Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
1 | P a g e
Fabrication and Characterization of Bulk Diamond
Radiation Detectors
Park Sun Myung (A0110404E)
April 3, 2017
Supervised by: Assoc Professor Andrew A. Bettiol
2 | P a g e
Abstract
Single-crystal electronic grade CVD diamonds possess outstanding electrical properties that
makes them suitable for radiation detector use. A diamond radiation detector was fabricated
using a coplanar interdigitated electrode structure optimised through computer simulations.
The detector was tested using a Pu-239, Am-241, Cm-244 triple alpha source. The charge
collection efficiency values measured exceeded the performance of a commercial silicon
surface barrier detector that was used as a reference by 7%. The leakage current measured was
low and is in good agreement with reported values. While the energy resolution for the diamond
detector was relatively worse than the energy resolution for the reference silicon detector, the
three alpha peaks were well-defined and can be improved by optimising the fabrication process.
CONTENTS
3 | P a g e
Table of Contents 1 Acknowledgements ............................................................................................................................ 4
2 Introduction ........................................................................................................................................ 5
3 Review of Diamond Detectors ........................................................................................................... 8
3.1 Diamond Synthesis ....................................................................................................................... 8
3.2 Properties of Diamond .................................................................................................................. 9
3.3 Current Research ......................................................................................................................... 11
4 Theory ............................................................................................................................................... 13
4.1 Introduction ................................................................................................................................. 13
4.2 Charge Transport......................................................................................................................... 15
4.3 The Shockley-Ramo Theorem .................................................................................................... 18
4.4 Calculation of Charge Collection Efficiency .............................................................................. 22
5 Detector Simulations ........................................................................................................................ 23
5.1 Introduction ................................................................................................................................. 23
5.2 Methodology ............................................................................................................................... 23
5.3 Simulation Results ...................................................................................................................... 26
6 Fabrication and Testing .................................................................................................................. 30
6.1 Detector Fabrication Procedure................................................................................................... 30
6.2 Fabrication Issues ........................................................................................................................ 33
6.3 Testing and Results ..................................................................................................................... 35
7 Conclusion ........................................................................................................................................ 41
7.1 Summary ..................................................................................................................................... 41
7.2 Future Research........................................................................................................................... 41
References ............................................................................................................................................ 43
Appendix .............................................................................................................................................. 45
ACKNOWLEDGEMENTS
4 | P a g e
1 Acknowledgements
I would like to express my deep gratitude to my supervisor, Assoc Prof. Andrew Bettiol
and co-supervisors Mi Zhao Hong and Tan Hong Qi for all of the advice and help they have
given over the past year. I have learned a lot from them in the past year, not just academically
but in other aspects of life as well. Without their guidance, I would not have the many
opportunities that await me after my graduation.
I also wish to thank my parents for the unwavering support throughout my life and
especially through university. I am truly grateful for the sacrifices they have made for me and
my brother. They are my sources of inspiration when the going gets tough. I would also like
to thank my brother for putting up with all my late nights doing work or otherwise.
Park Sun Myung
INTRODUCTION
5 | P a g e
2 Introduction
Radiation detectors are instruments that detect, track and measure ionising radiation.
Ionising radiation refers to all forms of radiation that are energetic enough to eject electrons
from atoms or molecules. Some examples include gamma rays, X-rays and extreme
ultraviolet of the electromagnetic spectrum, and alpha and beta particles emitted via
radioactivity. Radiation detectors have been playing important roles in science for more than
a century, having facilitated numerous scientific breakthroughs such as the discovery of
exotic particles. Detectors can be generally classified into 2 main types, gaseous ionisation
chambers and solid state detectors. Perhaps one of the most notable examples among the
former is the Geiger counter [1]: incident ionising radiation cause a series of electron
avalanche ionisation effects that produce an amplified, measurable electrical signal. While
Geiger counters are relatively cheap and reliable for general use, they can only detect the
presence of ionising radiation; they cannot resolve the energies of ionising radiation.
Solid state semiconductor detectors, on the other hand, can measure the particle's energy
in addition to its detection and are widely used in medical applications, spectroscopy and as
particle detectors. Most semiconductor detectors used today are made of silicon. Owing to
extensive research over the past few decades, silicon detectors are highly efficient and
consistent in particle detection [2]. However, there are some drawbacks. Silicon detectors
may require cooling systems to reduce electrical noise due to leakage currents, and they are
susceptible to radiation damage. Radiation damage degrades the performance of the detector
over time and shortens the detector’s effective lifespan.
In situations where such issues are rampant, diamond presents itself as a very promising
alternative [2]. Among its many outstanding properties, diamond has a large energy bandgap,
which results in less electrical noise and a higher signal-to-noise ratio. The relatively stronger
interatomic covalent bonding make diamond more resistant to radiation damage.
Furthermore, there has been great advancements in chemical vapour deposition (CVD)
growth processes for single-crystal diamonds. Thus, it is hardly surprising that there is
significant interest in the use of diamond as radiation detectors.
INTRODUCTION
6 | P a g e
Some select applications where diamond detectors display superior performance over
silicon detectors are in detectors for excimer lasers [3] and particle accelerators. Excimer
lasers are a type of ultraviolet laser. The relatively large bandgap of diamond renders it blind
to visible light, thus improving discrimination between deep UV light and visible light.
Diamond’s superior radiation hardness and operational lifetime over silicon makes it a good
alternative for both of applications. For case of particle accelerators, significant research is
being conducted to bring the performance of diamond detectors up to par or even outperform
current silicon detectors [2].
Now I will give a brief introduction of my project, leaving further details to the rest of
this report.
The aim of my project is the optimisation, fabrication and characterisation of coplanar
interdigitated electrode designs on bulk diamond detectors for the detection of alpha particles.
The optimisation of detector performance is performed using COMSOL and MATLAB
simulations of diamond detectors with various electrode configurations. The simulations are
based on a scanning microscopy technique known as ion beam induced charge (IBIC)
microscopy. The metric used for performance comparison is the charge collection efficiency
(CCE). Based on the simulation results and fabrication limitations, an optimal electrode
configuration is chosen for fabrication. Electrodes comprised of Cr/Au bimetal layer are
deposited onto a 4 × 4 × 0.5 mm CVD single-crystal electronic grade diamond sample
provided by IIa Technologies Pte Ltd. The diamond detector is then characterised using a
radioactive alpha particle source and compared with the performance of a commercially
available silicon detector (ORTEC surface barrier Si detector). Characterisation involves the
measurement of the charge collection efficiency (CCE) of the diamond detector versus
applied bias, leakage current, energy resolution and the signal-to-noise ratio (SNR).
To summarise, my project comprises of the following:
Optimisation of coplanar interdigitated electrode configurations on bulk single-crystal
diamond detectors through simulations
Fabrication of Cr/Au interdigitated electrodes on 4 × 4 × 0.5 mm single-crystal
diamond based on the optimal electrode configuration
Characterisation of the diamond detector with a radioactive alpha particle source
INTRODUCTION
7 | P a g e
Lastly, I will give an outline of this report. In section 3, I will give a comprehensive
review of diamond detectors and their properties, with some emphasis on the advantages they
hold over silicon detectors. Section 4 will cover the physics of particle detection in intrinsic
diamond detectors. Sections 5 and 6 form the main body of my thesis. Section 5 is dedicated
to the details of the simulations used in the optimisation of interdigitated electrode geometry
and the simulation results. Section 6 illustrates the fabrication process and the problems
encountered during the process and provides the characterisation results of the fabricated
diamond detector. Section 7 summarizes the findings, concludes the report, and considers
some possible future directions for subsequent research on diamond detectors.
REVIEW OF DIAMOND
8 | P a g e
3 Review of Diamond Detectors
3.1 Diamond Synthesis
Diamond is a remarkable material that has outstanding physical, optical, electrical and
chemical properties [2]. Being the hardest naturally occurring material, diamond is already
heavily relied upon for industrial purposes such as cutting or grinding. Diamond has the
highest thermal conductivity, about 5 times that of copper. Diamond also possesses a host of
other outstanding properties such as having a large band gap, high charge carrier mobility and
high resistivity that makes it an excellent candidate for detector use. However, progress in
this area had been hampered for several decades due to the following factors:
Expensive; single-crystal diamonds could only be sourced from natural diamonds,
limited by their size and quality
Difficult to produce high quality single-crystal diamonds with low impurities via
CVD growth processes
Difficult to pattern using conventional lithography techniques due to hardness and
chemical inertness
Advancements in chemical vapour deposition diamond growth processes in recent years
have made high quality single-crystal diamonds readily available [4]. Chemical vapour
deposition is a chemical process that involves the homoepitaxial growth of diamond on a
substrate under highly controlled conditions [5]. The source of carbon is a carefully
controlled gaseous mixture containing hydrocarbons such as methane. To grow single-crystal
diamonds, the substrate must be a single-crystal diamond as well. The gaseous mixture is
heated by microwaves, direct current, hot filament, lasers or other energizing sources to form
a plasma of temperatures above 2000 °C. This causes the thermal disassociation of hydrogen,
thereby producing carbon radicals that form single bonds with the carbon atoms on the top
layer of the substrate, which is kept at a much lower temperature. Although diamond is a
metastable state of carbon in the conditions above, CVD takes advantage of the fact that
diamond has a faster nucleation and growth rate than graphite. This enables the growth of
CVD diamonds under less extreme conditions than the high pressure high temperature
REVIEW OF DIAMOND
9 | P a g e
method. Dopants may be introduced in the growth process but there are no shallow dopants
available. The closest is boron with an acceptor level of 0.37 eV.
Single-crystal CVD diamonds are available in sizes of up to 8 × 8 × 2 mm, with a high
degree of control over the impurity levels. CVD diamonds can be classified into the
following categories:
Single-crystal CVD diamond
o Optical grade: Low absorption and birefringence diamond
o Detector grade: Ultrahigh purity for electronic and quantum states
Polycrystalline CVD diamond
o Optical grade: Engineered for far infrared laser optical applications
o Electronic grade: Ultrahigh purity material for large area passive electronics
o Thermal grade: High thermal conductivity diamond for heat spreading
o Mechanical grade: High strength diamond for precision machining
o Electrochemistry grade: Boron doped diamond for electrochemical
applications
The classification of diamonds above is obtained from The Element Six CVD Diamond
Handbook [5]. The diamond used for this project is detector grade single-crystal CVD
diamond supplied by IIa Technologies Pte Ltd.
3.2 Properties of Diamond
The electronic properties of diamond pertaining to radiation detection are listed in Table
1, in contrast to the electronic properties of silicon. As some of these properties are subject to
variability from diamond to diamond, they were obtained directly from IIa Technologies Pte
Ltd, the supplier for the diamond sample used in this project.
REVIEW OF DIAMOND
10 | P a g e
Table 1 Properties of diamond and silicon relevant to radiation detection
Property Diamond Silicon Reference
Band gap (eV) 5.47 1.12 [2]
Energy to form e-h pair (eV) 13.2 3.6 [6, 7]
Electron mobility at 300 K (cm2 V-1 s-1) 2200 1500 [8]
Hole mobility at 300 K (cm2 V-1 s-1) 1600 450 [8]
Saturation carrier velocity (× 107 cm s-1) 2.7 1 [8]
Breakdown field (V cm-1) 107 3 x 105 [8]
Resistivity (Ω cm) > 1012 2.3 x 105 [2]
Dielectric constant 5.7 11.9 [8]
Thermal conductivity (W m-1 K-1) 2200 148 [8]
Atomic displacement energy (eV) 43 13-20 [9]
Density (g cm-3) 3.5 2.33 [2]
Atom density (cm-3) 1.75 x 1023 4.83 x 1022 [2]
Due to the large band gap, diamond has higher resistivity and a higher breakdown electric
field. Therefore, diamond does not require a p-n junction under reverse bias to reduce dark
current under strong electric fields. Dark current refers to leakage current in the absence of
ionising radiation. It is caused by the random thermal generation of electrons and holes. This
makes diamond suitable for use at room temperature while many types of silicon detectors
require cooling as dark currents are a source of detector noise. The high thermal conductivity
of diamond also helps in dissipating heat away during operation. A consequence of the large
band gap however, is the higher minimum energy required to produce an electron-hole pair.
Radiation detection is dependent on the movement of electron and hole pairs produced by the
ionising radiation. They are separated under the influence of the electric field between the
electrodes. The Shockley-Ramo theorem shows that the movement of the charge carriers
induces an instantaneous current in the electrodes. This current is directly proportional to the
number of electron-holes pairs and their velocities. While the measured signal is smaller in
magnitude, it is more than compensated for by the relatively lower dark current in diamond
detectors compared to silicon detectors.
Diamond has high electron and hole mobility and high saturation velocity, resulting in
faster charge collection under a given electric field and thus faster detector response time.
Diamond also has a smaller dielectric constant than silicon, yielding a smaller capacitance as
an electronic device and therefore, better noise performance of the associated front-end
electronics.
The most notable quality of diamond is its radiation hardness. It arises from the high
atomic displacement energy, energy required to displace a carbon atom from its lattice site.
REVIEW OF DIAMOND
11 | P a g e
Significant research into radiation hardness was conducted by the RD42 collaboration at
CERN. Meier et al. [10] showed that CVD diamond detector performance is unaffected even
after exposure to 24 GeV/c and 500 MeV protons for fluences of up to 1×1015 particles/cm2
while the silicon diode showed signs of radiation damage above a fluence of 1.5×1014
particles/cm2. However, recent research by Zamboni et al. [11] and Grilj et al. [12] have
shown that the opposite outcome occurs for ionising particles in the low MeV range. In these
papers, it is said that diamond detectors had greater charge collection efficiency (CCE)
degradation than silicon detectors. CCE is an important measure of performance of a
semiconductor radiation detector. The above behaviour may be attributed to the smaller
inelastic particle-carbon cross section and the lighter nuclear fragments involved in secondary
collisions. Energy loss through inelastic collisions dominate in the high MeV/GeV range.
Nevertheless, much of the motivation for radiation hard diamond detectors originates from
high-energy physics research which often involves ionising particles of at least 100 MeV, a
region where diamond holds a clear advantage.
Figure 1 Simple diagram of a sandwich semiconductor detector. Blue region represents the
semiconductor detection region. Yellow regions represent the electrodes.
3.3 Current Research
Much research has been conducted for the use of diamond detectors for alpha detection.
Various diamond dimensions and electrode structures have been studied thus far. Galbiati et
al. [13] tested a 4.7 × 4.7 × 0.5 mm single-crystal high purity CVD diamond with
239Pu/241Am/244Cm triple α source. The electrodes were of the sandwich structure as shown in
Figure 1, with DLC/Pt/Au trilayer; DLC is diamond-like carbon, obtained via energetic
bombardment of carbon atoms on the diamond substrate. Calibration was done by assigning
REVIEW OF DIAMOND
12 | P a g e
5.15 MeV, 5.48 MeV and 5.80 MeV for the 3 respective alpha sources. The energy resolution
was found to be 64 keV, approximately 1.1% of the alpha energies. This proves that diamond
detectors have excellent energy resolution. 5 MeV alpha particles have a penetration depth of
about 10 μm in diamond. Tests by Galbiati et al. using the same diamond sample with TiW
electrodes and 90Sr beta rays have shown 100% charge collection efficiency above a certain
threshold bias voltage.
Dueñas et al. [14] performed similar characterisation tests with the triple alpha source.
The diamond sample used here is also a single-crystal high purity CVD diamond but of
smaller dimensions at 4 × 4 × 0.05 mm. The thinner configuration is expected to show better
performance as the charge carriers traverse a shorter distance towards the electrodes. This
reduces signal losses due to diffusion, charge trapping and recombination of the charge
carriers. Duenas et al. reported energy resolutions of 19 and 21 keV under negative and
positive biases respectively. This is about one-third of the energy resolution reported by
Galbiati. The lower energy secondary peaks corresponding to each of the 3 alpha sources
could be resolved clearly in the result by Dueñas.
The 2 papers above show that for bulk diamond detectors (thickness > 500 μm), the
performance of the sandwich structure is not optimal. As mentioned before, the penetration
depth of 5 MeV alpha particles is only about 10 μm. It naturally follows that we can consider
a coplanar electrode design for bulk diamond detectors. One common structure is the
interdigitated design which consists of alternating strips of positive and negative electrodes.
As both electrodes are on the top surface, they should be able to collect most of the charge
carriers produced near the surface. The small electrode gap also allows for fast charge
collection, improving the time response of the detector under high particle fluences. The
interdigitated structure is commonly used for UV and γ-ray detection as the photons generally
exhibit poor penetration through the metal electrodes [3].
There has not been significant research into the optimisation of the interdigitated design.
Values for electrode width and gap used by other researchers range from 20 to 200 μm with
little justification for the choice of their values. Thus, the aim of my project is the
optimisation of the interdigitated structure with respect to those 2 parameters. For the case of
alpha detection, there is no need to consider how much exposed diamond area is required
because alpha particles lose little energy when passing through the metal electrodes.
THEORY
13 | P a g e
4 Theory
4.1 Introduction
While certain minor details may differ for different types of detectors, the general
processes involved in radiation detection are the same. The following is the chronology of
radiation detection:
1) Ionising radiation impinges on the active region of the detector, releasing electron-
hole pairs along its path trajectory as it ionises the carbon atoms.
2) The electrons and holes are separated in the electric field produced by the
electrode. Holes migrate towards the anode while electrons migrate towards the
cathode.
3) During the charge carrier drift, some charge carriers may be trapped due to point
defects such as lattice vacancies and may contribute to a build-up of space charge.
This causes polarisation in the diamond. Some charge carriers may recombine
before reaching the electrodes.
4) The movement of these charge carriers along the electric field lines induce
instantaneous currents in the electrode, described by the Shockley-Ramo theorem.
The pulse is converted into a voltage pulse using an amplifier which in ideal
situations is proportional to the energy deposited by the ionising radiation.
Simulations on Stopping and Range of Ions in Matter (SRIM) show that the average
penetration depth of 5 MeV alpha particles is approximately 11.9 μm as shown in Figure 2.
The average straggle in the lateral directions is approximately 0.2 μm, a small number
relative to the penetration depth. Most of the energy (99.76%) is lost via ionisation while the
remainder is lost through secondary recoils and phonons.
THEORY
14 | P a g e
Figure 2 SRIM data for the penetration depth of 5 MeV alpha particles in diamond
A widely-used metric for comparing the performance of semiconductor radiation
detectors is the charge collection efficiency (CCE), which is defined as the ratio of the
amount of charge collected at the electrodes to the amount of charge created inside the
detector. It is related to the charge collection distance (CCD) which is the mean distance
travelled by the electrons and holes before recombination or trapping occurs.
𝐶𝐶𝐸 =𝐶𝐶𝐷
𝑑(4.1)
where 𝑑 is the distance between the electrodes in a sandwich structure. The above equation
would not hold for arbitrary designs as 𝑑 may vary. The 𝐶𝐶𝐷 is given by the following:
𝐶𝐶𝐷 = (𝜇𝑒𝜏𝑒 + 𝜇ℎ𝜏ℎ)𝐸 (4.2)
This equation is valid only if 𝐸 is uniform throughout the detector and both 𝜇 are constant.
For all cases where the CCD is greater than 𝑑, the CCE is taken to be 100% as all of the
charges would be collected at the electrodes.
The theoretical calculation of CCE requires knowledge on both the charge transport
mechanisms and the Shockley-Ramo theorem [15] that relates the charge of the electron-hole
pairs generated to the charge induced in the electrodes.
In the subsequent subsections, the theoretical backgrounds for charge transport, the
Shockley-Ramo theorem and the calculation of charge collection efficiency (CCE) are
discussed.
THEORY
15 | P a g e
4.2 Charge Transport
Charge transport in semiconductors [16] is governed by numerous factors. For simplicity,
this subsection covers the treatment of drift motion of charge carriers relevant to intrinsic
diamond semiconductors. Without loss of generality, the derivations are performed in one
dimension 𝑥.
At all temperatures above absolute zero, all charge carriers possess thermal kinetic energy
averaging at 𝑘𝐵𝑇/2 per degree of freedom. Due to interactions with the lattice atoms, charge
carriers have an effective mass 𝑚∗ that differs from their usual mass. In general, the effective
masses of electrons and holes are not equal. The average thermal velocity 𝑣𝑡ℎ can be obtained
from the following equation:
1
2𝑚∗𝑣𝑡ℎ
2 =3
2𝑘𝐵𝑇 (4.3)
where 𝑘𝐵 is the Boltzmann constant and T is the temperature.
At thermal equilibrium without any external electric field, there is no net current from the
thermal motion of charge carriers in any direction. The charge carriers undergo frequent
collisions with the lattice. An applied electric field introduces an additional drift velocity
component parallel or antiparallel to the electric field, depending on the charge carrier. The
electron/hole mobility 𝜇 is defined as the ratio of the average drift velocity to the electric
field strength 𝐸.
𝜇 =𝑣𝑑
𝐸(4.4)
The above equation assumes that the semiconductor is in a steady state, where some
momentum gained by the charge carriers due to the electric field is lost through collisions.
Consider the electron which has a charge of −𝑒. The average increase in momentum due to
the electric field between collisions is related to the average peak drift velocity of the electron
𝑣𝑝 just before collision by the following equation:
−𝑒𝐸𝜏𝑐𝑜𝑙 = 𝑚𝑒∗𝑣𝑝 (4.5)
where 𝜏𝑐𝑜𝑙 is the average time interval between collisions. However, initial drift velocity is
not necessarily zero. An accurate statistical treatment of the collision process yields that the
THEORY
16 | P a g e
equation for the average drift velocity 𝑣𝑑 is simply the same equation, with a simple but non-
trivial replacement of 𝑣𝑝 with 𝑣𝑑.
𝑣𝑑 = − (𝑒𝜏𝑐𝑜𝑙
𝑚𝑒∗
) 𝐸 = −𝜇𝑒𝐸 (4.6)
where 𝜇𝑒 is the electron mobility. The equivalent equation for holes with charge 𝑒 is
𝑣𝑑 = (𝑒𝜏𝑐𝑜𝑙
𝑚ℎ∗ ) 𝐸 = 𝜇ℎ𝐸 (4.7)
where 𝜇ℎ is the hole mobility. The average time interval between collisions 𝜏𝑐𝑜𝑙 are not
necessarily the same for electrons and holes, as was the case for their effective masses, due to
different collision cross sections and other factors. Therefore, electron and hole mobility are
in general not equal for a given semiconductor.
Another important process for charge transport is diffusion. Diffusion is a process where
particles move from a region of high concentration to a region of low concentration. Ionising
radiation create electron-hole pairs locally along their trajectories. Diffusion affects the
charge transport because the concentration of electrons and holes in those regions are higher
than that of the surrounding regions.
The current density associated with diffusion is called the diffusion current density 𝐽𝑑𝑖𝑓
and it is related to the electron and hole density gradients as follows:
𝐽𝑑𝑖𝑓,𝑒 = 𝑒𝐷𝑒
𝑑𝑛
𝑑𝑥(4.8)
𝐽𝑑𝑖𝑓,ℎ = 𝑒𝐷ℎ
𝑑𝑝
𝑑𝑥(4.9)
𝐽𝑑𝑖𝑓 = 𝐽𝑑𝑖𝑓,𝑒 + 𝐽𝑑𝑖𝑓,ℎ (4.10)
where 𝐷𝑒 and 𝐷ℎ are the diffusion coefficients for electrons and holes respectively, and 𝑛 and
p are the electron and hole concentrations respectively at a particular point in space.
THEORY
17 | P a g e
For a non-degenerate semiconductor (i.e. a semiconductor that is not heavily doped), the
equation known as Einstein’s relationship for electrons express the relationship between
diffusion coefficient for electrons and the electron mobility as follows:
𝐷𝑒 =𝑘𝐵𝑇
𝑒𝜇𝑒 (4.11)
and likewise, for Einstein’s relationship for holes,
𝐷ℎ =𝑘𝐵𝑇
𝑒𝜇ℎ (4.12)
The overall current densities for electrons and holes due to drift and diffusion in three
dimensions are given by
𝑱𝑒 = 𝑛𝑒𝜇𝑒𝑬 + 𝑒𝐷𝑒∇𝑛 (4.13)
𝑱ℎ = 𝑝𝑒𝜇ℎ𝑬 − 𝑒𝐷ℎ∇𝑝 (4.14)
where the first term is the contribution from drift and the second term is the contribution from
diffusion. The contribution from drift is derived from the simple relation that equates current
density to the product of the individual charge, velocity and density of the charge carrier.
The last two significant factors are charge carrier generation and recombination. For the
case of undoped diamond detectors, due to the large band gap, thermal generation of charge
carriers is negligible compared to the charge carrier generation arising from ionising
particles. Thus, generation can be expressed using a Dirac delta function. Recombination rate,
𝑈, is expressed as
𝑈𝑒 =𝑛
𝜏𝑒
(4.15)
𝑈ℎ =𝑝
𝜏ℎ
(4.16)
for electrons and holes respectively, where 𝜏 is the respective average lifetime before
recombination.
THEORY
18 | P a g e
The four factors discussed above are combined to form two continuity equations for
charge transport as shown below:
𝜕𝑛
𝜕𝑡=
1
𝑒∇ ∙ 𝑱𝑒 + 𝐺 − 𝑈𝑒
= ∇ ∙ (𝑛𝜇𝑒𝑬) + 𝐷𝑒∇2𝑛 + 𝐺 −𝑛
𝜏𝑒
(4.17)
𝜕𝑝
𝜕𝑡= −
1
𝑒∇ ∙ 𝑱ℎ + 𝐺 − 𝑈ℎ
= −∇ ∙ (𝑝𝜇ℎ𝑬) + 𝐷ℎ∇2𝑝 + 𝐺 −𝑝
𝜏ℎ
(4.18)
where 𝐺 is the generation rate of electron-hole pairs due to ionisation by ionising radiation. A
more advanced treatment of charge transport is possible, with additional factors such as
temperature and bias voltage dependence of the charge carrier mobility and trapping effects
due to defects. Further discussions will be made in the later sections.
4.3 The Shockley-Ramo Theorem
When ionising radiation is incident on the active region of the detector, the number of
electron-hole pairs generated is proportional to the energy deposited by the ionising radiation.
The movement of these charge carriers induce an instantaneous charge in the electrodes that
can be shaped and amplified into an output electrical signal. The tedious calculation of the
induced charge 𝑄 due to the moving charge 𝑞 was made simple by the Shockley-Ramo
theorem. The Shockley-Ramo theorem [15] allows us to predict the CCE with relative ease if
we know the initial and final positions of the moving charge.
The Shockley-Ramo theorem states that the charge 𝑄 induced in an electrode by the
moving charge 𝑞 moving from 𝒙𝑖 to 𝒙𝑓 is given by
𝑄 = −𝑞[𝜑0(𝒙𝑓) − 𝜑0(𝒙𝑖)] (4.19)
where 𝜑0(𝒙) is the electric potential present in the detector when the particular electrode has
unit potential while the other electrodes are grounded (i.e. zero potential).
THEORY
19 | P a g e
The review of the Shockley-Ramo theorem below is based on the conservation of energy.
Consider the detector illustrated in Figure 3a. It consists of several electrodes kept at constant
voltages 𝑉𝑗 (for 𝑗 = 1, 2, … , 𝑘). The electrodes enclose surfaces 𝑆𝑗 around the entire surface 𝑆
of volume 𝜏. The outermost boundary is surrounded by the electrode of voltage 𝑉𝑘. The
electric potential 𝜑(𝒙) in the detector satisfies Poisson’s equation with the given Dirichlet
boundary conditions. Thus, by the uniqueness theorem for Poisson’s equation, the electric
field 𝑬(𝒙) = −∇𝜑(𝒙) is uniquely determined.
Figure 3 The principle of linear superposition for electric potentials
By the principle of linear superposition, the electric potential for the case in Figure 3a is
the sum of the electric potentials for the cases in Figure 3b and 3c. For Case b, the electrode
potentials are kept while the space charges 𝜌(𝒙) and the moving charge 𝑞 are removed. For
Case c, both charges are kept while the electrodes are all grounded. Assuming that the
detector is a linear and isotropic medium, the electric displacement is given by
𝑫 = 𝜀𝑬 (4.20)
where 𝜀 is the dielectric constant. The energy density 𝑢 of the electric field is given by
𝑢 =1
2𝑬 ∙ 𝑫 =
1
2𝜀𝐸2 (4.21)
The total energy of the system can be obtained by integrating the energy density over the
entire volume 𝜏. By the conservation of energy, the total energy of the field can only be
changed by energy transfers between the field and the moving charge 𝑞, and between the
field and the power supply connected to the electrodes.
THEORY
20 | P a g e
We first consider the case in Figure 3c. The work done by the electric field on a moving
charge 𝑞 is
∫ 𝑞𝑬1′ ∙ 𝑑𝒙
𝒙𝒇
𝒙𝒊
(4.22)
where 𝑬𝟏′ is equal to 𝑬𝟏, excluding the electric field due to 𝑞. As the charge 𝑞 moves from 𝒙𝒊
to 𝒙𝒇, the amount of induced charges on the electrodes change. However, as all electrodes are
grounded, no work is done arising from the change in induced charges on the electrodes. This
means that only the electric field 𝑬𝟏′ does work on charge 𝑞. Any change in the energy of
charge 𝑞 comes from the change in the energy stored in the electric field
∫ 𝑞𝑬𝟏′ ∙ 𝑑𝒙
𝒙𝒇
𝒙𝒊
=1
2∫𝜀(𝐸1,𝑖
2 − 𝐸1,𝑓2 )𝑑𝜏
𝜏
(4.23)
where 𝑬1,𝑖 and 𝑬1,𝑓 are the electric fields of the case in Figure 3c at the initial and final
positions of charge 𝑞 respectively.
For the case in Figure 3a, since the electric field excluding the charge 𝑞’s own field is
𝑬 = 𝑬0 + 𝑬1′ , the work done on charge 𝑞 by the electric field is
∫ 𝑞(𝑬0 + 𝑬1′ )𝑑𝒙
𝒙𝒇
𝒙𝒊
(4.24)
and the work done on the induced charges ∆𝑄𝑗 on each electrode by the power supply is
∑ 𝑉𝑗∆𝑄𝑗
𝑘
𝑗=1
(4.25)
The expression for the conservation of energy in this case is
∑ 𝑉𝑗∆𝑄𝑗
𝑘
𝑗=1
− ∫ 𝑞(𝑬0 + 𝑬1′ )𝑑𝒙
𝒙𝒇
𝒙𝒊
=1
2∫𝜀(𝑬0 + 𝑬1,𝑓)
2− (𝑬0 + 𝑬1,𝑖)
2𝑑𝜏
𝜏
(4.26)
Here, consider Green’s first identity, a result derived from the divergence theorem,
∫(𝜑1∇2𝜑0 + ∇𝜑0 ∙ ∇𝜑1)𝑑𝜏𝜏
= ∮𝜑1∇𝜑0 ∙ 𝑑𝑺𝑆
(4.27)
THEORY
21 | P a g e
that applies to any arbitrary function 𝜑0 and 𝜑1. Note that ∇2𝜑0 = 0 as 𝜑0 is produced solely
by the electrode potentials, i.e. the space charges and charge 𝑞 do not contribute to 𝜑0.
Furthermore, 𝜑1 = 0 on the surface 𝑺 since all electrodes are grounded. Hence, for the cross
terms on the RHS of Equation (4.25),
∫𝑬0 ∙ 𝑬1 𝑑𝜏𝜏
= ∫∇𝜑0 ∙ ∇𝜑1 𝑑𝜏𝜏
= ∮𝜑1∇𝜑0 ∙ 𝑑𝑺𝑆
− ∫𝜑1 ∙ ∇2𝜑0 𝑑𝜏𝜏
= 0 (4.28)
⇒ ∫𝜀(𝑬0 + 𝑬1,𝑓)2
− (𝑬0 + 𝑬1,𝑖)2
𝑑𝜏𝜏
= ∫𝜀(𝐸1,𝑓2 − 𝐸1,𝑖
2 )𝑑𝜏𝜏
(4.29)
Combining Equations (4.28) and (4.22) with Equation (4.25), we have
∑ 𝑉𝑗∆𝑄𝑗
𝑘
𝑗=1
= ∫ 𝑞𝑬0 ∙ 𝑑𝒙𝒙𝑓
𝒙𝑖
= −𝑞[𝜑0(𝒙𝑓) − 𝜑0(𝒙𝑖)] (4.30)
This shows that the work done by the power supply is equal to the change in kinetic energy of
the moving charge 𝑞 due to the electric field produced by the electrodes 𝑬0 alone, with no
contribution from the space charges.
For the next step, we use an important result, which is that the induced charge ∆𝑄𝑗 is
independent of the potentials 𝑉𝑗 of each electrode [15]. Thus, in setting a unit potential for 𝑉𝑗
and zero potential for the other electrodes, we can easily obtain ∆𝑄𝑗 as a fraction of charge 𝑞.
Then by Equation (4.29),
∆𝑄𝑗 = −𝑞[𝜑0(𝒙𝑓) − 𝜑0(𝒙𝑖)] (4.31)
The value of 𝜑0 ranges from 0 to 1 inclusive. Therefore, ∆𝑄𝑗 ranges from 0 to −𝑞, and is
only dependent on the initial and final positions of charge 𝑞, in the absence of energy loss
through collisions within the detector. The 𝜑0 here is called the weighting potential and it is
unitless as it is a normalised quantity. Thus, the Shockley-Ramo theorem is proven.
THEORY
22 | P a g e
4.4 Calculation of Charge Collection Efficiency
The moving charges in semiconductor detectors are the electron-hole pairs generated by
the incident ionising radiation. Applying the Shockley-Ramo theorem, we have
∆𝑄 = −𝑁𝑒[𝜑0(𝒓ℎ) − 𝜑0(𝒓0)] + 𝑁𝑒[𝜑0(𝒓𝑒) − 𝜑0(𝒓0)] (4.32)
where 𝑁 is the number of electron-hole pairs generated, 𝒓ℎ and 𝒓𝑒 are the final positions of
the holes and electrons generated, and 𝒓0 is the initial position where the electron-hole pairs
were generated.
The CCE is then given by
𝐶𝐶𝐸 = −[𝜑0(𝒓ℎ) − 𝜑0(𝒓0)] + [𝜑0(𝒓𝑒) − 𝜑0(𝒓0)]
= 𝜑0(𝒓𝑒) − 𝜑0(𝒓ℎ) (4.33)
The CCE represents the fraction of charge collected. Multiple calculations may be
required due to diffusion, resulting in different final positions even if the charge carriers
started from the same initial position. Given that all electrons drift to the positive electrode
while all holes drift to the negative electrode, 𝜑0(𝒓𝑒) = 1 and 𝜑0(𝒓ℎ) = 0. Therefore, the
CCE would be 1. Losses due to recombination and trapping reduce the CCE value as the
weighting potential value at the final position would not be 1 or 0 for electrons and holes
respectively.
DETECTOR SIMULATIONS
23 | P a g e
5 Detector Simulations
5.1 Introduction
Due to the non-uniform nature of electric fields in detectors with coplanar interdigitated
electrodes, it is difficult to calculate the CCE by hand. Through computational methods, we
are able obtain the electric field profile and estimate the performance of a detector. Accurate
modelling results can be used to predict experimental outcomes and complement
experimental results through the precise control of certain variables, even those that may be
uncontrollable in a laboratory setting. The main aim of the detector simulations in my project
is to obtain an optimised interdigitated electrode structure for the diamond detector.
In this section, the methodology for detector simulations is first discussed, followed by
the simulation results and analysis.
5.2 Methodology
The diamond sample and interdigitated electrode structure was first modelled using
SOLIDWORKS, a 3D CAD software, as shown in Figure 4. Various dimensions were
considered for the diamond sample and electrode structure. The CAD file is then imported
into COMSOL Multiphysics software. COMSOL is a general-purpose modelling and
simulation software for physics-based systems. On COMSOL, the appropriate materials and
electric potentials were assigned (i.e. diamond for the cuboid, gold for the electrode surfaces
and electric potentials for each electrode).
Electrostatics module in COMSOL uses the finite element method (FEM) to obtain the
static electric potential profile in the diamond detector by solving the Poisson equation. FEM
is a numerical method for solving partial differential equations (PDE) with various boundary
conditions [17]. The system is first discretized into many small volume elements as shown in
Figure 5. The discretization allows for the approximation of the real solution for the PDE
through the selection of linear basis functions that are non-zero in their own selected elements
and zero elsewhere.
DETECTOR SIMULATIONS
24 | P a g e
Figure 4 3D CAD design of diamond detector modelled on SolidWorks
Figure 5 Discretization of diamond detector model into small volume elements
The electric potential value was obtained at each volume element in a regular grid of
1200 by 1200 by 100 elements. The data file was exported to MATLAB. The MATLAB code
is based the ion beam induced charge (IBIC) microscopy technique [18]. Highly focussed ion
beams are used to probe various samples to determine their characteristics. For radiation
detectors, IBIC is very useful as it is equivalent to being exposed to ionising particles from
DETECTOR SIMULATIONS
25 | P a g e
other sources. IBIC has the added benefit of the high focus of the ion beam, which allows
researchers to probe specific locations of the sample. Through this technique it is possible to
obtain the detector response to ionising particles at different parts of the detector. By
scanning the ion beam across the sample, we can obtain the variation in signals received from
the detector. After energy calibration of the detector signals, the CCE can be calculated at
each point.
The MATLAB code simulates the generation of charge carriers, charge transport and
calculates the CCE using the initial and final positions of the charge carriers as explained in
the Theory section.
For 5 MeV alpha particles, Stopping and Range of Ions in Matter (SRIM) simulation
results show that the penetration depth is approximately 10 μm. The Bragg peak for alpha
particles shows that alpha particles lose most of their energy towards the end of their
trajectory. Therefore, for computational efficiency, it is assumed that all electron-hole pairs
are produced at a depth of 10 μm from the top surface of the detector. SRIM simulation
results have also shown that alpha particles lose an average of less than 1% of their energy
when transmitted through 100 nm of gold. Thus, the energy losses while passing through the
electrodes were neglected.
The 3D electric field profile was obtained by applying the gradient function on the 3D
electric potential matrix, giving the 3D electric field matrix. The charge transport was
simulated by using simple finite difference scheme for electrons and holes separately. Each
time-step has an interval of 1 ps. The distance travelled by the charge carriers at each time-
step was calculated by simply taking the product of the drift velocity and the time interval of
1 ps. The velocity is governed by Equations (4.6) and (4.7), and it varies according to the
electric field strength and direction at each point along the charge carrier’s trajectory.
Diffusion effects were neglected for various reasons. A deterministic approach to
diffusion would give similar results because the average position of the “cloud” of diffused
charge carriers follows a similar trajectory to that obtained by the current formulation while
the Monte Carlo method would be computationally intensive. Furthermore, charge collection
is a short process given the high charge carrier mobility and the strong electric fields in
radiation detectors [19]. Therefore, diffusion is expected to be limited.
DETECTOR SIMULATIONS
26 | P a g e
The termination of charge transport is triggered by either of these two conditions: a) the
charge reaches the boundary of the diamond detector; b) a time interval of 10 ns has passed
from the moment of charge carrier generation. The second condition arises from the
approximate integration time of the preamplifier 142A used for the actual detector testing.
The CCE is then calculated based on the initial and final positions using Equation (4.33)
obtained from the Shockley-Ramo theorem.
5.3 Simulation Results
The simulations discussed in this report were done under 100V bias for electrode widths
of 20μm, 30μm, 40μm, 50μm and 60μm; for electrode gaps of 5μm and 10μm. The
simulation results for electrode widths of 20μm, 40μm and 60μm at 10μm electrode gap are
shown in Figure 6. The remaining results can be found in the Appendix. The CCE values at
every point are represented by a colour on the colour scale with dark blue corresponding to
zero CCE and dark red corresponding full CCE of 1. The x and y axes are swapped relative to
the original CAD design in Figure 4.
As shown in Figure 6, full CCE is achieved near the tips of the interdigitated electrodes.
This is an expected result because the electric field strength is highest near the tips where the
radius of curvature is small, allowing for fast and efficient charge collection. The CCE is also
relatively high directly under the electrodes and near the electrodes but falls off towards the
centre of the electrode gap. Preliminary simulations on electrode gaps of up to 150 μm have
shown the same trends. Relatively lower CCE were observed between electrodes for all
simulations. Therefore, our discussion here is limited to simulations with smaller gaps of
5μm and 10μm. The CCE values also fall off rapidly in the peripheral areas of the detector.
The electric field strength in those areas are weak as they are further away from the
electrodes and the change in electric potential is more gradual.
One downside of the simulations is that the CCE values under the electrodes are lower
than expected. Based on prior tests of the interdigitated electrode structure on a single-crystal
diamond provided by the same company, the CCE values under the electrodes were in the
region of 90-100%. The electrode width and gap for that detector were 20μm and 200μm.
The simulations are unable to replicate these CCE values accurately. However, the trends in
DETECTOR SIMULATIONS
27 | P a g e
the CCE agree with the experimental data; the CCE near the tips and under the electrodes is
relatively higher than the CCE in the electrode gaps. Therefore, the simulation results still
hold some merit in comparing the relative performance for different geometries.
Figure 6 Simulation of CCE for various interdigitated electrode geometries on a 1.8 × 1.8 × 0.05 mm
single-crystal diamond detector with 5 MeV alpha particles
DETECTOR SIMULATIONS
28 | P a g e
Figure 7 Relative average CCE against electrode width. Calculated from the central region of the
CCE maps for electrode widths from 20 μm to 60 μm, and electrode gaps of 5 μm and 10 μm
Due to computational constraints, the dimensions of the diamond detector used in the
simulations are 1.8 × 1.8 × 0.05 mm. The average CCE of the middle 8 electrode strips and
gaps, 300 μm in length, were calculated for comparison of detector performance. This
method was used as the middle areas are away from the high CCE at the electrode tips, for
the average values to be less skewed. Another justification for this is that if the detector
dimensions were scaled up to the actual 4 × 4 × 0.5 mm, most of the CCE values would be
similar to the middle areas rather than the surrounding areas where the tips are.
The CCE values were normalised by taking the highest CCE value, corresponding to the
simulation with 40 μm width and 5 μm gap, to be 1 and the others were scaled accordingly.
From the graph in Figure 7, the optimal electrode widths are clearly 40 μm for electrode gaps
of 5 μm, and 30 μm for electrode gaps of 10 μm.
Therefore, for the fabrication of the diamond detector, electrode width and gap of 40 μm
and 5 μm respectively were chosen initially. However, due to difficulties encountered during
testing, an alternative configuration of 40 μm electrode width and 10 μm electrode gap was
fabricated for testing. The detector with 5 μm electrode gap encountered air breakdown easily
and no alpha detection peaks could be resolved from the noise.
0.8
0.85
0.9
0.95
1
1.05
0 10 20 30 40 50 60 70
Rel
ati
ve
Av
era
ge
CC
E
Electrode width (μm)
5 micron separation
10 micron separation
DETECTOR SIMULATIONS
29 | P a g e
Before moving on to the next section, it should be noted that the optimal electrode
configuration obtained from simulations may differ from real detectors. The mismatch
between the absolute CCE values of the simulations and actual experimental data was
touched upon earlier. There are many other real factors that may affect CCE. These include
imperfect electrode contacts, Schottky barriers near the electrode contacts and polarization
effects in diamond.
The Schottky barrier is a carrier depletion region formed at a metal-semiconductor
interface. It is caused by the introduction of metal-induced gap states in the band gap of the
semiconductor near the metal. Fermi level pinning occurs, which is the alignment of the
Fermi level of the metal and the semiconductor, due to the tendency for electrons to occupy
the gap states. These metal-semiconductor junctions behave like p-n junctions, favouring
current flow in one direction over the other. However, if the Schottky barrier is low, there
may be no rectifying effect by the junction. These are called ohmic contacts as the IV curve
shows linearity for current in both directions, similar to Ohm’s law.
FABRICATION AND TESTING
30 | P a g e
6 Fabrication and Testing
6.1 Detector Fabrication Procedure
The materials used for detector fabrication are:
1) Single-crystal electronic grade diamond sample of dimensions 4 × 4 × 0.5 mm,
provided by IIa Technologies Pte Ltd
2) Blank photomask, which consists of a uniform layer of AZ1518, on a chromium-
glass mask
3) AZ1518 photoresist solution
4) AZ400K developer solution
5) Acetone, isopropanol (IPA) and distilled (DI) water for cleaning
The instruments used for detector fabrication are, the UV laser writer, spin coater, UV
exposure station and magnetron sputtering machine.
The main fabrication steps are listed below.
1) UV laser writing of the electrode design pattern on blank photomask
2) Spin coating of diamond sample with AZ1518 photoresist. The spin coating was
done in 4 steps with the following settings:
a. Duration: 15s, RPM: 800, RPM/s: 1000
b. Duration: 75s, RPM: 6000, RPM/s: 1000
c. Duration: 30s, RPM: 7500, RPM/s: 1000
d. Duration: 10s, RPM: 0, RPM/s: 1000
3) Baking at 90°C for 90s to harden the AZ1518 on the diamond sample
4) UV exposure of diamond sample with the patterned photomask placed over it with
a power setting of 200W for 90s
5) Photoresist development using AZ400K developer for 90s
6) Magnetron sputtering of 20 nm Cr, followed by 80 nm Au for the electrodes
7) Removal of remaining photoresist and unwanted Cr/Au on the diamond sample
using acetone
8) Mounting and wire bonding of the diamond sample on a holder
FABRICATION AND TESTING
31 | P a g e
Figure 8 Fabrication process of the diamond detector. (1) Spin coating of AZ1518 on diamond
sample; (2) UV exposure under patterned photomask; (3) Photoresist development using AZ400K;
(4) Magnetron sputtering of Cr/Au interdigitated electrodes
For the first step, the desired electrode design is drawn on AutoCAD, a computer-aided
drafting software. The design file is then exported to the UV laser writer software for laser
writing. The areas exposed to the laser are then removed by submerging the photomask in a
solution consisting of 1 part AZ400K and 4 parts DI water for 90 seconds while unexposed
areas remain intact as they are unaffected by the AZ400K.
Next, the diamond sample is placed in the spin coater and a drop of AZ1518, enough to
completely cover the top surface of the diamond. The above-mentioned spin coater settings
were used to obtain a thin layer of AZ1518 on the diamond. This is followed by baking the
diamond sample at 90°C for 90 seconds to harden the AZ1518.
The diamond sample is taped onto the patterned photomask to ensure that it is as close to
the pattern as possible. UV exposure is performed on the sample over the photomask. The
power setting used is 200 watts for a duration of 90 seconds. Following the UV exposure, the
diamond sample is submerged in 1:4 AZ400K developer/DI water solution for 90 seconds to
remove the UV exposed areas. The result is an inverse of the electrode pattern consisting of
AZ1518 on the diamond surface.
The diamond sample is placed in the magnetron sputtering machine. Magnetron
sputtering is a coating technique that uses the bombardment of ions to eject sputtering
material from a target onto a sample in front of it [20]. The vacuum chamber is first pumped
down to remove air molecules, followed by the injection of argon gas. A plasma is used to
ionise the argon atoms. A high voltage is applied to accelerate the argon ions to the target.
FABRICATION AND TESTING
32 | P a g e
The argon ions eject the sputtering material which coats the diamond sample placed in front
of the target. First, a 20 nm layer of Cr is sputtered at a rate of 0.5 nm per second, followed
by a 80 nm layer of Au at a rate of 1 nm per second. Cr is chosen for the diamond-metal
boundary because Cr forms carbides that help to bind the metal onto the diamond easily. An
imperfect metal contact may affect the detector performance. Au is chosen for the upper layer
because it is an excellent electrical conductor.
After sputtering, the diamond sample is washed with acetone to remove the remaining
photoresist and the unwanted Cr/Au coating on it. The diamond sample with the
interdigitated Cr/Au electrodes is mounted on a Teflon holder with an SMA connector using
double-sided tape. The electrodes are wire bonded to electrical copper contacts on the Teflon
holder as shown in Figure 9.
Figure 9 Schematic design of detector
Figure 10 Schematic of signal processing electronics
FABRICATION AND TESTING
33 | P a g e
6.2 Fabrication Issues
Two main issues were encountered during fabrication and pre-tests. They are uneven spin
coating and suboptimal photoresist developing duration. Due to the small size of the diamond
sample and the surface tension of the AZ1518, it was not possible to obtain an even layer of
AZ1518. The AZ1518 layer was thicker near the edges than at the centre. Despite numerous
trials, the optimal settings for the spin coater could not be found. Thus, Cr/Au could not be
properly deposited near the edges of the diamond sample. One electrode strip was
disconnected from the rest of the electrode as seen in Figure 11b. However, the effect on the
performance was expected to be minimal as each electrode consists of 30 strips. Optimisation
of the spin coater settings is essential for future tests as the suboptimal coating limits the
effective area that can be used on the diamond.
(a) (b)
Figure 11 Microscope images of the electrodes
Suboptimal photoresist developing duration caused minor defects in the electrode shapes
as shown in Figures 12a and 12b. From Figure 13, we can estimate the electrode width to gap
ratio. It is found to be approximately 3.40
0.75= 4.53 > 4 =
40
10. Further calculations show that the
electrode gap is approximately 9.04 μm and the electrode width is 41.96 μm. This shows that
the photoresist developing duration of 90 seconds was insufficient. Further trials are required
FABRICATION AND TESTING
34 | P a g e
to find the optimal developing duration in order to fabricate electrode shapes as accurately as
possible.
(a) (b)
Figure 12 Microscope images of electrode defects
Figure 13 Scanning electron microscope (SEM) image of electrodes
FABRICATION AND TESTING
35 | P a g e
6.3 Testing and Results
Characterization of the diamond detector was performed using a radioactive source. The
source used is a triple alpha source of Pu-239, Am-241 and Cm-244. The source was placed 7
mm directly below the diamond detector in a vacuum chamber at 10-1 mbar. The signal from
the detector was read out by a pre-amplifier and other associated electronics as shown in the
schematic diagram in Figure 10. Measurements were taken at various positive and negative
bias voltages for 10 minutes each. The gain and shaping time of the amplifier were set at 200
and 0.5 μs respectively. The spectrum data can be found in the Appendix. A commercial
silicon detector (ORTEC surface barrier Si detector) was also tested using the same
radioactive source with +40V bias with the amplifier set at 10, 20 and 50 gain.
Table 2 Alpha particle energy of the triple alpha radioactive source
Radionuclide Alpha particle energy (MeV) and probability
Plutonium-239 5.103 (11%), 5.142 (15%), 5.155 (73%)
Americium-241 5.442 (12.5%), 5.484 (85.2%)
Curium-244 5.763(23.6%), 5.806 (76.4%)
The positions of the three spectral peaks for every spectrum were determined using a
Gaussian fit. The commercial silicon detector was taken as a reference and assumed to be
operating at 100% CCE. An energy calibration of the spectrum channels was performed using
the 3 peaks in the spectrum for the silicon detector tested at 50 gain.
The energy required to generate an electron-hole pair is 13.2 eV in diamond and 3.6 eV in
silicon. Therefore, a calibration is required to compare the CCE of the diamond detector with
the silicon detector. Taking 𝐸 to be the energy of the ionising particle, it is related to the peak
channel by the following equation:
50×𝐸
3.6×100% = 𝑘 ∙ 𝐶ℎ𝑆𝑖 (6.1)
200×𝐸
13.2×100% = 𝑘 ∙ 𝐶ℎ𝐷 (6.2)
where 𝑘 is a scaling factor, and 𝐶ℎ𝑆𝑖 and 𝐶ℎ𝐷 are the channel numbers of the peak positions
at 100% CCE. The numbers 50 and 200 account for the difference in gain used. The data for
FABRICATION AND TESTING
36 | P a g e
the silicon detector tests at 10, 20 and 50 gain were used to confirm the linear scaling of the
gain. The channel number which represents 100% CCE for diamond detectors was calculated
using Equations (6.1) and (6.2).
𝐶ℎ𝐷 = 4 ∙3.6
13.2∙ 𝐶ℎ𝑆𝑖 (6.3)
Thus, the CCE values for the diamond detector for each of the three peaks in each spectrum
were calculated by the following ratio.
𝐶𝐶𝐸 =𝐶ℎ𝑒𝑥𝑝
𝐶ℎ𝐷
(6.4)
where 𝐶ℎ𝑒𝑥𝑝 is the channel number of the peak positions from the experimental data for the
diamond detectors.
It was observed that under the negative bias, due to errors arising from the signal
processing in the electronics, the position of the peaks did not match up with the peaks found
under positive bias. A reference pulser that produces fixed positive and negative voltage
pulses was used to calibrate and account for the differences in the signal processing under
positive and negative bias.
Figure 14 CCE values for diamond detector at different bias voltages. Blue data points represent
diamond detector of electrode width = 40 μm, gap = 10 μm. Red data points represent diamond
detector of electrode width = 20 μm, gap = 200 μm
FABRICATION AND TESTING
37 | P a g e
Figure 14 shows the CCE of the diamond detector under biases ranging from -20V to
20V, represented by the blue data points. The error bars were obtained by combining the
CCE data for the 3 alpha peaks for each bias voltage. The CCE data for biases of +3V to -3V
are unavailable because the 3 alpha peaks were unresolvable due to poor charge collection as
shown in the Appendix. The red data points represent CCE data for a prior test performed on
a previous detector fabricated using single-crystal electronic grade diamond provided by the
same supplier (IIa Technologies Pte Ltd). The electrode width and gap were 20 μm and 200
μm respectively and the radioactive source used was Po-210.
The data shows that maximum charge collection efficiency of 107% was achieved from
+5V/-5V onwards. This is depicted by the flat plateau at higher biases. This is much lower
than the operating voltage of the silicon detector (+40V) as well as the previous diamond
detector (~+50V/-50V). Due to the relatively smaller electrode gaps compared to the previous
diamond detector, strong electric fields could be obtained at a lower bias voltage. Therefore,
the detector can be operated using lower voltage equipment which are often cheaper and
easier to maintain. The CCE values for the diamond detector exceed 100% under both
positive and negative bias. This may be due to minor errors in scaling the silicon detector
data in order to compare the CCE values with the diamond detector data. Also, as the silicon
detector is not new, it may have experienced a slight degradation in performance.
Nevertheless, the high CCE values of the diamond detector should not be discounted.
Figure 15 Spectrum data for diamond detector at +20V bias
-10
0
10
20
30
40
50
60
70
80
0 1000 2000 3000 4000 5000 6000 7000
Co
unts
Channels
FABRICATION AND TESTING
38 | P a g e
Figure 16 Spectrum data for previous diamond detector at -150V bias
From the spectrum data for the diamond detector at +20V bias in Figure 15, the three
alpha peaks can be easily distinguished from each other, and there are no other significant
peaks present. On the other hand, for the previous diamond detector, two peaks are observed
on the right end of the spectrum. This is even though the radioactive source Po-210 used for
the previous diamond detector releases alpha particles of only one energy (5.3 MeV). The
two peaks correspond to two different CCE values. With this detector, it would be difficult to
calculate the energy of an unknown radioactive source due to the presence of multiple peaks.
The current detector would be a better choice as there is 1 well-defined peak corresponding to
the energy of the ionising particle. It is also observed that there is a non-negligible number of
counts registered on the lower end of the channel spectrum. Those counts may correspond to
alpha particles that are incident on the peripheral areas of the detector. As these areas are
further away from the electrodes, the electric field strength would be weaker. Furthermore,
there would be incomplete charge collection as some of the charge carriers would drift away
to the sides of the detector, depending on which electrode is closest. For example, if the
closest electrode is the positive electrode, it would force holes to drift towards the edge of the
detector. Another plausible cause would be the incidence of alpha particles on the aluminium
wire bonding, resulting in an unwanted electrical signal.
FABRICATION AND TESTING
39 | P a g e
Figure 17 Gaussian fitting of 3rd alpha peaks for the diamond and silicon detector respectively
Figure 17 shows the Gaussian fitting for the 3rd peak for the diamond detector at +20V
bias and the silicon detector at +40V bias. Both peaks were fitted using a sum of 2 Gaussian
functions, representing the 5.763 MeV (23.6%) and 5.806 MeV (76.4%) alpha particles
emitted by Cm-244. The equation for the full width at half maximum is
FWHM = 2√2 ln 2 𝜎 (6.5)
where 𝜎 is the standard deviation. Both Gaussian functions were made to share the same 𝜎
value as the FWHM should be the same for both peaks. The FWHM is commonly quoted as
the energy resolution of the detector. The energy resolutions for the diamond and silicon
detectors are 92 keV and 33 keV respectively. This is far from the energy resolution of 19
keV achieved by Duenas et al. [14] with a sandwich structure diamond detector. Further
improvements in performance are necessary to make this detector more reliable.
FABRICATION AND TESTING
40 | P a g e
Figure 18 Leakage current measurements of the diamond detector
Figure 18 shows the leakage current measured under various bias voltages. The leakage
current was measured using a RBD 9103 USB picoammeter. The symmetry of the IV curve
about the origin indicates an ohmic character for the electrodes. This shows that there is no
significant Schottky barriers at the metal-semiconductor junctions. The magnitude of the
leakage current is consistent with those reported by Pernegger et al [21]. Air breakdown was
observed at 250V for both positive and negative biases. This shows that the diamond detector
has a very wide operating voltage range of 5V-240V.
0
0.05
0.1
0.15
0.2
0.25
0.3
-250 -200 -150 -100 -50 0 50 100 150 200 250 300
Lea
kag
e C
urr
ent
(nA
)
Bias Voltage (V)
CONCLUSION
41 | P a g e
7 Conclusion
7.1 Summary
Using computer simulations, the optimal coplanar interdigitated electrode configuration
was found by varying the electrode width and separations. The single-crystal electronic grade
diamond detector fabricated using the recommended dimensions of 40 μm electrode width
and 10 μm electrode gap showed an improvement in performance over a previous design of
20 μm electrode width and 200 μm electrode gap. The CCE values were higher and the
energy resolution was better as well, as the previous detector suffered from having two
detection peaks from a single alpha source. The magnitude of the leakage current measured is
low and in good agreement with other reported values.
Although the CCE values for the diamond detector were relatively higher than the silicon
detector, the silicon detector had a better energy resolution of 33 keV compared to the
diamond detector’s 92 keV.
The fabrication process can be further optimised. Due to issues encountered during
fabrication, minor defects in the shape of the electrodes were found. Better performance is
expected if the entire top surface is covered by the interdigitated electrode structure as this
ensures strong electric fields throughout the entire volume of the diamond detector.
7.2 Future Research
The computer simulation code can be further improved upon to obtain more accurate
simulation results. The absolute CCE values reported by the simulation results are not in
good agreement with experimental data. However, the simulations can give the correct trend
in the relative CCE values under the electrodes and in the electrode gaps.
Optimisation of the fabrication process would be key in achieving reproducible results.
The defects in the electrodes cause variations in detector performance (usually detrimental
CONCLUSION
42 | P a g e
effects) and this makes it difficult to accurately assess the performance of a certain electrode
structure.
Another way of optimising interdigitated diamond detector performance would be to use
3D electrode structures, where the electrode strips are etched a narrow depth into the
diamond. With the sides of the electrode strips sputtered with electrodes, it may result in
better charge collection for charge carriers generated near the surface of the detector.
Improved performance has been reported by Forneris et al. [7]. However, optimisation of the
etching depth was not performed. This aspect could be explored through computer
simulations.
REFERENCES
43 | P a g e
References
[1] Frame P. W., A history of radiation detection instrumentation, Health Physics 87 (2): p. 111-135
[2] Mainwood A., Recent Developments of diamond detectors for particles and UV radiation,
Semicond. Sci. Technol. 15 (2000) R55-R63
[3] WhitField M. D., et al, Diamond photoconductors: operational lifetime and radiation hardness
under deep-UV excimer laser irradiation, Diamond and Related Materials 10 (3-7): p. 715-721
[4] Balmer R. S., et al., Chemical vapour deposition synthetic diamond: materials, technology and
applications, Journal of Physics: Condensed Matter, 2009. 21(36): p. 364221.
[5] The Element Six CVD Diamond Handbook,
http://e6cvd.com/media/wysiwyg/pdf/E6_CVD_Diamond_Handbook_A5_v10X.pdf, [Accessed 1st
April 2017]
[6] Forneris J., et al., Modeling of ion beam induced charge sharing experiments for the design of
high resolution position sensitive detectors, Nuclear Instruments and Methods in Physics Research
Section B: Beam Interactions with Materials and Atoms, 2013. 306: p. 169-175.
[7] Forneris J., et al., A 3-dimensional interdigitated electrode geometry for the enhancement of
charge collection efficiency in diamond detectors. EPL (Europhysics Letters), 2014. 108(1): p. 18001.
[8] IIa Diamond Properties, http://2atechnologies.com/2a-diamond-properties/, [Accessed 1st April
2017]
[9] Bruzzi M., et al., Advanced materials in radiation dosimetry, Nuclear Instruments and Methods in
Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment,
2002. 485 (1–2): p. 172-177.
[10] Meier D., et al., Proton irradiation of CVD diamond detectors for high-luminosity experiments at
the LHC, Nuclear Instruments and Methods in Physics Research Section A: Accelerators,
Spectrometers, Detectors and Associated Equipment, 1999. 426 (1): p. 173-180
[11] Zamboni I., et al., Radiation hardness of single crystal CVD diamond detector tested with MeV
energy ions, Diamond and Related Materials, 2013. 31: p. 65-71
[12] Grilj V., et al., Irradiation of thin diamond detectors and radiation hardness tests using MeV
protons, Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with
Materials and Atoms, 2013. 306: p. 191-194
REFERENCES
44 | P a g e
[13] Galbiati A., et al., Performance of Monocrystalline Diamond Radiation Detectors Fabricated
Using Ti/W, Cr/Au and a Novel Ohmic DLC/Pt/Au Electrical Contact, IEEE Transactions on Nuclear
Science, 2009. 56 (4): p. 1863
[14] Dueñas J. A., et al., Diamond detector for alpha-particle spectrometry, Applied Radiation and
Isotopes, 2014. 90: p. 177-180
[15] Zhong H., Review of the Shockley-Ramo theorem and its application in semiconductor gamma-
ray detectors, Nuclear Instruments and Methods in Physics Research Section A: Accelerators,
Spectrometers, Detectors and Associated Equipment, 2001. 463 (1-2): p. 250-267
[16] Leroy C., Principles of radiation interaction in matter and detection, (World Scientific
Publishing Co. Pte. Ltd., Singapore) 2016.
[17] The Finite Element Method. https://www.comsol.com/multiphysics/finite-element-method,
[Accessed 1st April 2017]
[18] Vittone E., et al., Semiconductor characterization by scanning ion beam induced charge (IBIC)
microscopy, Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with
Materials and Atoms, 2008. 266 (8): p. 1312-1318
[19] Vittone E., et al., Charge collection efficiency degradation induced by MeV ions in
semiconductor devices: Model and experiment, Nuclear Instruments and Methods in Physics Research
Section B: Beam Interactions with Materials and Atoms, 2016. 372: p. 128-142
[20] Magnetron Sputtering. http://www.hauzertechnocoating.com/en/plasma-coating-
explained/magnetron-sputtering/, [Accessed 1st April 2017]
[21] Pernegger H., et al., Charge-carrier properties in synthetic single-crystal diamond measured
with the transient-current technique, Journal of Applied Physics, 2005, 97 (7)
APPENDIX
45 | P a g e
Appendix
Figure 19 CCE data for 30 μm and 50 μm electrode width, with 10 μm electrode gap respectively
APPENDIX
46 | P a g e
Figure 20 Channel spectrum data for diamond detector at +2V and +3V
APPENDIX
47 | P a g e
Figure 21 Channel spectrum data for diamond detector at +4V and +5V
APPENDIX
48 | P a g e
Figure 22 Channel spectrum data for diamond detector at +10V and +15V
APPENDIX
49 | P a g e
Figure 23 Channel spectrum data for diamond detector at -2V and -3V
APPENDIX
50 | P a g e
Figure 24 Channel spectrum data for diamond detector at -4V and -5V
APPENDIX
51 | P a g e
Figure 25 Channel spectrum data for diamond detector at -10V and -15V
APPENDIX
52 | P a g e
Figure 25 Channel spectrum data for diamond detector at -20V