Development of a silicon diode detector for skin dosimetry in radiotherapy

Nikolina Vicoroski, Anthony Espinoza, and Mitchell Duncan

Centre for Medical Radiation Physics, University of Wollongong, Wollongong, NSW 2500, Australia
Bradley M. Oborn and Martin Carolan
Centre for Medical Radiation Physics, University of Wollongong, Wollongong, NSW 2500, Australia
Illawarra Cancer Care Centre, Wollongong Hospital, Wollongong, NSW 2500, Australia
Peter Metcalfe
Centre for Medical Radiation Physics, University of Wollongong, Wollongong, NSW 2500, Australia
Illawarra Health and Medical Research Institute – IHMRI, Wollongong, NSW 2500, Australia
David Menichelli
IBA Dosimetry GmbH, Bahnhofstr, 5 90592 Schwarzenbruck, Germany
Vladimir L. Perevertaylo
SPA BIT, Kiev 08296, Ukraine
Michael L. F. Lerch, Anatoly B. Rosenfeld, and Marco Petaseccaa)
Centre for Medical Radiation Physics, University of Wollongong, Wollongong, NSW 2500, Australia
Illawarra Health and Medical Research Institute – IHMRI, Wollongong, NSW 2500, Australia

(Received 7 February 2017; revised 14 June 2017; accepted for publication 28 June 2017;
published 12 August 2017)
Purpose: The aim of in vivo skin dosimetry was to measure the absorbed dose to the skin during
radiotherapy, when treatment planning calculations cannot be relied on. It is of particularly impor-
tance in hypo-fractionated stereotactic modalities, where excessive dose can lead to severe skin toxic-
ity. Currently, commercial diodes for such applications are with water equivalent depths ranging
from 0.5 to 0.8 mm. In this study, we investigate a new detector for skin dosimetry based on a silicon
epitaxial diode, referred to as the skin diode.
Method: The skin diode is manufactured on a thin epitaxial layer and packaged using the “drop-in”
technology. It was characterized in terms of percentage depth dose, dose linearity, and dose rate
dependence, and benchmarked against the Attix ionization chamber. The response of the skin diode
in the build-up region of the percentage depth dose (PDD) curve of a 6 MV clinical photon beam
was investigated. Geant4 radiation transport simulations were used to model the PDD in order to esti-
mate the water equivalent measurement depth (WED) of the skin diode. Measured output factors
using the skin diode were compared with the MOSkin detector and EBT3 film at 10 cm depth and at
surface at isocenter of a water equivalent phantom. The intrinsic angular response of the skin diode
was also quantified in charge particle equilibrium conditions (CPE) and at the surface of a solid water
phantom. Finally, the radiation hardness of the skin diode up to an accumulated dose of 80 kGy
using photons from a Co-60 gamma source was evaluated.
Results: The PDD curve measured with the skin diode was within 0.5% agreement of the equivalent
Geant4 simulated curve. When placed at the phantom surface, the WED of the skin diode was esti-
mated to be 0.075
0.005 mm from Geant4 simulations and was confirmed using the response of a
corrected Attix ionization chamber placed at water equivalent depth of 0.075 mm, with the measure-
ment agreement to within 0.3%. The output factor measurements at 10 cm depth were within 2% of
those measured with film and the MOSkin detector down to a field size of 2 9 2 cm2. The dose–re-
sponse for all detector samples was linear and with a repeatability within 0.2%. The skin diode intrin-
sic angular response showed a maximum deviation of 8% at 90 degrees and from 0 to 60 degree is
less than 5%. The radiation sensitivity reduced by 25% after an accumulated dose of 20 kGy but
after was found to stabilize. At 60 kGy total accumulated dose the response was within 2% of that
measured at 20 kGy total accumulated dose.
Conclusions: This work characterizes an innovative detector for in vivo and real-time skin dose mea-
surements that is based on an epitaxial silicon diode combined with the Centre for Medical Radiation
Physics (CMRP) “drop-in” packaging technology. The skin diode proved to have a water equivalent
depth of measurement of 0.075
0.005 mm and the ability to measure doses accurately relative to
reference detectors. © 2017 American Association of Physicists in Medicine [https://doi.org/10.1002/
mp.12469]

Key words: external beam radiotherapy skin dose, silicon detectors, skin dosimetry

5402 Med. Phys. 44 (10), October 2017 0094-2405/2017/44(10)/5402/11 © 2017 American Association of Physicists in Medicine 5402
5403 Vicoroski et al.: Silicon diode for skin dosimetry
1. INTRODUCTION
Quality assurance of radiotherapy treatments involves several
techniques and diversified approaches to verify the dose
delivered to the patient. Monitoring of the skin dose helps to
prevent and minimize the side effects associated with inevita-
ble irradiation of healthy tissues such as the skin. About 20%
of women undergoing 3D conformal radiotherapy for breast
cancer treatments develop acute skin toxicity which affects
dramatically the quality of life of the patient.1–3 ICRP defines
skin dose as the dose absorbed and measured at the water
equivalent depth (WED) of 0.07 mm which makes advance-
ment of skin dosimetry challenging.4
An ideal skin dosimeter should have extremely thin sensi-
tive volume covered with tissue equivalent build-up providing
a WED of 0.07 mm for skin dosimetry or even less if surface
dose measurement is required.
Although the reference instrument for surface dosimetry at
the air-to-water interface is the extrapolation ion chamber, it
is not used routinely for skin dosimetry.5 In this work, the
Attix ionization chamber (IC) is adopted as reference for
measurements in a phantom.6 Cylindrical IC are normally
unsuitable for this application because of their large diameter,
thick entrance window, and large beam perturbation.
Unguarded parallel plate ICs show an over-response at very
shallow depths due to the lateral secondary electrons scatter-
ing into the sensitive volume.7 For this reason, only large
diameter ring-guarded ICs such as the Attix ionizing chamber
(Model 449 CNMC Co. Nashville, TN – USA) are com-
monly considered as the reference instrument for interface
dose measurement. Although very accurate for in-phantom
measurements, the Attix IC can be used for skin dosimetry in
a phantom only; they are limited to measure field size beams
of 5 9 5 cm2 and above and they have been never manufac-
tured in array for skin 2D dose mapping.8
Radiochromic films have an excellent spatial resolution,
tissue equivalency, and can provide 2D dosimetric informa-
tion. However, as radiochromic film is a passive detector, it
does not allow for real-time measurements. Its response
may also be affected by a WED of measurements much
larger than 0.07 mm, improper handling, and scanner
performance.9,10
Thermal luminescence dosimeters (TLDs) have been
used11 clinically for skin dose measurements in their carbon-
loaded version. They are small in size and tissue equivalent,
however, they require a long series of pre- and postirradiation
processes and a refined calibration procedure to be readout
accurately. They are labor-intensive and not real-time dosime-
ters, rendering them impractical for use in real-time in vivo
skin dosimetry.12,13 Zhuang and Olch showed that a nano-dot
optically stimulated luminescence dosimeter can be adopted
as skin dosimeter with stabilization delay of less than 10 min
after irradiation. However, it has an effective WED of
0.8 mm which is larger than WED of 0.07 mm accepted for
skin dosimetry.14
Semiconductor detectors such as MOSFETs are able to
achieve excellent spatial resolution because of their high
Medical Physics, 44 (10), October 2017
5403
sensitivity and small sensitive volumes. The MOSkin, based
on the MOSFET technology and recently introduced by
CMRP, are also able to produce real-time readout and
provide a reproducible water equivalent depth of dose mea-
surement of 0.07 mm.15 They are the only detector for real-
time in vivo skin dosimetry currently available. MOSkin
has demonstrated good agreement with the Attix IC for
surface dose measurements and angular dependence in MV
and kVp photon beams.16 However, due to their limited
linear dynamic range (up to 50 Gy) they are not practical
to use in a 2D array.
Silicon diode detectors could represent a good candidate
to develop a new technology to measure in real-time in vivo
skin dose. The technology development of silicon diodes
makes them the standard reference dosimeter for a variety of
application in radiotherapy from quality assurance of stereo-
tactic modalities to in vivo dosimetry due to their small size
allowing for high spatial resolution and high sensitivity.17,18
However, current designs of silicon diodes die and their pack-
aging, which provides electronic equilibrium conditions for
MV field of interest, do not make them suitable for direct
skin dosimetry on medical linacs.
In this paper, a full characterization of a silicon epitaxial
diode detector for surface dosimetry in a phantom was per-
formed. It includes: (a) detector calibration, linearity and
reproducibility; (b) percentage depth dose; (c) field size
dependence; (d) angular dependence; (e) dose per pulse
dependence; and (f) radiation hardness. Comparisons were
made with Monte Carlo simulations, ionization chambers,
and film.
2. MATERIALS AND METHODS
Characterization of dosimeters is normally performed
where charged particle equilibrium (CPE) condition exists.18–21
However, characterization of a skin dosimeter should be
performed at the surface and at very shallow depth in the
build-up region, where strong electronic disequilibrium and
steep dose gradient occurs. Such conditions are typical for
in vivo skin dosimetry.16
2.A. Silicon diode detectors
Two types of epitaxial diodes were fabricated, and five
samples for each prototype were tested. The skin diode
(named Epi-5B), fabricated by Fondazione Bruno Kessler
(FBK - Trento, Italy) on an N-type silicon 5 um thick epitax-
ial layer grown on a 0.001 Ohm-cm N-type silicon substrate,
were originally designed for high energy physics applica-
tions. The detector structure is represented in Fig. 1 and is
composed of a boron implanted P+ sensitive area of
0.5 9 0.5 mm2, with a total size of the die of
0.75 9 0.75 mm2. The effects of the superficial leakage cur-
rent are minimized by using a guard-ring surrounding the
entire pixel sensitive area.
In order to estimate the effect of the thickness of the epi-
taxial layer (i.e., sensitive volume) on the WED of dose
5404 Vicoroski et al.: Silicon diode for skin dosimetry 5404

(a) (b)

FIG. 1. Simplified diagram (not to scale) of the skin diode Epi-5B topology (a) top view and (b) cross-section view. [Color figure can be viewed at wileyonlineli-
brary.com]

measurement, a second epitaxial silicon diode was fabricated
by CMRP on a silicon P-type 50 lm thick epitaxial layer
grown on a 0.001 Ohm-cm substrate (named Epi-1C –
Fig. 2). The Epi-1C has N+ phosphorus implanted active
area of 0.6 9 0.6 mm2 and overall die dimensions of
1.5 9 1.5 mm2. This diode has been adopted for the fabrica-
tion of the 11 9 11 2D diode array MagicPlate.22
Both epitaxial diodes have been embedded in kapton
pigtails with 0.6 9 3 mm2 cross section and 350 mm
length using CMRP proprietary “drop-in” packaging tech-
nology. This packaging technique is based on tab bonding
of the silicon die underneath a stack of polyamide and
aluminum layers is illustrated in Fig. 3. The build-up mate-
rial thickness above the detector sensitive layer is repro-
ducible and already proved to be able to provide a WED of
approximately 0.07 mm.16 The “drop-in” technology is
compatible with standard flexible printed circuit manufac-
turing technologies and avoids the use of high atomic num-
ber materials for packaging of semiconductor radiation
detectors, minimizing the perturbation of the beam. The
kapton pigtail outside of the diode is shielded by a thin
aluminum foil and grounded to minimize radiofrequency
interference.23 Further details on the “drop-in” packaging
technology can be found elsewhere.24

(a) (b)

FIG. 2. (a) Top view diagram of the Epi-1C diode topology and (b) cross-section view, not in scale. [Color figure can be viewed at wileyonlinelibrary.com]

Medical Physics, 44 (10), October 2017

5405 Vicoroski et al.: Silicon diode for skin dosimetry
FIG. 3. Simplified schematic of the kapton pigtail cross section accommo-
dating the skin diode die using the “drop-in” packaging technology. [Color
figure can be viewed at wileyonlinelibrary.com]
2.B. Attix ionization chamber
Ionization chambers are considered the reference tool for
calibration and comparison of other dosimeter detectors in
radiotherapy. The reason behind their popularity is due to
their small variation in response to energy, dose, dose rate,
and their high response reproducibility.25
However, measuring small or narrow fields becomes a
challenge when using ionization chambers because of the
lack of lateral electronic equilibrium and the volume of the
chamber perturbing the field.
The ionization chamber used in this work is the Attix
ionization chamber (Model 449 CNMC Co. Nashville, TN
– USA). The reason for using a parallel plate ionization
chamber over a cylindrical or Markus ionization chamber is
that the Attix ionization chamber is flat and thin with the
front electrode consisting of a window of Kapton 0.025 mm
thick including a conductive coating and only 2 mm separa-
tion between the electrode plates.26,27 This provides a suit-
able geometry for surface dosimetry, although at the
expense of the chamber’s directional dependence. The Attix
ionization chamber has a large guard-ring 13.5 mm wide
that reduces the over-response in the build-up region to less
than 1%.28
2.C. Electrometer
The extreme thin sensitive volume adopted for the skin
diode Epi-5B requires the use of a high sensitivity electrome-
ter. The multichannel electrometer used for this work is based
on the AFE0064 which is an analog front end designed and
commercialized by Texas Instruments29 to readout flat panel
amorphous silicon detectors. The AFE0064 is equipped with
an optional double sampler for the subtraction of the signal
baseline — a feature that has been adopted in this work for
accurate measurement of the charge collected in the volume
of 1.25 9 103 mm3 of the Epi-5B. The AFE0064 also has
the major advantage to be able to sample the detector only
Medical Physics, 44 (10), October 2017
5405
when the beam is on, by synchronization of the detector sam-
pling and the analog to digital conversion with the linac sync
signal available from the linac stand. This modality has the
double advantage of minimizing the effects of electronic
noise and parasitic currents, and managing the dead time by
reading out the data while the beam is off. It readouts the
integral charge on a 16 bit scale of 4.8 pC, which corre-
sponds to the measured charge resolution of 7 9 105 pC.
The electrometer is readout by a USB2.0 interface managed
by a graphical interface installed into a standard laptop. For
further details on the electrometer architecture please refer to
Ref. [30].
2.D. Response linearity
The detectors’ linear response is investigated under the
standard conditions of 6 MV photon beam with field size
10 9 10 cm2 and 1.5 cm depth in a Virtual Water phantom
(Standard Imaging Inc., Middleton, WI, USA). The diodes’
response was recorded from 50 cGy up to 500 cGy. Uncer-
tainty is calculated by two standard deviations over
three consecutive measurement of each of three samples
from the same production batch and packaged with the same
technology.
2.E. Irradiations and radiation hardness
characterization
All measurements were performed using the Varian Cli-
nac 21Xi medical linac at the Illawarra Cancer Care Centre
of Wollongong Hospital (NSW, Australia) 6 MV photon
beams with a flattening filter. The irradiation of the epitax-
ial diode probes for studying the radiation damage effects
on charge collection efficiency of the skin diodes has been
performed by at GATRI, a Co-60 source at the Australian
National Science and Technology Organisation (ANSTO,
Lucas Heights, NSW, Australia). The effect of radiation
damage on degradation of radiation sensitivity of Epi-1C
epitaxial diode is presented in Ref. [31]. As proved by
Bruzzi et al., passive epitaxial diodes are usually less
affected by accumulated dose due to the thin sensitive
volume.32
To stabilize the skin diodes response, they were preirradi-
ated with dose increments of 20 kGy up to a total accumu-
lated dose of 80 kGy with response measurement at each
irradiation step. The response of the skin diodes was mea-
sured under standard conditions (1 MU equals 1 cGy) by
delivering 100 MU on 6 MV photon beam, field size of
10 9 10 cm2, at depth of 1.5 cm in the Virtual Water phan-
tom at a source to surface distance (SSD) of 100 cm. Error
bars are calculated as one standard deviation over five repeti-
tions of the same measurement.
2.F. Percentage depth dose
The percentage depth dose (PDD) measurements are
obtained for both the silicon epitaxial diodes Epi-5B and Epi-
5406 Vicoroski et al.: Silicon diode for skin dosimetry
1C and compared to the Attix ionization chamber. All these
detectors were placed, one at a time, in a recess machined in
a Virtual Water (VW) slab to ensure consistency of the scat-
tering conditions. A 10 cm thick slab of 30 9 30 cm2 VW
phantom is used for backscattering and several slabs ranging
from 0.1 cm to 20 cm are used to obtain the depth dose pro-
files. The samples are irradiated with 100 MU of a 6 MV
photon beam with a 10 9 10 cm2 field size at 100 cm SSD
for each measurement.
Additionally, PDD were obtained for the skin diode Epi-
5B with the epitaxial layer of the skin diode facing down-
wards and the diode being exposed to the incident radiation
beam. This was conducted under the same experimental con-
ditions as when the epitaxial layer of the skin diode was fac-
ing up, to determine whether orientation of the detector
would affect the PDD.
Monte Carlo Geant4 version 10.0p0133 simulations were
performed to determine the dose distribution at the build-up
region generated by the specific Linac used for this experi-
ment. The beam 10 9 10 cm2, 6 MV x-ray, passing through
a 30 9 30 9 30 cm3 water equivalent block phantom was
simulated.
The source to surface distance (SSD) was 100 cm, and
the beams were fired from phase space files created from
an EGSnrc Monte Carlo based system that models the Var-
ian 2100C Linac at Illawarra Cancer Centre in Wollon-
gong (Australia). The model used to represent the Linac
and the phantom setup, along with the physics of trans-
port, has been already validated by previous experimental
results.34 The physics processes modeled in the simulation
are from the Geant4 electromagnetic standard physics
package and include photoelectric effect, Compton scat-
tering and gamma conversion (photons), ionization,
Bremsstrahlung, and positron annihilation. The particle
range cut was set to 0.1 mm and the electron/positron
maximum step length was 0.1 mm. Dose was scored
inside the phantom at a voxel size of 0.02 9 1 9 1 mm3.
Each simulation was split into 10 parallel jobs each with
unique seeds and the mean dose reported. The standard
deviation is taken across 10 simulations to evaluate uncer-
tainties. In total, 4 9 1013 primary histories are simulated
where the primary histories represent the number of elec-
trons hitting the x-ray target in the linac head. The dose
uncertainty is approximately
0.5% of the dose at 1.5 cm
depth at the beam central axis (CAX) for the 10 9 10 cm2
field size.
2.G. Output factor and surface field size
dependence
The output factor is defined as the ratio of dose in water
(Dw) for a given beam collimator aperture (A) at a reference
depth (d) to the dose at the same point and depth for the refer-
ence collimator aperture (Aref)35,36:
Dw ðA; dÞ
Output Factor ¼
Dw ðAref ; dÞ
Medical Physics, 44 (10), October 2017
5406
The reference field size for this study was 10 9 10 cm2,
which was measured at isocenter and at a depth of 10 cm
(SSD is 90 cm), with delivered 100 MU each time. The out-
put factor was measured by placing the preirradiated skin
diodes (60 kGy irradiation dose by Co-60) in the center of
the field for field sizes and varying the jaws’ aperture from
2 9 2 cm2 to 30 9 30 cm2. A careful alignment of the
detector was performed by maximizing the detector response
for the 2 9 2 cm2 beam by a scan in the SUP-INF and lateral
directions using the couch positioning system (spatial resolu-
tion of
1 mm). A measurement of the field factor at surface
of the phantom was also performed to evaluate the capabili-
ties of the detector to measure the surface dose as a function
of the field size defined by the jaws. The silicon epitaxial
diode was benchmarked against Gafchromic EBT3 film, the
Attix ionization chamber, and the MOSkin, which are fre-
quently used for interface measurements and described in
several studies.16,37–40
The films were calibrated using 1.5 9 1.5 cm² squares
and scanned using a flatbed scanner (Epson 10000XL scan-
ner; Epson America, Inc. Long Beach, CA, USA) 24 h after
irradiation to allow for postirradiation development. The
films were scanned in transmission mode with color correc-
tion turned off, at a resolution of 72 dpi and 48 bit RGB data
format. Analysis of three scanning repetitions for each film
placed with the same orientation were performed using Ima-
geJ 1.46r software (National Institute of Health, Bethesda,
MD, USA) and one standard deviation of the optical density
was used as uncertainty. A template was used to scan the
films at the center of the scanner bed to avoid nonuniformity
associated with the position of the film in respect to the scan-
ner lamp or sensor. Only the red channel was used for analy-
sis of a region of interest (ROI) at the center of the film. The
films used to perform the measurements have been pres-
canned to evaluate the background optical density and to take
care of the nonuniformities associated with the specific film
cut. Butson et al.10 provided in-depth details of the prepara-
tion and scanning procedure adopted for the analysis of the
EBT3 film.
2.H. Dose per pulse dependence
The measurement of dose per pulse dependence (DPP)
refers to the evaluation of the change of the detector sensitiv-
ity due to variation of the instantaneous dose rate in a pulsed
radiation beam typical of a medical linear accelerator.41 It
was reported by Rikner and Grusell that n-type silicon diodes
are more sensitive to dose per pulse variation than compared
to p-type silicon detectors, thus showing an increase in sensi-
tivity with increasing dose per pulse.41
Therefore, an evaluation of these characteristics is vital to
determine if the technology adopted for the Epi-5B skin
diode can be used for quality assurance on medical linac. The
DPP were performed by measuring the diode and the Attix
chamber’s response placed at a 1.5 cm depth and irradiated
with a 10 9 10 cm2 field of 6 MV photon beam at 600 MU/
min within the phantom, varying SSD from 100 cm up to
5407 Vicoroski et al.: Silicon diode for skin dosimetry 5407

250 cm. The ratio between the charge measured by the sili-
con diode to the charge measured by the Attix ionization
chamber represents the DPP dependence of the detector and
is normalized to the dose per pulse of 2.78 104 Gy/pulse
corresponding to the reference SSD of 100 cm. This method
avoids variation of the spectrum of the radiation and can be
used to characterize only the dose rate dependence assuming
that the ion chamber’s dose per pulse dependence is
negligible.42

2.I. Angular dependence

The angular dependence of the skin diode (Epi-5B)
response was evaluated in two different conditions. The con-
dition (a) corresponds to the detector placed in a cylindrical
PMMA phantom at 15 cm depth (aligned with the central
FIG. 4. Response variation of the skin diode (Epi-5B) as a function of accu-
axis of the phantom) and the detector placed at isocenter of a
mulated dose from a Co-60 gamma source.
10 9 10 cm2 field of 6 MV photon beam. 100 MU are deliv-
ered at each angle and the response normalized to the 0
degree, corresponding to the beam perpendicular to the
detector’s face-up configuration. The condition (b) corre-
sponds to the detectors placed at surface of a 30 9 30 cm2
Virtual Water phantom with 10 cm of backscattering. In this
configuration, the Epi-5B is benchmarked, after correction
for its intrinsic angular dependence, to the ion chamber Attix
using 10 9 10 cm2 field size 6 MV photon beam at SSD of
100 cm and to EBT3 film. The response of the film is sam-
pled at the center of the field and averaging the response from
9 9 9 voxels corresponding to an equivalent area of
2.7 9 2.7 mm2. The gantry has been rotated from  60o to
+ 60o in 15 degree increments and responses normalized to
the 0 degree gantry position.

3. RESULTS
3.A. Radiation hardness
Figure 4 shows the variation of the response of the n-type
skin diode (Epi-5B) as a function of the dose accumulated
from the Co-60 gamma source. As expected there is a
response decrease with accumulated dose; however, the
response becomes stable with a variation of approximately
0.05%/kGy after 60 kGy.
3.B. Dose linearity
Figure 5 displays the dose linearity of the accumulated
dose varying from 50 cGy to 500 cGy of the Epi-5B and
Epi-1C epitaxial diodes after preirradiation by 60 kGy.
Table I summarized the dose calibration factors from charge
to dose extracted from the linearity measurements. The con-
version factors demonstrate that despite the sensitivity of the
Epi-5B being very low in comparison with the thicker epitax-
ial diode Epi-1C, its reproducibility is still within 0.2% with
excellent linearity. The reproducibility of the diode detectors
is determined based on two standard deviations of five
repeated measurements of each accumulated dose.
FIG. 5. Dose linearity response for the Epi-5B (squared symbols) and Epi-
1C (dotted) epitaxial silicon diodes. [Color figure can be viewed at wileyonli-
nelibrary.com]
TABLE I. Dose calibration factors of the Epi-5B and Epi-1C.
Diodes Response (pC/cGy)
Epi-1C 115.18
0.14
Epi-5B 53.12
0.03
3.C. Dose per pulse dependence
Figure 6 demonstrates the dose per pulse dependence
(DPP) response of the skin diode Epi-5B normalized to the
dose measured by the ion chamber at 2.78 9 104 Gy/pulse
corresponding to a 6 MV photon beam, 600 MU/min, with a
field size of 10 9 10 cm2 and at 1.5 cm depth in solid water
and SSD 100 cm.
The skin diode exhibits a decrease of 50% in sensitivity
for a 10 times fold dose rate (Gy/pulse) reduction.

Medical Physics, 44 (10), October 2017

5408 Vicoroski et al.: Silicon diode for skin dosimetry
FIG. 6. Dose per pulse (DPP) response of the skin diode Epi-5B estimated at
Dmax (1.5 cm) in solid water with a 6 MV photon beam of 10 9 10 cm2
field size varying SSD from 100 cm to 250 cm.
3.D. Percentage depth dose
Figure 7(a) shows the comparison of the percentage
depth dose (PDD) measurements from surface to 20 cm
depth in Virtual Water with the skin diode and the Attix
chamber normalized to dmax for a 6 MV photon beam with
10 9 10 cm2 radiation field area. The observed maximum
difference between the PDD measured with the skin diode
and calculated with Geant4, excluding the outsider point at
1 cm depth, is within
2% with an agreement within
+ 0.5% at a water equivalent depth of 0.07 mm. The differ-
ence between skin diode and the uncorrected Attix cham-
ber response at 0.07 mm depth is approximately + 2%.
This difference reduces to 0.2% if the ion chamber
response is corrected for skin dosimetry, as suggested by
Gerbi and Khan implying an effective depth of measure-
ment for the uncorrected response of approximately
0.03 mm.27,28,43
Figure 8(a) displays the comparison of the PDD measured
in standard conditions between the thick epitaxial Epi-1C
diode and the ion chamber Attix in Virtual Water. The
observed maximum difference between the PDD measure-
ments is + 5.5% representing an over-response of the epitax-
ial Epi-1C diode on a surface of the phantom while its
agreement with the Attix ionization chamber at depths greater
than 5 mm is within 1.5%.
This measurement was performed to show how sensitive
the WED is to the thickness of the epitaxial layer and the fab-
rication of an effective skin diode, even if the diodes utilize
the same “drop-in” packaging technology. Another example
of the accuracy required in the design of a skin dosimeter
using silicon is given by the results showed in Fig. 8(b)
where the skin diode has been irradiated face-down. The Epi-
5B face-down over-responses with + 20% more dose
recorded at surface in comparison to the Attix ionization
chamber. Obtained results are expected due to the steep dose
gradient close to the surface.
Medical Physics, 44 (10), October 2017
5408
3.E. Output factor and surface field size
dependence
Figure 9(a) displays the output factor measured by the
skin diode and compared to EBT3 film and MOSkin detector
at 10 cm depth within the Virtual Water phantom. As
expected, MOSkin and EBT3 film agree within 1.2% from a
square field size (SFS) of 0.5 up to 30 cm, confirming the
results obtained previously for small field dosimetry stud-
ies.35,44 The skin diode in the face-up configuration has been
tested from a SFS of 2 up to 30 cm with an agreement within
2% in respect to EBT3 film. The lack of data below a SFS of
2 cm is due to a poor signal-to-noise ratio as the dose is low
at very small field sizes and at a depth of 10 cm in the solid
water phantom.
In addition, Fig. 9(b) shows the response’s dependence to
field size variations at surface of the Attix chamber and the
skin diode. The responses are normalized to a field size of
10 9 10 cm2 and WED of the Attix chamber adjusted using
extra thin sheets of kapton above the IC. The configuration
with 0.05 mm of build-up that is corresponding to a total
WED of 0.075 mm matches the skin diode response within
3% for all the field sizes investigated.
3.F. Angular dependence
The surface dose increases when beam’s incidence angle
increases (tangential beams); correct measurements of the
skin dose at different angles of beam incidence are valuable
for both treatment planning system (TPS) verification36 and
for studies correlating acute skin toxicity to dose delivered to
the skin in head and neck or breast cancer treatments. In order
to measure accurately the skin dose from tangential beams,
the intrinsic angular dependence of the skin diode must be
compensated for. The angular response of the skin diode in
face-up configuration was investigated by placing the device
in a central axis of a cylindrical PMMA phantom of radius
15 cm and aligned with the linac’s isocenter. Figure 10
shows the result for the intrinsic angular dependence of the
Epi-5B normalized to the response of the detector at gantry
angle 0 degree (beam incident normally). The behavior of the
device confirms what is expected from a planar device with a
very thin sensitive volume: the largest variation ( 8%)
occurs at 90 degree and the attenuation of the beam incident
from the backside of the detector is approximately  2%.
The correction factor for a tangential beam with angle of
60
degree, as expected in a 3D conformal treatment of breast
cancer, will be 5% at largest.
In order to evaluate the skin diode ability to measure skin
dose at various beam incidence angles, 100 MU of a 6 MV
photon beam, field size 10 9 10 cm2, was delivered with the
gantry position rotated from  60o to + 60o in increments of
15 degree. The skin diode and the EBT3 film were placed on
a surface of the phantom and their readings were normalized
to the reading obtained at 0o gantry angle with a fixed SSD of
100 cm. Figure 11 shows the comparison of the angular
dependence of the skin diode response (corrected for its
5409 Vicoroski et al.: Silicon diode for skin dosimetry 5409

(a) (b)

FIG. 7. (a) Percentage depth dose measured by the skin diode in comparison to the Attix ionization chamber with no build-up cap and no correction factors
applied; (b) Comparison of the same dataset measured with Attix and Epi-5B with the simulations performed by the means of Geant4 in water and represented in
logarithmic scale of the depth in water. [Color figure can be viewed at wileyonlinelibrary.com]

(a) (b)

FIG. 8. PDD measured by Epi-1C in comparison to Attix IC in a solid water phantom; (a) Epi-1C is “face-up” (b) Epi-5B is “face-down”.

intrinsic angular dependence), the film and the Attix ionizing
chamber, as the beam incidence angle increases. The mea-
sured dose increases due to Dmax shifting toward the surface.
4. DISCUSSION
The skin diode (Epi-5B) is manufactured on n-type epitax-
ial silicon substrate which shows a radiation damage effect as
variation of the detector response as a function of the accu-
mulated dose. Figure 6 shows that the response of the skin
diode decreases by a factor of 25% in the first 20 kGy of
accumulated dose. For higher doses the response variation is
within 2% only that is explained by the saturation of the dif-
fusion length of the minority carriers. This is in agreement
with previous results on thinned and epitaxial silicon detector
studies on substrate with similar resistivity.32,45
Dose rate dependence in silicon diodes can potentially
affect the capability of the detector to reconstruct accurately
dose profiles and depth dose distributions. The skin diode
shows a variation of less than 5% within the range between
104 and 2.78 9 104 Gy/pulse confirmed by the PDD mea-
surements that show an agreement within
1% compared to
the parallel plate ion chamber.
The Geant4 simulations of the PDD in build-up region have
been used to derive the effective WED of 0.07 mm and
0.03 mm of the skin diode and the Attix chamber, respectively.
The thickness of the sensitive volume of the detector in
step dose gradients is a major factor for accuracy in skin
dosimetry at water depth of 0.07 mm. The Epi-1C epitaxial
diode with 50 lm thick epitaxial layer was found to have a
WED of 1 mm, making this diode unsatisfactory for skin
dosimetry.

Medical Physics, 44 (10), October 2017

5410 Vicoroski et al.: Silicon diode for skin dosimetry
(a)
FIG. 9. (a) Output factor measured by MOSkin, EBT3 film, and skin diode Epi-5B at 10 cm depth in Virtual Water phantom and (b) field size dependence of the
skin diode placed at surface of the phantom in comparison with Attix with different build-up thickness. [Color figure can be viewed at wileyonlinelibrary.com]
FIG. 10. Normalized intrinsic angular response of the skin diode placed in
center of a cylindrical PMMA phantom of radius of 15 cm as a function of
the angle. Center of the phantom is placed at isocenter. Normal incident
angle is 0 degree.
The skin diode with epitaxial layer 0.005 mm has demon-
strated to be suitable for skin dosimetry with a WED of
0.075 mm. The angular response at surface of the phantom
agrees within 3% of relative variation between the skin diode
and EBT3 film with beam incident angles in the range
between
60 degree. The skin diode shows a minimal angu-
lar dependence comparable with the film response, but it also
measures the dose at the WED of 0.07 mm which cannot be
achieved by film. Film has a WED of dose measurement at
approximately 0.1 mm due to the polyamide protection layer
above the radiation sensitive polymeric substrate (0.05 mm
thick).19,43 The Attix IC response shows a large over-response
up to + 20% at
60 degree as confirmed by other studies.24
Skin dose increases also as a function of the field size and
depends on WED of measurement. In this sense, the skin
diode and Attix IC’s response with field size are different
Medical Physics, 44 (10), October 2017
5410
(b)
FIG. 11. Comparison of the normalized surface dose angular dependence
measured by the skin diode, EBT3 film, and Attix chamber at surface of a
Virtual Water phantom for a 6 MV photon beam. [Color figure can be
viewed at wileyonlinelibrary.com]
because the WED of the IC is about 0.025 mm, in contrast to
the WED of the skin diode, estimated to be approximately
0.075 mm.28 Adding two kapton sheets, each one 25 lm
thick, and matching field size response of the IC to the skin
diode confirms that the skin diode WED is approximately
between 0.07 and 0.08 mm, as estimated by the Geant4
simulations.
The use of the skin diode for low dose rate measurements
is challenging due to the trade-off between thin sensitive vol-
ume and signal-to-noise ratio. This was primarily observed
during output factor measurements, where measurements in
fields below 2 9 2 cm2 were impossible with the electrome-
ter adopted. Adopting a larger area for the p+ top junction
without enlarging the overall die’s size by reducing the exter-
nal guard-ring would double the sensitivity of the sensor
5411 Vicoroski et al.: Silicon diode for skin dosimetry
solving the issue of the signal-to-noise ratio. This will
increase the amount of charge collected without perturbing
the beam with a larger silicon die. Although variation of the
detector response as a function of temperature in a thin epi-
taxial diode is expected to be negligible because diffusion
length of the charge carriers is shorter than thickness of the
epitaxial layer, the skin diode response variation with tem-
perature will be investigated in a separate work. For a clini-
cal application of the skin diode also the electron
contamination effects due to different collimator configura-
tions and presence of the flattening filter will be investi-
gated and reported.
5. CONCLUSION
This work presents a feasibility study toward the develop-
ment of a silicon diode that can measure the skin dose accu-
rately in respect to clinical standards for QA of medical
linacs. The use of a diode for skin dose measurements is of
interest due to the possibility of developing an array of such
detectors for real-time, in vivo skin dose mapping in several
radiotherapy and diagnostic imaging modalities, where the
risk of acute skin toxicity is a serious issue for the patients’
quality of life. The skin diode is fabricated on a 0.005 mm
thick epitaxial silicon layer placed above 0.4 mm thick sili-
con substrate and embedded in a thin water equivalent pack-
age using “drop-in” packaging technology. The skin diode
has a WED of 0.075
0.005 mm, as recommended by the
ICRP and can be used to measure the skin dose delivered by
oblique radiation beams.
The skin diode can also be used for conventional QA
dosimetry in radiation therapy because it has a linear dose–
response over a dynamic range from 25 to 500 cGy and mini-
mal intrinsic angular dependence, in conjunction with the
ability to measure the PDD within 2% compared to IC.
The challenge to be addressed is the improvement of sig-
nal-to-noise ratio for low dose rate applications, which could
be achieved by increasing the sensitive area of the skin diode
and/or electrometer sensitivity, and could possibly lead to a
universal diode for QA in radiation therapy.
ACKNOWLEDGMENTS
Authors thank Dr Justin Davies for his support with irradi-
ating the detectors at the GATRI facility (ANSTO, Lucas
Heights, NSW, Australia). Thank you to FBK Foundation
Trento (Italy) for providing the n-type silicon epitaxial
samples for testing throughout the project.
CONFLICT OF INTEREST
No conflict of interest or financial involvement to be
disclosed.
a)
Author to whom correspondence should be addressed. Electronic mail:
marcop@uow.edu.au.
Medical Physics, 44 (10), October 2017
5411
REFERENCES
1. Fisher B, Anderson S, Bryant J, et al. Twenty-year follow-up of a ran-
domized trial comparing total mastectomy, lumpectomy, and lumpec-
tomy plus irradiation for the treatment of invasive breast cancer. N Engl
J Med. 2002;347:1233–1241.
2. Veronesi U, Cascinelli N, Mariani L, et al. Twenty-year follow-up of a
randomized study comparing breast-conserving surgery with radical
mastectomy for early breast cancer. N Engl J Med. 2002;347:1227–
1232.
3. Early Breast Cancer Trialists’ Collaborative Group. (EBCTCG), Effect
of radiotherapy after breast-conserving surgery on 10-year recurrence
and 15-year breast cancer death: meta-analysis of individual patient data
for 10 801 women in 17 randomised trials. Lancet. 2011;378:771–784.
4. ICRP Task Group. The biological basis for dose limitation in the skin. A
report of a Task Group of Committee 1 of the International Commission
on Radiological Protection. Ann ICRP. 1992;22:1–104.
5. Mellenberg DE. Determination of build-up region over-response correc-
tions for a Markus-type chamber. Med Phys. 1990;17:1041–1044.
6. Shani G. Radiation Dosimetry Instrumentation and Methods, 2nd Edn.
Boca Raton, FL: CRC Press LLC; 2001, pp. 205. ISBN 0-8493-1505-0.
7. Nilsson B, Montelius A. Fluence perturbation in photon beams under
nonequilibrium conditions. Med Phys. 1986;13:191–195.
8. Kawachi T, Saitoh H, Inoue M, et al. Reference dosimetry condition and
beam quality correction factor for CyberKnife beam. Med Phys.
2008;35:4591–4598.
9. Hardcastle N, Soisson E, Metcalfe P, Rosenfeld AB, Tome WA. Dosi-
metric verification of helical tomotherapy for total scalp irradiation.
Med Phys. 2008;35:5061–5068.
10. Butson MJ, Cheung T, Yu PKN. Scanning orientation effects on Gaf-
chromic EBT film dosimetry. Australasian Phys Eng Sci Med.
2006;29:281–284.
11. Ostwald PM, Kron T, Hamilton CS, Denham JW. Clinical use of carbon-
loaded thermoluminescent dosimeters for skin dose determination. Int J
Radiat Oncol Biol Phys. 1995;33:943–950.
12. Mijnheer B, Beddar S, Izewska J, Reft C. In vivo dosimetry in external
beam radiotherapy. Med Phys. 2013;40:070903.
13. Colussi VC, Beddar S, Kinsella TJ, Sibata CH. In vivo dosimetry using
a single diode for megavoltage photon beam radiotherapy: implementa-
tion and response characterisation. J Appl Clin Med Phys. 2001;2:210–
218.
14. Zhuang AH, Olch AJ. Validation of OSLD and a treatment planning sys-
tem for surface dose determination in IMRT treatments. Med Phys.
2014;41:081720.
15. Jong WL, Wong JHD, Ng K, et al. Characterisation of MOSkin detector
for in vivo dose measurement during megavoltage radiotherapy. J Appl
Clin Med Phys. 2014;15:120–130.
16. Kwan IS, Rosenfeld AB, Qi ZY, et al. Skin dosimetry with new MOS-
FET detectors. Rad Meas. 2008;43:929–932.
17. Jursinic PA. Angular dependence of dose sensitivity of surface diodes.
Med Phys. 2009;36:2165–2171.
18. AAPM Report No. 87. Diode in vivo dosimetry for patients receiving
external beam radiation therapy. Report of Task Group 62 of the Radia-
tion Therapy Committee; 2005: 1-64.
19. Meiler RJ, Podgorsak MB. Characterization of the response of commer-
cial diode detectors used for in vivo dosimetry. Med Dosim. 1997;22:31–
37.
20. Yorke E, Alecu R, Ding L, et al. Diode in vivo dosimetry for patients
receiving external beam radiation therapy. Report of Task Group 62 of
the Radiation Therapy Committee; 2005. Madison, WI.
21. Scalchi P, Francescon P. Calibration of a MOSFET detection system for
6-MV in vivo dosimetry. Int J Radiat Oncol Biol Phys. 1998;40:987–
993.
22. Wong JHD, Fuduli I, Carolan M, et al. Characterization of a novel two
dimensional diode array the “magic plate” as a radiation detector for
radiation therapy treatment. Med Phys. 2012;39:2544–2558.
23. Petasecca M, Cullen A, Fuduli I, et al. X-Tream: a novel dosimetry sys-
tem for synchrotron microbeam radiation therapy. J Instrum. 2012;
P07022:1–13.
24. Petasecca M, Alhujaili S, Aldosari AH, et al. Angular independent sili-
con detector for dosimetry in external beam radiotherapy. Med Phys.
2015;42:4708–4718.
5412 Vicoroski et al.: Silicon diode for skin dosimetry
25. Das IJ, Cheng CW, Watts RJ, et al. Accelerator beam data commission-
ing and procedures. Report of the Task Group 106 Therapy Physics
Committee of the AAPM, August; 2008.
26. Cross P. The use of parallel plate ion chambers to determine surface
dose of a 6 MV photon beam. Australas Phys Eng Sci Med.
1992;15:208–213.
27. Gerbi BJ, Khan FM. Plane-parallel ionization chamber response in the
buildup region of obliquely incident photon beams. Med Phys.
1997;24:873–878.
28. Gerbi BJ. The response characteristics of a newly designed plane-paral-
lel ionization chamber in high-energy photon and electron beams. Med
Phys. 1993;20:1411–1415.
29. AFE0064. Texas Instruments, 64 Channel Analog Front End for Digital
X-Ray Detector. http://www.ti.com/lit/ds/symlink/afe0064.pdf. Accessed
September 3; 2009.
30. Fuduli I, Newall MK, Espinoza AA, et al. Multichannel data acquisition
systems comparison for quality assurance in external beam radiation
therapy. Rad Meas. 2014;71:338–341.
31. Aldosari AH, Espinoza A, Robinson D, et al. Characterization of an
innovative p-type epitaxial diode for dosimetry in modern external beam
radiotherapy. IEEE Trans Nucl Sci. 2013;60:4705–4712.
32. Bruzzi M, Bucciolini M, Casati M, et al. Epitaxial silicon devices for
dosimetry applications. App Phys Lett. 2007;90:172109.
33. Agostinelli S, Allison J, Amako K, et al. Geant4-a simulation toolkit.
Nucl Instrum Methods Phys Res, Sect A. 2003;506:250–303.
34. Oborn BM, Williams M, Bailey M, Carolan MG. IMRT treatment moni-
tor unit verification using absolute calibrated BEAMnrc and Geant4
Monte Carlo simulations. J Phys Conf Ser. 2014;489:012020.
Medical Physics, 44 (10), October 2017
5412
35. Sauer OA, Wilbert J. Measurement of output factors for small photon
beams. Med Phys. 2007;34:1983–1990.
36. Khan FM. The Physics of Radiation Therapy, 3rd edn. Philadelphia: Lip-
pincott Williams & Wilkins; 2003.
37. Kwan IS, Wilkinson D, Cutajar D, et al. The effect of rectal heterogene-
ity on wall dose in high dose rate brachytherapy. Med Phys.
2009;36:224–232.
38. Hardcastle N, Cutajar DL, Metcalfe PE, et al. In vivo real-time rectal wall
dosimetry for prostate radiotherapy. Phys Med Biol. 2010;55:3859–3871.
39. Qi ZY, Deng XW, Huang SM, et al. Real-time in vivo dosimetry with
MOSFET detectors in serial tomotherapy for head and neck cancer
patients. Int J Radiat Oncol Biol Phys. 2011;80:1581–1588.
40. Alnawaf H, Butson M, Yu PK. Measurement and effects of MOSKIN
detectors on skin dose during high energy radiotherapy treatment. Aus-
tralas Phys Eng Sci Med. 2012;35:321–328.
41. Rikner G, Grusell E. Effects of radiation damage on p-type silicon detec-
tors. Phys Med Biol. 1983;28:1261–1267.
42. Wang Y, Easterling SB, Ting JY. Ion recombination corrections of ion-
ization chambers in flattening filter-free photon radiation. Appl Clin
Med Phy. 2012;13:3578.
43. Devic S. Accurate skin dose measurements using radiochromic film in
clinical applications. Med Phys. 2006;33:1116–1124.
44. Aldosari AH, Petasecca M, Espinoza A, et al. A two dimensional silicon
detectors array for quality assurance in stereotactic radiotherapy: Magic-
Plate-512. Med Phys. 2014;41:091707.
45. Petasecca M, Pignatel GU, Moscatelli F, Passeri D, Caprai G. Numerical
analysis of thinned silicon detectors. Nucl Instrum Meth A.
2007;572:319–320.