IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 54, NO. 4, AUGUST 2007 1151
Investigations on Photon Energy Response of
RadFET Using Monte Carlo Simulations
Peter Beck, Member, IEEE, Florian Bock, Helmuth Böck, Marcin Latocha, Robert A. Price, Member, IEEE,
Sofia Rollet, and Michael Wind
Abstract—We describe investigations of RadFET energy re-
sponse simulated with Geant4 and FLUKA2005 Monte Carlo
codes. An analysis of energy deposition is carried out for photon
irradiation with energies between 35 keV and 2 MeV. The ab-
sorbed dose in the silicon dioxide layer (few hundred nanometers)
is compared for both transport codes.
Index Terms—Basic mechanism, dosimetry, energy deposition,
RadFET, radiation detectors, radiation transport, simulation.
I. INTRODUCTION
M
ODELLING of effects in semiconductor components
caused by radiation exposure is becoming evermore im-
portant. On the one hand radiation hardness testing is expensive
in time and costs; therefore a partial replacement by radiation
simulations is of major interest. On the other hand modelling
assists the development of semiconductor devices, and allows
optimizing characteristics either of radiation hardness or ra-
diation sensitivity. Several Monte Carlo codes are available
for modelling irradiation measurements. For the validation of
the models, dosimetry in microstructures plays an important
role [1]–[4]. As an example of characterizing radiation effects
in semiconductors, we describe investigations of the RadFET
energy response calculated with Monte Carlo transport simula-
tion codes. The research has been carried out in the framework
of the EURADOS working group WG6 initiative and the
CONRAD (Coordinated Network for Radiation Dosimetry),
WP4: Uncertainty Assessment in Computational Dosimetry
[5], where simulation problems are presented to the dosimetry
community for the calculation of specified parameters and fur-
ther comparison.
1
A RadFET is a metal-oxide semiconductor
field-effect transistor (MOSFET) optimized for use as a sensor
of ionizing radiation [6]–[8]. The radiation sensitivity of a
RadFET is caused by manipulating the production process so
Manuscript received October 6, 2006; revised December 12, 2006. This work
was supported by the Austria Research Center.
P. Beck and S. Rollet are with ARC Seibersdorf Research, 2444 Seibersdorf,
Austria (e-mail: peter.beck@arcs.ac.at; sofia.rollet@arcs.ac.at).
F. Bock and M. Wind are with ARC Seibersdorf Research, 2444 Seibersdorf,
Austria, and also with the Atomic Institute of the Austrian Universities, Wien,
Austria (e-mail: florian.bock@arcs.ac.at; michael.wind@arcs.ac.at).
H. Böck is with the Vienna University of Technology, Atomic Institute, 1020
Vienna, Austria (e-mail: boeck@ati.ac.at).
M. Latocha is with ARC Seibersdorf Research, 2444 Seibersdorf (Austria),
and also with the Institute of Nuclear Physics, Polish Academy of Sciences,
31-342 Kraków, Poland (e-mail: marcin.latocha@arcs.ac.at).
R. A. Price is with the Department of Radiography, City University, London
EC1V OHB, U.K. (e-mail: r.price@city.ac.uk)
Digital Object Identifier 10.1109/TNS.2007.902350
1
http://www.eurados.org/conrad/conrad_overview.htm
that the device has a thick oxide-layer. Conventional radia-
tion-hard MOSFETs have an oxide layer with few nanometers
thickness [9]–[11]. The change of the dose response to ionizing
radiation can be related to the energy imparted in the gate oxide.
This makes the RadFET usable as a dosimeter. The absorbed
dose response of the RadFET dosimeter can be calculated by
simulation of the energy imparted in the gate oxide layer [12].
The main focus of this investigation is the comparison of
Monte Carlo transport simulation of the energy deposition in
the gate oxide for photon irradiation with energies between
35 keV and 2 MeV. All simulations performed for the RadFET
were done using Geant4.7.1 [13] and FLUKA (version 2005)
[14]. The RadFET geometry and material data were taken
from [5]. The energy response of the absorbed dose in the
silicon dioxide layer is compared for the transport simulation
codes. Both models describe the functionality and response
of the RadFET in the case of ionizing radiation exposure. In
particular, the comparison of the results of the two simulation
codes shows different capabilities and possible limitations of
these codes because of the fact that the sensitive volume in this
example is very small, pointing out the importance of a detailed
description of the energy loss fluctuations in thin layers [15].
Further details can be found in [16].
II. MONTE CARLO SIMULATION
For the simulations, the geometry of the RadFET can be
reduced to the most important parts schematically shown in
Fig. 1. The geometry of the RadFET in the proximity of the gate
oxide is shown together with the regions around it, modelled
as a stack of parallelepipeds, all with the same layer size but
different thickness (not in scale). The field oxide is positioned
left and right of the gate oxide. Its width and consequently
also the lateral dimensions of all the materials below and above
it will be varied during our investigations. Simulations are done
with and without the top layer which is representing the metal
lid of the RadFET. The device is called capped when the metal
lid is present; when no lid is used the device is called uncapped.
The RadFET simulation geometry is completely surrounded by
vacuum, and the source has exactly the same lateral extension
as the RadFET.
The coordinate system of the irradiation geometry in Fig. 2 is
consistent with the coordinate system in Fig. 1, thus the RadFET
is irradiated in the direction of the layers above the gate oxide.
In this investigation the field oxide thickness is 250 m.
For both Monte Carlo simulation codes, the thin gate oxide
layer represents a borderline of their validity, both in the energy
and thickness domain.
0018-9499/$25.00 © 2007 IEEE