
proteins and other biopolymers are modes for absorption in the frequency range
from 10 kHz to 100 MHz and beyond. In addition, smaller molecules and molecular
segments absorb at higher frequencies by this mode. The absorption of energy
into free rotational modes does not lead to structural changes in the absorbing
molecule or segment because it rotates rigidly. Through collisions with other
molecules in the biological system, this initial increase in rotational kinetic
energy will be dispersed throughout the various molecular motions of the system
(i.e., the energy will be thermalized). With this mode of absorption, functional
changes, if any, will result only from these local increases in temperature.
In a biological system at normal body temperatures, an individual molecular
species will absorb over a broad frequency range by this mode because of its
concurrent interaction with other molecules.
Since water, which has a characteristic frequency for free rotation near
20 GHz, is the major component of biological systems, a change in the rotational
energy of water molecules is the principle mechanism for absorption of microwave
radiation at frequencies > 2 GHz. (The frequency range for which this inter-
action is likely to be important is broadened by the interaction of absorbing
molecules with surrounding molecules.) At the lowest frequencies of interest,
changes in molecular rotational motion may also make a significant contribution
to the overall dielectric properties through the interaction of microwaves
with large molecules.
For many biological molecules, segments as large as many amino acids or
as small as a methyl group have some rotational freedom, but are constrained
3-59
-------
both by covalent bond(s) to the main structure of the molecule and by electro-
static interactions with other nearby molecules. This rotational freedom
provides a mode for the absorption of energy from an RF field. Although this
mechanism is similar to the preceding one, it has been suggested (Illinger
1970; Rabinowitz 1973) that it can result in changing the equilibrium position
of the rotating segment relative to the main structure of the molecule and sur
rounding molecules. This structural change will most likely be reversible
but could have consequences for biomolecular function. The importance of this
mechanism for a biological effect of RF radiation has not been substantiated
experiments Jly.
More complicated intramolecular motional modes involve rotation and
vibration, bond stretching, and twisting. These modes will also absorb energy
from an RF field and will have potentially similar effects on molecular
structure and function. In general, these more complicated modes are at fre-
quencies above the range of interest, but there is some suggestion that they
are important near 10 GHz. Prohofsky et al_. (1979) have calculated the
stretching modes for an artificial double-stranded DNA polymer containing one
strand of adenine that is hydrogen bonded to a strand containing only thymine.
They suggest that absorption of energy into these modes may affect DNA helix
melting and replication. The characteristic frequencies of these calculated
modes are, however, > 40 GHz.
The amount of energy
complex motional modes is
in molecular collision at
required to excite either
small with respect to the
biological temperatures.
rotational or the more
amount of energy exchanged
This implies that, unless
3-60
-------
there are special circumstances (IIlinger 1970; Rabinowitz 1973; Prokofsky
et aK 1979; Ginzburg 1968), the molecular configuration that results from
the absorption of energy from the RF field is not unusual. However, the
relative distribution of normally occurring configurations may be affected by
the absorption of energy. The effect produced may be only a relative change
in a normally occurring process, and may be reversible on the molecular level.
Reversibility for the molecular-level absorber, however, does not imply that
the resulting—if any—macroscopic biological effect is reversible. If the
unusual distribution of states does not return to the normal distribution as
quickly as one would expect from strictly thermal considerations, then these
effects will be much more important (Fermi et a2- 1965).
Another general mechanism for the interaction of RF radiation with bio-
logical molecules results from the field-induced migration of ions associated
with biopolymers. The migration may be of positively charged ions—like protons
or other cations—from one negatively charged site to another within the same
biopolymer (Kirkwood and Schumaker 1952) or from the polarization of the
ionic cloud that surrounds the biopolymers (Schwarz 1972). The distinction
between these two mechanisms has been made because in the case of the former,
the migration of the ion(s) may directly affect the function of the biopolymer
if the migrating ion(s) or one of the sites is directly involved in molecular
function. In both cases, there will be effects on the long-range interaction
between biopolymers. The minimum field strength needed for the first case can
be roughly estimated from the size and shape of biopolymers, and even for
5
rod-like structures, a field of at least 10 V/m would be needed. For the
second case, the field strength necessary is smaller and depends on the size
3-61
-------
of the surrounding ionic cloud. The frequencies where these processes are
important depends on the size and shape of the molecule (Pollak 1965). For
sonically fragmented DNA segments that have a rod-like structure, broad
absorption regions have been found near 10 kHz and 10 MHz (Takashima 1963;
Pollak 1965).
A mechanism has been described for the direct influence of an RF-electric
field on the configuration of biopolymers (Schwarz 1967). If the interaction
of the field with one configuration of a biomolecular system is much greater
than the interaction with other configurations, then—in the presence of that
field--a shift in the relative populations of the various configurations will
result. The difference in interaction energy may be due to differences in the
dipole moments or in the polarizabilities between configurations. This mecha-
nism has been demonstrated in the model protein poly(v-benzyl-L-glutamate), or
PBLG (Schwarz and Seelig 1968). In its random-coil conformation, PBLG has
essentially no dipole moment, whereas in its helical conformation it has a
large dipole moment (~ 10 Debye). The presence of a field of sufficient
strength can induce a transition from the coil to the helical conformation. In
PBLG this mechanism is responsible for absorption of energy near 1 MHz; for the
distribution of conformations to be affected significantly, the field must be
greater than 10 V/m. (If there were systems where the change in dipole moment
I
was larger, then the field needed would be proportionally smaller.) This change
in conformation is not unusual (although the relative amounts of the two con-
formations are altered) or irreversible, but, as previously stated, reversi-
bility on the molecular level does not necessarily imply that any resulting,
macroscopic, biological effect is reversible. For instance, the conformations
3-62
-------
may have differing biological activities, and the products of these activities
may remain even after the distribution of the conformations has returned to
equi1ibri um.
3.2.4.3 Unresolved Questions--
The mechanisms for the absorption of energy from an RF field previously
discussed have been demonstrated in solutions of biological molecules, and
the response of macroscopic biological systems to these fields is consistent
with these molecular mechanisms. The question that remains is: Are any of these
mechanisms likely to impose a potential hazard to human health? Certainly, if
enough RF energy is absorbed and converted to thermal energy, corresponding
biological effects ensue. Other mechanisms have been proposed for effects at
the molecular level. They require changes in molecular structure and function
as the result of absorption. It is extremely likely that some changes in bio-
molecular structure result when RF energy is absorbed. But are these changes
functionally significant? Some mechanisms for biological effects in these
circumstances have been proposed. They are plausible but often not quantita-
tive, and have not been demonstrated in biological experiments. Further
research to quantitate and consider these mechanisms experimentally is needed.
Since exposure to RF radiation will often lead to increased temperature, care-
ful studies are needed to separate effects that are not due to heating from
other concomitant changes in temperature.
There are two other general mechanisms at the supramolecular level that
need further discussion. A theoretical mechanism has been proposed for coherent
action over long distances in biological macromolecular systems that are far
3-63
-------
from thermal equilibrium (Frohlich 1968). This mechanism has been developed to
explain the extraordinarily high catalytic power of enzymes; one of its con-
sequences is an extremely long-range interaction between macromolecules. The
requirements of this model are that the macromolecules possess a metastable
3
excited state that has a large dipole moment (> 10 Debye) and that its polar
modes couple with elastic modes. This highly dipolar metastable state will be
stabilized by the migration of ions and by the structure of the water near
the macromolecular surface. One of the implications of this model for enzyme
action is that interaction with an EM field that is capable of supplying energy
above a critical rate will result in all of the energy going into a single
homogeneous electric vibration of this system (Frohlich 1975). This means that
all the other modes of the system will be at or near thermal equilibrium, while
that single mode will be far removed. Although it is unlikely that this redis-
tribution of energy will increase the rate at which the energy is absorbed by
the system, the biological consequences of putting all the energy into a
single mode that is related to the functional properties of the system could
be extraordinary. If that single mode relates to enzyme function or recognition,
it could greatly increase enzyme function until saturation is reached. The
possibility also exists of re-emission of large amounts of energy and action
at sites remote from the initial absorber. The specific biological consequences
of this model relative to RF-field effects on biological systems cannot be well
understood until detailed characteristics of particular biological systems are
included. Some data can be explained by this model (Grundler et al^ 1977); the
resultant i_n vivo vibrational spectral properties of biological systems also
have been discussed (Illinger 1982).
3-64
-------
One of the interesting and relevant consequences of Frohlich's (1968) model
is that the particular biological result of RF-field interaction with a bio-
logical system may be frequency dependent. It has been suggested that the
range of frequencies where this process is most likely to be significant is
probably > 30 GHz (Frohlich 1968, 1975). The actual range of applicability is
uncertain, and it also may be important at lower frequencies.
Frohlich's mechanism is general. Its applicability in real or model bio-
logical systems has not been demonstrated. For potential biological effects
from RF fields, it is a possible mechanism for grouping individual photons or
phonons with energies « kT (the average thermal energy). This process results
in the application of energy in a significant amount (> kT) at a single locus.
Effects resulting from this process could not be duplicated by adding the same
amount of energy to the system by a different process.
Another set of mechanisms at the supramolecular level has been proposed
recently. These mechanisms are theoretical means for the direct interaction
of RF fields with microscopic biological processes that depend on naturally
occurring electric potentials. In general, these mechanisms depend on a non-
linear response to an applied field by the cell membrane.
Barnes and Hu (1977) have proposed that the ion gradient across a cell
membrane can be altered significantly by applying PW radiation at 10 V/m
peak field strength or more. This theoretical view derives from consideration
of the balance between field-driven and thermal currents. Changes in the
time-averaged concentration gradient occur that depend on the square of the
3-65
-------
field applied across the membrane, but not on the frequency of the applied
field. There are larger terms that are frequency dependent but when averaged
over a complete cycle are zero. This mechanism has not been demonstrated in
a biological system.
Pickard and Rosenbaum (1978) considered a similar problem and came to the
same conclusion for the time-averaged concentration gradient across the mem-
brane. In their model, they included unidirectional ion channels through the
membrane, and they postulated that the new concentration gradient is achieved
by the field-induced movement of ions through these channels. They calculated
that, for this particular method of achieving the concentration gradient, the
frequency of the field must be < 200 MHz if the ions are protons, and < 10 MHz
for less mobile ions. Pickard and Barsoum (1981) have demonstrated activity
across the membrane of a plant cell that begins at the onset of an RF field
below 10 MHz. The immediate response to the field indicates that the activity
is not thermally induced. The activity demonstrated is consistent with an
RF-induced DC potential across the membrane. This potential could result
from any nonlinear response of the membrane to the RF field, and it is not
clear that any particular mechanism for that response has been demonstrated.
More data are needed to demonstrate a realistic microscopic model for this
process. It is difficult to proceed from these mathematical models and this
single experimental demonstration to an assessment of the importance of this
mechanism. The induction of DC or extra-low-frequency potentials in biological
systems by an RF field could provide significant mechanisms for the biological
effects of RF radiation.
3-66
-------
3.3 EXPERIMENTAL METHODS
Claude M. Weil
Joseph S. Ali
3 3.1 Exposure Methods Used in Biological Experimentation
In this section the most commonly used exposure methods employed in RF
biological effects research are reviewed, and the inherent advantages and dis-
advantages of each discussed. Researchers have used a diversity of exposure
methods, the chosen method having depended on the specialized application
involved, the existing facilities available for experimentation, and the cost
of setting up new facilities. This diversity may have created difficulties
in reproducing biological effects studies, which have, in turn, fueled con-
troversies on several reported effects. Whereas an effect may have been noted
by one group who employed a particular form of exposure system, a second group
that carefully reproduced the experiment in every detail but for the exposure
method might not see the reported effects. It is possible that basic dif-
ferences in the exposure environment may, in some cases, explain this lack of
reproducibi1ity.
In much of the literature, effects have been traditionally reported as a
function of incident power density, the independent variable. As pointed out
by several prominent workers (King et aK 1970; Johnson 1975), power density
is an inadequate and inappropriate independent variable for two reasons:
1) Biological effects are logically related to localized or whole-
body RF absorption, which is in turn a nonlinear and complex
3-67
-------
function of frequency vs. body size, body shape, and orientation
to field vectors, etc. (See § 3.2, RF Field Interactions with
Biological Systems.)
2) Although meaningful for certain plane-wave-exposure situations,
such as free-field or transverse electromagnetic (TEM)-mode
cells, the power density concept is no longer applicable in
most other commonly used exposure systems with complex irradia-
tion fields. Consequently, meaningful comparisons of the inci-
dent field levels in different types of exposure systems cannot
be made on this basis.
For these reasons, it must be concluded that the various exposure methods
are unique; experimental results obtained in different exposure environments
cannot be compared except perhaps on a whole-body-averaged SAR basis, and
reproducibility of experimental data is not guaranteed unless exactly the same
exposure methodology has been used. Even if equivalency of whole-body-averaged
SAR is achieved, there may still be differences in biological outcomes for ex-
posures performed in multipath fields (e.g., a cavity) when compared to
plane-wave fields, which have the same carrier frequency and modulation para-
meters.
Although many of the specialized requirements of a particular experiment
are unique, some requirements are common to almost all experiments. For
example, there is a critical need in virtually all work for temperature and
humidity regulation of the environment in which animals, biological specimens,
etc., are located. This ensures that the ambient air temperature and humidity
are properly controlled, thereby minimizing the possibility of unwanted vari-
ability in the experimental protocol. Two other important requirements are
common to the majority of experiments involving whole-body or unrestricted
irradiation of animals: (1) an animal population large enough to allow for
3-68
-------
statistically reliable conclusions to be drawn, and (2) an equally large popu-
lation of control animals that must be housed during exposure periods under
conditions that are, in every way, virtually identical to those of irradiated
animals except for the presence of RF energy. For certain critical experi-
ments, there may also be a need for a population of passive or reference
control animals that are maintained under normal conditions in their vivarium
during the tenure of a study. Much of the early experimental work on RF bio-
logical effects frequently contained no provision for controls, involved
inadequate numbers of experimental subjects, and lacked adequate provision
for environmental control, thus yielding data of questionable validity. It is
only within the past decade or so that researchers have succeeded in correcting
most of these experimental deficiencies in exposure protocol.
Existing exposure methods have been reviewed by Weil (1977) and Ho et al^.
(1976); the latter paper discusses exposure techniques that have been used at
the FDA's Bureau of Radiological Health. The commonly used methods are
categorized into two groups and considered separately: (1) free-field, and
(2) enclosed. In the former category, the exposure fields are generated
within a reflection-free environment by a radiating antenna or applicator,
whereas in the latter category, the exposure fields are those excited within
a structure of reflective or conducting walls.
3.3.1.1 Free Field—
Free-field exposure methods were historically the first to be used in
such experimentation and remain the most commonly used techniques. These
methods involve placing the experimental subject(s) in the near-field or
3-69
-------
far-field region of a radiating antenna, such as a horn, parabolic dish, or
diathermy applicator. For exposures in the far-field region, all unwanted
scattering of RF energy from the surrounding room structure, etc., must be
eliminated. Otherwise, the power density levels at the subject's location are
not well established. This requires using microwave-absorbing material that
wholly or partially surrounds the experimental subjects, creating what is
termed an anechoic chamber. (The term "anechoic," meaning "without echoes,"
has been borrowed from acoustics research.) The design of anechoic chambers
is a specialized art normally performed by commercial manufacturers of absorb-
ing material (e.g., Emerson and Cuming, Canton, MA). The anechoic properties
of such chambers, which are defined by the characteristics of the "quiet zone"
where the experimental subjects are placed, is a function of the reflective
properties of the absorbing material with which the chamber is lined. (Such
properties refer to the degree with which RF energy is reflected by the
material, rather than absorbed.) Reflectivity is proportional to the incident
wavelength so that the performance of an anechoic chamber will markedly
deteriorate as the frequency of operation is lowered. A minimum reflectivity
of 20 dB (1 percent) is usually considered essential; the frequency at which
this value is reached then defines the lower frequency limit of operation for
the chamber. Anechoic chambers perform well at frequencies in the microwave
region of the spectrum (above 500 MHz). However, satisfactory performance
at lower frequencies or longer wavelengths requires using large blocks of
absorbing material formed into long pyramids or cones. Consequently, a
large structure is needed to house such a chamber, so that the cost of
such a facility rapidly escalates to the point of nonfeasibi1ity. For all
3-70
-------
intents and purposes, it is impractical to contemplate building anechoic
facilities that operate at frequencies below ~ 250 MHz.
Some of the most extensive anechoic chamber facilities devoted to RF
biological effects work are to be found at the Department of Microwave Research
of the Walter Reed Army Institute of Research, Silver Spring, MD. Similar con-
ventional anechoic facilities are found elsewhere; e.g., at the Bureau of
Radiological Health in Rockville, MD, at the U.S. Environmental Protection
Agency in Research Triangle Park, NC (Figures 3-15 and 3-16), and at the U.S.
Air Force School of Aerospace Medicine in San Antonio, TX.
The principal advantages—applicable to exposures in only the far-field
zone of an antenna—of the free-field exposure method are:
0 The incident power density to which the subject is exposed can
be well defined to an accuracy of ±1 dB (±20 percent) or less.
This makes replication of experiments performed under free-field
conditions fairly straightforward.
• Depending on the directionality of the radiating antenna, the
exposure area where the irradiated subjects are placed is
large, and the irradiation is nearly uniform (±10 percent
or less).
The principal disadvantages of free-field exposure methods are:
0 The facilities are costly and often require extensive floor
space.
0 Very high-power RF sources are needed to obtain the necessary
power density levels. Consequently, many free-field facilities
employed in bioeffects research operate on either of two ISM
3-71
-------
Figure 3-15. EPA 2450-MHz anechoic chamber facility, or 2.45-GHz far-field
exposure facility. Horn antenna (top), temperature control
chamber (center), and one air duct (bottom left) are visible
in the shielded anechoic room (Blackman et al. 1975, Elder and
Ali 1975).
3-72
-------
BCflOUVt POIERSfNSf
CONTROL 4 SENSE LINES
¦CROMVe
POIER
CONTROLLER
¦CROVAVC POlER CO BUM 0
CONTROL I SENSE LINES
VARIAN
¦iCflonvE
GENERATOR
mm km
ioi« ca
IICROf AVE POtER SENSE
DATA
ACQUISITION
ST3TEI
ENVUtONBENTAl.
PARAMETER
COWNOS
ENVIRONKNTAl
PARAMETER
SENSE
ENVIRONMENTAL
CONTROL
PACKAGE
I
AMIIUL _
EXPOSURE
* CHAJtSER'
/ ANtCNOlC
CHAMBER
PHYSIOLOGICAL OATA
Figure 3-16. EPA 2450-MHz anechoic chamber facility: diagram of the
microwave exposure facility (Blackman et a]L 1975; Elder
and Ali 1975).
frequency assignments used for microwave heating, 2450 MHz or
915 MHz, where high power sources are commercially available at
relatively low cost. This is why 2450 MHz has become a standard
microwave frequency for bioeffects research.
• Free-field systems are an excellent simulation of the idealized
exposure concept visualized in many protection standards, but
these systems do not represent the real-life exposure situation
encountered by humans.
• The whole-body-averaged SAR cannot be measured directly in a
free-field situation. Reasonably accurate estimates of the SAR
can be made by calorimetric measurements on animal carcasses.
(See § 3.4, Dosimetric Methods.)
Various attempts have been made to modify conventional free-field exposure
methods to ameliorate the cost and space problems associated with them. A
study (Bassett et al^. 1971) performed at the Georgia Institute of Technology
developed several such techniques. One of these employs the absorber-lined
3-73
-------
horn antenna, which enables a subject to be placed much closer to the horn
aperture than is the case with a conventional horn and yet to remain within
the far-field zone of the radiator. This means that a lower power source can
be used to create the same power density at the irradiated subject. One of
these absorber-lined horns has been used for several years at the National
Institute of Environmental Health Sciences in Research Triangle Park, NC
(Figure 3-17). Another technique involves the compact range (Figure 3-18),
which uses a feed horn to irradiate a large parabolic reflector. The test
subject is exposed in the collimated beam region of the reflector. The advan-
tage of this system is that it creates a much larger test volume of nearly
uniform field intensity with associated decrease in the anechoic facility size,
as well as reduced requirements for absorbing materials. The compact-range
technique is used at the Naval Aerospace Medical Research facility in
Pensacola, FL.
etfecjive
K . BADIATOB
APCfitURE
or Aesoapifjr
MAIfKIAl
Zl
-» 30" H
SCClION A a'
17
Figure 3-17. Diagram of absorber-1ined horn (Bassett et ah 1971).
3-74
-------
TEST
VOLUME
rt*t«
Figure 3-18. Diagram of a point-source compact range (Bassett et al^ 1971).
As discussed earlier, a critical need in whole-body exposure of animals
is to expose large numbers of animals. Consequently, in many experiments, un-
restrained animals have frequently been placed inside an anechoic chamber in
a closely spaced matrix for simultaneous exposure. This practice tends to
obviate the principal advantage of free-field exposure methods, because the
power density to which the animal is exposed can no longer be accurately
defined due to RF-energy scatter from one animal to another. The degree of
scatter is dependent on much the same factors as is absorption (see § 3.2.1)
and will vary randomly as the animals constantly alter their relative posture
and orientation. Interanimal scatter is most serious when the frequency of
exposure lies in the resonant and supraresonant region. Several solutions to
this problem have been devised:
(a) Sequential exposures of single animals in the same free-field
facility
(b) Miniature anechoic chambers for exposing animals individually
(c) Separating the restrained animals so that they are spaced one
or more wavelengths apart
3-75
-------
The first solution is obviously costly and time-consuming, and may not be
acceptable in certain critical biological experiments requiring all animals
to be exposed at exactly the same age or the same time of day, etc. The
second solution has been advanced by Guy (1979) at the University of
Washington, Seattle. He has described a system of 16 miniature anechoic
chambers designed for chronic low level (< 10 mW/cm ) exposure of rabbits and
rodents at 2450 MHz; eight of these chambers are energized, while the other
eight serve as control chambers (Figure 3-19). A similar chamber has been
constructed at EPA (Figure 3-20). Such a system has, of course, the advantage
of ensuring total RF isolation between animals, and, although costly, it may
be more economical in cost and power requirements than is the conventional ,
anechoic chamber.
The third solution has been tried by a few workers and appears to be a
reasonable compromise between the conflicting requirements of adequate isola-
tion and low cost. Oliva and Catravas (1977) have described a method for
simultaneously exposing 10 animals in a conventional anechoic chamber with
minimal field interaction between animals (Figure 3-21). The animal cages are
located on the antenna beam's three-dimensional contour of constant power
density, with cages separated by ~ 2.5 wavelengths at 2450 MHz. The authors
claim that the average variation of power density from one cage location to
another does not exceed ±5 percent. The major disadvantage of this technique
is that its implementation may require a large exposure volume, which, in turn,
creates costly problems of ensuring satisfactory environmental control.
D'Andrea et aK (1979, 1980) at the University of Utah have developed an
3-76
-------
TAPEREO RADIATION CHAMBER
FRONT VtEW E-PLANE
TAPEREO RADIATION CHAMBER
SIDE VIEW H-PLAfC
.336.
ll 43cm SQUARE
MUFFIN FAN
20cm tq HORN
MOUNT CENTER-
ING STRIP twooa
1375 em
DISTANCE FROM
HORN TO SPY-12
TIPS
FXR
S60I8
ADAPTER
032 cm
095cm HARD0OARO
plywooo
15 2 cm
644 HORN
914 cm
STrROrOAM
FLOOR
16 cm
TOP VIEW)
609
cm - * I 27cm
TYP ALUM
ANGLE
SUBJECT
CENTER
92 7 cm
31 7 cm
' SPY-12
L - 609 cm—
755cm «|
ALL ALUMNUM
STRIPS a ANGLES
ASSEMLED WITH
032cm POP RIVETS
PLYWOOO HORN MOUN
{WITH 2 AUGMENT
CENTERNG STRPS)
SITS ON CHAMBER
0 32cm(OI25n)
HARDBOARD
DOOR
755 cm X 609 cm
254 cm X 76cm X
203 cm STYROFOAM
BRACE BONDED TO
AN-77 WITH
, SILASTIC
ADHESIVE .
SEPARATION
0l6cm X
2 54 cm ALUM
STRIPS
904cm
609cm
609 cm
508cm iquve
095 PLYWOOO
40W CAMELA6RA
LAMPS
4 EACH MOUNTED 4cm
FROM EACH EDGE
>95 cm PLYWOOO
PLYWOOO PARTS
GLUED 8 NAILED
032cm (0375n)
HAR060AR0
AN-77
ABSORBER
508cm (20in)
I 27 cm (05W
ALUMINUM
ANGLE
.SPY-12 ABSORBER,
30 5~"^
SQUARE AIR
INTAKE HOLE
a) Construction details of tapered exposure
chamber
b) Construction details of tapered exposure
chamber
1/2 SCALE
-20*
2cm
GLASS
ROOS !
^-*•1032 cm
WEWEO
NYLON IINE-*
c) Construction details of cage for housing
rat in tapered exposure chamber
Figure 3-19. Miniature anechoic chamber facility (Guy 1979).
3-77
-------
Figure 3-20. Photograph of tapered exposure chamber at EPA facility.
3-78
-------
METERS L_
DISTANCE FROM ANTENNA
_ 0 1 13 14 15 16 >7
FEET | 1 i / 1 1 1 1 |—
% LOCUS OF EQUAL POWER
DENSITY (colcoloted)
I I CAGE ON TALL PEDESTAL
~ CAGE ON MEDIUM PEDESTAL
¦i CAGE ON SHORT PEDESTAL
OSr^
SUPPORTING TABLE
1—41.5
METERS
a) Locus of equal power density in the H plane through
the axis of the transmitting antenna.
u>
i
METERS L
13
-// 1-
—LOCUS OF EQUAL POWER DENSITY
~ CAGE ON ANTENNA AXIS
DISTANCE FROM ANTENNA
14 15 16 17
~
CAGE INTERMEDIATE DISTANCE
FROM ANTENNA AXIS
| CAGE FURTHEST FROM
I ANTENNA AXIS
UPPORTING TABLE
I— - 1.0
METERS
b) Locus of equal power density in the E plane through the
axis of the transmitting antenna.
c) Multiple-animal array for equal power density microwave
irradiation.
Figure 3-21. Facility for simultaneous exposure
(Oliva and Catravas 1977).
of
10 animals with minimal inter-animal interaction
-------
alternate method involving a monopole radiator on a vertical ground plane.
The animals are located next to the ground plane on a circular locus that is
centered at the monopole (Figure 3-22). Each circular array contains 10 ani-
mals that are separated by ~ 5.5 wavelengths at 2450 MHz, and by 2.1 wave-
lengths at 915 MHz. This system offers the advantage of reducing both space
and RF-power requirements. Furthermore, the presence of a ground plane makes
this system more representative of the real-life exposure situation for humans.
On the other hand, the exposure characteristics of this system may not be
comparable to a conventional anechoic chamber facility, so that caution is
needed when comparing data.
For situations involving the exposures of only one subject at greatly
increased power density values, D'Andrea et cH. (1977) and Hagmann and Gandhi
(1979) have discussed a modification of this system in which a corner reflector
is placed behind the monopole. By placing the subject 1.5 wavelengths from
the corner point, it is possible to obtain an SAR enhancement of more than 20
times relative to that obtained without the reflector.
3.3.1.2 Enclosed Systems-
Enclosed exposure systems are those in which the EM fields are contained
within a conducting structure. These fall into two basic subcategories, which
are considered separately: (a) transmission line systems, and (b) cavity
systems. The principal advantages of enclosed systems are as follows:
• By suitable monitoring of the incident, reflected, and transmitted
power in the exposure system, more accurate SAR estimates can be
made for the exposed subject than is possible in free-field
3-80
-------
GROUND
PLANE
MONOPOLE
ANTENNA
STYROFOAM
MOUNTS a
RAT CAGE
WALK-ON
ECCOSORB
PYRAMIDAL
ECCOSORB
3 ) Representation ot the microwave-exposure chamber
<4.73
Left chamber
Mean mW/cm
5.42 ± .21
Right chamber
2
Mean mW/cm
4.85 ± .27
SAR -rnW/y
1.23 + .25
(n = 3)
b)
Distribution of power densities measured at each rat's location within the Plexiglas holding cages for both
sides of the microwave exposure chamber in the absence of the rats ( 2450 MHz)
Figure 3-22. Monopole-over-ground plane irradiation facility
(D1Andrea et a2. 1979, 1980).
3-81
-------
facilities. The accuracy in estimating SAR depends in part on
how much base-line power loss there is for the unloaded system
and to what extent this is affected by the presence of the ex-
posed subject.
• The cost and space requirements of such a system are usually
lower than those for free-field methods, depending on the
number of identical systems needed. Also, much smaller and less
expensive power sources will suffice to establish the same field
intensities.
• Better control of RF absorption in the irradiated subject is
possible.
The principal disadvantages are as follows:
• With the exception of TEM-mode transmission lines (to be explained),
the fields are complex and not plane-wave equivalent, thereby
making comparison with free-field data more difficult.
• With the exception of multi-modal systems (also to be explained),
the exposure space available inside such systems is limited.
Although the available space may be adequate for small animals
such as rodents, it may not be possible to fit a rabbit or monkey
in it. Thus, the variety of species that can be employed in an
existing facility of the enclosed type is restricted.
• The field distribution within the exposure system is rarely as
uniform as that found in a free-field facility. The field dis-
tribution is well defined for single or dominant mode operation,
but undefined for multiple-mode operation.
Enclosed systems have been the preferred exposure method for samples,
specimens, jn vitro preparations, etc., since these materials are small enough
to fit easily inside such systems. Many experiments have been described,
too numerous to survey here, in which different types of waveguide, TEM-mode
lines, and cavities have been used to irradiate such preparations. In recent
3-82
-------
years, some enclosed-type systems have been specifically designed for whole-
body irradiation of animals. These are discussed here in more detail. Owing
to the need to define accurately the field level to which the animal is exposed,
dominant-mode operation of these devices has been preferred over multi-mode
operation. However, this puts an upper limit on the frequency of operation,
depending on the size of the system needed to expose an experimental animal.
For example, most of the dominant-mode, enclosed systems that have been used
to expose rats operate at frequencies below 1000 MHz; for mice, which require
a smaller exposure volume, single-mode operation to ~ 3000 MHz is possible.
Since free-field methods of exposure are largely impractical at these lower
frequencies, as discussed earlier, enclosed systems do provide experimenters
with virtually the only practical and economical method of exposing animals
*
in the HF, VHF, and lower UHF segments of the RF spectrum. A few large-scale
systems have been built to operate in the dominant mode at HF and lower VHF
frequencies; these are large enough to accommodate a group of animals within
the exposure area, and the animal-to-animal interaction problem is minimized
because exposures are conducted in the subresonant region where interanimal
scatter is minimal. However, at higher frequencies such methods become imprac-
tical, and it becomes necessary to provide an identical exposure system for
each animal to be irradiated. Each exposure device is fed from a single
high-power source by a power dividing network. One of the basic electrical
problems associated with any enclosed exposure system is that, when loaded,
there is considerable reflection of RF energy from the load, which propagates
via feed lines back to the RF power source. This characteristic creates
particular problems for multiple systems fed by a single power source because,
3-83
-------
without sufficient isolation between each exposure device, energy is continu-
ously reflected from one system into the others, causing unwanted fluctuations
in the RF energy incident to each system. These problems can be solved by
installing isolators in each feed line, thereby preventing the reflected energy
from reaching the dividing network and the RF power source. The two categories
of enclosed system are discussed separately.
Transmission lines—In these systems, an EM wave propagates down the
line from source to termination. When the line is terminated in its
characteristic impedance, only the outwardly propagating wave exists with no
reflected wave present. (The characteristic impedance is a basic electrical
parameter that is defined by the line's configuration and dimensions.) However,
if the line is terminated in a short circuit or an open circuit, all of the
energy is then reflected at the line's termination. The two systems of
traveling waves together create a standing wave that possesses successive
maxima of electric and magnetic fields (E and H, respectively) spaced a quarter
of a wavelength apart. This property is a useful feature that has been used
by some workers to expose specimens or samples to isolated E or H fields,
depending on the location along the line. The characteristics of a shorted or
open transmission line are similar to those of the resonant cavity, and are
discussed below.
Transmission-line devices can be subcategorized into TEM-mode or wave-
guide. In TEM-mode lines—coaxial air lines, parallel-plate (strip) lines,
etc.—the dominant mode of operation is a propagating TEM mode, provided the
3-84
-------
line is properly terminated in its characteristic impedance. In TEM-mode
operation the E- and H-field vectors are mutually orthogonal, and lie only in
the plane transverse to the direction of wave propagation. This is seen to
be virtually identical to a plane wave, so that it is possible to simulate
free-space conditions inside such lines, provided that the exposed subject does
not occupy more than about a third of the line's cross-sectional area. (If
the one-third restriction is exceeded, ground-plane effects are observed.)
Because of its circular cross section, the coaxial air line has been used only
with fluid or precut solid samples; the primary application has usually been
in making dielectric measurements. A sophisticated coaxial air-line system,
designed to expose cell cultures to high field strengths, has been described
by Guy (1977) (Figure 3-23). Parallei-piate systems consisting of a con-
ductor located over a ground plane or between two ground planes are better
suited to animal exposures since they provide a rectangular volume where ani-
mals can be exposed. D'Andrea and Gandhi (1976) have described a simple
parallel plate system used in the range 200 to 500 MHz (Figure 3-24). This
system is equivalent to a large scale version of a microstrip line with air
dielectric. Capacitive-plate devices, which are equivalent to a short length
of parallel plate that is terminated in an open circuit, have been used for
exposing i_n vitro preparations to isolated electric fields. This type of
simple device has been incorporated in a unique HF-band (10- to 40-MHz) ex-
posure system developed by the National Bureau of Standards (Figure 3-25).
Termed a near-field synthesizer, it is capable of separate E- and H-field ex-
citation, the latter generated by special loop inductors (Greene 1976). By
altering the spatial position of the loop with respect to the E vector and by
changing the phase relationship between the signals exciting the plates and the
3-85
-------
FIRST SURFACE
MIRROR
THERMOGRAPH
CURRENT SENSING
PROBE.
-CULTURE SAMPLE HOLDER
•VOLTAGE SENSING PROBE
PROBE
LOOP
MATCHING
SECTION
VECTOR VOLTMETER OR
NETWORK ANALYZER
CONSTANT TEMPERATURE
LIQUID SOURCE 9 PUMP
DIGITAL VOLTMETER
TEMPERATURE MONITOR
REFLECTED
POWER
FEED BACK
LOOP
DIRECTIONAL
COUPLER
POWER
METER
POWER
SOURCE
INCIDENT
POWER
a) Complete system for exposing cell cultures to EM fields.
8mm
• 334cm
ORING
7mm
j3mrrv
ZOUTLET
225cm
INLET
385 cm
384cm
246cm
E3 STAINLESS STEEL
^ BRASS
R— CROSS-LINKED STYRENE
HUH TEFLON
b) Cross-sectional view of assembled transmission-line cell-culture sample holder and
heat exchanger. (A section of the outer conductor is machined to 7.70-cm ID so a
7.71-cm diameter dielectric support can be firmly locked into place after being pressed
in while the conductor is heated.)
Figure 3-23. Coaxial air-line system for high power exposures of cell
cultures (Guy 1977).
3-86
-------
If
Load
Radiation Chambor
Figure 3-24. Parallel-plate (microstrip) exposure system (D1Andrea et aL 1976).
BALANCED
FEED
TO PARALLEL
PLATES
~~r~
MATCHING
NETWORK
REFLEC-
TOMETER
LOWER
PLATE
MATCHING
NETWORK
4 -GAP
LOOP
INDUCTOR
MATCHING
NETWORK
REFLEC-
TOMETER
RF POWER
AMPL
NO 1
DOUBLER
AMPL
NO 1
DRIVER
NO 1
DRIVER
NO 2
D0U8LER
AMPL
NO 2
RF POWER
AMPL
NO 2
COMMON _
OSCILLATOR
Figure 3-25.
Block diagram of the complete RF near-field synthesizer,
showing all principal components including RF power
sources.
3-87
-------
loop, it is possible to simulate the complex field environment existing in the
near-field of an HF source. Two of these systems are presently being used;
one is at the U.S. Air Force School of Aerospace Medicine, TX, and the other
at the National Institute of Occupational Safety and Health laboratory in
Cincinnati, OH.
Triple-plate lines, in which a conductor is mounted between two ground
planes, have been used by a few workers. These have the advantage of allowing
for simultaneous irradiation of two animals while maintaining good isolation
between them. In a more popular version of this system, the open sides of the
structure are closed off with two side plates so that the center plate is
entirely surrounded by a rectangular ground structure. These are commercially
available in different sizes (from Instruments for Industry, Inc.,
Farmingdale, NY) and have been termed "Crawford cells" by the manufacturer,
after M. L. Crawford (1974) of the National Bureau of Standards, Boulder, CO,
(also available from Narda Microwave Corp., Plainview, NY). This author
prefers the term "rectangular strip line." In addition, a few large-scale
facilities of this type have been built for various specialized exposure
applications at relatively long (£ 3 m) wavelengths. The largest of these is
at the Defence Research Establishment in Ottawa, Canada, and is being used to
measure human RF absorption in the HF band. A smaller facility has been used
at the U.S. Air Force School of Aerospace Medicine for exposing infrahuman
primates and phantoms to 10- to 50-MHz radiation (Allen et aK 1976). The
U.S. Environmental Protection Agency has also built a Crawford-type structure
(Figure 3-26) for exposing 20 rats to 100-MHz radiation.
3-88
-------
:0
Figure 3-26. EPA 100-MHz rectangular strip line or Crawford cell.
3-89
-------
The second category of transmission-line devices, waveguides, cannot sup-
port the propagation of a TEM mode so that the incident radiation is inevitably
more complex than that in a TEM-mode line. However, the lack of a center con-
ductor means that there is more exposure space available in waveguides than in
the corresponding TEM line. Multiple systems that use commercial waveguide
systems have not been applied much for animal whole-body irradiation because of
high cost. Ho et aJL (1973) have described a single, environmentally con-
trolled system employing standard S-band rectangular waveguide that has been
used in studies of mice at 2450 MHz. Ho et aK (1976) have also described a
system of six waveguides fed from a common 2450-MHz source that provides simul-
taneous irradiation of six mice. Besides cost, another problem associated
with using standard waveguides for whole-body irradiation is the high mismatch
condition created by the presence of an animal in the waveguide. As discussed
earlier, this necessitates the use of costly isolation circuitry. This problem
has, to a certain extent, been eliminated in a specialized waveguide system
developed by Guy and Chou (1976) at the University of Washington in Seattle.
Guy's system is specifically tailored for chronic whole-body irradiation of
rats for extended periods of time with known and reproducible dosimetry. It
consists of a number of 20-cm (8-in) diameter circular waveguides (Figure 3-27),
economically constructed from galvanized wire mesh and short lengths of brass
tubing, in which a circularly polarized TEX1 dominant mode is excited at
915 MHz. Figure 3-28 is a photograph of this type of facility. The use of
circular polarization (see § 3.1, Electromagnetic Field Theory) has two major
advantages:
1) The wide variations in absorption due to constant changes in the
animal's orientation are partially reduced so that the energy
3-90
-------
RG 58
CABLE
r40 25cm 3S
APERTURE FOR
FOOD DISPENSER
SHORTING PLATE,
63mm SQUARE
MESH
49cm
111 cm
II lem
MOUNTING
BOLT
•-CONTACTING
FINGERS
2-5cm LONG SECTION
8* BRASS TUBING
SECTION
A- A
127cm CHA
tubing
723 cm LONG
^ jy—j-• 27cmOIA.
XJL-/ TUBING
2032 cm-I SflcmLONG
HYBRID
Figure 3-27. Physical details of the exposure chamber of the circularly
polarized 915-MHz waveguide facility (Guy and Chou 1976).
coupled to the animal load remains relatively constant. Note
that, whereas in a linearly polarized free-field of perfect
field uniformity, the average SAR can vary by ±5-8 dB
(±50 percent) at or near resonance, in this system, the average
variation of SAR does not exceed ±0.4 dB (±8 percent), even
though the incident field uniformity is far from constant.
2) The hybrid coupler used to establish the circular polarization
in the waveguide provides a way to isolate reflected from
incident energy. This is done by dumping most of the unwanted
reflected energy into a load connected to the hybrid, rather
than allowing it to return through the power dividing network to
the source. This effectively reduces the mismatch normally
encountered with loaded waveguides (mean voltage standing wave
ratio of less than 1.5:1) and permits parallel operation of mul-
tiple units with less isolation circuitry.
Guy et aK (1979) have described a modification of this waveguide system
such that it operates in a multimoded field configuration at 2450 MHz. Though
it is difficult to define quantitatively the incident energy levels in these
3-91
-------
u>
I
LO
ro
Figure 3-28. Photograph of exposure chamber with associated instrumentation for the 970-MHz
circularly polarized waveguide facility at EPA.
-------
units because of the complexity of the fields, the absorption characteristics
have been well defined for rats of varying weights. An extensive system of 100
of these units has been installed at Guy's laboratory (Guy et al^ 1980) for
conducting a chronic radiation study in which rats will be continuously
irradiated over a 3-year life span. This study underlines a major advantage of
Guy's waveguide systems: animal subjects are able to live continuously in
their exposure system with minimal disturbance to their normal living patterns
and without artifactual perturbation. Consequently, the system is ideal for
the kind of chronic, long term studies that are now being emphasized.
Cavities—The resonant cavity, which is the microwave analog of the
lumped-element tuned circuit, consists of a cylindrical, rectangular, or
square box with conductive walls. By suitable means, a system of standing
waves may be excited within the cavity. By adjusting the cavity dimensions,
the system can be made to resonate at the frequency of excitation, such that
intense field levels are created within the cavity. The use of the resonant
cavity as an exposure device has both a major advantage and disadvantage. A
cavity system represents the most efficient way to couple RF energy into a
subject per watt of input power; this constitutes its principal advantage.
The primary disadvantage is that the field structure within a resonant cavity
is probably the most complex to be found in any exposure device. Although some
researchers have argued that the field structure within a resonant cavity,
particularly when operated multimodally, is a good simulation of the complex
multipath environment in which humans are often exposed, it is unlikely that
the human exposure environment ever approaches this level of complexity. For
the reasons already discussed, the field complexity existing within a cavity
3-93
-------
makes it difficult to compare experimental data obtained using a particular
cavity with those obtained using free-field exposures or even another cavity.
Indeed, there are some basic differences between a free-field exposure system
and a resonant cavity. For a resonant cavity that is empty (unloaded), or
that contains a biological load small in comparison to the cavity's volume
(minimal loading), the fields in the cavity are only static and do not pro-
pagate. As such, the energy within is only stored. When the loading is
greater (i.e., when the exposed subject occupies an appreciable fraction of
the cavity's volume) the fields are hybrid, containing both stored and propa-
gating energy. Because the fields are static, minimally loaded cavities
are useful for simulating a quasi-static interaction in which object size is
small relative to wavelength (i.e., a long-wavelength type exposure). This
characteristic has been used by Guy et al. (1976), who designed a large
rectangular cavity that could be excited either separately or jointly in the
TMno or TE-^q2 modes at frequencies of ~ 147 MHz (Figure 3-29). This system
has been used to study absorption distributions in scaled-down phantom models
of man independently exposed to electric or magnetic fields, depending on the
mode of excitation. Single-mode circular cavities have also been used by
Edwards and Ho (1975) for exposing the head of a monkey at 385 MHz. Guy and
Korbel (1972) have discussed the dosimetric aspects of a rectangular cavity
used to multiply expose rats at 500 MHz and have sought to emphasize some of
the problems associated with cavity exposure systems. Recently, Spiegel et al_.
(1980b) described an interesting variable-volume cavity that contains movable
walls such that the cavity can be tuned across a wide frequency range. The
system is capable of simultaneously exciting two independent cavity modes and
can be used to synthesize complex EM fields.
3-94
-------
EXPOSURE CHAMBER
2 58m
HP 608
VHF OSCILLATOR
LINEAR AMPLIFIER
(5 W)
ATTENUATOR
(2 W)
—
POWER SPLITTER
—
PHASE SHIFTER
1
1
PREAMPLIFIER
(90W)
preamplifier
{ 90 W)
1
1
HIGH POWER
AMPLIFIER
(500 W)
RESONANT p
HIGH POWER
AMPLIFIER
(500 W)
CAVITY
a )Exposure of phantom scale model of man m a resonant cavity. b ) Block diagram of resonant cavity driving system.
Figure 3-29. VHF resonant cavity facility (Guy et al^. 1976).
Multimodal cavities, such as microwave ovens, have been used frequently
for experimental purposes because of ready availability and modest cost
(Justesen et aK 1971). Where ovens have been used, they have been extensively
modified to achieve the experimenter's objectives. Multimodal cavities have
often been preferred over single-mode types because of the larger exposure
volume available and the more uniform field distribution. The latter is
achieved by a mode stirrer that helps average-out the field variations within
the cavity. The degree of field uniformity achieved in these devices seems to
depend greatly on its design. Heynick et aK (1977) have described a multi-
modal cubical cavity suitable for irradiating nonhuman primates at 2.45 GHz
(Figure 3-30). Each unit possesses its own regulated power source. Twelve of
these units have been used at SRI International, Menlo Park, CA, to chronically
irradiate a group of squirrel monkeys as part of an EPA-sponsored project.
Each cavity housed two animals plus offspring.
3-95
-------
Microwave cavity with door open, showing mode
stirrer, iris, radiopaque windows, and waste-collection tray.
Dielectric cage.
Figure 3-30. SRI multimodal cavity facility for primate irradiation
(Heynick et al. 1977).
3.3.1.3 Conclusions and Unresolved Questions-
There are diverse exposure methods used in biological effects experimen-
tation, most of which differ considerably in their field characteristics.
Since each has its own particular advantages and disadvantages, the choice of
which system to use has depended on the application. For experiments involving
whole-body irradiation of animals, there is a need for some kind of stan-
dardized method of exposure on which most researchers can agree. Although
such ideas have been proposed and discussed in the past, they are difficult to
implement because many exposure requirements are conflicting, and, therefore,
no one exposure system can satisfy all of the various requirements. However,
some agreement on standardization should enable experiments to be more readily
3-96
-------
replicated and effects more readily confirmed. Further efforts are needed to
this end.
During the past decade, there has been significant improvement in the
exposure facilities available for biological experimentation. Many of these
facilities have succeeded in overcoming the deficiencies inherent in many of
the older exposure devices. Exposure systems specifically designed for chronic
irradiation studies must also be developed further, particularly to reduce
their cost without significantly sacrificing performance.
There also is a serious need for further studies that compare absorption
and biological end points for different forms of exposure. This problem can
be summarized as follows: For the same experiment performed in different ex-
posure systems involving plane-wave, complex or multi-path fields, are there
differences in biological outcomes even where the same whole-body-averaged SAR
is maintained throughout?
3.3.2 Animal Holders
Any experimental irradiation of nonanesthetized animals requires using
some form of holder or restrainer to keep the animal in a location of known
field strength or power density. For multiple animal exposures, it is also
necessary to keep the animals separated from each other. From a dosimetric
point of view, the optimal exposure condition is one in which the animal is
constrained in one orientation and posture. This way, both whole body and
regional dosimetry could be determined with satisfactory precision. However,
3-97
-------
complete restraint of an animal is considered by most biologists an excessive
stress and, therefore, unacceptable. Consequently, the commonly accepted
exposure environment represents a reasonable compromise between these conflict-
ing demands: the holder is designed to restrain the animal, allow it to change
posture and orientation, and also allow for a small degree of lateral and ver-
tical movement.
3.3.2.1 Perturbation by Restrainers—
One of the important criteria in designing animal holders is that they be
constructed of materials that cause the least perturbation of incident fields
possible. This means that cages can be built only from nonmetallic materials.
Various types of low-loss dielectric materials are available in rigid and
semirigid form, and have been the material of choice. The most popular of
these has been acrylic plastic (polymethyl methacrylate) because of its optical
transparency, superior mechanical strength, and ease of machining. Some ex-
perimental and analytical studies have been performed to determine the degree
to which these materials perturb the incident field (Weil 1974; Lin et al^.
1977). These studies showed that, under certain conditions, these materials
can indeed cause serious field perturbations, implying that, in reality, the
power density of the field to which the animal is exposed is not known precisely.
Using foamed polystyrene instead of acrylic is a possible solution. Foamed
polystyrene has a relative permittivity value close to unity and, therefore,
creates only minimal field perturbation. There are, however, two disadvan-
tages associated with polystyrene materials: They are optically opaque, so
that animals cannot be observed in their cages; they also are not strong and
can be readily gnawed by rodents. Coating the material with a substance to
3-98
-------
which the rodents are averse, such as quinine, has been tried by a few workers
(Catravas 1976) to reduce gnawing. However, this practice adds an unwanted
additional insult to the experimental subjects. Nonetheless, foamed polysty-
rene does have application for studies involving short-term exposures.
A more recent paper by Ho (1978) questioned the significance of the
incident field perturbations created by holders made of acrylic materials. In
an analytical study involving a two-dimensional model that contained a cylinder
of muscle-equivalent material mounted concentrically inside an acrylic
cylinder, Ho showed that despite a significant perturbation of the incident
field in the empty holder, there was relatively little perturbation in the
whole-body-averaged SAR and the SAR distribution inside the muscle-equivalent
cylinder when compared with that which exists when no holder was present.
Ho, therefore, concluded that large field perturbations created by animal
holders do not automatically create a correspondingly large perturbation in
the animal's absorption rate because the presence of a mass of lossy dielectric
(i.e., the animal) tends greatly to damp-out the standing wave patterns
created by the acrylic holder. Further experimental observations are needed
to confirm this conclusion.
As part of a behavioral investigation, Gage et aK (1979) investigated the
influence of two different holders on the whole-body SAR of rats and mice at
2450 MHz. One holder was a cuboidal container made of foamed polystyrene, and
the other was an acrylic cylinder. They concluded that there was a significant
alteration of SAR values for both rats and mice due to the presence of the
acrylic cylinder when the animals were oriented parallel to the E-field. This
3-99
-------
effect was considerably more pronounced in the mouse, probably because 2450 MHz
is a frequency close to resonant absorption for the E-parallel orientation of
the mouse.
3.3.2.2 Feeding and Watering During Exposure—
Experiments involving chronic, long-term RF irradiation of animals usually
require that some arrangements for food and water be made during irradiation.
Provision of food does not present difficulties if the standard form of dry
food pellets fed to laboratory animals is used, since these absorb little RF
energy. However, providing drinking water or liquid nutriment does present
some problems because water is a comparatively good conductor and, therefore,
absorbs energy. Two methods of supplying water have been used, and each has
associated problems.
In the first, the water reservoir is placed in the exposure system with
the animal. The major disadvantage here is that, since the water supply is
located in a high-field-strength environment, it will absorb RF energy and
will perturb surrounding fields. These factors will create uncertainty in
both dosimetric and densitometric measurements. Furthermore, it is conceivable
that, if the field levels are intense enough, the water temperature could
become sufficiently high that the animal would refuse to drink. Consequently,
this approach is not recommended.
The second method involves placing the reservoir outside the exposure
system in a region of little or no field strength, and supplying water to the
animal via a supply tube or pipe. This method is superior, but it, too, has
3-100
-------
associated pitfalls which, without special precautions, can introduce serious
experimental artifacts. As Guy and Chou (1976) have pointed out, the water
tube or pipe provides a conductive pathway for microwave currents. When the
animal attempts to drink, a circuit is closed between the animal, which is at
high potential, and the water reservoir, which is at zero or ground potential.
At this time, RF currents will pass from the animal to the watering system;
because the contact point area is small and involves sensitive tissue, a high
current density may exist in the animal's tongue. Since the animal will find
this condition aversive, it will probably refuse to drink. Therefore, the RF
current flow through the water supply system must be interrupted. Three
methods of interruption have been devised: (a) When the animal interrupts a
beam of light using a tongue-licking operant, a small bolus of fluid is injected
into the animal's mouth (King et aL 1970); (b) the animal holder is equipped
with a small water trough which is continuously supplied through a water drip
system; and (c) an RF-choke assembly is placed between the animal and the water
reservoir to effectively decouple the animal from the water supply when drink-
ing (Figure 3-31). In the choke assembly design (applying only to enclosed sys-
tems), water is supplied through copper tubing that forms the center conductor
of a shorted quarter-wave section of a triaxial air-line. To prevent the animal
from coming into contact with the copper tubing, a short section of glass or
acrylic tubing is mounted at its tip. The choke assembly effectively provides
a high impedance at the point of animal contact, thereby reducing the RF current
in the watering system to near zero.
3-101
-------
BOUNCE
BOTTLE
Ri »l 75cm
Ri • 075cm
Rj ¦ 475cm
METAL
WALLS
EXPOSURE
CHAMBER
WALL
(SCREEN)^
GLASS
'TIP
plastic
'cage wall
Figure 3-31. Water-supply system for exposure chamber (Guy and Chou 1976).
3.3.3 Densitometric Instrumentation
Densitometry is the measurement of RF-field strength and is usually
expressed in units of equivalent plane-wave power density. Dosimetry is the
measurement of absorbed energy, or the rate of energy absorption by some object
in an RF field. In this section, the strength or intensity of an RF field
are quantified by measuring a suitable field variable. The question of what
constitutes a suitable variable has been addressed (Bowman 1970). For a
linearly polarized, plane-wave, EM field in free space--and, for practical
purposes, in air--the following relationships exist between the electric field,
the magnetic field, the power density, the electric energy density, and the
magnetic energy density:
3-102
-------
§ = Z0 (3-12)
Z = (^)V2 (3-13)
0
E2 2
S = T = ZoH (3-14)
o
U = 1/2 (e E2 + |j H2) = U,- + Uu (3-15)
o o t n
UE = UH (3-16)
where E = electric field, V/m
H = magnetic field, A/m
ZQ = characteristic or intrinsic impedance of free space, 377 ft
= permeability of free space, H/m
eQ = permittivity of free space, F/m
2
S = power density, W/m
3
U = total energy density, J/m
3
= electric energy density, J/m
3
U^| = magnetic energy density, J/m
The quantities and eQ are properties of the medium and are scalar constants
for free space. Equation 3-12, therefore, states that the magnitudes of
E and H are related by a constant factor ZQ for free space. Equation 3-14
relates the power density S to the individual field components E and H for
the basic plane wave. Equation 3-15 is a general relationship for the total
energy density U of an RF field, and it defines the individual electric- and
magnetic-field energy densities UF and IL.
3-103
-------
The relationships between field variables for a plane wave are simple,
so that the measurement of any one field parameter allows calculation of the
others. Although power density is a variable of interest in biological effects
studies, it is particularly difficult, if not impossible, to measure directly
because it is a time-averaged vector cross product that, in general, is not
expressed as simply as Equation 3-14. It may, however, be indirectly evaluated
for plane waves by measuring the E or H field, as indicated by Equation 3-14.
o
This measurement, when converted to units of power density (W/m ) using
Equation 3-14, is referred to as the "equivalent plane-wave power density."
Above 1 GHz, it is usually not necessary to measure both the E and the H fields
(Bowman 1970), whereas below ~ 300 MHz, the H field becomes increasingly impor-
tant, and an independent measurement of H is necessary (Asian 1979; Lin et aK
1973).
Power density is, however, inappropriate to use in evaluating near fields
of antennas or fields near reflecting surfaces because intense fields can exist
in these regions while the power density is low or even zero. A plane wave
normally incident on a conducting surface can create standing waves with large
E and H values but very low power density. Since power density is a measure
of the power flow across a unit area, the net power flow of a plane wave
normally incident on a conducting surface and reflecting back is zero (or small
in practical cases), and hence the power density is zero. The E and H fields
will now be in a standing-wave pattern with amplitudes twice those of the
fields of the incident wave.
3-104
-------
In the near field of sources or for complex RF fields, the E and H fields
are not related by Equation 3-12; however, Equation 3-15, which relates the
individual E and H fields to their respective energy densities, is always true.
Therefore, measurements of both the E and H fields are required to evaluate the
total energy density where Equation 3-12 does not apply.
Bowman (1970), Swicord (1971), and others have described the desirable
characteristics of an instrument for quantifying hazardous EM fields. The
instrument and probe should: 1) measure in terms of energy density and respond
only to the variable being measured, 2) be small to permit a high degree of
spatial resolution and thus be useful in small volumes, and 3) should have an
isotropic response and be insensitive to field polarization. The probe should
also cause little scattering of the field and operate over a broad frequency
range. The instrument and probe should be battery powered, lightweight, and
rugged, and should read either peak or average values and have a dynamic range
of at least 20 dB (one hundredfold) without the need to change probes. The
readout instrument should be direct reading and free from susceptibility to
RF interference.
Many approaches have been used to develop field probes that meet these
criteria. The most successful and enduring designs have centered around the
thin-film thermocouple or bolometer sensor/antenna and the crystal detector/
antenna. The first type of sensor is a square-law-responding device that
senses a temperature change due to the absorption of energy. The second type
uses a semiconductor diode that produces a voltage or current related to the
strength of a field variable. The diode can respond as either a square-law
3-105
-------
detector or a linear detector, depending on the region in which it is operating.
Some commercially available electric-field probes are shown in Figure 3-32.
' ".A
Figure 3-32. Samples of commercially available survey meters for measuring
RF electric-field strength. From left to right the units shown
are: The Narda Microwave Corporation Model 8315A meter with
a Model 8321 isotropic probe; the General Microwave Corporation
Model 481A meter with a Model 81 probe; and the Holaday Indus-
tries, Inc., Model HI-3001 meter and probe.
3.3.3.1 Electric-Field Probes--
The NBS EDM-2 electric energy density meter (Belsher 1975) was developed
for accurate measurement of occupational exposure to EM fields from 10 to
3-106
-------
500 MHz. This probe uses three short orthogonal dipoles terminated by crystal
rectifiers to detect the RF field, and it produces an isotropic response
within ±1 dB. The DC signal appearing on the diodes reaches the meter's
electronics through RF-transparent high-resistance lines. The meter displays
electric energy density in units of microjoules per cubic meter. Energy
density was chosen as the field variable to display because of the ambiguity
of power density in the near field of radiating elements.
The meter has a dynamic range of 50 dB in nine ranges from 0.003 to
3
30 pJ/m . Although the probe's detector diodes are used in both the square-
law and linear regions, a square-law response is achieved over most of the
range by a varistor network. The meter is calibrated at specific frequencies
within the FCC ISM bands at 13.56, 27.12, and 40.68 MHz. The frequency
response is flat within ±1.0 dB over the 10- to 500-MHz operating range. One
of the improvements in the EDM-2 device over previous designs is reduced
temperature-induced drift. The reduced drift is achieved by loading the
dipole-detector diodes with equivalent load diodes. This technique has re-
duced the temperature response to < ±0.7 percent/°C in the 15 to 35 °C range.
The field probe design provides a rapid response time to permit observation of
the modulation of the field and the measurement of peak values. The rise and
fall times (10 to 90 percent of the signal and 90 to 10 percent of the signal,
respectively) are both < 0.6 ms on the most sensitive range.
A commercial version of the NBS electric-field probe is marketed by
Holaday Industries (Edina, MN). This field-strength meter (Model HI-3001)
operates over the frequency range 0.5 to 1000 MHz and is accurate within
3-107
-------
±1.0 dB over this range. The meter has an isotropic response within
±1.0 dB and has peak-hold and read circuitry to capture the highest reading
observed. The response time is 1.5 s. Measuring range of the meter is 1 to
3160 V/m in six ranges on an analog meter. An audible tone, with a pitch
proportional to the meter reading, is provided to prevent operator over-
exposure in survey applications.
In 1980, the General Microwave Corporation (Farmingdale, NY) introduced
the RAHAM Model 4A radiation-hazard meter. This instrument is a broadband
crystal-detector design having a frequency coverage of 200 kHz to 26 GHz, and
a display calibrated in units of equivalent plane-wave power density ranging
2
from 0.001 to 20 mW/cm . This meter has an isotropic response, is battery
powered, and has a response time of 1.5 s.
For use in the near field, Rudge (1970) constructed an electric-field
probe using two short crossed-dipole antennas. This meter was designed to
operate from 915 to 2450 MHz in proximity to RF emitters. The probe detected
the electric field in two perpendicular planes, and it was calibrated in power
density terms. Rudge carefully analyzed sources of error in the design of
near-field probes. His study included the effects of field curvature and
antenna backscatter, particularly with regard to multiple coupling and impe-
dance variation, both of which affect the calibration of the field probe. The
major source of scattering was the probe's antenna, particularly when the
lengths of the dipole wings were on the order of one-fourth wavelength. The
prototypal probe achieved a dynamic range of 47 dB with a square-law response.
This was accomplished by adding fixed values of load resistances at increasing
power levels to keep the detector diode operation in the square-law region.
3-108
-------
2
Sensitivity was reported at 300 nW/cm with a probe antenna of 8-mm overall
dipole length.
Bassen et ah (1975) developed a miniature broadband probe. Although the
probe was designed for free-space operation, the ultimate goal was to develop
an implantable isotropic probe for biological effects research (Bassen et ah
1977a). Using 3-mm dipole elements, the free-space probe achieved a rela-
tively flat frequency response over the range of 915 MHz to 10 GHz. After
some refinements were incorporated (Bassen 1977; Bassen et ah 1977b), the
probe was tested over a frequency range of 200 MHz to 12 GHz. The improved
version had a 30~dB dynamic range with a 20-(jW/cm sensitivity, isotropic
response to ±2 dB, a detection bandwidth of 2 kHz, and a spatial resolution of
~ 3 mm.
The requirement of small size was imposed on the implantable-probe design
to ensure minimal alteration of the field when the probe is used relatively
close to sources and dielectric boundaries. At the same time, the constraint
of small dipole size has the further advantage of extending the frequency
response of the probe into higher frequencies. With 3-mm dipoles, theoreti-
cally, resonance should occur at 50 GHz. The probe uses dipole elements
terminated with three zero-bias Schottky diodes for a maximal square-law
response, arranged in an "I"-beam configuration. High-impedance leads are
connected to the diodes and routed along the probe's body to an optical tele-
meter. A fiber-optics cable then routes the digitally encoded field-strength
signal to a digital receiver that is calibrated to display the equivalent
3-109
-------
plane-wave power density. A commercial version of the miniature field probe
is marketed by Electronic Instrumentation and Technology (Sterling, VA).
An electric-field sensor (EFS) and readout system capable of operating
over the frequency range 10 kHz to 200 MHz has been developed by Instruments
for Industry, Inc. (Farmingdale, NY) and was described by Ruggera (1976). The
EFS-1, -2, and -3 are linearly polarized, square-law electric field detectors
with an analog readout calibrated in volts per meter. The EFS-1 is intended
for CW fields, the EFS-2 for pulsed fields, and the EFS-3 for both types of
fields with the addition of a peak-reading provision. The dynamic range is 1
to 300 V/m with antennas of different lengths. An optical transmitter is used
when remote readout is required, eliminating field perturbations introduced by
metallic cable. An isotropic monitor has also been designed (RHM series) that
essentially encloses three EFS-series sensors in one box. Readings from each
sensor can be summed (by the square root of the sum of the squares of the
field strengths) to measure the total electric field, or a particular sensor
can be read directly when field polarization is a variable of interest.
Another probe design uses thin-film thermocouple arrays acting as both
the dipole antenna and the detection element. The dipole elements are lossy;
therefore, their temperature will increase proportionally with the square of
the tangential electric field. The dipole elements can be thought of as the
series connection of a large number of tiny thermocouple dipoles. This detec-
tion method provides true square-law response and produces signals propor-
tional to the average energy density of the electric field in the measured
volume.
3-110
-------
The Narda Microwave Corporation (Plainview, NY) produces a wide selection
of RF-radiation monitoring instruments, such as the Model 8606 broadband
isotropic radiation monitor. This instrument uses three orthogonal, thin-film
thermocouple sensors that respond to the square of the electric field. Two
probes are available that have the combined dynamic range of 0.02 to
2
100 mW/cm over the frequency range 0.3 to 26 GHz. The probes have isotropic
response within ±0.5 dB except when the electric field is aligned with the
handle axis. The General Microwave RAHAM Model 3, another electric-field-
sensitive meter with thin-film thermocouple sensors, performs comparably to
the Narda 8606 system. The Narda 8100 series meters were designed earlier
(Asian 1970) for measuring the leakage of microwave ovens, dryers, and medical
equipment, which operate at 915 and 2450 MHz. The probe used with the 8100
series meters is not isotropic but must be oriented with the probe handle
perpendicular to the wave front. Three probes are used to measure across a
2
dynamic range of 0.02 to 200 mW/cm . The probes contain two orthogonal,
thin-film vacuum-evaporated thermocouple elements that function both as
antennas and detectors. When the probes are used at 915 MHz, an adapter is
attached to the end of the probes, increasing the effective length of the
dipoles. This increase in length compensates for the reduced efficiency of
the antenna at 915 MHz.
3.3.3.2 Magnetic-Field Probes--
As discussed previously, measurement of only the electric field for
frequencies below ~ 300 MHz does not fully characterize the hazard or heating
potential of an RF field, because the electric and magnetic fields are not
related by the free-space intrinsic impedance at frequently encountered
3-111
-------
distances from radiation sources. In the mid-1970's, NIOSH supported an NBS
effort to develop a magnetic field probe that would operate in the 10- to
300-MHz band. Two probes were developed (Greene 1975a,b) that consist of
small single-turn, balanced loop antennas of 10- and 3.16-cm diameter. The
dynamic ranges of the two probes are 0.5 to 5.0 A/m and 5.0 to 50 A/m,
respectively. The probes are capable of measuring the magnetic field within a
few centimeters of an RF-radiation source, in the near field. The loop
antennas are constructed with two short gaps; one gap is terminated with a
silicon diode that detects the induced RF voltage, and the other gap is ter-
minated with a capacitor to bypass RF currents without short-circuiting the DC
output. The DC signal is then transmitted over NBS-developed conducting
plastic lines to an electrometer voltmeter. One of the inherent problems of
loop antennas is that the response is proportional to frequency, making these
antennas particularly sensitive to errors if harmonics (integral multiples of
the fundamental frequency) are present in the field under study. Greene also
shows that both the electric-dipole response of the loop and the partial-
resonance effect increase with frequency. An experimental method is provided
to minimize the error due to electric-field sensitivity, and plots are pro-
vided for the two probes to indicate worst-case measurement error vs. fre-
quency due to the electric-dipole response. The self-resonant frequencies of
the two loops were computed to be 280 and 760 MHz for the 10- and 3.16-cm
loops, respectively, and a correction curve was provided to estimate the in-
crease in response due to this effect.
Asian (1976) reported the development of a magnetic-field-radiation
monitor that has a frequency response within ±1.0 dB over the frequency range
3-112
-------
of 10 to 200 MHz. Two probes, the Narda Corporation Models 8631 and 8633, are
2
used to provide a full-scale measurement range of 0.2 to 100 mW/cm . (Sensi-
2
tivity extends to 20 mW/cm on the Model 8631 probe.) These probes have an
isotropic response (±0.5 dB) and accept all polarizations. Each probe has
three orthogonal coils, with each coil having two turns. The coils are ter-
minated with thin-film thermocouple elements. High-resistance lines are used
to route the DC signal to a preamplifier in the handle of the shielded probe,
and the preamplifier output is connected via shielded cable to an indicating
meter. A partial solution to the problems of frequency dependence of loop
antennas was achieved by designing the coils to be series resonant slightly
below the low-frequency end of the operating band. The response of the coils
increases at frequencies above 200 MHz, as the electric-dipole effect begins
to govern the response.
Magnetic-field probes of various designs were reviewed by Ruggera (1976),
many of which were commercially available and others of which were experi-
mental. Ruggera described electric-field-shielded loops constructed with
semirigid coaxial transmission line from 2 to 6 cm in diameter. Ruggera also
described an isotropic, electric-field-shielded, three-loop, orthogonal,
magnetic-field probe that has a cross-polarization rejectivity of ~ 20 dB from
40 kHz to 50 MHz. Signals from the probe are recovered with terminated,
coaxial cables. In addition, the author described a three-axis, magnetic-
field-measuring instrument developed by Southwest Research Institute. The
original design was plagued by cable-pickup problems, which limited the usable
frequency to 50 MHz. Refinements by the FDA Bureau of Radiological Health
3-113
-------
(BRH) that incorporated the optical telemetry systems (see § 3.3.3.1) used on
the BRH miniature electric-field probe led to a flatter frequency response
over the extended frequency range of 150 kHz to 150 MHz.
In 1981 Holaday Industries introduced a new magnetic-field probe for
their 3000 series of field-strength meters. The Model STH-01 probe has a fre-
quency response of 5 to 300 MHz within ±2 dB. The uniformity of response
improves to ±1 dB over the range of 10 to 200 MHz. When used with the HI-3001
2 2
meter (§3.3.3.1), full-scale sensitivity of the probe is 0.1 A /m in the low
2 2
range and 1.0 A /m in the high range. The sensor consists of three ortho-
gonal loops terminated with detector diodes that rectify the RF voltages
induced in the loops. The sensors can withstand an 800-percent overload above
the full-scale field strength without damage.
3-114
-------
3.4 DOSIMETRIC METHODS
James B. Kinn
Dosimetric methods are used to determine where and how much energy is
absorbed by a biological target. The methodology is divided into two areas:
(1) whole-body dosimetry, dealing with the integrated or spatially averaged
SAR for the entire animal; and (2) regional dosimetry, dealing with the SAR's
or internal field strengths in a specific site of the biological target.
3.4.1 Whole Body Dosimetry
Whole-body dosimetry is performed using either a power difference or a
calorimetric method. The power-difference method is limited to closed exposure
systems such as waveguides, TEM-mode transmission cells, and coaxial air lines.
In this method, power meters are attached to the exposure systems by directional
couplers to measure the incident, reflected, and transmitted power through a
section of the closed system containing the biological target. The power
absorbed by the biological target is then measured by subtracting the values
of the reflected and transmitted power from the incident power. In the case
of exposure systems using live animals, the values of those three measurements
will vary as the animals move, because absorption is a function of the animal's
orientation relative to field polarizations. This makes it desirable to have
a data collection system that can integrate the measured power difference and
then determine an average over a period of time, usually the exposure duration
(Christman et a^. 1974; Ho and Edwards 1977b). The advantage of this method
is that it provides on-line instantaneous measurements of absorbed power. A
3-115
-------
disadvantage is that its optimum application is in single-animal exposure;
simultaneous exposures of multiple animals in the same closed system would
result in average dose rate values for the group, rather than for an indi-
vidual animal. A second disadvantage is that, under exposure conditions where
the RF energy absorption rate is very low, it may be impossible to measure the
power absorption, because the measurement error in the power meter exceeds the
computed power difference.
Calorimetric methods use the thermalization of the absorbed power as a
measure of SAR. Under the tacit assumption that all the RF energy absorbed by
a biological target is converted to heat, the calorimetric method measures the
heat added to the target. The approach is to irradiate an animal carcass with
a short exposure at high power to reduce heat loss errors due to conduction
and convection during the exposure. The thermalized energy added to the
carcass is measured with the calorimeter. Calories per unit time per unit
mass are then converted to W/kg, the SAR for this high incident power density.
The SAR for any other power density is then scaled from this measured value in
direct proportion to the ratio of power densities. Three calorimetric methods
that have been used in whole body dosimetry are gradient-layer calorimetry
(Gandhi et a^. 1979), Dewar-flask calorimetry (Blackman and Black 1977), and
twin-well calorimetry (Hunt and Phillips 1972; Kinn 1977). All three methods
use a freshly killed animal carcass that has been briefly exposed to high-
power RF fields and thus raised to a higher temperature. It has been assumed
that the dielectric properties of the carcass do not differ significantly from
those of a living animal.
3-116
-------
The gradient-layer calorimeter measures the rate at which heat passes
from the heated animal through the walls of the calorimeter to the room air.
It is accurate, but it takes several hours to make the measurement.
The Dewar-flask calorimeter method uses the difference between the aver-
age temperature of the animal carcass before and after rapid heating with RF
radiation to determine the heating rate and thus the dose rate. The average
temperature of the animal carcass is determined by placing the carcass and a
coupling fluid, usually water, into a Dewar flask and measuring the equilib-
rium temperature of the mixture. From the theory of mixtures and a knowledge
of the average specific heat capacity of the animal, the average carcass
temperature is determined. The average temperature of the heated carcass is
determined similarly, and the SAR is computed. Although the method is simple
by design, a disadvantage is that a long time is required to measure equilib-
rium temperature. Furthermore, the average specific heat capacity of the
carcass must be known, and, because the body's composition is complex, this
capacity cannot be precisely determined.
The twin-well calorimeter method uses pairs of carcasses of equal mass.
One carcass of a pair is heated with RF radiation; the heated and unheated
carcasses are each placed in a well of the calorimeter, and the calorimeter is
allowed to return to equilibrium. The twin-well calorimeter measures the
increment of heat introduced by the exposure. The SAR is then calculated from
the measured caloric increase, carcass mass, and exposure time. This method
does not require the specific heat capacity of the carcass but requires 4 to
16 h to make a measurement. A recent improvement to the method has reduced
3-117
-------
the measurement time to 15 to 30 min and increased measurement accuracy by
controlling the calorimeter with a microprocessor system. This arrangement is
the same as the standard setup except that the unexposed carcass is in contact
with a heating element (Figure 3-33). The microprocessor program controls the
rate at which heat is applied to the unexposed carcass. Using the twin-well
calorimeter as a "thermal" balance, the program controls the heating rate so
that equilibrium is established in as short a time as possible. All of the
power applied to the heating element, and thus to the animal carcass, is inte-
grated and divided by the mass of the carcass to determine the SAR.
Figure 3-33. Microprocessor-controlled twin-well calorimeter
3-118
-------
3.4.2 Regional Dosimetry
Regional dosimetry is performed to determine either the energy absorption
rate or the field strength at a specific location within an animal or tissue
preparation. The average field strength within the animal can be measured
directly or inferred from the SAR. Direct measurement of field strength
is made with an implantable probe containing one or more miniature diodes
mounted on a dielectric substrate and attached with high resistance leads to
a voltage-reading device (Bassen et aK 1977b). The voltage across the diode
is a measure of the field strength at the diode, provided the animal tissue
material in which the probe is implanted does not differ significantly in
dielectric properties from the material used in the calibration procedure
(Bassen 1977). The useful frequency range over which such field strength
measurements can be made is 100 to 12,000 MHz. The reduced sensitivity at
lower frequencies and the greatly reduced field strength in biological targets
at the higher frequencies delineate the useful frequency range. Indirect
measurement of internal field strength is derived from the SAR value. The
heating rate is proportional to the square of the electric-field strength.
Measurement of regional SAR can also be accomplished by temperature probes
- or by thermographic imaging. Temperature probes are small (1-mm diameter) and
are designed to be implanted in the site of interest without altering the EM
field in that area (Rozzell et aH. 1974; Bowman 1976). An animal or preparation
is briefly exposed to RF fields of high intensity, and the temperature rise is
measured. The rate of temperature rise, together with the amount of heat loss,
3-119
-------
is then used to compute the SAR at a specific site in an absorber, such as a
rat brain.
The technique described above is a special case of heating and cooling
curve analysis, i.e., estimating the initial rate when the rate of heating is
much higher than the rate of cooling. The analysis of the complete heating
and/or cooling curve is especially useful in cases where the RF energy absorp-
tion is of the same order of magnitude as the cooling rate of the animal or
preparation. In this procedure, the temperature is recorded continuously
during exposure and/or after exposure until a steady state or equilibrium
temperature is reached. Analysis of either curve yields the cooling rate,
which is directly related to the SAR. The heating curve also contains the
cooling rate that can be determined by the appropriate analysis; therefore,
the SAR can be determined from either a heating or a cooling curve (Allis
et al_. 1977).
Thermographic imaging also requires very high power exposures. This
method uses a thermographic camera containing a mirror scanning system, a
liquid-nitrogen-cooled detector, and a data collection system usually
interfaced to a minicomputer (Guy et aK 1977; Kantor and Cetas 1977). Speci-
mens are prepared by cutting a frozen carcass into halves, and supporting
each half in an RF-transparent medium such as urethane foam. The surface of
the cut plane is covered with either a plastic film or polyester "silk" screen
material to hold each half of the specimen in its foam support. The "silk"
screen material allows electrical contact between the two halves when they are
readjoined during exposure. Unlike plastic film, which prevents induced
3-120
-------
current from flowing across its surface, the "silk" screen material allows
current flow, and the model may be placed in a complex field of unknown polari-
zation. The assembled specimen is allowed to equilibrate to room temperature
(25 °C). Thermographic pictures are taken of the animal specimens at the
plane of the cut before and after exposure, and the SAR is computed from the
point-to-point temperature change during exposure. The SAR values are then
used to produce a contour plot of iso-SAR lines or averaged values over a
specific area.
Scaled-down or full-size phantoms made with materials having dielectric
properties similar to those of biological tissues are also used to determine
the SAR distribution in a biological specimen, such as man and small animals.
Animal models have been used for the investigator's convenience. Scaled
models of man (Guy et aH. 1976; Gandhi et aK 1977) have been used to simulate
human exposures under varying conditions. This involves frequency scaling,
in which a combination of higher-frequency radiation with a smaller-than-1ife-
size model is used to simulate the absorption characteristics of man when
exposed to lower-frequency radiation. The technique also requires suitable
scaling for the dielectric parameters of the tissue-equivalent materials used
in the model. The validity of the models used to date is open to some
question, because the phantom "equivalent" model is filled with a homogeneous
material whose complex dielectric constant is the estimated average for the
whole body; a more realistic model would attempt to simulate the inhomogeneous
tissue structure of the body. The dielectric material normally used in human
modeling is gelled water whose conductivity has been adjusted with sodium or
3-121
-------
potassium chloride. The phantom is made in two halves, and the thermographic
method described above for animal specimens is used.
3.4.3 Unresolved Questions
Refinements are lacking for models of the inhomogeneous tissue structure
of man and animals, and the equivalency of these models to the actual bio-
logical target has not been validated. In simultaneous exposure of multiple
animals, to what extent do animal movements create uncertainty in the SAR
estimates? How much target separation is required before this uncertainty
falls within acceptable limits? As the sophistication of research into
RF-induced biological effects improves, and as the sites for many of the
reported effects—particularly those associated with the nervous system—
become better isolated and defined, there will be an ever increasing need for
localized dosimetry of greatly improved spatial resolution. Finally, although
it is well understood that RF energy is converted to heat in a biological
target, the question remains: Are there transient "field-specific" effects
not explicable by a temperature change?
3-122
-------
SECTION 4
EFFECT OF RF-RADIATION EXPOSURE ON BODY TEMPERATURE
4.1 THERMAL PHYSIOLOGY
Christopher J. Gordon
4.1.1 Temperature Regulation
Almost all animal species, including vertebrates and invertebrates, are
capable of sensing and responding to changes in environmental temperature.
The ability to maintain a constant body temperature independent of ambient
temperature, termed homeothermy, is restricted to humans and other mammals as
well as most birds. Reptiles, amphibians, fish, and invertebrates generally
have a body temperature similar to ambient temperature, and are thus termed
poikilotherms (i.e., having changeable temperature).
Below the point of protein denaturation, temperature has a direct effect
on the rate of biochemical reactions and thus affects the rate of physiologic
processes. The effect of temperature on biological reactions is described by
the Q1q parameter, a dimensionless number equal to the increment in the reaction
rate for a 10 °C increase in temperature:
4-1
-------
R9 10/T9-T-.
Q">=
or
log Q1q = (10/T2 - Tj) (log R2 - log R1)
where R^ = reaction rate at T-^
R2 = reaction rate at T2
T = temperature
Thus, a Q1q of 2 means that the reaction rate doubles with a 10 °C increase in
temperature, a of 3 means a tripling of the rate, and so forth. For
biological reactions, the Q-^q generally ranges between 2 and 3.
The activity of poikilotherms is dictated by environmental temperatures.
At low temperatures enzymatic activity is decreased, causing reduced muscle
activity, active transport, etc. Hence, poikilotherms generally have a narrow
range of ambient temperatures (~ 20 to 40 °C) where normal life functions can
take place. On the other hand, many homeotherms, by maintaining a relatively
constant body temperature, can function normally at ambient temperatures far
below 20 °C and occasionally above 40 °C. Of course, to maintain a constant
body temperature at extremely high and low ambient temperatures requires
additional metabolic energy.
There are three major physiologic systems employed by homeotherms when
subjected to changes in ambient temperature: (1) the neurally activated
behavioral and autonomic motor systems, (2) the neurally activated
4-2
-------
hypothalamohypophyseal system, and (3) the morphology and growth system. The
following is a brief explanation of each physiological system. (1) During a
thermal stress, thermal receptors in the skin, central nervous system (CNS),
and other deep body areas are activated. The neural activity is integrated
predominantly in the preoptic area and anterior hypothalamic area of the brain
stem where the proper motor signal originates to either generate or dissipate
heat (Figure 4-1). Cold-exposure may result in vasoconstriction of peripheral
blood vessels, shivering, and selection of a warmer environment. Heat exposure
may result in vasodilation of peripheral blood vessels, sweating or panting,
and selection of a cooler environment. (2) If the thermal stress continues
for at least several minutes, the CNS responds by changing blood hormone
levels. In both heat and cold stress there is an increased release of adreno-
corticotropin from the pituitary, which in turn increases blood corticoste-
roids. Cold stress facilitates the increased secretion of thyroxine from the
thyroid gland and epinephrine from the adrenal medulla. Heat stress facili-
tates the release of vasopressin, the water-conserving hormone. (3) Prolonged
thermal stress for weeks and months will promote morphologic changes designed
to reduce the impact of the thermal stress. For example, raising animals in
cold environments will result in shorter appendages, increased fur density,
and cutaneous fat deposition, which reduces the rate of heat loss.
4.1.2 Ambient Temperature vs. RF-Radiation Exposure
The physical mechanisms underlying the regulation of a constant body
temperature during normal temperature or RF-radiation exposure can be described
4-3
-------
TRIGEMINAL NUCLEUS
SPINAL CORO THALAMUS
I „ I
WARM :
PERIPHERAL
CUTANEOUS RECEPTORS
COLD »
DEEP BODY
WARM AND COLD
RECEPTORS
PREOPTIC/ANTERIOR HYPOTHALAMUS
J
CNS WARM-SENSITIVE NEURON
W
A
CNS COLD-SENSITIVE NEURON
MOTOR
OUTPUTS
HEAT DISSIPATION
PERIPHERAL VASODILATION
PANTING/SWEATING
BEHAVIORAL SELECTION OF A
COOLER ENVIRONMENT
HEAT GENERATION/RETENTION
PERIPHERAL VASOCONSTRICTION
SHIVERING THERMOGENESIS
- NONSHIVERING THERMOGENESIS
BEHAVIORAL SELECTION OF A
WARMER ENVIRONMENT
.SENSORY
INPUTS
Figure 4-1. Simple neural model of thermoregulation in a mammal for predicting the motor responses
to short-term (i.e., hourly) changes in ambient and/or body temperature. Activation of
the warm-sensitive pathway leads to an increase in heat dissipatory mechanisms, whereas
activation of the cold-sensitive pathway leads to an increase in heat generating/
retaining mechanisms. This information is taken from several models proposed by Hammel
(1968) and Hensel (1973).
-------
using a simple model (Figure 4-2). Under steady-state thermoneutral
conditions, the direction of heat transfer is from the inner core (e.g., core
or rectal temperature, T ) to the cooler peripheral shell (e.g., skin and
subcutaneous temperature, T ^). In the absence of blood flow the transfer of
heat between the core and periphery is approximately equal to the intershell
temperature gradient multiplied by the heat conductivity of the tissues. Heat
exchange between the skin and environment occurs through convection, radiation,
and evaporation. Under most conditions the transfer of heat within the central
and peripheral shells is greatly influenced by the circulation of blood.
Raising the temperature of either the peripheral or ambient shell impedes the
transfer of heat and results in an increase in T .
c
Either an elevation in ambient temperature or exposure to RF radiation
may lead to an increase in body temperature; however, the route of heat trans-
fer from the two heat sources varies greatly. For example, raising ambient
temperature reduces the temperature gradient between T . and T , resulting in
S K a
an elevated T The higher peripheral shell temperature, which in turn re-
duces the T -T . gradient, causes an increase in T . In contrast, RF-radiation
L bl\ L
exposure can increase the temperature of the T and T . without affecting T
C S K o
by direct deposition of energy in the peripheral and/or core shells.
4.1.3 Mechanisms of Heat Gain During RF-Radiation Exposure
The coupling of electromagnetic (EM) energy into biologic subjects exposed
to RF radiation was discussed in detail in § 3.2, RF-Field Interactions with
4-5
-------
PERIPHERAL TEMPERATURE
SKIN TEMPERATURE
J r— 50
(—50
—45
—45
CORE TEMPERATURE
U
o
HYPERTHERMIC
UJ
— 40 §
—40
NORMOTHERMIC
K
<
DC
UJ
a.
5
UJ
UJ
CD
2
<
— 30
—30
GRADIENTS GRADIENTS
Figure 4-2. Simple model of temperature gradients in a homeotherm under
conditions of thermal neutrality, normal heat stress (a), and
heat stress due to RF-radiation exposure (b). At a thermoneutral
ambient temperature, the skin temperature T ^ is maintained
at approximately 33 °C. (a) Raising ambient temperature to
35 °C results in an elevated T which in turn causes an
increase in core temperature T . (b) Exposure to RF
radiation causes an elevation in the peripheral tempera-
ture without affecting T and promotes an elevated T .
a C
4-6
-------
Biological Systems. To introduce the following material, some of the important
features of that discussion will be reiterated and summarized here.
The penetration of RF energy into biological tissues is dependent on the
wavelength of the incident energy. Longer wavelengths can penetrate deeply
into living tissues, but the shorter wavelengths found in the microwave region
of the spectrum cannot penetrate deeply. For example, at 2450 MHz (wavelength
of 12.5 cm) the RF energy is absorbed within approximately the first 4 to 5 cm
of muscle tissue, assuming that the RF radiation is incident on an object that
is large in comparison to this wavelength (such as a human). Similarly, RF
energy of still higher frequencies (i.e., in the millimeter wave region of the
RF spectrum) will not penetrate more than 1 to 2 mm in skin tissue.
This difference in penetration poses a major problem for comparing the
effects of RF radiation of very different wavelengths on thermoregulatory
function in different species. For example, when a 20-g mouse or a 70-kg man
is exposed to RF radiation of millimeter-size wavelengths or to infrared (IR)
radiation, the energy is deposited in the first 1 to 2 mm of the skin in a
basically similar fashion. However, at 2450 MHz a mouse is comparable in size
to the wavelength so that a resonant absorption condition exists, resulting in
efficient energy coupling with deep penetration and nonuniform internal energy
deposition (i.e., EM radiation "hot spot" formation). On the other hand, when
a human is exposed to the same 2450-MHz radiation, the energy will be peripherally
deposited within a few centimeters of the body's surface in that area facing
the radiation source.
4-7
-------
Notwithstanding these problems, some attempts have been made to compare
thermoregulatory function in different species when subjected to RF radiation
of the same frequency. For example, deLorge (1979) compared the power density
at 2450 MHz needed to induce a 1 °C elevation in the rectal temperature of
rats, squirrel monkeys, and rhesus monkeys. Using a semi-log plot of power
density vs. body weight (Figure 4-3), deLorge extrapolated the data on rats
and subhuman primates to the case of a 70-kg man. He determined that a value
of 92 mW/cm is needed to raise the rectal temperature of man by 1 °C at
120
CM
100
E
o
§
E
80
>
K
CO
60
z
LU
Q
CC
1 I 1
40
UJ
§
o
a.
20
i i i iiiiii i
I I Mill I
I I I llll
Y = 24.88 LOG X + 45.82
— r = 0.975
MAN—
— SQUIRREL
MONKEY ^0^
RHESUS
MONKEY
^ RAT
mi i
I I I llll I
I I I llll
0
0.1
1.0 10
BODY WEIGHT, kg
100
Figure 4-3. Power densities at 2450 MHz necessary to raise the rectal tempera-
ture by 1 °C in 60 min for the rat, squirrel monkey, and rhesus
monkey (deLorge 1979).
4-8
-------
2450 MHz. This comparison is probably acceptable because, for the four species
considered, a supraresonant kind of EM interaction takes place at this fre-
quency (see § 3.2). This means that the coupling of RF energy into the animal
is less than optimal (i.e., the coupling is of intermediate efficiency), and
most of the energy is peripherally deposited, although less so for the rat
than for the man. If the same interspecies comparison were to be performed
under conditions in which the RF-radiation coupling is optimal (i.e., the
resonant absorption case), then the results will likely be very different
(i.e., the power density values needed for a 1 °C elevation in rectal tempera-
ture may well be significantly reduced by a factor of 8 or more). Tell and
Harlen (1979) have also pointed out that if a human were exposed to 2450 MHz
at a power density of approximately 90 mW/cm , irreversible local peripheral
tissue damage will likely result without any change in rectal temperature.
When there are supraresonant-type interactions, the absorption coeffi-
cient is roughly constant and is ~ 0.5. Under these conditions, the RF-energy
absorption depends only on the geometric cross-sectional area of the exposed
animal. Assuming a spherical shape of radius r for the animal, it can be
2/3
shown that the cross-sectional area-to-volume ratio is proportional to r
(Kleiber 1961). This means that the area-to-volume ratio as well as the
area-to-mass ratio will decrease as the animal increases in size. Consequently,
the absorbed energy per unit of body mass (i.e., whole-body-averaged specific
absorption rate, SAR) decreases with increasing size. This general rule of
thumb is valid only for interaction conditions of the supraresonant type, and
it is not valid for resonant- and subresonant-type conditions.
4-9
-------
SAR at 2450 MHz as a function of body weight with the body axis aligned
with the electric field is shown in Figure 4-4. By taking the threshold power
density for a 1 °C rise in temperature (Figure 4-3) and converting to SAR, it
is predicted that an inverse relationship exists between body weight and SAR
to evoke a 1 °C rise in temperature. For example, a power density of
2
33 mW/cm for a 0.3-kg rat translates to an SAR of 7.9 W/kg; whereas, for a
70-kg man, a power density to evoke the same 1 °C rise in temperature,
92 mW/cm^, translates to an SAR of 2.1 W/kg.
10H
LOG Y= -0.432 log X-0 844
r=0 987
in
CM
••
••
CO
001
1000
100
0 01
Body Weight (kg)
Figure 4.4 SAR as a function of body weight.
4-10
-------
4.1.4 Local Thermal Responses: Effect of RF-Radiation Exposure on Blood Flow
Interest during the early 1940's in using short-wave diathermy (13 to
43 MHz) as a therapeutic agent prompted research into the effect of RF radiation
in producing localized changes in peripheral blood flow. Near-field diathermic
application with capacitance plates (SAR not determined) was shown to produce
a more than twofold increase in blood flow of an exposed limb of a dog (Wakim
133
et aK 1948) and of a human (Abramson et aK 1957). Using the Xe clearance
technique, McNiven and Wyner (1976) found that localized exposure to 2450 MHz
caused a nearly fourfold increase in blood flow of the vastis lateralis muscle
in humans. Although it was not possible to determine the SAR, this study is
important because it contains some of the few data collected concerning RF
effects on blood flow in humans.
RF-radiation-induced increases in peripheral blood flow can be attributed
to a direct effect of heat on the caliber of arterioles, or to an indirect,
neurally induced, vasodilation via the activation of peripheral and deep-body
thermal receptors. The increase in femoral blood flow during RF-radiation
exposure (2450 MHz and 27.3 MHz) is similar in dogs with intact and denervated
limbs (Siems et aK 1948). Hence, RF radiation can directly affect peripheral
vascular resistance; however, the fact that this response occurs in the
denervated limb does not preclude a neural response to RF radiation under nor-
mal circumstances.
Adair and Adams (1980a) found that RF-radiation exposure of the squirrel
2
monkey to 2450 MHz at 8 to 10 mW/cm (SAR=1.5 W/kg) promoted vasodilation in
4-11
-------
the tail and foot without any change in rectal temperature. Furthermore, it
was shown that an equivalent power density of IR heat, which has a much shorter
wavelength and is absorbed on the skin surface, was ineffective in promoting
vasodilation. Exposing the squirrel monkey to RF radiation promotes vasomotor
responses similar to direct heating of the preoptic anterior hypothalamus, an
extremely thermal-sensitive area of the brainstem considered an integrative
center for the control of body temperature. Hence, it was concluded that
RF-induced vasodilation was caused by the activation of warmth-sensitive
neural sites in and/or outside the CNS, which in turn affect the central
neural control of heat-dissipating motor outputs, including the dilation of
the peripheral vasculature.
4.1.5 Whole-Body Thermal Response to RF-Radiation Exposure and Dependence on
Ambient Conditions
Figure 4-5 displays an idealistic metabolic response of a homeotherm to
changes in ambient temperature. The range of ambient temperature where metabolic
rate remains at a stable and minimal level is defined as the thermoneutral
zone. As ambient temperature decreases in the thermoneutral range, the rate
of heat loss increases, and the homeotherm responds with an increased peripheral
vasoconstriction. If ambient temperature decreases to a point where the
animal's physical mechanisms for restricting heat flow are maximized, then
metabolic heat production (mostly from shivering) is elevated to match the
heat loss. The upper critical temperature defines the upper limit of the
thermoneutral zone where metabolism starts to rise with increasing ambient
temperature. Heat stress increases metabolism directly by raising the
4-12
-------
FAILURE OF
TEMPERATURE
REGULATION
<
cc
y 2
_i
O
CO
<
I- 1
"It
FAILURE OF
TEMPERATURE
REGULATION
1~k 140
RANGE OF NORMOTHERMIA
BO DY_J EMP ERATUR^E ~
LOWER UPPER
CRITICAL CRITICAL
TEMP TEMP
THERMONEUTRAL ZONE
38 o
36
CC
D
I-
<
a.
34!
>
o
32 O
m
10 15 20 25 30
AMBIENT TEMPERATURE, °C
35
40
Figure 4-5. Idealistic response of a homeotherm's metabolism and body temperature
to changes in ambient temperature. The upper and lower critical
temperatures greatly depend on the animal's size and adaptability
to warm or cold environments. When the limits of normothermia are
exceeded, death from hypo- or hyperthermia is imminent.
temperature of peripheral tissue. A 10 °C increase in tissue temperature
results in an approximate doubling of the oxygen consumption rate. (See
discussion above.) Heat stress also raises the oxygen requirement by evoking
an increase in ventilation. Upper and lower critical temperature and profile
of metabolism (i.e., shape) vary tremendously with species.
When a homeotherm is exposed to a thermal stress, whether it be an elevated
ambient temperature or an RF-radiation exposure, if the heat gain from the
environment plus basal metabolic heat exceeds the animal's capacity for heat
dissipation, there will be an increase in body temperature.
4-13
-------
Above the lower critical temperature, homeotherms rely on evaporative
water loss (EWL) to dissipate excess heat. A major route of EWL is the respi-
ratory passages where relatively dry air is saturated with water and exhaled
with the amount of heat loss equal to the quantity of water evaporated multi-
plied by the heat of vaporization (580 cal/g). The major route of EWL in
humans is sweat vaporization from the skin. The effectiveness of this mech-
anism depends upon relative humidity and ambient temperature. Increasing the
relative humidity of the ambient air lowers the rate of evaporization.
Since a homeotherm's metabolic rate is dependent on ambient temperature,
the response to RF radiation also depends on environmental temperature. For
example, in mice held at 24 °C (lower critical temperature = 29 to 30 °C), ex-
posure to 2450 MHz at 20 to 30 W/kg results in a depression in metabolic rate
(Ho and Edwards 1977a). Apparently, when metabolic rate is elevated above
basal conditions during cold exposure, the heat absorbed during RF-radiation
exposure can be substituted for metabolic heat, and therefore results in a
lowering of metabolism.
The pre-exposure heat balance of the animal should be considered when
measuring thermoregulatory responses to RF radiation. Ho and McManaway (1977)
used direct calorimetry to measure the heat dissipation of mice before and
after 30 min of exposure to 2450 MHz at 24 °C ambient temperature. The mice
could absorb from 0 to 25 W/kg with no detectable rise in heat dissipation.
Above 25 W/kg there was a clear increase in the rate of heat loss. At 20 °C
ambient temperature, evaporative heat loss (EHL) in mice is stable between
SAR's of 0 and 29 W/kg. Above 29 W/kg, EHL increases with SAR with a slope of
4-14
-------
0.6 (i.e., above threshold, 60 percent of the absorbed RF-heat load is
dissipated by evaporation; Figure 4-6a). It is expected that the threshold
for EHL activation is reduced at higher temperatures. For example, various
exposure durations of 2450 MHz at 49 W/kg elevate EHL but only after a distinct
threshold is exceeded. Furthermore, the percent of the heat load dissipated
by evaporation increases with ambient temperature (Figure 4-6b).
When a homeotherm is pushed into a positive heat balance (e.g., heat gain
from metabolism and environment exceeds heat loss) during RF-radiation exposure,
there are three distinct phases that describe the change in deep-body
temperature. Michaelson et al_. (1961) exposed unanesthetized dogs to 2800-MHz
pulsed-wave (PW) radiation at 100 to 165 mW/cm while recording changes in
rectal temperature. During the first 25 min of exposure, there is a 1 °C
increase in deep-body temperature, followed by a stabilized period of hyper-
thermia lasting 30 to 40 min. A similar body-temperature rise followed by a
stabilization period has been observed in the rhesus monkey (Krupp 1977). If
exposure continues, the mechanisms of heat dissipation apparently become
exhausted, and body temperature rises to dangerously high levels. Sustained
body temperatures of 42 to 45 °C are considered lethal. The maintenance of
adequate body-water levels is critical in regulating stable body temperature
during prolonged RF-radiation exposure. The body temperature of dehydrated
animals will increase much more than that of hydrated ones during such
exposure (Michaelson et al^. 1961).
In addition to ambient temperature and body-water levels, the relative
humidity is another critical determinant of an animal's body-temperature
4-15
-------
-pi
I—»
cr>
04
? 03
O
c
0 1
1 1 1 1 1 1
1 1 1
1 1
i i i | 1 1 1
i 1 i i
•
— Ta = 20 °C
—
/ • —-
• /
•
/ •
—
/ •
—
•
• /
•/
•
/ •
•
/ •
-L \
•• •
1 i i i i 1
i i i
•
1 1 l
' £' '
//
i/i i i i i i
, I , ,
16
<
>
-o
O
u
>
H
<
m
8 I
m
>
10 20 30
SPECIFIC ABSORPTION RATE, W/kg
Figure 4-6a. Relationship between SAR at 2450 MHz
and EHL in male mice maintained at
20 °C ambient temperature (Gordon
1982a).
>40
• EVAPORATION
¦ THRESHOLD
120
£ 40
AMBIENT TEMPERATURE °C
Figure 4-6b. Effect of ambient
temperature on the
sensitivity of EHL
in mice exposed to
various RF heat loads
at 2450 MHz. Note that
threshold is inversely
related, and the sensi-
tivity (percent heat
evaporated) is directly
related to ambient
temperature.
-------
response to RF-radiation exposure. A temperature-humidity index (THI) has
been developed that defines ambient conditions important to heat balance
during RF-radiation exposure (Mumford 1969; Rugh et al^ 1974). The THI is
calculated from the following equation:
THI = 1.41T + 0.1RH + 30.6
where T = ambient temperature (°C)
RH = relative humidity (percent)
Using THI1s between 50 and 95, Rugh et aK (1973) found that the average
lethal dose (J/g) for male mice decreased linearly with an increasing THI
(Figure 4-7). Raising the THI decreases the rate of heat loss in two ways:
(1) The body temperature-ambient temperature gradient is reduced, decreasing
the rate of heat loss by convection and radiation; and (2) the elevated relative
humidity reduces the vaporization of water from the respiratory passages.
Hence, the animal's net heat balance increases with increasing THI, thereby
reducing the lethal dose for RF-radiation exposure.
4.1.6 Heat Stress and the General Adaptation Syndrome
Since the majority of biological effects from exposure to RF radiation
can be attributed directly or indirectly to tissue heating, a brief discussion
on the health effects of heat stress is necessary. Exposing a mammal to heat
stress, either by elevating ambient temperature or by local heating of the
4-17
-------
hypothalamus (Gale 1973), leads to an activation of the hypothalamic-pituitary-
adrenal-cortex axis with a resulting elevation in blood glucocorticoids. This
100
0)
CO
Ik
50
LU
100
50
70
90
60
80
TEMPERATURE-HUMIDITY INDEX (THI)
Figure 4-7. Effect of an increasing THI on the lethal dose of RF radiation
(2450 MHz) in mice (Rugh et a±. 1973).
response is part of a series of physiological events known as the General
Adaptation Syndrome, since it occurs with non-heat-related stress as well.
Increased glucocorticoids (e.g., Cortisol and cortisone) in the blood can
/
promote an array of physiological and pathophysiological changes (Harper 1975)
such as those listed:
4-18
-------
e Elevation in blood levels of glucose, fatty acids, and amino
ac i ds
e Reduction in muscle protein synthesis
• Reduction in the inflammatory response
9 Suppression of immune system
e Increased gastric acidity
« Osteoporosis (weakening of bones)
o Elevated blood pressure
Not all of these physiological changes have been shown to occur with heat
stress; however, any one of the above symptoms could be manifested, since heat
stress can evoke release of the glucocorticoids.
4.1.7 Thermal Physiology of Humans
In view of the paucity of studies on human physiology during RF-radiation
exposure, this report has centered primarily on the responses of laboratory
mammals. One must be cautious in relating measured biological effects in
experimental mammals to man, however, because the thermal physiology of humans
and other mammals may differ substantially. The following discussion briefly
summarizes the major physiological responses of humans to heat stress.
A major difference between humans and most experimental mammals is the
ability to secrete sweat on the skin. Under resting, thermoneutral conditions
there is a certain amount of insensible EWL through the skin, which accounts
for about 8 W of heat. Approximately 25 percent of the total metabolic
4-19
-------
heat is lost through evaporation at rest. With thermal stimulation the amount
of heat lost by evaporation increases tremendously. The partitioning of heat
loss at various ambient temperatures in humans is listed in Table 4-1.
TABLE 4-1. PARTITIONING OF HEAT LOSS IN HUMANS
AS A FUNCTION OF AMBIENT TEMPERATURE*
Ambient
Temperature
(°C)
IR
(%)
Convection
(%)
Evaporation
25
67
10
23
30
41
33
26
35
4
6
90
*Folk 1974.
Sweating from the skin of man and other primates can be evoked by heating
the skin or thermally sensitive sites in the nervous system (Figure 4-8a). It
is interesting to compare the relatively high threshold of sweating at low vs.
high skin temperatures (Figure 4-8b). This comparison may be relevant to some
frequencies of RF radiation that heat predominantly the inner tissues and
organs without warming the skin. Such latencies may affect the organism's
ability to respond to RF heating.
It is important to understand the thermoregulatory responses of humans
during exercise, since it represents one of the few cases where internal tem-
perature can increase without rising ambient temperature and, in this respect,
is similar to that of RF heating. Humans are well adapted to dissipate excess
heat loads from exercise. Examples of thermoregulatory responses during two
4-20
-------
-42
0 85-
-40
RH 20%
0 65-
-38
Right
Forearm
-36
0 45-
< 0 35-
Heating
t 0 25-
0 15-
Left/
Forearm
TIME, mm
Figure 4-8a. Effect of increasing skin temperature on
sweating from the forearms of a human
(Elizondo 1973).
400-
300-
Skin Temperatures, "C
39,38,37,36,35.34,33
U
2
I-
<
LU
5
200-
100-
33 32 31 30 29
Skin Temperatures, "C
364 36 8 37 2 37 6
CRANIAL INTERNAL TEMPERATURE, °C
Figure 4-8b. Effect of skin temperature on the
threshold internal cranial temperature
for activation of sweating in humans
(Benzinger 1969).
4-21
-------
levels of exercise are shown in Figure 4-9. A work load of 40 W causes an
increase in metabolism from 1.5 to 4.2 W/kg and a concomitant 0.34 °C rise in
tympanic temperature; the temperature of the tympanic membrane in the ear is
considered to be a good indicator of the internal temperature of the head.
A 90-W work load increased metabolism to 7.0 W/kg and raised tympanic
temperature by 0.96 °C.
-1.0
Ambient Temperature = 30°C
o>
C/5
CO
O
_l
Metabolism
< 4-
LU
X
o
-0.6 z
2
<
-0.4 -o
m
to
Evaporation
_i
O
00
<
I-
-0.2 a
m
LU
Temperature
-0
Recovery
Work at 90 W
Work at 40 W
Rest
120
140
20
60
TIME, min
100
Figure 4-9. Effect of exercise on heat loss, metabolic rate, and tympanic
temperature of humans (Chappuis et al. 1976).
4-22
-------
It is interesting that the rise in metabolism from peddling the ergometer
and the heat load normalized to time and weight from RF-radiation exposure
have the same dimensions (W/kg). Of course, the two forms of heat cannot be
considered to evoke the same physiologic effects. However, measuring the
rise in body temperature, evaporation, and other physiological parameters
during an exercise-induced heat load may aid our understanding of human response
to RF radiation. It is clear from the data in Figure 4-9 that healthy humans
can tolerate metabolic heat loads of 7.0 W/kg without significantly stressing
the thermoregulatory system. However, it remains to be shown if, under similar
circumstances, humans can tolerate RF- heat loads of the same magnitude.
There are numerous reports on the effects of chronic and acute heat
stress on man. A few of the symptoms that occur with extreme heat stress
(Folk 1974) are listed below:
• Initial increase in plasma volume as fluid shifts from the
interstitial spaces into the vasculature. Several days of
heating increases the number of red blood cells. Hence, blood
viscosity can be elevated, which in turn increases the work load
on the heart.
• Initial exposure to heat results in a transient hypoglycemia.
• Food consumption is initially elevated, but anorexia is associ-
ated with prolonged heating. Spontaneous activity is also
reduced with heat stress.
• As the heat stress progresses, dehydration will occur along with
an elevation in body temperature.
• Pulse rate is increased.
• Urine volumes are reduced, and the urine is more concentrated.
4-23
-------
Ellis et ah (1960) have provided the following criteria for the termi-
nation of heat exposure in humans during intermittent periods of work:
« A pulse rate of 160 beats/min after working in the heat
• A pulse rate of 140 beats/min before starting work
• A rectal temperature of 39.2 °C or higher
0 A rising pulse rate during the rest period
« Complaints of unpleasant symptoms (e.g., cramps or faintness)
4.1.8 Thermoregulatory State and Physiological Responsiveness to RF Radiation
A recent investigation by Smialowicz et al. (1982) has employed a unique
form of cold stress as an assay for detecting the physiologic effects of low
level RF-radiation exposure. Smialowicz et ah found that mice injected
intraperitoneally with 5-hydroxytryptamine (5-HT) became hypothermic, the
decrease in body temperature being inversely related to ambient temperature
(Figure 4-10). The 5-HT-induced hypothermia is attenuated in a dose-related
2
response by exposure to 2450-MHz radiation at 1 to 10 mW/cm (Figure 4-10).
The linear regression of colonic temperature after 5-HT vs. ambient tempera-
ture calculates to a slope of 0.30 °C colonic temperature/°C ambient
temperature. Plotting the 5-HT hypothermia against RF-radiation power density
calculates to a slope of 0.15 °C colonic temperature/1.0 mW/cm . Dividing the
ambient temperature response by the RF-radiation response we obtain:
4-24
-------
38
36
34
o
o
lu 32
CC
D
<
CC
30
1
1 1
1 1 1 1
1 6
—
°^
o
—
o
§
—
'V '
ooo
\
©
o
—
©
1
1 1
1 1 1 1
1
16
20 24 28 32
AMBIENT TEMPERATURE, OC
34 a
§
2 4 6 8
RF RADIATION POWER DENSITY, mW/cm2
b
Figure 4-10.
(a)
(b)
Effects of 5-HT injections on mice (Smialowicz et al. 1980a).
Linear regression of colonic temperature after an intraperitoneal
injection of 5-HT at various ambient temperatures. The drop in
body temperature increases with decreasing ambient temperature.
Linear regression of 5-HT-induced hypothermia for various power
densities at 2450 MHz. Similar to ambient temperature, the mag-
nitude of hypothermia is less with an increasing power density.
4-25
-------
q 20 °C colonic temp
1.0 °C ambient temp
= 2 mW/cm2/°C
i 25 °c colonic temp
1.0 mW/cm2
2
That is, a 2-mW/cm increase in the power density of RF-radiation exposure
is equivalent to a 1 °C increase in ambient temperature. This is one of the
very first attempts to equate ambient temperature and RF-radiation exposure.
Most importantly, this quantitative comparison had to be performed in cold-
stressed animals.
4.1.9 RF-Radiation Exposure and Behavioral Temperature Regulation
Adair (1980b) trained squirrel monkeys to regulate their ambient tempera-
ture by controlling the influx of warm or cold air into an environmental
chamber. Under normal conditions the monkeys selected an ambient temperature
of 34 to 36 °C. The selected ambient temperature significantly decreases when
2
the animals are exposed to an incident power density of 6 to 8 mW/cm (SAR =
1.1 W/kg). IR radiation of equal power density elicited much smaller changes
in the selected ambient temperature. It was concluded that RF-radiation-
induced absorption activates thermoregulatory behavior by stimulating internal
thermal receptors.
4-26
-------
Stern et aL (1979) trained cold-exposed rats to press a bar for heat
from an IR lamp. When exposed to 2450-MHz radiation at a power density as low
o
as 5 mW/cm (SAR = 1 W/kg), the rate of bar pressing for heat decreased.
Furthermore, the rate of bar pressing is inversely related with the RF-
radiation power density. The power density levels that elicited changes in
the thermoregulatory behavior had no effect on deep-body temperature.
Whiptail lizards will move into a microwave field of 93 mW/cm at 2450 MHz to
maintain their body temperature above ambient temperature (D'Andrea et aK
1977). However, the same animals will maintain an even higher body
temperature when allowed to move under an IR lamp.
4.1.10 Unresolved Questions
There are two principal schools of thought in physiologic research:
(1) the reductional approach and (2) the holistic approach. Reductionalism
determines many properties of an organism by closely examining individual
cells, tissues, or organs under i_n vitro conditions, albeit most of the data
must be greatly modified or discounted for predicting activity of the cellular
elements when they are incorporated into the whole animal. For example, in
the field of thermal physiology it is well known that isolated tissues tolerate
higher temperatures than organs, which in turn tolerate higher temperatures
than the whole organism. The holistic approach to physiology research relies
on the classic concept that the whole is greater (or occasionally lesser) than
the sum of its parts. These philosophic principles must be considered in an
assessment that relies on animal experiments to determine the effects of
RF radiation on man.
4-27
-------
It has been argued, since rectal temperature may rise 2 to 3 °C with a
fever or strenuous exercise, that a 1 °C rise in temperature during RF-
radiation exposure can certainly not be unhealthy (Tell and Harlen 1979).
This is an example of a break between the reductionalist and the holist; the
former attempts to equate two entirely different biologic perturbations (fever
or exercise and RF-radiation exposure) that evoke similar effects (increase in
body temperature), whereas the latter examines each of the two treatments as
unique to the whole organism. During a fever there is a perturbation in the
thermoregulatory center in the hypothalamus, causing the subject to vasocon-
strict and shiver violently, resulting in a regulated increase in body tempera-
ture. During exercise humans and other mammals undergo an initial period of
peripheral vasoconstriction. Also, during exercise the thermoregulatory
system's sensitivity is increased in anticipation of an elevated thermal load.
On the other hand, with RF-radiation exposure, body temperature will increase
above the subject's preferred body temperature. The subject responds by
increasing the activity of thermoregulatory mechanisms of heat dissipation
(see Figure 4-1). By comparison, with fever, the subjects increase their
preferred body temperature instead of preventing the rise. Hence, equivalent
rises in body temperature from fever vs. RF-radiation exposure differ greatly
from a physiologic standpoint. In the former, the system is regulating toward
an increase in temperature, whereas in the latter case the system is
regulating against an increase. Clearly, internal heating of normal subjects
with RF radiation is unusual for the whole organism. The increase in body
temperature is not regulated, the body temperature is forced to a higher
level, while thermal dissipating motor outputs (Figure 4-1) are activated to
counteract the heat load.
4-28
-------
From these basic principles of physiologic research it is apparent that
information is lacking in the following areas:
e The equivalence of heat loads from normal ambient heating and RF
heating. The study by Smialowicz et a^. (1980) has scratched the
surface of this problem (Figure 4-9}. We know from this study
that, for the mouse exposed to 2450 MHz, 2 mW/cm2 is proportional
to a 1 °C increase in ambient temperature. It is important to
know what this relationship is for larger mammals. Then it would
be possible to determine an equivalence between ambient tempera-
ture and RF radiation exposure for humans.
o The manner in which neural elements of the thermoregulatory
system respond to RF heating. It would be interesting to con-
trast the neural responses during RF radiation exposure and
normal ambient heating.
© Measurement of the SAR of RF radiation while mildly perturbing a
physiologic system (Tables 4-2 and 4-3). Few of these study data
are available; furthermore, there are few data from studies using
identical SAR's for normal heating and RF heating while measuring
the magnitude of the physiologic response.
4-29
-------
TABLE 4-2. SUMMARY OF STUDIES CONCERNING RF-RADIATION EFFECTS ON THERMOREGULATION
Effects
Species
Frequency
(KHz)
Exposure Conditions
Intensity
(mW/cra2)
Duration
(total mm)
SAR
(W/kg)
Reference
•C-
i
o
Inhibits metabolic rate without
changing rectal temperature
Stimulates heat dissipation in
mice
Vasodilation and hyperthermia
Increases core and skin
temperature
Behaviorally selects a
cooler ambient temperature
Male and female mice of varying
size have a lethal dose of
approxinately 40 J/g
This dose level along with
electric foot shock is
synergistic in promoting
increases in core temperature
Increases temperature of heart-
liver area
House
House
Dog
Rhesus
monkey
House
Rat
House
2450 (CW)
2450 (CW)
2800 (PW)
26
Squirrel 2450 (CW)
monkey
2450
2450 (PW)
NA
NA
165
500, 750,
or 1000
6-8
30
30
120 to 180
360
10
3 to 8
1
1 6, 5 5,
10 4, 23 6,
or 44 2
24 6, 30 7,
38 8, or 40 1
1-2
~ 80
2 and 10
2450
260
30 mW/g
Ho and Edwards (1977a)
Ho and HcHanaway (1977)
Hichaelson et a^ (1961)
Frazer et a^ (1976)
Adair and Adams (1980b)
Rugh (1976b)
Justesen et a^ (1974)
Rotkovska et al (1973)
(continued)
-------
TABLE 4-2. (continued)
Exposure Conditions
Effects
Species
Frequency
(MHz)
Intensity
(mW/cm2)
Duration
(total mm)
SAR
(W/kg)
Reference
Core temperature rises rapidly
initially then stabilizes
Reduces the rate of bar-pressing
for heat when exposed to a cold
environment
Dose-related increase in core
temperature
Increase in body temperature
and decrease in serun cortico-
steroids
Optimal frequency for raising
core temperature
Increase whole-body evaporative
heat loss
Rhesus
monkey
Rat
Rat
Rat
Rat
Mouse
15, 20,
or 25
2450 (CW)
2450 (CW)
2450 (CW)
600 (PW
and CW)
2450 (CW)
760-1260
13-60
20
20
NA
180
15
120
460
55
90
NG
10
2-9 6
5
5
29
Krupp (1977)
Stern et a^ (1979)
Lotz and Michaelson (1978)
Lu et al 1977
D'Andrea et a^ (1977)
Gordon (1982a)
*
,„NA = Not applicable, because exposures were made in a waveguide system
NG = Not given
-------
TABLE 4-3. SUMMARY OF STUDIES CONCERNING RF-RADIATION EFFECTS ON BLOOD FLOW
Exposure Conditions
Effects
Species Frequency
(MHz)
Intensity Duration
(rft//cm2) (total mm)
SAR
(W/kg)
Reference
Increases femoral blood flow
Diathermy more than doubles
femoral blood flow
Causes rapid increase in heart
apex during ventral exposure
Increases tail and foot
temperature with no change
in deep body temperature
Promotes vasodilation in
vessels of skeletal muscle
Increase in blood flow of
extremities
Increases femoral blood flow
along with muscle, subcutaneous,
and skin temperature
Dog 2450 and
27 (CV)
Dog 27
Dog 2450 (CW)
Squirrel 2450 (CW)
monkey
Human 27
Human 2450 (CW)
Dog 3000
NA
NA
NG
8-10
NG
NG
NG
20
15
15 to 140
20 to 34
1 to 30
15
NG
NG
NG
1 5
NG
NG
NG
Siems et a^ (1948)
Wakim et a^ (1948)
Harks et aj (1961)
Adair and Adams (1980a)
Abramson et al (1957)
Gersten et a2 (1949)
Leden et al (1947)
*NA = Not applicable
**NG = Not given
-------
4.2 NUMERICAL MODELING OF THERMOREGULATORY SYSTEMS IN MAN AND ANIMALS
Ronald J. Spiegel
4.2.1 Heat-Transfer Models
It is important to develop accurate mathematical models to predict the
absorption of electromagnetic energy and the thermal response of biological
subjects. Mathematical modeling can provide a method for gaining qualitative
and quantitative information that may be valuable in explaining and under-
standing many of the effects measured in experimental animals. More im-
portantly, experimentation on human subjects is often unethical. Thus, to
extrapolate RF-field results in experimental animals to equivalent effects in
man, one must know thoroughly the dosimetry for both species. This can be
achieved by the application of a sophisticated and realistic numerical simu-
lation of the energy absorption and subsequent thermal response of subjects
exposed to RF fields.
Several investigators have developed mathematical models to calculate the
thermal response of the human body when subjected to different environmental
conditions or levels of exercise. The same basic approach can be taken to
develop a model to simulate the effect of RF radiation. This model, however,
must take into account that RF fields deposit energy nonuniformly in the
human. The simple one-dimensional heat transfer models used in the past will
not accurately simulate this condition. A short review of these previous
4-33
-------
attempts to model man in one dimension, however, will provide good background
for developing a more general model.
One of the earliest attempts at modeling the human body was made by
Pennes (1948), who developed a cylindrical model of a human limb. This model
was first used to simulate the human forearm but was later applied to the
general case of limbs. The following factors were included in this model:
(1) radial conduction, (2) metabolic heat generation, (3) convection to the
blood, and (4) environmental exchange by convection, radiation, and evaporation.
Machle and Hatch (1947) introduced the concept of a core-and-shel1 model
by comparing measured values of rectal and skin temperatures representing the
core and shell temperatures used in the model. Empirical correlations for
radiation, convection, and evaporation were experimentally developed for in-
clusion in this model. A modification by Kerslake and Waddell (1958) extended
the model to include the case of complete skin wetness due to sweating.
Wyndham and Atkins (1960) further extended the core-and-shel1 model by
introducing several concentric cylinders representing the different body
layers. This model used a finite difference technique to solve a set of
resulting first-order differential equations by an analog computer.
A completely rigorous analytical approach was taken by Hardy (1949), who
applied the laws of thermodynamics and heat transfer to the human system.
This analysis included radiant exchange with the environment, thermal
4-34
-------
conduction through concentric cylinders, both natural and forced convection,
and evaporation from the skin and lungs. The analysis was accompanied by
experimental verification of several of the calculated responses.
Wissler (1961, 1964) modified the models of Pennes and Wyndham and Atkins
and combined them to obtain a model of the entire human body. This model sub-
divided the body into six elements: head, torso, two arms, and two legs.
Each of these elements was assumed to have the following: (1) a uniformly
distributed metabolic heat generation, (2) a uniformly distributed blood
supply, (3) a composition of homogeneous materials, and (4) a geometry of
isotropic cylinders. The effects of heat loss through the respiratory system
and countercurrent heat exchange between the arteries and veins were also
included.
In contrast to the models based on concentric cylinders, Crosbie et a]_.
(1963) used an infinite slab model to represent an element of the body. These
investigators argued that, by applying this geometry, the physics and
physiology of the human system could be better understood and that many
uncertainties in previous analyses could be clarified. This model was pro-
grammed on an analog computer and verified by experimental observation.
Smith and James (1964) developed another analog model to study thermal
stress in man. This model had the following characteristics: (1) metabolic
heat production in the working muscles, (2) muscles insulated by a layer of
subcutaneous tissue, (3) blood flow from the muscles to the skin, and
(4) blood flow between different elements of the body and heart. This model
4-35
-------
considered radial conduction through three concentric cylinders and counter-
current exchange between the arteries and veins, and it was verified experi-
mental ly.
The next major effort at modeling the entire human body was made by
Stolwijk and associates at the John B. Pierce Foundation Laboratory. The
initial effort by Stolwijk and Hardy (1966) was a model composed of three
cylindrical segments, one each for the head, trunk, and extremities. The
trunk was divided into three concentric layers: skin, muscle, and core. The
head and extremities were divided into two concentric layers: skin and core.
In this work, the concept of the body being composed of a controlled system
and a controlling system was suggested. These investigators also did a
rigorous review to determine accurate thermal properties of the constituents
of the human body. This model was then programmed for analysis by an analog
computer and compared to experimentally developed parameters.
The 1966 Stolwijk-Hardy model was expanded (Stolwijk and Cunningham 1968;
Stolwijk 1969, 1971; Stolwijk and Hardy 1977) to include six segments: head,
trunk, arms, hands, legs, and feet. All of these segments were composed of
four layers: skin, fat, muscle, and core. The geometry of each was cylindri-
cal except for the head, which was spherical. This model was programmed for
analysis by a digital computer and included high metabolic heat production,
sweating, blood flow to all layers, and convective and radiant exchange with
the environment. Stolwijk's model is the most comprehensive program to date,
4-36
-------
and several investigators have adapted it to special cases; e.g., Montgomery
(1972; 1974a,b; 1975) used the model to study the effects of man immersed in
water.
4.2.2 RF-Radiation/Heat-Transfer Models
As discussed earlier, there exist reasonably realistic RF-energy depo-
sition models (see § 3.2.3, Analytical and Numerical RF Electromagnetic In-
teraction Models) and heat transfer models (see above discussion) for the
human body. However, few attempts have been made to combine the two models to
predict the body's thermal response under exposure to RF fields. In one
/
study, Emery et aK (1976) determined the thermal effects of a uniform depo-
sition of RF energy for a one-dimensional model of heat conduction. Although
this model may yield realistic values for whole-body temperatures, the heating
pattern produced by nonuniform deposition of energy may deviate substantially
from that produced by uniform absorption; this deviation may occur if various
parts of the body (head, arms, legs, etc.) selectively absorb energy from the
incident field because of whole- and partial-body resonance. In another
study, Guy et aj. (1978) used thermographic determinations of the distribution
of RF energy in phantom models of man, which were subsequently used to provide
input for Emery's one-dimensional thermal model. While this method certainly
accounted for the nonuniformity in the RF-energy deposition, it apparently did
not account for heat flow along the major axis of the body. This is important,
because the primary nonuniformity in the RF-energy deposition will occur along
the body's major axis, since the body is much longer than it is thick. Thus,
this model will tend to overestimate the temperature profile in the body
4-37
-------
because the model allows heat flow to occur only from the core to the skin.
In reality, as a result of localized RF-energy deposition, heat flow must also
occur along the major length of the body.
The most general attempt to date has been the work of Spiegel et aK
(1980a), who have used block models to calculate the RF-energy deposition in
the body with a two-dimensional extension of Stolwijk's model to determine the
resulting thermal response of the body. This model allows heat flow from the
core to the skin as well as along the major axis of the body. The authors
used a transient heat conduction model with internal heat generation and heat
dissipation. The internal heat generation is caused by metabolism and
absorption of RF energy. The internal dissipation is caused by convective
exchange with the cardiovascular system and a combined convective and radiant
exchange with the surrounding environment at the surface of the skin. The
equation simulating the response is, therefore,
pc |I = V (kVT) + i (QEM + QM - QE - QR) (4-1)
where p = tissue density
c = tissue specific heat
T = local tissue temperature
t = time
V = tissue volume
k = tissue thermal conductivity
QEM = electromagnetic energy deposition
4-38
-------
QM = metabolic heat generation
QE = evaporative heat dissipation in the skin
QR = respiratory heat loss in the lungs
To solve Equation 4-1, the body is divided into several finite elements
and this relationship is applied to each element or node. The body is repre-
sented by 15 segments, with each segment subdivided into 4 concentric layers —
core, muscle, fat, and skin. The head is modeled by a sphere; the neck,
hands, and feet are approximated as single cylindrical segments, the arms and
legs are each divided into four cylindrical segments, and the trunk is divided
into three cylindrical segments. The radius and length of each of these
cylindrical segments are based on dimensions for a standard man (Diffrient et
aL 1974). Calculations of heat capacitance, thermal conductance, and density
are based on the type of tissue, the surface area, and the volume for each
segment and layer.
In this model, the time and spatial derivatives are represented by finite
difference approximations, and the resulting system of equations is solved by
an iterative procedure in which the initial temperatures are used to compute
the temperatures a short time later. These new temperatures are then used to
compute the temperatures at the new time and so on until thermal steady-state
conditions are reached.
Figure 4-11 shows the relationship of the various thermal conductances
for a typical segment that includes four nodes in the finite element analysis.
4-39
-------
OEM
OEM
QEI\»
OEM
QR
QM
KA
QM
KA
QE
QM
KA
QM
KA
KC
KC
KC
CORE
MUSCLE
-vw-
FAT
¦M/V"
SKIN
KA
KA
BF
BF
BF
" i "I I
KA
CENTRAL
BLOOD
POOL
Figure 4-11. Block diagram for one segment of the thermal model.
Each lumped conductance represents heat exchange with the other nodes. The
quantities KC and KA represent the radial and axial conductances, respectively,
the values of which are determined by layer and segment geometry and by tissue
thermal conductivity (Carslaw and Jaeger 1959). Heat exchange between the
skin and surrounding environment by convection and radiation is represented by
the quantity H, and it is determined in the standard engineering manner as
described by Stolwijk (1971). The BF terms designate the amount of heat
exchanged by each node and the central blood pool, and they are a function of
blood flow. Blood flow rates to all segments, except the skin, are set to
basal values (Stolwijk 1971). The skin blood flow rate is controlled by
vasodilatation, which is a function of the temperature difference between the
local skin temperature and the skin set-point temperature, as well as the
difference in temperature between the hypothalamic temperature and its
set-point temperature. It is believed that all tissues respond to local
4-40
-------
temperatures in excess of 40 °C by increasing blood flow. However, the lit-
erature does not appear to provide enough information to include this response
for all 100 compartments of the model.
The other heat exchange modes shown in Figure 4-11 include metabolic heat
production, QM; heat loss by sweat evaporation on the skin, QE; respiratory
heat loss in the lungs, QR; and electromagnetic heat input, QEM. The respira-
tory loss is determined according to Stolwijk, and the model assumes that the
expired air has come to thermal equilibrium with the upper trunk core tempera-
ture and is saturated. This term enters the equation for thermal balance
(Equation 4-1) only at the node that represents the core of the upper trunk
segment. Several mathematical models have been proposed to calculate the heat
lost by sweating (Emery et al_. 1976). All are empirical and attempt to best
fit experimental data for various conditions, such as for a sedentary subject
or for various levels of exercise. This model is based on the detailed work
of Stolwijk (Stolwijk 1971, Stolwijk and Hardy 1977). Evaporation basically
is controlled by the sweating rate. As with the rate of blood flow, the term
that simulates sweating is a function both of the temperature difference
between the local skin temperature and the skin set-point temperature and of
the difference between the hypothalamic temperature and its set-point tempera-
ture. The body develops its own source of heat as a result of metabolic heat
production, QM. For a sedentary subject, this value can be at the basal level
or can be accelerated by shivering. Shivering is initiated when the body is
subjected to a low ambient temperature (< 28 °C for a nude subject) and
experiences a chill. Since the air temperature for this study is always in
4-41
-------
excess of 28 °C, the effects of shivering and vasoconstriction are not
included in the analysis. The RF energy input term, QEM, is calculated by a
block model comprised of 180 cubical cells of various sizes (Hagmann et al.
1979a).
4.2.3 Numerical Results
To illustrate the model used by Spiegel et a^. (1980a), Table 4-4 shows
the resulting steady-state temperature distribution in a 70-kg, 170-cm-high
human exposed to a plane wave at frequencies of 80 and 200 MHz. The electric
field vector is oriented parallel to the major axis of the body (i.e., E
polarization). Whole-body resonance occurs for the 80-MHz field, and partial-
body resonance occurs in the arms for the 200-MHz field. For the 80-MHz case,
2
an incident power density of 10 mW/cm is used (SAR = 2.25 W/kg). For the
2
200-MHz field, incident power densities are 10 and 32.5 mW/cm (SAR's = 0.58
and 1.9 W/kg, respectively). As mentioned above, the RF-energy deposition
term, QEM, is inserted into Equation 4-1 in a manner prescribed by the
180-cell model.
Note that the distribution of temperatures is quite different in the two
cases. For the 80-MHz field, a thermal hot spot (a temperature of around 41.6
°C) is generated in the lower thigh; for the 200-MHz field the hot spot tends
to occur in the arms. In addition, note that for the 200-MHz case there is an
elevated temperature (40.6 °C) in the neck. Another difference is that the
2
hot spot occurs for an incident power density of 10 mW/cm for the 80-MHz
2
field, while 32.5 mW/cm was required to produce a hot spot for the
4-42
-------
TABLE 4-4. STEADY-STATE TEMPERATURES (°C) IN A HUMAN BODY
AFTER EXPOSURE TO 80- AND 200-MHz RF FIELDS
80 MHz
200
MHz
10 mW/cm2
0
10 mW/cm
32.5 mW/cm2
Segment
No.
Muscle
Core
Muscle
Core
Muscle
Core
Head
1
36.3
37.4
36.1
37.0
37.0
37.2
Neck
2
38.3
38.9
37.7
38.2
40.6
40.6
Upper
trunk
3
37.7
37.3
36.9
36.9
37.6
37.1
Middle
trunk
4
38.8
37.5
37.2
37.0
39.3
37.2
Lower
trunk
5
38.9
37.8
37.1
37.0
38.4
37.2
Upper
humerus
6
34.5
35.9
35.3
36.5
37.0
38.4
Lower
humerus
7
34.6
36.0
36.9
38.0
41.2
42.9
Upper
forearm
8
34.4
35.7
36.3
37.5
39.8
41.4
Lower
forearm
9
34.0
35.0
35.1
36.1
36.9
38.4
Hand
10
36.4
36.6
36.0
36.2
36.4
36.6
Upper
thigh
11
39.5
40.3
35.7
37.0
35.8
37.1
Lower
thigh
12
40.3
41.6
35.5
36.7
35.7
37.0
Upper
calf
13
39.9
41.2
35.7
36.9
36.7
37.9
Lower
cal f
14
38.4
40.0
35.5
36.4
36.7
37.9
Foot
15
36.4
36.8
35.6
35.8
35.9
36.1
4-43
-------
200-MHz case. This occurs because the whole-body SAR is at least four times
greater for the 80-MHz field.
As another useful method to present the data, Figure 4-12 graphically shows
the tifte it takes to obtain a hot spot for a given level of incident power
density level for the two frequency conditions. The production of a hot spot
can be viewed in terms of a strength-duration model; for example, Figure 4-12
shows that the time required to create a hot spot in the lower thigh for an
2 2
80-MHz field takes only a few minutes at 50 mW/cm , but at 10 mW/cm it will
take an hour or more of exposure. In addition, different minimum levels of
power density are required to produce hot spots at various body locations.
For instance, the rectal temperature profile has an asymptote at a level just
2
above 35 mW/cm , but the lower thigh core temperature asymptote occurs around
10 mW/cm . For 200-MHz radiation, the lower humerus core temperature was con-
sidered, and the intensity-duration curve for this case is shown by the dashed
curve.
Based on these results, and as might be expected, the most critical tem-
perature elevations occur at whole-body resonant frequencies. In fact, the rise
in temperature is sufficient to produce a hot spot (41.6 °C) in the legs for an
2
incident power density of 10 mW/cm . Based on the calculated results, it should
again be emphasized that the increased blood flow response for tissue tempera-
tures in excess of 40 °C was not included in the model. Therefore, the model
most-likely overestimates the magnitude of the temperature rise in the thigh.
4-44
-------
100H
—— Exposure at 80 MHz
Exposure at 200 MHz
90-
80-
70-
60-
E 50-
Rectal Temperature
40-
Lower Humerus Core Temperature
30-
20-
Lower Thigh Core Temperature
10-
150
100
200
50
DURATION EXPOSURE,mm
Figure 4-12. Incident power density vs. exposure duration to obtain a hot
spot (41.6 °C).
4.2.4 Unresolved Questions
Although the combined RF-heat-transfer model reported here is realistic
enough to predict gross effects and trends, the model could be further refined.
The most obvious refinement is the extension of the two-dimensional heat
4-45
-------
transfer model to a three-dimensional simulation. While this extension is
numerically straightforward, problems might arise in computer storage require-
ments and in the computational time needed to generate the heating patterns.
In addition, altered tissue blood flow for temperatures in excess of 39 °C has
not been implemented in the model. While this modification would be simple,
its implementation would require experimentation to quantify the altered blood
flow rates.
Another problem is that little is known about the effects of RF energy on
the predominant heat-dissipating mechanisms of sweating and vasodilatation.
While the empirical models yield results that compare well with experimental
measurements from a human subject placed in a room at a controlled temperature
and relative humidity, it is certainly open to question whether good agreement
between measured and calculated results would likewise be obtained when the
thermal load is caused by an RF field. This problem could be resolved by
verifying the model experimentally using animal models.
4-46
------- |