THE DESIGN AND CONSTRUCTION OF A MINIATURE
DOSIMETER FOR THE STUDY OF THE EFFECTS
OF AIR CAVITIES IN RADIATION THERAPY
.E HOPE PATRICIA McKETTY
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
The author wishes to acknowledge with gratitude the help and
guidance given by her chairman and the members of her supervisory
committee in her research and writing of this manuscript. She would
also like to thank Dr. Million of the department of Radiation Therapy
and Dr. Carroll of the department of Nuclear Engineering Sciences for
providing funds to procure some of the equipment used in this research.
The author also wishes to thank the National Fellowships Fund for
assistance in defraying the cost of the typing of this manuscript.
Finally, thanks to Mrs. Janet Eldred for typing the manuscript, to
Mr. Wesley Bolch for his help with the graphics and to Dr. Robert
Luthman for his help with the photography.
TABLE OF CONTENTS
ACKNOWLEDGEMENTS . . . . . . .
LIST OF TABLES . . . . . . . . . . . .
LIST OF FIGURES . . . . . . . . . . . .
ABSTRACT . . . . . . . . . . . . . .
1 INTRODUCTION . . . . . . . .
2 BACKGROUND AND RELATED RESEARCH . . .
Types of Dosimeters . . . . . .
Ionization Chambers . . . . .
Thermoluminescent Dosimeters . . .
Solid State Dosimetry . . . . . .
PN Junctions . . . . . . . .
Use of PN Junctions . . . . . . .
Fabrication of PN Junctions . . . . .
P-I-N Detectors . . . . . . . .
Fabrication of P-I-N Diodes . . . . .
Methods of Detection and Measurement .
Uses of P-I-N Diodes . . . . . .
Problems Encountered in the Use of P-I-N
and PN Junctions . . . . . . .
Air Cavities . . . . . . .. . .
Objectives of Research . . . . . . .
3 THE DESIGN AND CONSTRUCTION OF THE DOSIMETER
AND PERIPHERAL APPARATUS . . . . . . .
Description of Ideal Dosimeter . . . . .
Prototype I . . . . . . . . .
Prototype II . . . . . . . . . .
Peripheral Apparatus . . . . . . . .
. . 9
. . 10
. . 12
. . 15
. . 16
. . 20
. . 21
. . 27
TABLE OF CONTENTS (Continued)
Amplifier Circuit ... ... . . . . . . .. 50
Operational Amplifier . . . . . . ... 50
Feedback Capacitor . . . . . . ... 53
Reset System . . . . . . . . ... .57
Circuit Construction . . . . . . ... 58
Display . . . . .. . . . . . . 59
4 DESCRIPTION AND RESULTS OF TESTS PERFORMED ON
THE DOSIMETERS . ........ . . . . ... 63
Tests on the Dosimeters . . . . . . .. 63
Reproducibility . . . . ...... ... ... 63
Linearity . . . . . . . . ... . 63
Temperature Dependence . . . . . ... 63
Directional Dependence.. . . . . . 71
Construction Parameters Affecting
Response . . . . . . . .... . 74
Energy Dependence . . . . . . . ... 75
Dose Rate Dependence . . . .... ... 76
Tests on Peripheral Equipment . . . . . ... 77
Cable Effect . . . . . . . .... .77
Position of Amplifier . . . . . ... . 78
Gain Settings . .. . . . . . . .... 78
5 DETERMINATION OF THE EFFECT OF AIR CAVITIES . . .. 79
Depth Dose Measurements . . . . . ... 80
Backscatter Measurements . . . .. ... . 93
Effect of Width of Air Cavity . . . . ... 103
Measurements in Anatomical Phantom . . .. 111
6 CONCLUSION AND DISCUSSION . . ... . . ... . 114
Physical Factors . . . . . . . ... 114
Use of Dosimeter. .. ... . .. .. . . 115
Air Cavity Measurements . . . . . ... 116
REFERENCES . . . . . . . . . . . . . 121
BIOGRAPHICAL SKETCH .. .. . . . . . . . . .... 128
LIST OF TABLES
1 Summary of Measurements with Commercial Diodes .
2 Gain Characteristics of the Amplifier System . .
3 Correction Factors for Increased Response of
Dosimeter 2 to Temperature . . . . ... .
4 Directional Sensitivity Measurements for
Dosimeter 2 . . . . . . . . . .
5 Dose Rate Dependence of Dosimeter 2 . . . .
6 Comparison of Silicon Diode and Ionization Chamber
Central Axis Depth Dose Measurements . . . .
7 Comparison of Change in Percentage Depth Dose
Caused by Air Cavity in a 6 x 6 cm2 6Co Field .
8 Comparison of Change in Percentage Depth Dose
Caused by Air Cavity in a 6 x 6 cm2 17 MeV
Photon Field . . . . . . . . . .
9 Results of Measurements in an Anatomical
Phantom . . . . . . . . . . .
. . 34
. . 56
. . 71
. . 74
. . 94
. . 110
. . 111
. . 112
LIST OF FIGURES
1 Diagram of a p-i-n Chip . . . . . . ... 18
2 Diagram of Idealized Detector . . . . .... 37
3 Diagram to Illustrate Arrangement of p-i-n Chips in
Prototype 1 .. .. . . . . . ..... ... . 39
4 Photograph of Prototype 1 . . . . . . ... 42
5 Photograph of Prototype 2 . . . . . .... 46
6 Diagram of Requirements for Displaying Signal ..... 47
7 Typical Integrating Circuit . . . . . . . 49
8 Operational Amplifier Integrating Circuitry . . .. 52
9 Photograph of Amplifier and Display System . . . 62
10 Graph to Demonstrate Linear Response of Dosimeter 1
to Radiation . .. . . . . . . . . . 65
11 Graph to Demonstrate Linear Response of Dosimeter 2
to Radiation . .. . . . . . .... ... . 67
12 Temperature Dependence of Dosimeter 2 . . . ... 70
13 Directional Sensitivity of Dosimeters 1 and 2 .... 73
14 Depth Dose Measurements for 6Co ... . . . 82
15 Depth Dose Measurements for 60Co . . . . .. . 84
16 Depth Dose Measurements for 6Co .. . . . .. . 86
17 Depth Dose Measurements for 8 MeV Photons . . ... 88
18 Depth Dose Measurements for 8 MeV Photons . .... .. 90
19 Depth Dose Measurements for 17 MeV Photons . .... . 92
LIST OF FIGURES (Continued)
20a Diagram of Phantom Used for Backscatter
Measurements . . . . . .. .
20b Diagram of Phantom Used for Backscatter
Measurements . . . . . . .
20c Diagram of Phantom Used for Backscatter
Measurements . . . . . . .
21 Results of Backscatter Measurements .
22 Results of Backscatter Measurements .
23a Diagram of.Phantom Used for Air Cavity
Measurements . . . . . . .
23b Diagram of Phantom Used for Air Cavity
Measurements . . . . . . .
24 Results of Air Cavity Measurements . .
25 Summary of Air Cavity Measurements . .
S .. .. 96
S. . . 97
S. . . 98
. . . 100
S. . . 102
S. . . 105
. . 107
. . . . 109
. . . 119
Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the Requirements
for the Degree of Doctor of Philosophy
THE DESIGN AND CONSTRUCTION OF A MINIATURE
DOSIMETER FOR THE STUDY OF THE EFFECTS
OF AIR CAVITIES IN RADIATION THERAPY
Marlene Hope Patricia McKetty
Chairman: Lawrence T. Fitzgerald
Major Department: Nuclear Engineering Sciences
A miniature dosimeter utilizing silicon p-i-n chips was designed
for performing in vivo measurements in radiation therapy. The dosimeter,
which is less than 2 mm in diameter, can be inserted into patients and
phantoms for rapid measurements. An important advantage of this type of
dosimeter over thermoluminescent dosimeters and miniature ionization
chambers is the instant readout and the elimination of a waiting period
Dosimeters which have pn junctions are useful for detecting alpha
particles and fission fragments but are usually unsuitable for photons
because insufficient energy is deposited in the depletion region to pro-
duce an adequate signal. By using p-i-n diodes rather than diodes with
pn junctions, electrons and photons can be .detected since sufficient
energy can be deposited in the intrinsic region to produce a measurable
Different arrangements of the p-i-n chips were explored in an
effort to obtain minimal detector size, directional independence and a
large signal. The physical parameters of the dosimeters were examined.
These include temperature effect, directional sensitivity, effect of
encapsulating material, energy and dose rate dependence and linearity
For initial measurements in the study, the dosimeter, which was
operated in the photovoltaic mode, was used with a Keithley 616 digital
electrometer. Once the magnitude of the signal was established, an
electrometer was designed and built specifically for use with the diode
dosimeters. The electrometer allowed a voltage proportional to the
dose received by the dosimeter to be integrated and then displayed on
a digital panelmeter. A voltage divider network incorporated in the
integrating circuit was used to vary the effective feedback capacitance
and, therefore, the sensitivity of the system.
The dosimeters, which were calibrated against a National Bureau of
Standards calibrated Victoreen ionization chamber, were used to investi-
gate some of the effects of air cavities in a radiation treatment,
volume. These include change in backscatter and build-up characteristics.
Geometrical as well as anatomical phantoms of tissue equivalent material
were used for measurements. Cobalt-60 gamma rays and 8 and 17 MeV
photons from a linear accelerator were used as the radiation source for
the various measurements. Measurements were taken to determine the
variation in dose caused by changing the dimensions and geometry of an
In patient treatment planning, the presence of an air cavity in a
treatment volume is not considered when dose distributions are
determined. The measurements with the miniature dosimeter indicated
that an air cavity had a greater impact if a volume were being irradi-
ated with 60Cobalt rather than 17 MeV photons. As the dimensions of the
ated with Cobalt rather than 17 MeV photons. As the dimensions of the
treatment field and the length of the air cavity were increased, the
discrepancy between measured dose and calculated dose became more marked.
Since air cavities are often in the treatment volume of head and neck
cancers, these measurements suggest that the use of high energy photons
may be advantageous in obtaining a more uniform dose distribution.
In the treatment of cancer patients with radiation, a major problem
is the accurate determination of the dose at different points in the
treatment volume. The objective of treatment planning in radical radio-
therapy is to design a plan in which a tumorcidal dose is given to the
malignant tumor without surpassing the tolerance of the normal tissue.
The greater the dose received by the tumor, the greater is the chance
of destroying the malignant cells. A complete treatment plan should
provide an accurate representation of the doses that will be received
in an irradiated area. The methods of collecting the depth dose data
that are used for treatment plans are very precise; therefore, the basic
data with which one begins are accurate. However, the accuracy of the
depth dose data collected in a homogeneous medium, usually water, is
reduced when it is applied to a nonhomogeneous one.
It is difficult to ascertain the degree of accuracy needed in
radiotherapy. Direct experimental evidence relating degree of precision
and result of radiation therapy is unavailable. Therefore, one must
rely on data extrapolated from animal studies or examine clinical data.
With clinical data one must relate tumor control probability, or the
frequency with which normal tissue tolerance is exceeded, to the total
dose. These methods are inadequate; however, for lack of better methods
they are still in use.
Herring and Compton (1971) stated that the dose at the tumor or
other critical volume should be known to within 5 percent or greater
accuracy. This value was determined by the development of a mathe-
matical model which depicted the response of tumors to a course of
radiotherapy. The model illustrated that small variations in the radia-
tion dose had a large influence on the probability of cure. If the dose
is not known to this accuracy, there is an increased probability of
necrosis or lack of local control. Shukovsky (1970) demonstrated the
large effect of variations in dose of 5 percent or more on the local
control of tumors in the head and neck region. Fowler (1963) also
agrees that the accuracy for radiotherapy should be no worse than 5
percent but an accuracy of 2 percent is desirable. Dosimetric errors
should be small compared with individual clinical variations.
Overall, many aspects for obtaining accurate dose measurements
have been refined. However, there are still some aspects of measure-
ments which need considerable improvement. One aspect is the measure-
ment of doses at a specific point in a treatment volume with a dosimeter.
A high degree of accuracy is essential when measuring the change in dose
over a small area or for in vivo studies but there are few types of
dosimeters available for these measurements. There are often rapid
changes in the dose distribution depending on the anatomy of the irradi-
ated area and the type and energy of radiation. A well known example is
the build-up of dose from a Cobalt-60 (60Co) beam in soft tissue, the
dose changes from 30 percent of the maximum to 100 percent in the first
5 mm and then decreases slowly. Another area which requires improvement
is the correction of dose for the presence of internal and external
inhomogeneities. On entering the body, a radiation beam often does not'
pass through a solely homogeneous medium but may pass through air cavi-
ties as well as through different types of tissue (e.g., lung, bone, fat
and muscle), each of which affects the beam differently. The dose dis-
tribution is, therefore, different from what is shown on a treatment
plan where it is assumed that the beam is passing through a homogeneous
medium and that there is always sufficient scattering material.
A miniature dosimeter could be used for checking the doses at
different points in the treatment volume, especially in areas where
there is a known discrepancy between the dose received and the dose
shown on a treatment plan. This discrepancy is often manifested clini-
cally, e.g., a reaction may be more severe than would be expected for a
BACKGROUND AND RELATED RESEARCH
Types of Dosimeters
The type of dosimeters most commonly used in radiation therapy are
ionization chambers which use the detection principle of gas ionization.
They are used for routine calibrations of therapy machines and for depth
dose and exposure measurements. Ionization chambers will continue to be
used because they can be made energy independent over a wide range and
they provide the most reliable data. Basically, an ionization chamber
consists of a small volume of gas and a pair of electrodes, one of which
is positive, the other negative. The wall of the chamber is lined with
material that serves as one electrode and the other electrode passes
down the middle of the chamber. When the air becomes ionized by radia-
tion the charges produced are collected at the electrodes. Under
proper conditions, e.g., at saturation voltage, the charge collected is
directly proportional to the amount of radiation hitting the ionization
chamber. The charge liberated in the chamber produces a voltage drop
between the electrodes and is measured with an electrometer. Errors
caused by attenuation of the radiation in the wall of the chamber may
be minimized by special design and calibration of the ion chamber. The
chamber wall and electrodes are of different chemical nature from the
air volume and, therefore, will contribute greater or lesser amounts of
photoelectrons than air. Thus, there is a dependence of calibration on
wall material and on the beam spectrum. This energy dependence is mini-
mized by making the wall relatively small and of a material similar to
air. Thewall is usually made of bakelite coated with graphite.
Victoreen R-meters are standard equipment in most radiotherapy
departments. Chambers of different sensitivities and with walls of
different thicknesses are used to measure low, medium and high energy
photons. According to Johns and Cunningham, "The Victoreen is capable
of great precision, but it is useless to measure exposure at a 'point'
because of its size and should certainly.not be used to measure expo-
sures in body cavities" (1969, p. 238). Apart from the fact that the
size of these dosimeters prohibits their insertion into body cavities,
they only measure the mean value of a radiation beam at any point of
The design of miniature ionization chambers to measure exposures
in body cavities has been described (Skoldborn, 1959). Several require-
ments were specified for the first such chambers to be made. The
radiation measurement requirements are
1. air equivalent material around the measuring volume,
2. energy independent sensitivity,
3. registration of radiation independent of direction of rays and
4. tissue equivalent material in the entire chamber in order not
to disturb distribution of rays.
The electrical requirements are
1. high insulating properties and tolerance to radiation of
insulating material used,
Victoreen R-meters are manufactured by the Victoreen Instrument
Company, Cleveland, Ohio.
2. lack of dielectric after-effect in insulating material and
3. sufficient capacitance to permit measurements within the
intended range at a suitable voltage.
The measuring requirements are
2. ease in handling of the chamber which should not be fragile
3. insensitivity of the chamber material to changes in temperature
between 15 and 40 C.
Most of these requirements were met in the design and construction of
Alderson Sievert ionization chambers which are supposed to approximate
point detectors. This type of chamber is 2 cm in length and 5 mm in
diameter with a measuring air volume of 25 mm3. It is designed for
measurements of doses ranging from one roentgen up to 300 roentgens.
Sievert chambers have been used for dose measurements in phantoms
and in patients. Although they have provided a useful service, Sievert
chambers are not really adequate for routine patient dosimetry. One
disadvantage is that they are not really small enough to be easily in-
serted in patients for in vivo dosimetry. Another disadvantage of these
chambers is that they should only be exposed a maximum of once daily.
After exposure, before the dose can be read on the electrometer, a wait-
ing period of about one hour is needed. The use of these chambers was
discussed briefly by Alderson et al. (1962) and more extensively by Dahl
and Vikterlof (1960). McKetty (1975) described the use of Sievert cham-
bers in a RANDO phantom to determine if the depth dose calculated at
certain points in a mantle field was the same as the measured dose.
These miniature ionization chambers are no longer manufactured in the
Thermoluminescent dosimeters (TLDs) have been used for patient
dosimetry for the past three decades. Daniels (1950) proposed the fact
that thermoluminescence could be used as a method of dosimetry since
the amount of light emitted from certain materials on heating after
irradiation was proportional to the amount of radiation received.
Because of their small size TLDs could be used to approximate point
detectors and have been used for measuring doses in vivo and whenever a
small dosimeter is imperative. Cameron et al. (1968) have done exten-
sive research with lithium fluoride (LiF) dosimeters and this resulted
in the development of several types of LiF dosimeters, in particular
TLD 100 which is manufactured by Harshaw Chemical Company for routine
dosimetry. Lithium fluoride is also available as loose powder, as
powder sealed in tubes or as plastic (Teflon) sealed rods or discs
ranging in size from about 1 mm to several cm.
There are several advantages in using LiF as a dosimeter. These
1. the relative energy independence,
2. the good directional uniformity,
3.. minimal fading,
4. its small size and inert nature,
5. a response that is independent of dose rate and
6. the fact that the effective Z is close to that of tissue, thus
causing the response per roentgen to be similar to that of soft tissue.
Lithium fluoride shows a linear response with dose up to about 700 rads,
above this level the response is supralinear. The disadvantages of LiF
1. the change in sensitivity with reuse,
2. the long preparation time needed before measurements,
3. the delay in getting a readout after exposure and
4. the destructive readout.
The accuracy of dose measurements with TLDs is at best 5 percent. They
lack the reproducibility needed to be accurate enough for exact dosage
Worsnop (1968) has described a study whose purpose was to deter-
mine whether the dose distribution that is actually delivered to
patients is the same as the dose prescribed. An Alderson RANDO phantom
containing LiF dosimeters distributed in a specific pattern was sent to
15 different institutions to be irradiated under certain conditions.
The institutions were participating in a national clinical study.
Hendee (1966) described a comparison of beta depth dose data obtained
with LiF dosimeters and by film and ionization chamber techniques.
Intracavitary measurements utilizing LiF have been discussed by several
researchers (Cameron et al., 1961; Naylor, 1967).
Despite the useful roles played by miniature ionization chambers
and LiF dosimeters in patient dosimetry, there are still many areas in
dosimetry which are beyond the scope of these detectors. Fowler (1963)
states that of the three main dose ranges requiring measurements it is
in the dose range used for radiotherapy of a few rads to several thou-
sand rads that small, reliable and convenient measuring devices are not
yet available. Several clinical applications in which other methods of
detection and dosimetry were used can now be performed with semicon-
ductor devices (Hertz and Gremmelmaier, 1960; Abson et al., 1968; Baily
and Norman, 1962). Solid state dosimetry may be able to solve some of
the problems that remain.
Solid State Dosimetry
Solid state dosimeters use the detection principle of electrical
conductivity and behave as solid state ionization chambers. A temporary
change in electrical conductivity occurs in semiconductor materials when
they absorb ionizing radiation--this change is a function of the
absorbed dose rate. Usually when semiconductor detectors are used in
medicine and biology, they are used as particle detectors and nuclear
spectrometers rather than as dosimeters. Marcus (1973) described the
use of silicon probes for physiological studies in the human body and
presented some of the reasons for their success and failure as radiation
detectors. Most procedures that use the probes for the detection of x
and y rays are unsuccessful, some success is reported, however, in the
detection of very low energy x rays as from plutonium localized in
lymph nodes and of g particles. Lauber and Wolgast (1972) also dis-
cussed the construction and properties of some miniature detector probes
for measurement of and y tracer activity in vivo.
The properties that are desirable for semiconductor detectors and
dosimeters have been discussed by several authors (Friedland and
Zatzick, 1967; Hendee, 1970; Fowler, 1963; Friedland and Katzenstein,
1973). The observed effect should be independent of the type of radia-
tion and the dose rate. It should vary linearly with the energy
deposited and be independent of temperature. The detector should have
a short time constant and variable sensitive depth. It should be small
enough to allow for convenient insertion into the body and have a
protective coating thick enough to safeguard against attack from body
fluid but thin enough to enhance sensitivity. It should be operated at
low potential to minimize danger to the patient.
Silicon and germanium are the two most important semiconductor
materials. They both have four electrons located in the valence band
at absolute zero temperature. When the temperature is increased some
of the valence electrons will acquire enough energy to cross the for-
bidden band or energy gap and enter the conduction band, thereby leaving
a hole in the valence band. The hole is considered as having a positive
charge. The free electrons and holes will act as current carriers if
there is an applied voltage. The magnitude of this current which is
called the drift current depends on the resistance of the semiconductor
material and the number of electron hole pairs that are thermally
generated. There is also a surface leakage current. The energy gap
for silicon is approximately 1.1 ev. Certain impurities are often
added to semiconductor material in a process called doping in order to
increase either the electron or hole conduction. Whereas an n type
impurity has five valence electrons and will result in a semiconductor
with excess electrons, a p type impurity has three valence electrons
and will lead to the formation of excess holes.
The design of most semiconductor devices is based on the pn
junction, but there is also the p-i-n structure. A pn junction is
simply the transition from a p type material to an n type material
within a semiconductor crystal. Diffusion of majority carriers across
the junction causes a build-up of charge near the junction. This region,
which is called the depletion region or space charge region,has a posi-
tive charge on the n type side and a negative charge on the p type side.
In the absence of an applied voltage there is no net current across the
junction. However, if a voltage source is applied with the negative
terminal attached to the n type material and the positive terminal
attached to the p type material (forward bias) majority carriers will
move across the junction, i.e., holes will move from the p type side and
electrons from the n type side. If instead there is a reverse bias, the
majority carriers are pulled away from the junction and the depletion
area is widened. The only current that flows is due to minority
carriers. When the depletion region is subjected to ionizing radiation
new electron hole pairs are generated which can then be collected by
the potential across the semiconductor. These pairs form a current
which is similar to an ionization current in an ion chamber. The number
of charge carriers produced by radiation in the depletion region and
collected per unit time is proportional to the radiation energy absorbed
in the region and not to the number of absorbed photons. The following
equation is obtained
Nd = (AR gi/p) (1 exp (-pw)) (1)
Nd is the number of radiation produced charge carriers,
AR is the irradiated silicon surface area in cm ,
gi is the generation rate of electron hole pairs in cm3
w is the width of the depletion region in cm and
p is the linear attenuation coefficient of the radiation
in silicon in cm .
The properties of a pn junction structure are determined by the
width of the depletion region and the carrier lifetime (T) in the base
material. The depth of the depletion region is proportional to
( (pV) where p is the base resistivity and V is the.applied bias
voltage, g = 1/2 for n type silicon and B = 1/3 for p type silicon.
The average energy (E) required to form an electron-hole pair in
silicon is 3.5 ev. This is approximately 1/10th of the amount of energy
needed to produce an ion pair in.an air-filled ionization chamber where
30-40 ev are needed. Brown (1961) pointed out that this low value for
C is attractive in a counting device because of the reduction in the
statistical fluctuations in the number of pairs produced. Because
silicon is approximately 1800 times as dense as air, the number of
electron pairs formed per unit volume is 18,000 times greater than in
an air-filled ionization chamber. These factors enable solid state
dosimeters to be of small volume. Baily et al. (1962) noted that the
advantages of using high density material for radiation detection had
led to extensive research on semiconductors, but the manufacture of
satisfactory detectors was a comparatively recent achievement. The
charge collection time in a semiconductor is much less than in an ioni-
zation chamber because of the greater charge carrier mobility and the
shorter distance for transit. Semiconductor dosimeters are more rugged
than ionization chambers and can be made into shapes to suit particular
Use of PN Junctions
PN junctions are used in diodes and in photovoltaic cells. Diodes
are normally used as rectifiers in electronic circuits but because of
their ability to produce a current when exposed to ionizing radiation
they can be utilized as dosimeters. Photovoltaic devices are used to
convert solar energy into electrical energy but can also be used as
dosimeters. Several authors have recommended the use of silicon detec-
tors for x and y ray dosimetry (Baily and Norman, 1963; Calkins, 1962;
Jones, 1963). The feasibility of using such dosimeters has been
substantiated (Scharf and Sparrow, 1966; Scharf, 1960; Scharf, 1967).
When silicon diodes and photovoltaic cells are used as dosimeters
several problems must be considered. Trump and Pinkerton (1967) list
some of these problems as
1. energy dependence of calibration in terms of absorbed dose,
2. effect of high atomic number material often used in mounting
the silicon wafer and
3. optimum electrical readout.
Other problems are the dependence of the signal on environmental
factors rather than solely on dose and angular and directional
Calkins (1962) discussed the fact that in low energy beams, energy
absorption as a function of beam energy was different in photovoltaic
cells and tissue. This is because of the difference in atomic number
of silicon and tissue. He indicated that for photons of energy greater
than 0.5 MeV, the photovoltaic cell reading could be converted into
absorbed dose by a constant multiplication factor. This is because
absorption is mainly by Compton interaction and, therefore, depends on
electron density per gm which is the same for silicon and tissue.
Trump and Pinkerton (1967) have reported the use of a commercially
available pn junction diode as a radiation probe for an automatic
isodosimeter. They measured dose distributions produced by high energy
radiotherapy sources, namely, 60Co y rays and 6, 10, 15 and 20 MeV
electrons. In 1962 Stanton and Lightfoot had reported that pn junction
diodes should be better suited for in vivo measurements than conven-
tional radiation detectors but that there was a lack of experience in
their use. They tested gallium arsenide diodes but concluded that
silicon diodes would show a smaller energy dependence because of their
lower atomic number. As early as 1958 Moody et al. reported the
response of silicon pn junctions to x and y rays. They determined that
a calibrated photovoltaic cell connected to a sensitive voltmeter by
shielded electrical leads were all that was required to build a y ray
dosimeter which would have a wide range of response. More experience
in the use of these dosimeters has been gained.
Raju (1966) reported the use of commercially available pn junction
diodes in determining beam profiles and depth dose distribution of a
cyclotron beam. Guldbrandsen and Madsen (1962) also experimented with
commercially available diodes and concluded that they fulfilled the
usual requirements of dosimeters, although dose rate measurements at
small energies were always more uncertain than those at higher energies.
Wright and Gager (1973) used a diode-scanner-recorder system to monitor
some operating parameters of a linear accelerator. They also used it
for measuring central axis depth dose data and off-axis data (Wright
and Gager, 1977). Because of problems that resulted from nonuniform
energy response of the diode, they developed an energy compensating
shield for the diode from a high Z material (Gager et al., 1977). The
shielded diode could then duplicate.the response of a Farmer probe.
Jones and Schumacher (1975) also developed an instrument utilizing
diodes to measure the output, energy and symmetry of a linear accelera-
tor beam. This device consists of a block of wood containing two
silicon diode radiation detectors along the central axis. The output
of each diode was amplified and displayed on a digital panel meter.
Parker and Johnson (1969) described a silicon pn junction detector
designed for checking the radiation dosage to the rectum and bladder of
patients with gynaecological cancer who are undergoing intracavitary
irradiation. Whelpton and Watson (1963) used pn junctions supplied as
silicon solar cells to measure the build-up curve from 60Co Y rays in
air. They found that the curve obtained using this method was similar
to that obtained using thin walled ionization chambers. Whelpton and
Watson (1963) designed a probe which was approximately quality indepen-
dent over the range 1.0 mm Cu HVT to 13 mm Cu HVT.
Fabrication of PN Junctions
PN junctions may be formed by two methods, surface barrier and
diffusion. Diffusion junction devices are formed by diffusing a suitable
impurity into the base material, that is, by diffusing a shallow layer
of a donor (or acceptor) material into high resistivity p type (or n
type) silicon. Goulding (1964) reports that phosphorus is usually used
as the diffusant and p type silicon as the bulk material. The diffused
layer is about 0.5 micron in thickness. The depletion layer forms in
the bulk p type silicon when a positive voltage is applied to the n
type layer. The depth of the depletion regions is given by the following
W = 0.32 (pV) micron (2)
W = depth of depletion layer,
p = resistivity of the p type material and
V = potential.
Ziemba et al. (1962) describes a conventional pn junction detector as
having four regions,
1. a thin phosphorus or lithium diffused layer,
2. a space charge region,
3. an undepleted silicon layer and
4. a boron diffused or aluminum alloyed layer which provides an
ohmic contact to the p type material.
Surface barrier detectors function in a similar manner to pn
junction detectors formed by diffusion but they are cheaper and easier
to construct (Parker and Morley, 1966) and more stable in their
properties. Because of their availability in a needle probe geometry,
they are well suited for many dosimetric applications in medicine.
Basically surface barrier detectors consist of high resistivity n type
silicon on one surface of which is a very thin p type layer formed by
spontaneous oxidation. Thus, a pn junction is formed close to the
Until fairly recently the most successful solid state radiation
detectors were surface barrier and pn junction silicon counters.
Whereas these detectors were adequate for alpha particles, low energy
protons, deuterons and fission fragments, they were unsuitable for use
with electrons, photons and minimum ionizing particles. This is
because insufficient energy was deposited in the shallow depletion
region to produce an adequate signal.
When semiconductor detectors are used for x and y rays the maximum
amount of energy should be deposited in the charge depletion region
(Baily and Mayer, 1961). That is, the photons should not be absorbed
until they have arrived at the depletion region and should be absorbed
before penetrating the depletion region. This may be achieved by
either increasing the atomic number of the semiconductor material or
increasing the depth of the depletion region. The first method is
achieved by using germanium (Z = 32) doped with gallium or zinc instead
of silicon (Glos, 1964). The latter method is used most often and can
be accomplished by either increasing the bias voltage or increasing the
resistivity of the semiconductor material. If the bias voltage is
increased the noise is increased and this is undesirable. One method
of increasing the resistivity is by diffusing lithium into silicon to
form a lithium drifted silicon detector which is also called a p-i-n
junction detector. Thus, the development of the p-i-n structure enables
solid state semiconductor detectors to be used successfully for x and y
ray and electron dosimetry.
A p-i-n junction consists of a wafer of extremely high resistivity
silicon on one surface of which is a thin p type diffusion and on the
opposing surface a thin n type diffusion (Figure 1). The high resis-
tivity material in the middle is the intrinsic (I) layer and it can
sustain a high electric field. The width of the intrinsic layer deter-
mines the electrical properties of the junction. The space charge
region will extend throughout the entire intrinsic region when a few
volts reverse bias is applied (approximately five volts) of if there is
Figure 1. Diagram of a p-i-n Chip
a high degree of compensation in the I region. The operating potential
can, therefore, be quite low.
If the space charge region extends throughout the I region there
is no variation of capacitance (C) with voltage, since the capacitance
is constant. At low voltage, however, the p-i-n detector behaves like
C a 1/V1/2 or C a 1/V1/3 (Ziemba et al., 1962). Since the capacitance
is quite low, p-i-n junctions can be used in applications requiring high
sensitivity. Mayer (1962) reported the results of an investigation to
determine the junction characteristics as a function of width of the
intrinsic region by the use of capacitance and current measurements.
The width (W) of the depletion region is given by
W = kA/4rC (3)
k = dielectric constant,
A = junction area and
C = junction capacitance.
P-I-N diodes may be characterized in several ways. The quantum
efficiency Nq refers to the number of electrons produced per photon.
The speed of response depends on what areas are irradiated, the amount
of reverse bias applied and the load resistance.
In order to detect both short and long wavelength photons, a diode
should have a thin p layer and a thick depletion area. This is because
long wavelength photons are absorbed near the surface while short wave-
length photons penetrate more deeply. The structure of p-i-n diodes
allows both long and short wavelengths to be detected in the same
Fabrication of P-I-N Diodes
Ziemba et al. (1962) described four methods for making p-i-n
detectors each of which has its advantages and disadvantages. These
1. paint on technique,
2. single alloyed-single diffused layer,
3. double diffusion technique and
4. lithium ion drift technique.
The last method was first described by Pell (1960) and is the most
popular of the methods. It allows semiconductors of low purity to be
used rather than intrinsic material and provides the capability for
controlling device geometry. By this process lithium is diffused into
p type silicon, until an n-p junction is produced internally. A reverse
bias is then applied to the junction. The electric field in the deple-
tion area will move the positively charged Li+ ions from the lithium
rich side to the lithium deficient side in a reasonable length of time
at sufficiently high temperatures (approximately 1750 C). The amount
of drifted lithium adjusts itself to exactly compensate the negatively
charged acceptors in the bulk material. In this way an intrinsic region
is produced. Blankenship and Borkowski (1962) improved on the lithium
ion drift technique. Using their method a thin dead layer of low
resistance can be formed on the n side of a p-i-n diode. Ammerlaan
and Mulder (1963) reported a very detailed description of the procedure
to be used for preparing p-i-n detectors.
Methods of Detection and Measurement
PN and p-i-n junctions can be used for measurement of radiation in
one of three ways. In order of sensitivity these are pulse counting,
voltage measurement and current measurement.
Pulse counting. In pulse counting a high reverse bias is applied
to enable radiation generated carriers to form a pulse. The amount of
charge collected in each of the pulses is proportional to the energy
deposited in the depletion area. The counting rate depends on the
photon flux and energy, the applied junction potential and the thickness
of the depletion layer. The charge collection time (T) is given by
T = W2/1V
= carrier mobility,
W = depth of depletion area and
V = applied voltage.
The pulse size is not greatly affected by changes in bias or temperature.
The pulses are usually amplified by low noise, charge sensitive
Although pulse counting is the most sensitive of the measurement
methods, it is only useful for dose rates ranging from 10 prad/hour to
100 rad/hour (Baily and Kramer, 1963). Since the dose rates generally
encountered in radiation therapy are higher, this method of operation
would be unsuitable for dosimeters. PN junction detectors which
operate by pulse counting could find useful application in health
physics, when used in conjunction with well designed amplifiers
(Jones, 1962). Baily and Hilbert (1966) reported a study in which
p-i-n junctions were used as pulse counters for y ray dosimetry. Pulse
counting methods have been used for Y rays ranging in energy from 0.1
to 1 MeV (Jones, 1962).
Voltage measurements. Voltage measurements may be used for dose
rates ranging from 1 mrad/hour to 100 rads/min. This method of measure-
ment, which is the most simple, is an analog of the photovoltaic effect
in solar cells. The photovoltaic measurements of p-i-n and pn junctions
used for x and y ray dosimetry are of greater sensitivity than measure-
ments of solar cells (Scharf and Sparrow, 1964). This is due to the
increased resistivity and collecting volume of the dosimeters. In the
photovoltaic mode of operation there is no applied bias. When the
detector is irradiated, the radiation produced carriers in the depletion
region and any minority carriers that reach the junction by diffusion
are acted on by the junction field. A photocurrent is generated in the
reverse direction of the junction as electrons are moved to the n side
and holes to the p side. Thus, a voltage difference is produced between
the opposite sides of the diode. This photovoltage biases the junction
in the forward direction. The induced potential may be read out
directly with a high (10 MQ) impedance voltmeter, i.e., the open circuit
voltage is measured.
The relationship between the voltage and current is given
kT I I
V kT In (g I + 1) (5)
V = open circuit voltage produced by irradiation,
k = Boltzman constant,
T = absolute temperature,
q = electronic charge,
I = generated photocurrent,
I = photovoltaic output current and
I = diode saturation current.
Measurements have been performed to determine the relationship
between open circuit voltage and a number of parameters (Scharf, 1960;
Baily and Kramer, 1963; Scharf and Sparrow, 1964). These parameters
include dose rate, temperature, wavelength, energy, width of the in-
trinsic region and angle of incidence of the radiation. At low dose
rates the voltage response is approximately linear, but at higher dose
rates the response becomes logarithmic. Equation 6 shows the relation-
ship between open current voltage and dose rate.
V 2kT In K D (6)
K = a constant of the detector and
D = dose rate.
Whelpton and Watson (1963) found a linear response of open circuit
voltage up to a dose rate of 750 R/min of Co y rays in a pn junction,
while Baily and Norman (1963) found linearity up to a dose rate of
400 R/min using a p-i-n diode. The voltage response is highly dependent
on temperature, it decreases nearly exponentially with increasing
temperature. The wavelength dependence of the open circuit voltage is
due to variations in the carrier generation rate. Since the variation
rate is small for energies greater than 200 kev, the voltage
measurements become wavelength independent above this energy (Parker
and Morley, 1966).
Current measurements. If a detector operates in the photovoltaic
mode an alternative to measuring the voltage is to measure the photo-
current generated. As stated previously, the photocurrent generated by
radiation is proportional to the intensity of the radiation absorbed.
It consists of the drift current (I ) of electron-hole pairs produced
in the depletion region and the diffusion current (I )D of minority
carriers that reach the depletion area by diffusion. If the load
resistance across the detector terminals is zero, the short circuit
current is equal to these two currents which flow through the external
circuit in the reverse direction. If the load resistance is greater
than zero, there will also be a diffusion current (Ij)Di in the forward
direction. This current is produced through the junction by the photo-
voltage in an attempt to restore the equilibrium state and causes the
total photocurrent to decrease. The net diffusion current will be
(I )Di (I.) .. A sensitive galvanometer is used for measuring
gDi j Di
current. The short circuit current may be determined by extrapolating
photovoltaic current measurements to zero load resistance, but generally
is approximated from measurements at small load resistances.
Several studies have been performed to determine the relationship
between photocurrent and temperature. Scharf and Mohr (1971) found the
temperature dependence of short circuit currents was positive and nearly
linear in pn junctions but nonlinear and negative for p-i-n junctions
and surface barrier type detectors. Klevenhagen (1973, 1977) also
observed positive and negative temperature dependence of short circuit
current and that the magnitude of this dependence varied from detector
to detector. Petushkov and Parker (1973) also noticed anomalies and
suggested that dosimeters should be zeroed at the operating temperature.
Scharf and Mohr (1971) explained that the variation seen by different
investigators of short circuit current with temperature in different
detectors was due to the influence of the strongly temperature dependent
junction current (I.)Di and to the different values of series and
junction resistances. If there is no junction current, the short cir-
cuit current is independent of temperature.
Current measurements may also be performed when an external reverse
bias is applied. This is called the photodiode mode of operation. As
in the photovoltaic mode the current generated by radiation in the
reverse direction of the detector is proportional to dose rate. The
total current measured when the detector is irradiated consists of the
generated photocurrent and the dark current produced by the bias
voltage. If the dark current and total current are measured, the gen-
erated photocurrent can be calculated. The dark current is composed of
thermally induced carriers and surface leakage. As the bias voltage is
decreased the ratio of radiation generated current to leakage current
Studies have shown that photocurrents measured in the photodiode
mode increase with increasing temperature (Scharf and Sparrow, 1966).
The energy dependence is similar to that exhibited in the photovoltaic
mode. Current measurements may be used for dose rates above 1 mR/min.
The choice of a measurement technique will depend on the sensi-
tivity required and the dose rate of the radiation.
Uses of P-I-N Diodes
P-I-N diodes are used mainly as detectors and dosimeters.
Detectors can be made in a variety of shapes and sizes; thus, there is
a high degree of versatility in the measurements that can be obtained.
Baily and Kramer (1963) report that some detectors can be used for
animal and human implantation since they can operate without an applied
voltage or with the exposed portions being held at ground potential.
Baily and Norman (1962) fabricated small cylindrical p-i-n radiation
detectors for internal dose measurements and for use with internally
administered radioisotope tracers. Such detectors can be used to
detect tagged tumor tissues during brain surgery. When these detectors
are used for tracer applications the pulse counting mode of operation is
used and, therefore, a bias voltage has to be applied. A preamplifier,
amplifier and scaler must be used.
The response of p-i-n junctions to f rays emited from 32P and
Th was investigated by Baily and Hilbert (1965). The response which
was determined for various source geometries, detector sizes and energy
distributions of the emitted electrons was compared with the dose rate
measured by extrapolation chamber techniques. West et al. (1962) de-
scribed the reliability they found in using p-i-n detectors as B ray
spectrometers. Greening et al. (1969) used a lithium-drifted silicon
detector for the measurement of energy fluence rates of low-energy
x rays and found that it was as precise as a gas ionization chamber.
However, because the leakage current is greater, the minimum fluence
rates which can be measured is greater. P-I-N detectors are particu-
larly useful for recording pulsed radiation fields.
Besides being used as radiation detectors and dosimeters, p-i-n
junctions have been used as light detectors since silicon diodes are
also sensitive in the optical portion of the spectrum.
Problems Encountered in the Use of P-I-N and PN Junctions
Several precautions must be taken when using p-i-n junctions for
dosimetry. Since diodes are light sensitive, for accurate radiation
dosimetry the junctions should be protected from light. Whelpton and
Watson (1963) used black PVC tape for this purpose, whereas Raju (1966)
coated the diodes with Kodak dull black brushing lacquer.
Bloch and Worrilow (1969) described the use of polyfoam as thermal
shielding for a detector in order to minimize the drift in reverse
current due to thermal changes. Although the radiation induced current
does not vary with temperature, the leakage current varies strongly
with temperature. Drifting of this leakage current will give rise to
errors especially at low dose rates.
In order to prevent electrical leakage from these semiconductor
dosimeters, very thin insulating coatings may be used. Examples of
these insulators are epoxy resins and silicon monoxide.
Wavelength dependence of pn and p-i-n junctions is not a major
problem if high-energy radiation is being measured. Parker and Morley
(1966) found that detectors could be regarded as wavelength independent
for exposure measurements of radiation whose energy spectrum is above
Thomas and Shaw (1978) stressed the need for caution in the use of
diode detectors because of the problems that could result from variation
of response with angular orientation. If the semiconductor material is
not axially mounted in a detector, there may be a directional response.
Semiconductor materials are very sensitive to radiation damage
which produces imperfections in the crystal lattice. Vacancy-
interstitial pairs known as Frenkel defects may be formed as silicon
atoms are displaced from their equilibrium sites. P-I-N diodes have a
thick depletion area and increased sensitivity to radiation and,
therefore, have decreased resistance to radiation damage., In these
diodes radiation damage causes a decrease in charge carrier lifetime
and mobility. Loferski and Rappaport (1958) determined that minority
carrier lifetime (T) was more sensitive to radiation damage than con-
ductivity or mobility. They did direct measurements of T and determined
the minimum energy needed to produce a Frenkel defect. Coleman and
Rodgers (1964) reported that p-i-n junctions are more susceptible to
radiation damage than pn junctions because of the low fields at which
they are operated and the increased collection time and decreased
collection efficiency. The concentration and type of impurity in the
compensated material may also affect the sensitivity of p-i-n junctions
to radiation damage. Rosensweig (1962) noted that when a silicon solar
cell was exposed to electrons and y rays of energy greater than 300 kev
or to heavy particles, radiation damage caused the diffusion length to
decrease. The diffusion length is the average distance travelled by a
minority carrier prior to recombining with a majority carrier. Dearnley
(1964) pointed out that if radiation damage caused a change in detector
response this would lead to a misinterpretation of data.
A final problem to be considered is the energy dependence due to
the difference in Z of silicon and soft tissue. This is important for
low-energy radiation where the photoelectric effect is the principal
mode of interaction.
Despite the problems involved in using semiconductor detectors,
the advantages make them very attractive for in vivo dosimetry. To
reiterate, these include the small size, high sensitivity per unit
volume, the low operating voltage, the mechanical strength, ease of
construction, instantaneous readout and the simplicity of electrical
In vivo dosimetry in radiation therapy is needed in several parts
of the body. If a treatment field is regular, i.e., of normal rectan-
gular shape, the dosimetry procedures are fairly simple and
straightforward. Isodose curves for the given field size are used and
corrections are made for any curvature on the surface. This procedure
may not result in a precise estimation of the dose in a patient because
of the presence of differing types of tissues in the treatment volume.
This problem is compounded even further in irregular fields and fields
treated by rotation therapy.
Many of the fields used when treating head and neck cancer with
external radiation are regular but because of the variation of tissue
across this area the treatment plans may not provide adequate accuracy.
Air cavities, e.g., sinuses and trachea, may play an important role in
decreasing this accuracy. Several studies have been performed to deter-
mine the influence of air cavities on the dose distribution of x and y
rays as well as electrons. One of the older studies was performed by
Epp et al. (1958). They investigated the ionization build-up at the
surface of a lesion located in an air cavity of the upper respiratory
system when irradiated with a Co beam which first passed through the
air cavity. They simulated this condition by measuring the response of
a small flat ionization chamber as the thickness of the front wall was
varied from nearly zero thickness to equilibrium thickness. The con-
clusion was that unless a radiation field was bigger in all directions
than the air cavity by at least the electron range in tissue, the
absorbed dose at the inner surface of the air cavity may be less than
equilibrium value. The dose to a lesion located on the inner surface of
an air cavity may be reduced by as much as 10 percent if electronic
equilibrium is not attained.
Epp et al. (1977) performed similar studies utilizing 10 MeV x rays
from a linear accelerator. They found that the lack of electronic
equilibrium at the surface of an air cavity was more severe for higher
energy rays. Another study of this type was performed by Koskinen and
Spring (1973). They used LiF teflon dosimeters, which were 20-28 pm
thick, to study the dose distribution of .Co radiation in a phantom
simulating the upper respiratory air passages. They devised a formula
for calculating the decrement in dose at the surface of an air cavity.
Nilsson and Schnell (1976) expanded on the previous measurements. They
used 10 pm thick LiF dosimeters to measure the dose in a phantom in
which air cavities of different sizes were placed. The radiation source
used was a 6 MeV linear accelerator and a 42 MeV betatron. Nusslin
(1975) studied the effects of air cavities on the dose distributions of
a high energy electron beam from a betatron, and Samuelsson (1977)
studied the influence of air cavities on the dose distribution of 33 MeV
These types of measurements have particular significance in the
treatment of cancer of the larynx. If a parallel opposed pair treatment
technique is used to treat a laryngeal lesion, one of the beams will
pass through an air cavity before arriving at the lesion, and if adequate
precautions are not taken underdosing may result in the transition zone
between air and tissue. Bagshaw et al. (1965) found that the treatment
plan of choice for most patients with laryngeal cancer was opposed
Morrison (1971) observed that laryngeal cancer is usually diagnosed
at an early stage and, therefore, has a good chance for cure. However,
Lederman (1971) reported that despite its accessibility and ease of
examination, the radiation treatment of laryngeal cancer has not been as
successful as desired. Most of the tumors are on the vocal cord which
has a poor lymph drainage and, therefore, a low rate of spread to lymph
nodes. Tumors in the supraglottic and subglottic regions have a much
greater tendency to lymphatic spread. With the use of supervoltage
radiation in particular Co, rather than orthovoltage there has been
gradual improvement in cure rate, in addition to the advantage of less
discomfort suffered by the patient and minimizing of reactions.
It is unlikely that there is any advantage to be gained in using
radiation that is more penetrating than 60Co for treatment of laryngeal
cancer. Morrison and Deeley (1962) noted that the only possible ad-
vantage might be a different physical distribution of the radiation. If
8 MeV radiation from a linear accelerator is used, the tissues between
1.3 and 2.1 cm below the skin receive the peak dose, and since this is
the depth of the vocal cord, the dose received would be homogeneous in a
vocal cord lesion and fall off towards the normal tissue.
An analysis of cases of laryngeal cancer which were not cured by
radiation therapy suggested that underdosing may be a major cause of
these failures. This reinforces the need for better dosimetry. A study
performed by Spring et al. (1972) illustrated that there was a correla-
tion between recurrence of laryngeal cancer and the treated volume.
Another study (Salmo et al., 1973) narrowed down the analysis to include
stage 1 patients only. The conclusion was that larger treatment fields
gave a greater probability of cure without recurrence.
A dosimeter which is more accurate than TLDs and smaller than an
ionization chamber would be useful in the continuation of studies of
the effect of air cavities in the treatment with radiation of laryngeal
Objectives of Research
The overall objective of this research was the design and construc-
tion of a miniature dosimeter to be used for measuring doses under
several conditions that are of clinical importance in radiation therapy.
In particular, a study will be made of the effects of air cavities in
the treatment with radiation of head and neck cancers.
At the present time, Nuclear Instruments manufactures a dosimeter
utilizing diodes but such a dosimeter is too large for the application
needed. Basically what is required is a miniaturization of a diode
dosimeter, but with a change in scale has come a change in resources.
THE DESIGN AND CONSTRUCTION OF THE
DOSIMETER AND PERIPHERAL APPARATUS
In order to determine the feasibility of this project initial
tests were performed on commercial diodes. The diodes used were
RCA 30808, Unitrode UM 9442 and 125-8. The RCA 30808 was an n type
silicon p-i-n photodiode with a 5 mm2 photosensitive area packaged in a
TO-5 can. The Unitrode 9442 is also packaged in a TO-5 can, with the
p-i-n chip bonded to an alumina substrate. The 125-8 was an axially
leaded device. These diodes were exposed to 150 and 250 KV x rays,
filtered with 0.5 mm Cu from a Siemens unit, in order to obtain an
estimate of the current generated .by radiation. The diodes were con-
nected via coaxial cable to a Keithley 616 digital electrometer and
dark current and photocurrent measurements recorded. Of the three
diodes tested the RCA 30808 was the most sensitive and the Unitrode
125-8 least sensitive. A rough check of angular dependence was also
performed. Measurements were taken when the diodes were rotated
through 900 intervals. All three diodes showed some degree of varia-
tion in sensitivity between angles. Table 1 is a summary of the
With the RCA 30808 and the Unitrode 125-8 diodes the maximum dif-
ference in readings at different angles was approximately 22 percent and
with the UM 9442 the maximum difference was 14.5 percent. Such a large
difference would be unacceptable in a dosimeter. Since it would be
Table 1. Summary of Measurements with Commercial Diodes
A e 150 KV, 15 mA 250 KV, 15 mA
Angle Filter 0.5 mm Cu Filter
(degrees) 0.5 mm Cu Filter 0.5 mm Cu Filter
Photocurrent Measurements in nAmp
Unitrode UM 9442
Angular Dependence of Diodes
Unitrode UM 9442
NOTE: All measurements were taken using a field
a focus skin distance (FSD) of 60 cm.
size of 6 x 6 cm2 and
extremely difficult to position the dosimeter with the same orientation
each time, a dosimeter is required which shows the same response to
radiation from any direction.
Measurements were made with the p-i-n junction in two different
orientations--parallel to the beam and perpendicular to the beam. It
was noted that the readings were very different in the two configura-
tions, e.g., in the Unitrode 125-8, a measurement of 2.5 nAmps was made
for a given dose when the diode was perpendicular to the beam compared
to 0.462 nAmp when the diode was parallel to the beam.
Another method of dosimetry considered was cadmium telluride
(CdTe). Although CdTe can detect radiation efficiently without the use
of photomultipliers its use has been limited to uptake measurements of
radiopharmaceuticals (Walford and Parker, 1972). Ideally a detector
should have a tissue equivalent response and the main disadvantage of
using CdTe for dosimetry is the large difference between its effective
Z (Z = 50) and that of tissue (Z = 7.8). Cadmium telluride is more
energy dependent than silicon, however, because of its resistance to
physical injury and ultimately increased reliability and its ability to
be fashioned into a miniature detector CdTe may prove to be a good
choice for a dosimeter. The evaluation of a CdTe dosimeter would be a
complete area of investigation and was not pursued in this study.
Description of Ideal Dosimeter
As discussed previously some of the properties of the ideal
1. small size,
2. lack of angular dependence,
3. lack of temperature dependence,
4. ease of insertion prior to measurement and ease of removal,
5. ease of setting up and taking a measurement and
6. accurate and reproducible measurements.
One of the aims of this research project was to design a dosimeter with
as many of these ideal properties as possible, and which is small
enough for insertion.
The primary limiting factor in constructing the dosimeter as
desired was the lack of sufficient funds. The secondary factors were
the availability of small diodes and coaxial cable. Once the feasi-
bility of using silicon diodes was determined it was essential to find
a manufacturer who could supply extremely small p-i-n chips. Although
several companies manufacture diodes utilizing chips, few were able to
supply individual chips, especially of the small size needed. Unitrode
Corp. could supply chips in two sizes, namely 0.062 in. square and 0.03
To overcome the problem of angular and directional dependence the
sensitive area of the silicon detector should have a coaxial geometry as
in Figure 2. This would allow the most compact sensitive volume, and
regardless of the direction of the radiation beam the response would be
Since p-i-n chips are not available in such a configuration, this
would entail having the chips specially manufactured in this
configuration. However, this would prove extremely expensive and was
beyond the budget of this project; therefore, alternate methods of
achieving this requirement had to be used.
Figure 2. Diagram of Idealized Detector
Since several chips were to be used in the construction of a
dosimeter, the characteristics of each should be checked individually.
Ideally before each chip is used in the dosimeter it should be irradi-
ated and the signal examined. The response to factors such as tempera-
ture should also be determined for each chip. The manufacturer of the
chips suggested that they should be checked for leakage at room tempera-
ture before and after heating them up to 1250 C for 168 hours. Several
chips should be examined and those with similar response used. Several
dosimeters should also be made for testing. Once each dosimeter is
calibrated the same tests should be performed on each to determine if
there is any variation in the response and thereby check the accuracy.
Since the sensitivity of the dosimeter may decrease with cumulative
radiation damage, calibrations should be repeated periodically. The
cost of carrying out all these procedures was prohibitive; therefore,
many of them were omitted.
Since the accuracy needed in radiotherapy is 5 percent, the
dosimeter has to be of even greater accuracy since there are several
other sources of error. The signal to noise ratio should be better
than 100:1. The diode dosimeters that are currently manufactured com-
mercially are intended for use on the surface and utilize a single chip.
In this design in order to have a large signal and equal directional
sensitivity several chips were used.
In order to simulate a coaxial geometry of p-i-n diode the first
dosimeter constructed was made with three chips positioned on the cir-
cumference of a supporting substrate as shown in Figure 3. The chips
Figure 3. Diagram to Illustrate Arrangement of p-i-n Chips in
used were 0.062 in. square. The supporting rod was made of aluminum and
served as one of the electrodes. The signal from each chip was con-
nected in parallel. The entire structure was surrounded with potting
compound which supplied mechanical strength and protection and was
placed in polyethylene tubing. The signal was led through a 2 ft long
miniature coaxial cable manufactured by Microtech. The dosimeter was
assembled by Eltec Instruments, Inc. The diameter of this dosimeter
was 3 mm and it was 1.1 cm in length. Kodak dull black lacquer was
used to make the device light-tight. However, initial measurements
indicated that the device was extremely sensitive to light and electri-
cal interference. Since this detector would not be used in patients,
materials were then used which would be undesirable if the dosimeter
were to be used clinically. The dosimeter was painted with Television
Tube Koat and this was connected to ground potential. This served to
make the device insensitive to light and static electrical interference.
This first prototype which is shown in Figure 4 was not enclosed
in a stainless steel tube in order to reduce the number of variables in
the measurement and to better evaluate the performance of the diode.
The tubing would also serve to attenuate the beam and act as build-up.
Several tests were performed to check different physical parameters
and the measuring device used was a Keithley 616 digital electrometer.
The diode was used in the photovoltaic mode with zero bias voltage and
the short circuit current measured. The reproducibility was checked as
was the linearity. The directional sensitivity and temperature depen-
dence were other characteristics of the dosimeter that were measured.
These measurements are of particular importance since the dosimeter is
being designed for in vivo studies. The response to temperature has
Figure 4. Photograph of Prototype 1
" .I .. .".. .'.*"..I *| Ti* s ' I T i iniT I (TITI i T i T i il il
1.. 21 2 -4 -5 6 7 8 9 .''O --1 0 11 1 2 '1 1 4 ""
Sli li 1 i Ii ll it i li i l i i -il i ii il IIij ,il ,l h ,, ,l,, ,IIIn I i n I . I I' J ..i. A .. u... 1 ... ..
been seen to vary for different diodes and should be determined for
each dosimeter. The procedures for and the results of the measurements
are discussed in Chapter 4.
There were several problems that had to be overcome in the con-
struction and use of this dosimeter. Since the dosimeter was intended
for interstitial use, the use of inert non-toxic materials was a re-
quirement that had to be fulfilled. The design called for the dosimeter
to be encased in thin stainless steel hypodermic tubing; however, pro-
visions had to be made in the event that the needle should be ruptured
at its insertion or removal. An inert potting compound was used around
the chips. Another potential problem is the lack of tissue equivalence
of the materials in the dosimeter. It is desirable that the effective
Z of the dosimeter be close to that of tissue; however, since the dosi-
meter will be used at energies where Compton interaction predominates,
this is not an essential factor provided a proper calibration is
obtained. The problem of cumulative radiation damage may be minimized
by regular periodic calibration of the dosimeter. Thus, if there is a
change in sensitivity, a new calibration factor will be established.
The tests of the first prototype indicated there was some amount
of angular dependence. The maximum difference was 12 percent. This
could have been due to the use of a defective p-i-n chip or to the im-
proper placement of the chips. Since it was not possible to check the
response of each chip individually, a different arrangement of the
chips was used in the next dosimeter.
In the second dosimeter the six chips used were each 0.03 in.
square, and again they were connected in parallel electrically; however,
they were placed side by side. Unitrode Corp. (the manufacturers of
the chips) indicated that the chips would not exhibit angular or direc-
tional sensitivity. The dosimeter which is shown in Figure 5 was
encased in 14 gauge hypodermic stainless steel tubing. The criteria
for the choice of the probe material were strength, hygiene and ease of
insertion rather than tissue equivalence. The response of this dosi-
meter is discussed in Chapter 4.
The basic requirements for displaying the signal from the dosimeter
are an amplifier with an integrating feedback circuit, a switch for
resetting after each measurement, an analog to digital converter and a
display panel (Figure 6). The current generated in the dosimeter by
radiation is extremely small and should be amplified prior to display
to give statistically significant results.
The actual measurement and amplification of the radiation generated
current was performed with an operational amplifier operating in inte-
gral mode with a feedback capacitor. An operational amplifier (op amp)
consists of several transistor amplifiers connected in series but has
the advantage over transistors of being able to provide a higher gain.
Operational amplifier circuits were originally developed to perform
mathematical operations for analog computers using feedback but now
have numerous applications, among which are instrumentation, use in
Figure 5. Photograph of Prototype 2
i TT!'I' 1 IT"i 11 i!1 ir I1I~ll I iiI 1iir If rIT 1111 111, 11 11,I' I i' ily i i i 1 1VIIIIII I T TI T
mm 1, :-31TK4 -T5 M6 `7 S ,,, 9 ---10 1 1 1,2 1 3 1'4 "I`
INTEGRATING A/D DIGITAL
AMPLIFIER CONVERTER DISPLAY
Figure 6. Diagram of Requirements for Displaying Signal
control systems and in regulating systems. The main properties of op
1. infinite voltage gain,
2. zero output impedance,
3. high input impedance,
4. capability of maintaining the above properties over a particu-
lar frequency range,
5. zero voltage across input terminal and
6. absence of changes with environmental conditions, e.g.,
These features are typical of an ideal amplifier. In reality an op amp
does not possess these features but is close to achieving some of them.
A typical op amp will have high gain, high input impedance, low output
impedance and wide bandwidth. Since all parameters cannot be optimized,
the choice of a suitable op amp depends on its application. In this
application the current was integrated over the time of exposure and the
amplifier produced an output voltage which is proportional to the inte-
gral of the input voltage. Figure 7 shows the circuit of a typical
integrator using an op amp. This circuit incorporates a voltage divider
network which will be discussed later.
The gain between input and output is largely independent of the
gain of the amplifier but is determined by other elements in the
circuit. The output of the amplifier is fed back to the negative or
inverting input through the feedback capacitor. Since the rate of
change of output voltage is proportional to the input voltage, the basic
equation that describes the functioning of an integrator is
Typical Integrating Circuit
V V dt
o RC i
V is the output voltage,
V. is the input voltage and
C is the feedback capacitance.
For this application an op amp designed for electrometer applica-
tions would be especially useful. This circuit should give accurate
results when measuring small currents, i.e., currents in the nanoampere
range. An important op amp parameter for this application is the input
offset voltage. This is the voltage that must be applied across the
inputs of an op amp or to some other point specified by the manufacturer
to ensure that the output is zero when there is no signal input source.
Sufficient voltage can be applied to null out an unwanted signal, e.g.,
the dark current of the dosimeter. Another important parameter is the
input offset current which is the difference between the bias currents
going to the inputs. The circuit used for measuring the signal is
shown in Figure 8.
In the first circuit an Intersil ICH 8500A amplifier was used.
This was expected to be ideal for this application since it is designed
for measuring currents in the pico-ampere range. However, because of
problems with the offset voltage null the circuit was redesigned and a
Figure 8. Operational Amplifier Integrating Circuitry
Datel AM 490-2 op amp was used. This amplifier is chopper stablizied
and, therefore, has the advantage of having low input offsets and
temperature coefficients. It can also be used successfully for pico-
ampere level signals. The method of zeroing the offset voltage recom-
mended by the manufacturer was used with a few changes. Although the
Datel AM 490-2 showed greater stability than the ICH 8500A, there was
still a problem of reducing the dark current reading by using the input
offset null adjustment. The method for zero adjustment suggested by
the manufacturer is applicable if the inverting input is very close to
ground potential. An offset current is injected into the inverting
input by means of a resistor connected to a potentiometer. However,
the silicon p-i-n chips had a greater resistance than could be success-
fully handled by the circuit.
In the final circuit a Teledyne Philbrick Model 1702 op amp was
used. A 50 KQ potentiometer was used to zero the voltage offset.
Since the feedback network is the main determinant of the char-
acteristics of the circuit, the type and value of the feedback capacitor
used is important. As shown in the circuit diagram negative feedback
was used, the signal going to the negative input. Previous measurements
with the dosimeter utilizing a Keithley electrometer indicated that the
signal was in the nanoampere range. The time in which integration
would occur would be a typical treatment time of about 200 sec. Since
the current I is 2 x 10-9 amp and the rated amplifier output is T 10 V
dV 10 V 0.05 V/sec
dt 200 sec
But I ,
C = 2 x 109 amp = 0.04 pF.
A low loss capacitor was used. Because the value of the feedback
capacitance is fairly large, the shunt or stray capacitance (Cs) did
not affect the values obtained in further calculations and was omitted.
The capability of adjusting the sensitivity of the amplifier system
was needed. One method of achieving this is by the use of different
capacitors and switching to the one required for a particular
measurement. A simpler method and the one that was used in this device
was a voltage divider. This network served to alter the effective feed-
back capacitance. A simple voltage divider network incorporated in an
integration circuit was shown in Figure 7.
Since I = V (Ohm's law)
and IR I (i.e., the current flowing into R1 and R2
V V1 V1
Solving equation (7) for V1, it is seen that
V = R2+ (8)
1 R + R2
If = R2/(R + R)
then V = R V,
1 E 0,
then I = -C R (9)
Solving equation (9) for V leads to
V 1 Idt. (10)
The gain (A) of the voltage divider network is V /V where
V 1 I dt (11)
n C J
A = (12)
S1 I dt
which ultimately reduces to
C RI + R2
A = -
RE C R2
Thus, by selecting the appropriate values of R1 and R2 the gain and,
therefore, the sensitivity of the amplifier system can be regulated.
The gain becomes unity when R2 = R1 + R2.
The effective feedback capacitance (CE) can be calculated
E = C ( R + R
In this circuit a fixed feedback capacity of 0.04 pF was used and the
total resistance of the voltage divider network was 20.35 KQ. Table 2
lists the gain characteristics of the system.
Table 2. Gain Characteristics of the Amplifier System
Gain Added Resistance Effective Feedback
Switch R2 (KQ) Each Position A Capacity
Position (KQ) (pF)
7 20.35 9.95 1.0 0.04
6 10.40 7.74 2.0 0.02
5 2.655 1.52 7.7 0.005
4 1.135 0.582 17.9 0.002
3 0.553 0.205 36.8 0.001
2 0.348 0.1 58.4 0.0007
1 0.248 0.248 82.0 0.0005
When the input goes to zero the output voltage does not go to zero
but holds the accumulated integral. Before each measurement a means of
discharging the capacitor is essential to establish an initial condition.
A switch is a simple means of achieving this. In the reset mode the
initial condition is established.
d Vo 0
The switch is closed prior to t = 0, just after it is opened the voltage
across the capacitor is V i.e., the value of Vo established in the
reset mode serves as the initial condition for the integration time
An electronic means of switching was used in the first circuit,
namely a CMOS quad bilateral switch. In essence only a single switch
is needed but because of the difference in voltage on either side of the
switch at the input and output of the amplifier and the increased chance
for leakage, a T network was used in which the output of one switch was
connected to the input of the other and this junction connected to
ground potential across a resistor. Thus, two of the switches in the
4066 package were used. When the switch is open during integration the
voltage on the switch is at ground potential. At the time of reset,
15V is applied to the control of the switch which then discharges the
capacitor. This method of switching was not used in the final circuit
because the integrated circuit (IC) was destroyed when the signal was
In the final design an electromechanical method of switching was
used, namely a reed relay. Basically, a reed relay consists of a coil
which supplies a magnetic field to a metallic switch. Since the switch
is made of magnetic material, it is controlled by the coil. The ad-
vantage of using a reed relay switch for the reset is the simplicity
with which it can be placed in the circuit. The type used was IC
compatible with axial leads. If the current being measured was smaller
than about 10-9 amp, this method would be unsuitable since the switching
noise would be too large for the feedback capacitor used.
After an exposure the output of the amplifier should ideally remain
constant until the circuit is reset and the capacitor is discharged,
prior to the next measurement. There are several factors, however,
which will cause this value to change with time. These factors include
amplifier voltage offset, switch leakage, amplifier input leakage,
capacitor leakage and dielectric absorption and most significantly in
this device the dark current of the dosimeter. Therefore, a means of
nulling these effects or keeping them as stable as possible is needed.
If the leakage is stable then it can be subtracted out of the signal.
In the circuit with the AM 490-2 slight changes were made in the values
of the components suggested by the manufacturer for the offset voltage
adjustment to compensate for these other effects.
All the components for the circuit were placed on a small printed
circuit board. Additional components not previously mentioned include
capacitors for bypassing the positive and negative voltage supply to
ground. They serve to protect against high frequency noise.
Special precautions were taken with grounding and protection of the
input of the amplifier. In order to prevent ground loops a common
grounding point was used for the amplifier and digital panel meter. A
guard ring was placed around the inverting input of the AM 490-2 and
attached to ground. Grounded shields were used around the signal cable
and the other apparatus. In order to avoid leakage paths at the input
pin, the wires that had to be connected to the input were connected
directly rather than by the circuit board. A socket with Teflon inserts
was used for the operational amplifier.
Since when in use the amplifier is in the radiation therapy treat-
ment room, it is subject to scattered radiation although it is not in
the direct beam. In order to prevent ionization in the air surrounding
the amplifier, thereby increasing the signal, the housing for the cir-
cuit board was filled with Ceresin wax. The housing was simply an
aluminum box connected to ground potential. It served as a shield
against radio frequency (RF) and electromagnetic interference. In order
to make the amplifier system as compact as possible, the power supply
was placed outside the treatment room in the housing with the digital
panel meter (DPM). The input voltages, the output of the amplifier and
the reset signal were led through a shielded cable to the DPM. This
cable could be disconnected from both the amplifier and DPM. A short
miniature coaxial cable carried the signal from the dosimeter.
The analog to digital converter and display panel used was a
Fairchild Model 53 three and one half digit panel meter. This was
housed in a metal cabinet that could be easily positioned for
measurements (Figure 9). The DPM was powered directly by AC voltage.
Since the maximum output voltage of the DPM was 200 mV and the output
of the op amp is approximately 10V, a voltage divider network was used.
As mentioned previously, the power supply for the op amp was also placed
in the cabinet for the DPM. A 1.5 amp fuse was used as a protective
measure in the device. A reset button was placed on the front panel of
the cabinet which on depression would energize the reed relay and dis-
charge the capacitor resulting in a reading of zero. An on-off switch
for regulating the power to the device was also placed on the front
Figure 9. Photograph of Amplifier and Display System
r -.r -~
", -,J "i ----- .........
.:o -. ;or
.- - .S :
. . -. .
S;4 .. ._
DESCRIPTION AND RESULTS OF TESTS
PERFORMED ON THE DOSIMETERS
On exposing dosimeter 1 to Y rays from a 6Co treatment unit which
had a dose rate of 107.6 rads per min, a current of 2.95 nAmp was
generated. This was approximately 196 times greater than the dark
current which was 0.015 nAmp. Dosimeter 2 generated a current of 2.19
nAmp when exposed to the y rays and had a dark current of 0.011 nAmp.
Tests on the Dosimeters
The method of checking the reproducibility was to simply take 10
measurements consecutively using the same set-up each time. The maximum
variation between any two readings was 0.56 percent.
Each dosimeter was exposed to increasing amounts of radiation and
the photocurrent measured. The results are shown in Figures 10 and 11.
There was an exact linear relationship between dose and generated photo-
current for all the doses examined.
For the temperature dependence measurements each dosimeter was
exposed to y rays while immersed in a water bath whose temperature
Figure 10. Graph to Demonstrate Linear Response of Dosimeter 1 to Radiation
DOSIMETER I RESPONSE VS. TIME OF EXPOSURE TO 6CO BEAM
of Exposure Response
15- 7x 108
46 Ox 10-8
61 -0 10 "
122.4 152.4 1824 212.4 242.4
TIME OF EXPOSURE (SEC)
Figure 11. Graph to Demonstrate Linear Response of Dosimeter 2 to Radiation
DOSIMETER 2 RESPONSE VS. TIME OF EXPOSURE TO 6CO BEAM
STime of Exposure Response
o 32-4 sec 63 x 10 C
z 62-4 12-3 x 10C
92-4 18-1 x 10 C
. 20 122-4 24-0x 108C
z 152-4 29-8 x O8 C
3 0 182-4 35-7 x 10- C
TIME OF EXPOSURE (SEC)
324 62.4 92.4 1224 152.4 182.4 212.4
was controlled. The temperature of the water was allowed to increase
gradually from 230 C up to 400 C. For each of the temperatures for
which a measurement was taken, a reasonable length of time was allowed
to lapse between the water attaining the temperature and the time of
measurement. This was to ensure that the dosimeter had attained the
same temperature as its environment. At each temperature the dosimeter
was exposed to the same amount of radiation and several measurements
were taken at each point. The complete experiment was performed several
times to check the reproducibility. It was noted that erroneous results
were obtained if the water was not stirred constantly. Figure 12
demonstrates graphically the results of some of these measurements.
Both dosimeters showed a positive temperature dependence, i.e.,
as the temperature increased the response to radiation also increased.
For a 30 C rise in temperature the reading of dosimeter 2 increased by
approximately 2.2 percent. Because there is a definite reproducible
relationship between temperature and dosimeter response, it was possible
to calculate a series of correction factors, one of which could be
applied depending on the temperature of the dosimeter. Thus, there
would be a factor for room temperature and another for body temperature.
These correction factors for dosimeter 2 are shown in Table 3.
Klevenhagen (1978) found that it was possible to stabilize the
detector response over a wide temperature range by.means of selected
load resistance in the external circuit. This method was not tried.
Figure 12. Temperature Dependence of Dosimeter 2
TEMPERATURE DEPENDENCE OF DOSIMETER 2
WHEN EXPOSED TO A 60CO BEAM
340 25*C 333 mV
280C 341 mV
31C 348 mV
37C 365 mV
40C 372 mV
TEMPERATURE (C )
25 28 31 34 37 40
Table 3. Correction Factors for Increased Response
of Dosimeter 2 to Temperature
Temperature Correction Factor
250 C 1.0
280 C 0.977
310 C 0.957
340 C 0.928
370 C 0.912
400 C 0.895
For these measurements each dosimeter was positioned in a cylin-
drical lucite phantom which could be rotated through 3600. The long
axis of the dosimeter was perpendicular to the axis of the beam. At
450 intervals, the dosimeter was exposed to a set amount of radiation
from a 60Co treatment unit and the response recorded. Measurements were
also taken at 300 intervals. As discussed previously, the variation in
response at different angles for dosimeter 1 was as great as 12 percent,
but this may have been due to a defective p-i-n chip. For dosimeter 2,
the maximum variation was 2 percent. Figure 13 demonstrates graphically
the result of directional sensitivity measurements for both dosimeters
1 and 2, when readings were taken at 450 intervals. Table 4 shows the
results for dosimeter 2 when readings were taken at 300 intervals.
Measurements were also taken to compare the response of the dosi-
meter if the radiation beam were parallel to the long axis as opposed to
perpendicular. In both configurations it was necessary to have the same
amount of build-up, i.e., expose the dosimeter so the p-i-n chips were
Figure 13. Directional Sensitivity of Dosimeters 1 and 2
DIRECTIONAL SENSITIVITY MEASUREMENTS
FOR DOSIMETER I AND 2
* DOSIMETER I
a DOSIMETER 2
Table 4. Directional Sensitivity Measurements for
Relative Scale Reading
(Average of three measurements)
at the same depth in a phantom. Since the sensitive area of the dosi-
meter is not at the tip of the needle but is below potting material, it
was necessary to estimate the amount of phantom material which would be
equivalent to.the potting material. This thickness of phantom material
was used above the needle when measurements were taken with the sensi-
tive area perpendicular to the beam. The difference in response was
Construction Parameters Affecting Response
The amount of attenuation and the amount of build-up provided by
the encapsulating material were determined indirectly in depth dose
measurements. It was especially important to determine the effect of
the stainless steel tubing of dosimeter 2. The encapsulating material
of dosimeter 2 was seen to be equivalent to about 5 mm of tissue.
The results and their implications are discussed more extensively
in Chapter 5 with the depth dose measurements.
In order to determine the energy dependence of dosimeter 2, measure-
ments were performed using the different energies for which the dosimeter
would be used clinically. The energies used were 1.25 MeV Y rays from a
60Co unit and 8 and 17 MeV photons from a Philips SL 75-20 linear
accelerator. The dose rate was 360 rads per min for the 8 MeV photons,
440 rads per min for the 17 MeV photons and 107.6 rads per min for the
6Co source. The dosimeter was also exposed to a radium source whose
average energy is approximately 0.8 MeV. Any differences in the calibra-
tion factors are due mainly to the effect of the encapsulating material
on the dose absorbed rather than energy dependence of the silicon chips.
The actual response of semiconductor materials varies linearly with. the
energy deposited and is independent of the type of radiation which
deposits the energy.
For calibration for the 60Co radiation the miniature dosimeter was
compared with a National Bureau of Standards (NBS) calibrated Vittoreen
R-meter Model 621 using the standard protocol established by SCRAD
(1971). Measurements were taken with the Victoreen at a 5 cm depth in
a water phantom using a 10 x 10 cm field and the dose rate and dose
determined. Then, using the same geometry, measurements were performed
with the miniature dosimeter.
For the 8 and 17 MeV photons, measurements were taken with a
Capintec 0.6 cc air equivalent ionization chamber attached to a Keithley
616 digital electrometer and a Keithley 6169 ion chamber interface.
The Capintec had been previously cross calibrated with the NBS calibrated
Victoreen R-meter. The procedure established by SCRAD was again used
with the appropriate CX factors. The calibration factor for the 2Ra
source was obtained by comparing the reading of the Victoreen Model 131
with that of the miniature dosimeter when exposed under similar
conditions. The following are the calibration factors for the miniature
0.132 for 60Co y rays
0.111 for 8 MeV photons
0.108 for 17 MeV photons
0.135 for 226Ra.
By multiplying the scale reading of the display system by the appro-
priate calibration factor, the dose in rads is obtained. If the factors
are normalized to the Co calibration factor, this would give the
0.8 MeV 1.02
1.25 MeV 1.00
8 MeV 0.856
17 Mev 0.856
Dose Rate Dependence
Measurements were taken with dosimeter 2 to determine if the
response was dose rate dependent. The dosimeter was exposed to the
same given dose of radiation (50 rads) at different dose rate settings.
Both 8 MeV and 17 MeV photons from the Philips SL 75-20 linear accelera-
tor were used. The dose rate of the 8 MeV photons was varied from 140
rads per min up to 420 rads per min, and the dose rate of the 17 MeV
photons was changed from 100 to 300 rads per minute. The results are
shown in Table 5. Although there was an increase in scale reading with
increased dose rate, the change was minimal, for 17 MeV it was 1.1
Table 5. Dose Rate Dependence of Dosimeter 2
Ds R e Relative Scale Reading
radss per minute)
8 MeV Photons
17 MeV Photons
Tests on Peripheral Equipment
Tests were performed to determine if there was a change in the
dosimeter response when the cable was irradiated. This change would be
analogous to the "stem effect" seen in ionization chambers. The dosi-
meter was irradiated using a series of field sizes that ranged from
4 x 4 cm to 4 x 20 cm At each field size two sets of measurements
were taken, one with the cable in the beam and the other with the cable
outside of the beam. The response was the same whether or not the
cable was irradiated.
Position of Amplifier
An amplifier system should ideally be as close as possible to the
detector to eliminate noise and degradation of the signal. However,
it should not be in the path of the radiation beam. Measurements were
taken to determine how much effect the position of the amplifier had on
the dosimeter reading. Measurements were taken first with the amplifier
12 in. from the dosimeter and then with the amplifier outside of the
treatment room, the signal being carried to the amplifier through a
100 ft cable. There was no change in the response.
Measurements were taken at the different voltage divider gain
settings on the amplifier to check that the gain that was calculated was
actually obtained. The dosimeters were exposed to a given dose at dif-
ferent settings and the readings compared. The calculated and measured
values agreed within 1 percent.
DETERMINATION OF THE
EFFECT OF AIR CAVITIES
The problems in dosimetry and treatment planning caused by the
presence of air cavities in the treatment volume have been discussed by
several workers. These problems were discussed as they related to
treatment with Co (Epp et al., 1958; Koskinen and Spring, 1973), with
10 MeV photons (Epp et al., 1977) and with 33 MeV photons (Samuelsson,
1977). The dosimeters used in these studies were thin walled ionization
chambers and TLDs. In this study in addition to 6Co y rays, 8 and 17
MeV photons from a linear accelerator were used as the radiation source.
Both dosimeters used were made from silicon p-i-n chips, but dosimeter 2
was enclosed in a stainless steel needle whereas dosimeter 1 was not.
These dosimeters had the advantage of providing immediate readout. There
was no waiting period between measurements. As soon as a measurement
was taken, the diode dosimeter could be repositioned for additional
measurements. It was fairly simple to position the dosimeters in the
phantoms for the measurements. When TLDs are used they have to be
handled with forceps, and if they are paired it is necessary to know and
maintain the identity of each pair. Sievert chambers should only be
exposed to radiation a maximum of once daily. These problems are elimi-
nated with the use of the silicon diode dosimeters.
Depth Dose Measurements
The first set of measurements taken in this series were "build-up"
measurements, i.e., the miniature dosimeters were used to measure depth
doses. Using a 10 x 10 cm2 field size and a SSD of 80 cm, central axis
depth doses were measured for 6Co, for 8 and 17 MeV photon measurements
a FSD of 100 cm was used. Depth dose measurements were also taken for a
6 x 6 cm2 field. The phantom used for these measurements consisted of
6 x 6 in.2 slabs of polystyrene, each slab was 1/16th in. thick. Full
backscatter was provided for all the measurements and there was phantom
material on either side of the dosimeter to eliminate air spaces.
Although it is desirable when measuring depth doses on an x ray machine
to determine the dose at a depth and at the reference point simultane-
ously, this was not done in these measurements.
The results of the depth dose measurements are shown in Figures 14-
19. The point of maximum electronic build-up (Dmax) is 0.5 cm for 60Co,
2 cm for 8 MeV photons from the SL 75-20 linear accelerator and 3 cm
for 17 MeV photons from the same machine. These values are obtained if
measurements are taken with an extrapolation chamber or build-up chamber.
In these miniature silicon dosimeters the materials surrounding the
p-i-n chips, namely the potting epoxy, polyethylene tubing and stainless
steel tubing, provide build-up and reduce the thickness of phantom
material that must be used to give full build-up. From the measurements
with dosimeter 2 on the Co unit it was apparent that full build-up was
provided by the encapsulating material. The readings decreased steadily
as the depth of the dosimeter in the phantom increased, indicating that
the encapsulating material was equivalent to at least 5 mm of tissue.
With dosimeter 1, however, the readings increased rapidly at first and
Figure 14. Depth Dose Measurements for 6Co
DEPTH DOSE MEASUREMENTS FOR 60CO WITH DOSIMETER I
w 420 "
405 Field Size 6x6cm2
0 .79 1.58 2.37 3.16 3.95
DEPTH OF DOSIMETER IN CM
Figure 15. Depth Dose Measurements for 60C
Figure 15. Depth Dose Measurements for Co
DEPTH DOSE MEASUREMENTS FOR 60CO WITH DOSIMETER I
.79 1.58 2.37 3.16
DEPTH OF DOSIMETER IN CM
Field Size IOx 10cm
Figure 16. Depth Dose Measurements for 60
Figure 16. Depth Dose Measurements for Co
DEPTH DOSE MEASUREMENTS FOR 60CO WITH DOSIMETER 2
385 Field Size IOx 10 cm
0 .79 1.58 237 3.16 3.95
DEPTH OF DOSIMETER IN CM
Figure 17. Depth Dose Measurements for 8 MeV Photons
DEPTH DOSE MEASUREMENTS FOR 8 MeV PHOTONS
WITH DOSIMETER I
_j 180 -
170 FIELD SIZE I0 x 10 cm
FSD 100 cm
0 .79 1.58 2.37 3.16 3.95
DEPTH OF DOSIMETER in cm
Figure 18. Depth Dose Measurements for 8 MeV Photons
DEPTH DOSE MEASUREMENTS FOR 8 MeV PHOTONS
WITH DOSIMETER 2
175 FIELD SIZE 10 x 10 cm
0 FSD 100 cm
0 .79 1.58 2.37 3.16 3.95
DEPTH OF DOSIMETER in cm