Radiation Tests of Semiconductor Detectors

Valery Chmill

KUNGL TEKNISKA HOGSKOLAN¨ Fysiska Institutionen Stockholm 2006

Valery Chmill: Radiation Tests of Semiconductor Detectors

Abstract

This thesis investigates the response of Gallium Arsenide (GaAs) detectors to ionizing irradiation. Detectors based on π-ν junction formed by deep level centers doping. The detectors have been irradiated with 137Cs γ−rays up to 110 kGy, with 6 MeV mean energy neutron up to approximately 6 · 1014 n/cm2, with protons and mixed beam up to 1015 p/cm2. Results are presented for the effects on leakage currents and charge collection efficiencies for minimum ionizing electrons and alpha particles. The signal from minimum ionizing electrons was well separated from the noise even after the highest delivered exposures and the diodes are thus still operational as detectors. Saturation of the effects of radiation damage is observed in both the I-V characteristics and charge collection efficiency measurements. The requirements for detectors e.g. at present and planned hadron colliders is very high in terms of radiation hardness. Detectors for tracking applications close to the interaction point will receive charged particle doses in the range of 110 kGy and fast neutron fluences of 1014 n/cm2 during the lifetime of an experiment. In this thesis it is confirmed that GaAs detectors are radiation resistant to neutron irradiation for fluences up to 1015 n/cm2 and that GaAs detectors are feasible as inner trackers. Most of this work was performed in the framework of the RD8 collaboration at CERN.

°c Valery Chmill 2006 TRITA-FYS-2006:45 ISSN 0280-316X ISRN KTH/FYS/–06:45–SE ISBN 91-7178-401-2

Contents

1 Introduction 3 1.1 Background ...... 3 1.2 Aims of Work ...... 4 1.3 Properties of Semiconductor ...... 5 1.4 Band Structure ...... 5 1.5 Charge Carrier Mobility and Drift Velocity ...... 8 1.6 Classiﬁcation of Defects in Semiconductor ...... 8 1.7 Author’s Contribution ...... 10 1.7.1 List of publications included in thesis...... 11 1.7.2 List of publications not included in thesis. . . . 12

2 Description of Investigated Structures 14 2.1 Technology ...... 15 2.1.1 Diﬀusion ...... 15 2.1.2 Epitaxy ...... 16

3 Response of GaAs detectors to Ionizing Irradiation 19 3.1 Measuring bench ...... 21 3.2 Response of the GaAs samples to radioactive sources . 22 3.3 Signal formation mechanism and estimations of charge collection time ...... 25 3.4 Temperature dependence of the properties of π-ν structures ...... 31 3.5 γ–ray irradiation of GaAs structures ...... 34

4 Neutron Irradiation of GaAs Structures 37 4.1 Description of the experimental method ...... 37 4.2 Irradiation induced defects and test structures properties 39

1 2 Contents

4.3 Sensitivity of GaAs structures to minimum ionizing beta particles ...... 44 4.4 Conclusion ...... 47

5 Irradiation in Mixed Beam at CERN 48 5.1 Introduction ...... 48 5.2 Investigated structures ...... 49 5.3 Experimental details ...... 50 5.4 Exposure in mixed beam, basic results ...... 51 5.5 Signal estimation ...... 56

6 Microstrip Detectors Test 57 6.1 Main parameters of the microstrip detectors ...... 58 6.2 Experimental set-up ...... 60 6.3 The beam test results ...... 61 6.4 Investigation of radiation hardness ...... 65

7 GaAs detectors for α-particles and heavy ions spectrometry 68 7.1 Introduction ...... 68 7.2 Detector manufacture technology ...... 69 7.3 Electrical characteristics ...... 69 7.4 Detector performance for α-particles ...... 72

8 X-ray Irradiation of Silicon Detectors 76 8.1 Experimental layout ...... 77 8.2 Results and discussions ...... 79 8.3 Summary ...... 86

9 Summary and Outlook 87 9.1 Comparison with prediction from Local Charge Neutral- ity model ...... 87 9.2 Rate of Irradiation ...... 89 9.3 Conclusion ...... 90

10 Acknowledgments 92 Chapter 1

Introduction

1.1 Background

The operation of semiconductor detectors in harsh radiation environments depends strongly on the radiation hardness of the device. One such environment is the CERN Large Hadron Collider (LHC). Here the inner tracker detectors close to the interaction point will receive charged particle doses in the range of 110 kGy and fast neutron ﬂuences of 1014 n/cm2 during the lifetime of an experiment. Applications with harsh radiation environments exist in a wide range of ﬁelds from experiments with high charged particle multiplicity to proton and nuclei irradiation therapy. The basic damage mechanisms in semiconductors are conveniently separated into two categories:

a) Bulk eﬀects, which are manifestations of the displacement of atoms from their normal sites in the crystal lattice.

b) Surface eﬀects, which are long-term ionization eﬀects in the in- sulators connected with the active semiconductor region.

Modern electronics is based on semiconductor structures formed by the introduction of impurities into semiconductors, thus creating centers with shallow energy levels (see later in this chapter). The Siberian Institute of Technology in Tomsk, Russia, has over the last twenty years adopted a non-traditional approach to create device structures through doping semiconductors with impurities which create centers with deep

3 4 Chapter 1. Introduction energy levels. The resulting, so called π-ν structures enabled new devices, since these structures allows to form high ohmical intrinsic-like GaAs material. In this thesis, the effects of bulk damage due to gamma, neutron, proton and pion irradiation on the properties of GaAs detectors in terms of current as a function of voltage (I-V), capacitance as a function of reverse bias voltage (C-V) measurements and charge collection efficiency degradation are experimentally investigated. The thesis also includes a study of recovery effects of a Si-detectors by voltage training from X-ray irradiation.

1.2 Aims of Work

The aims of this work are to enhance the understanding of GaAs with deep level dopants from an experimental physics point of view and to study the radiation hardness of GaAs for high energy physics applications. More speciﬁcally this involved:

• investigation of diﬀerences between industry-standard, SI-U (Semi- Insulating, Undoped), LEC (Liquid-Encapsulated Czochralski [1]) and epitaxially-grown GaAs layers

• provide input to optimize fabrication technology for simple pad and microstrip detectors production

• evaluation of the radiation hardness of GaAs detectors to charged particles, neutrons and gamma

• testing of commercial and laboratory pad and microstrip detectors to LHC speciﬁcation with LHC speed read-out electronics

• development of methods for enhancing the detection eﬃciency for ionizing radiation.

• investigate the feasibility of GaAs detectors as α-particles spectroscopy detectors 1.3. Properties of Semiconductor 5

1.3 Properties of Semiconductor

Figure 1.1: Unit cube of GaAs crystal lattice [2].

The first synthesization of GaAs was reported in 1929 by Gold- schmidt, and the first reported electronic properties of III-V (3 and 5 valence electrons respectively) compounds as semiconductors appear in 1952 [3]. Two sublattices build up the crystal, each of them consists of a so called face centered cubic lattice (fcc). The off-set between the two lattices is equivalent to half of the diagonal of the fcc cube. This structure is normally referred to as zinc blende. All semiconductor compounds from group III-V and II-VI have the same crystalline structure; Si and Ge (group IV) have the same crystalline structure as diamond. Both the diamond lattice and the zinc-blende lattice are cubic lattices. Figure1.1 shows a unit cube for GaAs. In Table 1.1 a list of fundamental properties and characteristics is given for GaAs and Si [4].

1.4 Band Structure

Electrons in isolated atoms can have only certain discrete energy values. When isolated atoms are put together in a crystal, the electrons are not restricted to single energy levels, but ranges of energies are allowed. 6 Chapter 1. Introduction

Property Units GaAs Si

Crystal structure Zinc blende Diamond Lattice constant [A]˚ 5.6533 5.43095 Density [g/cm3] 5.32 2.328 Atomic density [atoms/cm3] 4.5·1022 5·1022 Molecular weight 144.64 28.09 El.-hole pair creation energy [eV/pair] 4.3 3.7 Dielectric constant 12.85 11.9 Band gap at 300K [eV] 1.42 1.14 Intrinsic carrier conc. [cm−3] 1.7·106 1.45·1010 Electron mobility (undoped) [cm2/V·s] 8500 1500 Hole mobility (undoped) [cm2/V·s] 400 450 Melting point [oC] 1238 1415

Table 1.1: Room-temperature properties of GaAs and Si.

Those bands are referred to as valence and conduction bands (Fig- ure 1.2).

Figure 1.2: Energy band diagram for GaAs [5].

These bands are separated by an energy band gap, which is the main parameter for the semiconductor material. At low temperatures, there is little thermal energy available to push valence electrons across this gap, and the semiconducting material acts as an insulator. At higher temperatures though, the ambient thermal energy becomes suﬃcient 1.4. Band Structure 7 to force electrons across the gap, and the material will conduct elec- tricity. The probability of an electron having enough energy to make the transition is given by the Fermi distribution function 1.1.

1 f(E) = (1.1) eE−EF /kT + 1

The Fermi level shown on Figure 1.2 is the energy level at which the probability is equal to one half. For intrinsic semiconductors, the Fermi level is in the center of the band gap. For undoped GaAs, the energy band gap at room temperature is 1.42 eV and for Si 1.14 eV. The energy band diagram is referenced to the vacuum potential [6,7].

Figure 1.3: Energy band structure of Si and GaAs [5].

GaAs is a direct band gap semiconductor, Figure 1.3. This means that transitions between bands require only a change in energy, and no change in momentum. Si is an indirect band gap semiconductor and transitions between bands requires phonon assistance. 8 Chapter 1. Introduction

1.5 Charge Carrier Mobility and Drift Ve- locity

One of the main advantages of GaAs is that carrier mobility of GaAs can be as much as six times greater than that of Si at typical field strengths [8]. During the time between collisions with other carriers and ions, the carrier will achieve a velocity that is a function of the electric field strength. This drift velocity (ν) is proportional to the applied qτc electric field (E) and can be written as: ν = −( m∗ )E, where q is ∗ charge electron, τc is the mean free time between collisions and m is qτc the electron effective mass. The proportional constant m∗ is defined as charge carrier mobility (µ). The mobility is related to the mean time between collisions, determined by lattice scattering and impurity scattering. The term scattering refers to the physical process in which particles, such as electrons, are deflected haphazardly as a result of collision. With rising temperature the lattice scattering increases and dominates at high temperature. The mobility thus decreases with increasing temperature and the impurity scattering becomes less significant [6].

1.6 Classiﬁcation of Defects in Semicon- ductor

Crystals inherently possess imperfections, or crystalline defects. The defects can be classiﬁed into point defects, line defects, two-dimensional defects and three-dimensional or volume defects. The point defects usually involve isolated atoms in localized regions of a crystal. Line defects, on the other hand, involve rows of atoms, and typical exam- ples of line defects are dislocations. The electron states introduced by such surfaces are usually called ”surface states” rather than defect states. Volume defects in a crystal are also known as bulk defects, which include voids of extrinsic and intrinsic point defects. 1.6. Classiﬁcation of Defects in Semiconductor 9

The crystal point defects in semiconductor can be classiﬁed as follows:

• Vacancy: created by a missing atom.

• Interstitial: atom occupying interstitial site (is located in a non- lattice site).

• Substitutional: impurity atom replacing the host atom.

• Antisite: a special kind of substitutional defect when a host atom occupies the site of another host atom.

• Frenkel defect pair: a complex formed by the host atom displaced from a lattice site to a nearby interstitial site (vacancy- interstitial).

• Impurities: defects involving foreign atoms (i.e., impurities) are referred to as extrinsic defects.

All the listed eﬀects are so called point defects, which are important since they strongly impact the performance of any semiconductor device. For example they determine the dark current as well as the bias voltage for any detector [9–11]. Intrinsic defects in GaAs include both arsenic and gallium vacancies. The eﬀect of these vacancy defects has been observed to be neutral [12], deep donor-like, and deep acceptor-like [13]. A donor-like defect located at the middle of the energy band gap is the so called EL2 defect [14]. This defect is important since it enables a conversion of p-type GaAs to semi-insulating. The EL2 defect is more frequent in material grown from an arsenic rich melt. The electrical properties of the semiconductor will change depending on the impurity. Figure 1.4 shows the energy band diagram of GaAs with the impurity additions. The impurities will shift the Fermi level from the center of the band towards these impurity levels. Doping impurities e.g. Cu, Se are referred to as shallow level impurities accordingly [15–17] and impurities with energies in the center of the band gap e.g. Fe, Cr as deep level centers. Device performance is degraded by deep impurities since the charge carrier lifetime will be shorter. 10 Chapter 1. Introduction

Figure 1.4: Energy band diagram of GaAs with impurities [5].

GaAs resistivity can be controlled by precision doping with a deep- level impurity like Fe, Cr that has a conductivity type opposite to n-type impurities or defects introduced during growth. The importance of semi-insulating GaAs is that devices made of it by direct ion implantation are self-isolating, so they become well suited for detector fabrication. Tomsk GaAs researchers have entered the terms π-type and ν-type conductivity layers in analogy with p−-type and n−-type to distinguish the deep level centers doping from shallow level impurity doping. En- ergy zone diagrams of π-ν structures presented on Figure 2.2.

1.7 Author’s Contribution

The research presented in this thesis was performed at the Institute for High Energy Physics (Russia), at CERN (Switzerland) and at Kungl. Tekniska H¨ogskolan (Sweden) during the years 1993-2005. The thesis is based on work listed in Refs. [18–25]. For seven years I was heavily involved in the design and implemen- tation of beam tests of prototype Gallium Arsenide (GaAs) pad and microstrip detectors where the goal application was the inner detectors of the ATLAS experiment at CERN, Geneva. This work was performed in the RD8 collaboration. I took a major part in the analysis of the 1.7. Author’s Contribution 11

data from these beam tests. Different combinations of detectors and electronics were evaluated in parallel. I was also involved in measurements on proposed front-end electronics and in particular the optimization of the design of the preamplifier. In this case the foreseen application was not only high energy physics experiments like the ATLAS detector but also other imaging applications. The electronics was of both hybrid types with discrete components mounted on a PCB and integrated analogue circuits with 64 parallel channels. The Siberian Physics Technical Institute in Tomsk has manufac- tured the GaAs detectors described in this thesis. This institute has developed a new technology for GaAs detector fabrication based on growing the GaAs crystal in a magnetic field resulting in an improved crystal with less defects. I have supplied this group with experimental results for optimizing the inner structure of the GaAs detectors. I also evaluated the possibilities for the use of GaAs in the high intensity mixed beam employed by the CHORUS and NOMAD experiments at CERN [23]. In 1999-2001 I investigated possibilities of using GaAs detectors compensated with deep level centers for alpha particle and heavy ion spectroscopy. The plan was that if the other requirements were fulfilled the detectors could, due to their radiation hardness, be used in environments where other materials would not work. The result was promising as presented in [22] and the detectors may in the future be used in these applications. I did a significant amount of the writing for articles [18–25] and was corresponding author of all of them. If not explicitly stated otherwise I performed all preparations for irradiation experiments and provide data collection and most of analysis myself.

1.7.1 List of publications included in thesis. 1. [18] Chmill V.B., Chuntonov A.V., Sergeev V.A., Smol A.V., Tsyupa Y.P., Vorobiev A.P., Gordienko A.I., Potapov A.I. An exploration of GaAs structures for solid state detectors. Nucl. Instr. and Meth., A326: pp.310-312, 1993.

2. [19] Chmill V.B., Chuntonov A.V., Smol A.V., Vorobiev A.P., Tol- 12 Chapter 1. Introduction

banov O.P., Koretskaya O.B. Exploration of GaAs structures with π-ν junction for coordinate sensitive detectors. Nucl. Instr. and Meth., A340: pp.328-340, 1994.

3. [21] Chmill V.B., Chuntonov A.V., Krupnyi G.I. Rastsvetalov Ya.N. Radiation resistance of GaAs structures based on π-ν junctions. J.Phys. D: Appl. Phys. 28, , pp.559-564, 1995.

4. [20] Chmill V.B., Chuntonov A.V., Khludkov S.S., Smith K.M., Tol- banov O.P., Vorobiev A.P. Radiation hard microstrip detectors based on gallium arsenide. Nucl. Instr. and Meth., A379: pp.406- 408, 1996.

5. [22] Chmill V.B., Chuntonov A.V., Falaleev V.V., Smith K.M., Smol A.V., Vorobiev A.P., Tolbanov O.P. The Result of GaAs Irra- diation at Mixed Beam Neutrino Cave CERN. Fifth Interna- tional Workshop on Gallium Arsenide Detectors and Related Compounds (Cividale del Friuli, Italy June17-20 1997) Nucl. In- str. and Meth., A410: pp.54-60, 1998.

6. [23] Chmill V.B., Chuntonov A.V., Smol A.V., Vorobiev A.P., Tol- banov O.P., Smith K.M. Particle detector based on GaAs. Radi- ation hardness and spatial resolution. Nucl. Instr. and Meth.,A 409: pp.247-250, 1998.

7. [24] Chmill V., Chuntonov A., Kholodenko A., Vorobiev A., Tsyupa Y., Porohovnichenko L., Potapov A., Klamra W. Investigation of epitaxial GaAs charged particle detectors. Nucl. Instr. and Meth., A438, pp.362-367, 1999.

8. [25] Chmill V. X-ray Irradiation of Silicon Detectors. Submitted to Nucl. Instr. and Meth., 2006.

1.7.2 List of publications not included in thesis. 1. Chmill V.B., et al. Radiation Hard Microstrip Gallium Arsenide Detectors. Proceeding of the Third International Workshop on Gallium Arsenide and Related Compounds. S.Miniato, Tuscany, Italy, March 21-24 1995. World Scientiﬁc Publishing Co Pte Ltd. 1.7. Author’s Contribution 13

2. M. Lundqvist, B. Cederstrom, V. Chmill, M. Danielsson and D Nygren. Computer simulations and performance measurements on a silicon strip detector for edge-on imaging. IEEE Trans. Nucl. Science 47:(4), pp. 1487-1492, 2000.

3. M. Danielsson, H. Bornefalk, B. Cederstrom, V. Chmill, B. Hasegawa, M. Lundqvist, D. Nygren and T. Tabar, Dose-eﬃcient system for digital mammography, Proc SPIE Physics of Medical Imaging, 3977, San Diego (2000)

4. Mats Lundqvist, Bjorn Cederstrom,Valery Chmill, Mats Daniels- son and Bruce Hasegawa, Evaluation of a Photon-Counting X-ray Imaging System, IEEE Transactions on Nuclear Science, Vol 48 (4) (2001)

5. M. Lundqvist, D. Bergstrom, B. Cederstrom, V. Chmill, A. Chuntonov, M. Danielsson and M. Aslund, Physical Evaluation of a Prototype for the Sectra Microdose Mammography System, Proceedings 6th Internationa Workshop on Digital Mammography IWDM2002, Bremen, 2002, ed. Heinz-Otto Peitgen

6. Danielsson M, Cederstroem B, Chmill V, Lundqvist M, Aslund M. Measurements on a full-ﬁeld digital mammography system with a photon counting crystalline silicon detector. SPIE Phys. Med. Imaging 2003; 5030:547-552.

7. R.A. Achmadullin, V.V. Artemov, V.F. Dvoryankin, G.G. Dvoryank- ina, Yu.M. Dikaev, M.G. Ermakov, O.N. Ermakova, V.B. Chmil, A.G. Holodenko, A.A. Kudryashov, A.I. Krikunov, A.G. Petrov, A.A. Telegin and A.P. Vorobiev. Photovoltaic x-ray detectors based on epitaxial GaAs structures. Nucl. Instr. and Meth., A554, pp.314-319, 2005. Chapter 2

Description of Investigated Structures

A microstrip semiconductor detector has to contain a high-resistivity layer. In case of Si, depleted by external bias voltage, a p–n structure with an impurity concentration of 1011–1013 cm−3 has been taken as the basis. GaAs can not be grown as pure as Si, at least not yet, and the Tomsk engineers have chosen another way of making the concentrations of donors and acceptors equal to each other – compensated GaAs. This choice has substantiated the high radiation hardness. For the test structures, the GaAs detectors were fabricated both from laboratory material and from industrial material. The GaAs was grown in a magnetic ﬁeld by the Bridgeman [26] method in the Siberian Institute for Physics and Technology, Tomsk (basically a controlled freezing process taking place under liquid - solid equilibrium conditions). The industrial material was produced following the Czochral- ski [1] method (the crystal is pulled from the melt). We have low- resistivity n–type GaAs in industrial case and high-resistivity for laboratory material.

14 2.1. Technology 15

2.1 Technology

There are two technologies used by the Tomsk group to form high- resistivity layers: diﬀusion and epitaxy.

2.1.1 Diﬀusion Industrial material produced following the Czochralski method was taken as a wafer. Sn was chosen as the donor, and Cr as the acceptor. Initial concentration of the donor was of the order of 1017 cm−3. Ac- ceptor impurities were doped at the depth of 150 µm (see Figure 2.1).

Figure 2.1: Distribution of shallow donors Nd(Sn), shallow acceptors + Na(Zn) and deep acceptors Nt(Cr). Energy diagram for the n -π-ν-n structure. 16 Chapter 2. Description of Investigated Structures

Cr accepts electrons from Sn, otherwise, the acceptor compensates the donor, thus, the high-resistivity region of 109 Ohm·cm is formed at a depth of 70 µm (full thickness of sensitive layer ∼100–150 µm). That structure has been named as p-π-ν-n. The π-layer is weak p-type and the ν- one is weak n-type. Both layers can be considered as pseudo- intrinsic. Advantages:

large surface area – up to 5 inch in diameter.

Disadvantage:

chromium solubility, as well thickness of sensitive layer are limited: 150 µm at 1017 cm−3 impurity.

2.1.2 Epitaxy

Advantages:

resistivity distribution of sensitive region is even.

Disadvantage:

very small surface area can be made – no more than 10 mm2.

Two main structure types were investigated: p-π-ν-n and n+-π-ν-n. The contact layer was made of either n+– or p– type GaAs. On the n-type substrate the high-resistivity layer was grown by liquid-phase epitaxy with Fe as a dopant. The layer thickness was about ∼150 µm and the resistivity 107 Ohm·cm. The dopant concentration gradient, specified during the doping process in the π– and ν– regions induces an internal electrical field. In a structure without an applied biasing field, the average strength of the electrical field is 350 V/m, whereas in the space charge region (SCR) of the π-ν junction it can reach 104 V/m. The energy zone diagrams for the samples are given in Figure 2.2. 2.1. Technology 17

Figure 2.2: Energy zone diagrams.

The n+-π-ν-n detector has two back-to-back junctions: n+-π and π-ν. Here ”forward” and ”reverse” bias voltages refer to the π-ν junction. For the test p-π-ν-n structures, the GaAs used was the industrial material produced by the Czochralski method. In the industrial substrates the high-resistivity layer of 70 µm thickness was prepared by thermo-diﬀusion of chromium as a dopant. Inside the high-resistivity layer, by means of controlled doping, a π-ν junction was formed at a depth of 20 − 30 µm. 18 Chapter 2. Description of Investigated Structures

Bridgeman Czochralski Epitaxy Thermo-diﬀusion T n+, p γ n+, p n p(Fe) p(Cr) p n+(Cr) mix p(Cr) strip n+(Cr) α p+(pure)

Table 2.1: Type of tested GaAs detectors, material, conducting layer and main dopant for the diﬀerent measurements. T is temperature dependance, γ, n and p are the behavior for γ, neutron and proton irradiation, mix is a mixed particle beam study, strip is a microstrip detector test and α is a spectrometry test.

The investigations made for the diﬀerent structures, i.e. initial material, technology process of the π-ν layer and the type of the main dopant are summarized in Table 2.1. It is important to note, that the type of contact layer does not play a main role in the case of a correct choice the of concentration of compensating dopants. Using p+ as the top contact layer results in less thickness of dead layer compare to using n+, but on the other hand an n+ layer is the easiest way of making ohmical contact. Chapter 3

Response of GaAs detectors to Ionizing Irradiation

The investigated samples were plates of n–type GaAs with surface area varying from 2×2 to 5×10 mm2 and 200−300 µm thickness. By means of thermo-diffusion or epitaxy, either gas-phase or liquid-phase, a high- resistivity 100 ± 50 µm thick layer with resistivity 105 − 107 Ohm·cm has been formed on these plates. Inside this layer, through controlled doping with iron group elements, a π-ν junction which penetrates 70 ± 20 µm into the layer is formed. The dopant concentration gradient, specified during the doping process in those π– and ν– regions induced an internal electrical field.

19 Chapter 3. Response of GaAs detectors to Ionizing 20 Irradiation

Figure 3.1: Forward and reverse I-V branch of the samples. The lines are drawn to guide the eye.

A typical I-V characteristic for the samples with the p–type contact layer is presented in Figure 3.1. The I–V of the samples has a clear diode shape with the forward and reverse currents differing by a factor of 104. The samples have an S–shaped reverse I–V branch with the switch-over voltage specified at the manufacturing stage. More details about studies of I-V characteristics can be found in Refs. [15–17]. The integrated π–ν junction in the structures ensures the smallness of reverse bias leakage currents, about 1 − 5 nA per sample, at a reverse bias voltage of 200 V, a characteristic difference between these structures and those with a p–i–n (i–intrinsic) junction or a Schottky barrier. 3.1. Measuring bench 21

3.1 Measuring bench

The characteristics of GaAs structures were investigated by means of a measuring bench with two independent electronic channels: Channel one — a charge sensitive amplifier; an amplifier-shaper with an active filter Polon 1101 (Ortec) and a self-triggered peak sensitive pulse analog-digital converter (ADC) ADC 712 . The calibration of the channel is approximately:

Q(N) = 230 · N + 250 [e−], (3.1) where Q is the charge in electrons collected at the input of the amplifier, and N is the ADC channel number . The optimum Gaussian shaping time for the output signal ts = 1 µs, was chosen such as to minimize the noise in the channel. At an equivalent detector capacity the noise at the input of the amplifier is approximately 1600 electrons (FWHM). Channel two — a current sensitive amplifier ”Garantiya” [27], a ”strobable integrator” analog-digital converter LeCROY 2249A and trigger electronics. The gate length required to trigger the ADC was 40 ns. The equivalent noise charge at the input of the amplifier was 3500 electrons (FWHM). The calibration of the channel is approximated by: Q(N) = 400 · N + 4100 [e−], (3.2) All the electronics except the trigger system is industrial and made in CAMAC standard. The information was read out by a standard computer through a CAMAC crate controller. Chapter 3. Response of GaAs detectors to Ionizing 22 Irradiation

3.2 Response of the GaAs samples to radioactive sources

As a result of the repeated investigations of diﬀerent structures with doping of the active layers by dopants with deep levels (Fe, Cr), n+-π- ν-n and p-π-ν-n structures have shown the best signal to noise ratios and the highest charge collection eﬃciency. Figure 3.2 shows the charge spectra of the signals from a radioactive 106Ru β–source, (3.55 MeV end-point energy) obtained with the p-π-ν-n structure using the charge sensitive channel.

Figure 3.2: Spectra of the signals from p-π-ν-n structure registered in the charge sensitive channel for diﬀerent bias voltage.

The signal due to the β–particles can be clearly detected above the noise peak even at zero bias. Clearly, the bias voltage of a sample does not affect the charge collection efficiency. This shows that the Space Charge Region (SCR), as well as the regions of π– and ν– layers, are fully depleted by the intrinsic electric field. Another indication of this fact is the weak dependence of the structure capacitance on the reverse bias voltage, see Figure 3.3. 3.2. Response of the GaAs samples to radioactive sources23

Figure 3.3: C–V of the p-π-ν-n sample (at the frequency 1 MHz).

Figure 3.4: Spectrum of the signals from the n+-π-ν-n structure, exposed to 2 MeV β–particles, registered in the charge sensitive channel; the full line represents the ﬁtted Landau distribution for the ionization losses in the thin layer.

The response of n+-π-ν-n structures to monoenergetic β–particles with energies between 0.6 MeV and 3.1 MeV was explored in detail in Chapter 3. Response of GaAs detectors to Ionizing 24 Irradiation

Ref. [28]. Of greatest interest are the experimental results of the exposure of samples to 2 MeV electrons ( minimum ionizing particles) for which the energy loss can be approximated by the Landau distribution [29] for ionization losses in a thin layer of matter. Such a distribution in the 150 µm thick n+-π-ν-n high-resistivity sensitive layer is illustrated Figure 3.4. For minimum ionizing particles, the most probable ionizing energy loss in GaAs is 6.48 MeV/cm. The generation of an electron–hole pair requires on average 4.2 eV. Therefore, 23000 pairs are expected in the 150 µm layer of the sample active region. From Figure 3.4, the most probable value of the collected charge is 22000 electrons, which corresponds to a charge collection eﬃciency of more than 95%.

Figure 3.5: Spectrum of the signals from the n+-π-ν-n sample exposed to radioactive 241Am α–particles, registered in the charge sensitive channel.

Conventionally, the charge collection eﬃciency in Si and Ge detectors is estimated from α–particle radiation. However, the high density of charge released at a depth of 5 − 10 µm from 5 MeV α–particles in GaAs results in a high density of pairs of charge carriers and, as a consequence, in noticeable recombination. For this reason the charge collection eﬃciency is not more than 20% in this case (Figure 3.5). 3.3. Signal formation mechanism and estimations of charge collection time 25

In particular samples the thickness of the upper contact layer exceeds 25 µm which is enough to stop α–particles. Nevertheless, we have detected signals due to the ionization by α–particles. Figure 3.5 shows the spectrum of the signals from 241Am α–particles (5.5 MeV) due to the ionization of the n+-π-ν-n structure.

3.3 Signal formation mechanism and estimations of charge collection time

The model of the signal formation mechanism [28] in the semiconductor structures is based on the following assumptions:

• the problem is one dimensional;

• the electric ﬁeld in the sample should be constant;

• the track ionization density should not change;

• eﬀects due to diﬀusion and charge carrier capture are negligible;

• no plasma eﬀects are observed;

• the currents of electrons and holes are uncorrelated.

Then the problem is reduced to the motion of the charge column in the fully depleted layer of the p-i-n structure. If the proper field strength in the structure is 350 V/m, then the charge collection time, calculated by the drift model, should not exceed 500 ns. However, in analysis of the response of the structures to β–particles some signals were longer that 800 ns, whereas with 241Am α–particles the charge collection period increased to even 1 µs, which exceeds the estimate obtained with the drift model. To explain this effect, a plasma model has been applied whose es- sential points are: in the sensitive layer a charged particle (electron, in our case) leads to a high density of electron-hole pairs ∼ 1017cm−3. For the track diameter one may choose 0.5 µm, which is of the order of magnitude of the maximum δ–electron path length. For a track ionization density of 177 pairs/µm the necessary condition for the existence of a plasma in a semiconductor should then be fulfilled [30], since at Chapter 3. Response of GaAs detectors to Ionizing 26 Irradiation the given concentration of carriers the penetration depth of the internal electrical field of the structure into the plasma is small compared to the length of the plasma track. Clearly, the duration of the signal from the structure will be extended by the plasma lifetime, tpl. In other words, it will become longer by the time period starting when a particle enters a detector until the plasma state has decayed and all the carriers start to move under the action of the electrical field. To estimate the time tpl, a model has been chosen [31] which exploits the ambipolar character of diffusion. Underlying the model were the assumptions that a particle track has a cylindrical geometry and that in their motion away from the track the current carriers generate a current limited by the space charge. Ã ! 1 3 3QoqenlA 1 tpl = 3 2 2 · ; where, (3.3) 32π µ(²²o) Da E

Qo — means the charge due to ionization, qe — is the elementary electrical charge; nl — is the linear density of the generated charge; µ — is the mobility of carriers; ²²o — are the dielectric constant; Da — the ambipolar diffusion coefficient; A — the initial cross-sectional area of the track; E — the electrical field strength.

Figure 3.6 shows the calculated charge collection time, taking into account the plasma effect(plasma time) and drift time as function of built-in voltage. For the calculations, the extraction of 50000 electrons (the average value of the Landau distribution) was assumed. The integrated potential difference was assumed to vary between 0 and 50 mV for the 150 µm sensitive layer. Present-day technology allows to attain potential differences up to 1 V. At this voltage for the same number of electrons (50000) the charge collection time will be reduced to 40 ns. The investigated samples have a built-in field voltage of 40–50 mV. The charge collection time versus the extracted charge at this potential difference is presented in Figure 3.7. 3.3. Signal formation mechanism and estimations of charge collection time 27

Figure 3.6: Estimated dependence of the charge collection time taking into account the plasma eﬀect for the release of 50000 electrons. The lines are drawn by interpolating spline.

Figure 3.7: Estimated dependence of the charge collection time taking into account the plasma eﬀect at the integrated potential diﬀerence 50 mV as function of the numbers of injected electron. The lines are drawn by interpolating spline. Chapter 3. Response of GaAs detectors to Ionizing 28 Irradiation

To get an experimental veriﬁcation of the increase in the charge collection time (> 500 ns) due to the plasma lifetime in the structures, the samples were exposed to monoenergetic 2 MeV and 0.6 MeV β–particles at diﬀerent bias voltages. To get monoenergetic electrons from the radioactive source, a setup based on a permanent magnet was used, described schematically in [28]. In experiments with monoenergetic beams, a 5 × 10 mm2 sample of the p–π–ν–n structure was used having a p–type contact layer in the form of a lattice.

Figure 3.8: The charge spectra of the signals from the p-π-ν-n sample exposed to monoenergetic 0.6 MeV β–particles at various bias voltages. Every ﬁfth data point is labelled. The lines drawn connecting every data point.

Figure 3.8 shows the spectra of the signals obtained after exposure of the sample to 0.6 MeV monoenergetic electrons at bias voltages between 0 and 300 V. The exposure to 0.6 MeV electrons is characterized by multiple scattering in the 150 µm thick sensitive region. The signal charge spectrum of these particles do not follow the Landau distribution for ionization losses in a thin layer because of the accompanying events with a large number of extracted charge carriers. The shape of the distribution is practically unaltered at bias voltages above 100 V. 3.3. Signal formation mechanism and estimations of charge collection time 29

Minimum ionizing particles in GaAs, 2 MeV electrons, lead to a track ionization density which is much lower than that due to ionization by 0.6 MeV electrons. Figure 3.9 shows the signal spectra from the structure exposed to 2 MeV electrons at bias voltages from 0 to 300 V. As can be seen, at voltages above 30 V the form of the distribution practically does not change. The output pulse base width after the ampliﬁer is about 100 ns at 0 V bias. The width decreases for increasing bias but is due to the read out electronics limited to >20 ns. The signal distribution together with a Landau ﬁt is shown in Figure 3.10 for a bias voltage of 100 V.

Figure 3.9: The charge spectra of the signals from the p-π-ν-n sample irradiated by 2 MeV monoenergetic β–particles at diﬀerent bias voltages. Every ﬁfth data point is labelled. The lines drawn connecting every data point.

Clearly, as the density of the carriers in the track increases, the plasma decay time is increased and, as a consequence, a higher electric field should be applied to the structure to make the charge collection time smaller than the gate length (40 ns) in the current sensitive channel. The typical increase of the charge collection time with the linear density of the extracted charge, together with the strong dependence on the electric field voltage, are consistent with the model predictions of Chapter 3. Response of GaAs detectors to Ionizing 30 Irradiation the influence of the plasma on the charge collection time. Calculations which take into account only the carrier drift in the depleted detector region surely underestimate the charge collection time.

Figure 3.10: The charge spectrum of the signals from p-π-ν-n samples irradiated with 2 MeV by β–particles, registered in the current sensitive channel at 100 V bias.

The conclusion from the above is that in GaAs structures with an integrated π–ν junction an increase of the external electric field should reduce the charge collection time due to the decrease of the plasma lifetime. This does however, not increase the charge collection efficiency. In high-resistivity GaAs the carrier lifetime τo may be as large as 10−4 s, so that all the carriers may be collected without a bias, since τo > tpl and τo > tdrift. This has been verified by numerous measurements [28,32] carried out without a bias voltage. A correct description of the charge collection mechanism requires also that carrier recombination in the plasma column along the particle track should be taken into account. The high carrier density, ∼ 1015 cm−3, increases the probability of radiative recombination of the carriers [33]. GaAs is known to be a straight-band semiconductor in which ”band to band” – type recombination does not involve phonons and is there- 3.4. Temperature dependence of the properties of π-ν structures 31 fore accompanied by the emission of light quanta. The radiative recombination lifetime, estimated in the Van Roosbroek and Schoklye −9 model [34], was τr < 10 s, whereas estimates for the nonradioac- −8 tive recombination lifetime gave τo ∼ 10 s. The recombination in the track channel should lead to spontaneous reradiation with hν ≤ Eq (Eq being the band gap width). If, as supposed in [28], radiation due to recombination were selectively absorbed in the π–ν junction, then the newly produced pairs would appear away from the track of the particle, which, in turn, would lower the density of nonequilibrium carriers in the track. As a result this mechanism, together with the electric field and diffusion, would reduce the plasma lifetime. Presumably, however, due to the sharp increase of the optical absorption coefficient in the strong electric field region (the Franz–Keldysh effect) [35] the radiation is absorbed selectively in the ambipolar diffusion field region inside the plasma column, generating new electron-hole pairs. This mechanism of charge transfer due to photoelectron transformations in GaAs may be observed for samples with surface contact layer thicker than 20 µm where the α–particle stops (see Figure 3.5). At a bias voltage of 100 V, the signal duration from these structures was shorter than 20 ns (limited by the capability of the electronics). It follows from the above that structures with p–type GaAs strips may be utilized as coordinate detectors in high luminosity experiments. For the case of irradiation by 2 MeV β–particles (minimum ionizing particles) the signal-to-noise ratio was equal to 8.5, while the detection efficiency approached 100% for a threshold of the trigger electronics at 3σ (noise). Structures with an n+–type contact layer may be used for detectors in experiments with relatively low luminosity, since such structures have charge collection times between 150 and 1000 ns depending on the extracted charge value. In this case an external electric field can help to reduce the charge collection time.

3.4 Temperature dependence of the properties of π-ν structures

To optimize the characteristics of semiconductor detectors in the operating temperature range 10 to 70 oC, the process parameters at the Chapter 3. Response of GaAs detectors to Ionizing 32 Irradiation manufacturing stage of n+-π-ν-n and p-π-ν-n structures have been revised. Samples of the structures have been obtained whose properties seem optimal for the given temperature range. Investigations were made how noise characteristics (Figure 3.11), signal peak position (Figure 3.12) and signal-to-noise ratio (Figure 3.13) under exposure to 106Ru β–particles varied with temperature. In all ﬁgures triangle is the n+–contact and square is the p–contact.

Figure 3.11: The temperature dependence of the noise in GaAs samples. Fit1 and Fit2 are line ﬁts.

Figure 3.11 shows the measured noise as function of temperature for bouth structures. With increasing temperature from −10 oC to +40 oC, the noise is reduced due to the reduction of the resistivity of π and ν regions due to the more eﬃcient compensation role of the doping atoms (thermo-ionization). Above +50 oC the noise in the sample increases, because decreasing of resistivity due to depletion of the doping admix- tures and the scattering of phonons of the crystal lattice. 3.4. Temperature dependence of the properties of π-ν structures 33

Figure 3.12: Temperature dependence of the signal peak position in GaAs samples.

Figure 3.13: The signal-to-noise ratio vs. temperature in GaAs samples. Chapter 3. Response of GaAs detectors to Ionizing 34 Irradiation

The location of the minimum noise value relative to the temperature axis can be optimized by varying the compensating admixture concentration, so that the signal-to-noise ratio is maximized in the operating temperature range (Figure 3.13).

3.5 γ–ray irradiation of GaAs structures

GaAs samples of n+-π-ν-n and p-π-ν-n structures were also exposed to 137 Cs γ–rays, Eγ = 661 keV, at a dose rate of Pγ = 6.074 rad/s. The absorbed dose was accumulated in stages, with measurements of parameters of the samples carried out in the intervals between. The dark current, noise and the charge collection eﬃciency were measured using a 106Ru radioactive beta source. The results of the measurements, presented in Figures 3.14, 3.15 and 3.16 show that an absorbed dose up to 110 kGy (11 Mrad) leads to a slight decrease of the charge collection eﬃciency. The signal-to-noise ratio increases by 10–30% when the absorbed dose increases to 110 kGy. The decrease of the noise can be explained by thermo-normalization of GaAs structure, impurity atoms obtains the right places and became compensating centers.

Figure 3.14: The noise of GaAs samples vs. absorbed dose of 137Cs γ-rays for two diﬀerent samples. The straight lines are ﬁt to the data. 3.5. γ–ray irradiation of GaAs structures 35

Figure 3.15: The signal peak position vs. absorbed dose of 137Cs γ-rays for two diﬀerent samples. The straight lines are ﬁt to the data.

Figure 3.16: The SNR vs. absorbed dose of 137Cs γ-rays for two diﬀerent samples. The straight lines are ﬁt to the data. Chapter 3. Response of GaAs detectors to Ionizing 36 Irradiation

Conclusion

The high charge collection efficiency (> 95%) that is independent of the external bias voltage, by virtue of the complete depletion of high- resistivity sensitive layers, may be achieved due to the high level of compensation in the starting material and the presence of an internal electric field. Low-level noise is also attainable at a bias voltage due to S–shaped reverse I–V branch of the π-ν junction. These characteristics give detectors based on of p-π-ν-n structures significant advantage compared with available detector prototypes based on the p-i-n junctions and Schottky barriers using semi-insulating GaAs [36–39]. The studies of the radiation resistance of the samples have revealed that for a gamma dose of 110 kGy only a very minor deterioration is observed. The research into the mechanisms of charge collection and pulse formation processes allow to conclude that at external bias voltages above 100 V the charge collection time is independent of the extracted charge value and that the pulse length in that case does not exceed a few nanoseconds. It is important to note that detectors based on p-π-ν-n and n+- π-ν-n structures might operate, in principle, without an external bias voltage under the action of only the internal electric field. In this case however, one should keep in mind the dependence of the pulse length on the created charge value. Chapter 4

Neutron Irradiation of GaAs Structures

Introduction

In this chapter the results of experimental investigations of the effect of very large neutron fluences 1015 n/cm2 on the characteristics of GaAs detector structures with π-ν – junctions are presented. The sensitivity of this type of structure to ionizing radiation was described in the previous chapter. The degradation of the GaAs structures is measured in terms of the reduction in charge collection efficiency and signal to noise ratio measured for minimum ionizing beta particles from a radioactive 106Ru source and from changes in I–V and C–V electrical characteristics.

4.1 Description of the experimental method

The p-π-ν-n structures used in this test was fabricated from the Bridge- man method of crystal growth in a magnetic ﬁeld and from the industrial GaAs material produced following the Czochralski method. On Bridgeman material, a high-resistivity 107 Ohm·cm ∼ 150 µm thick layer was grown by epitaxy method with iron doping. In the case of industrial substrates the high-resistivity layer of 70 µm thick was prepared by in-diﬀusion of chromium as a dopant. The contact layer was made of n+ or p – type GaAs.

37 38 Chapter 4. Neutron Irradiation of GaAs Structures

Neutron irradiation of the prepared samples was carried out at the I-100 proton linear accelerator laboratory of the Institute for High En- ergy Physics (Protvino). The characteristics of the proton beam from the accelerator are given in Table 4.1.

Parameter Value Energy 100 [MeV] Beam current 50 [mA] Pulse length 100 [µs] Frequency 0.5 [Hz]

Table 4.1: Parameters of the I-100 proton beam. The beam size and its transverse shape (circular or elliptical) depends on the tuning of the accelerator and varies from 5 to 50 cm2. The beam is made up of 0.6 ns bunches at 6 ns intervals.

For the irradiation of the samples, a target of aluminium of total absorption length for the 100 MeV protons was used. The GaAs samples to be irradiated were located on the beam axis at a distance of 0.1 m from the target. The main characteristics of the radiation at this point are given in Table 4.2.

Average Neutron ﬂuence Gamma dose Absorbed dose neutron energy En > 6 MeV [Rad/proton] [Rad], 20 min [MeV] [n/cm2] after beam oﬀ

23 ± 3 (1.2 ± 0.2) · 10−4 (4.7 ± 0.9) · 10−13 (1.2 ± 0.4)

Table 4.2: Main characteristics of the irradiation volume. Right column shows absorbed gamma dose for the 20 min waiting time due to safety regulation.

The GaAs samples were irradiated by neutrons with a rate of (4–5)·1012 n/cm2 per hour. The exposure time was typically two hours. The samples were exposed to neutron irradiation - two commercial samples and two fabricated at the Tomsk laboratory. In order to sim- ulate more precisely the actual conditions to be expected in practical use of the detectors, one sample of each type was irradiated under a 4.2. Irradiation induced defects and test structures properties 39 reverse bias voltage of 300 V, since at an electric ﬁeld of more than 1 V/µ m of thickness, defect migration mechanisms and formation of more complex defects can become more signiﬁcant [40]. The irradiation was carried out in stages, with the response of the samples to beta particles from a 106Ru β-source being checked at each stage.

4.2 Irradiation induced defects and test structures properties

The present understanding of radiation-induced defects in low resistivity GaAs crystals is reviewed in references [41–44].

Figure 4.1: Resistivity of GaAs material as a function of radiation ﬂux by fast protons, neutrons and electrons. (no is an initial density of electrons in GaAs.) Open symbols are earlier measurements by Tomsk group. Solid circles are results from one of the neutron irradiated samples. The lines are drawn to guide the eye.

When the low resistance GaAs material are exposed to high energy ionizing radiation [41], compensation of the electrical conductivity has been demonstrated. This compensation is found not to depend on 40 Chapter 4. Neutron Irradiation of GaAs Structures either the type of initial conductivity or on single impurities, but does depend on the trapping of free charge carriers by radiation-induced defects. Figure 4.1 shows the change in resistivity of GaAs samples sub- jected to irradiation by fast protons, neutrons and electrons [45]. The resistivity tends to a maximum value of 109 Ohm·cm independently of the type of irradiating particle. The increase in specific resistance is due mainly to the reduction in the density of free charge carriers. In our case, the initial compensation means that there is no further increase in specific resistance up to some limit of the neutron dose from the I-100 accelerator. For our samples of GaAs which have been compensated by iron, the typical resistance only begins to show an increase for neutron fluences in excess of 1015 n/cm2 (Figure 4.1).

Figure 4.2: Forward I–V characteristic as a function of neutron irradiation. Regions I, II and III correspond to diﬀerent current mechanisms.

The I-V characteristic variation with neutron irradiation was studied using π-ν-n structures. The forward I-V characteristic is illustrated in Figure 4.2. The ohmic region of the characteristic (I) can be observed for both forward and reverse bias voltages of less than 0.1 V. The minority carrier lifetime, τ0, in the space charge region has been measured from the slope of this part of the I-V curve and found to 4.2. Irradiation induced defects and test structures properties 41

Figure 4.3: Reverse I–V characteristic as a function of neutron irradiation. Regions I, II and III correspond to diﬀerent current mechanisms.

be independent of the irradiation and lies in the range 0.5 − 1 ns. In the recombination region (II) the forward characteristic changes signif- icantly. When the dose increases, it is observed that this part tends to disappear, i.e. recombination in the space-charge region becomes negligible in spite of the widening of this area of the volume charge. The third region (III), connected with the double injection of current in the π–layer, has also changed: with increased dose this part of the current becomes smaller and, ultimately, we see the transformation to an ohmic state. Estimation shows that when the dose increases, the specific resistance of the thin π–layer also grows in proportion up to 108 Ohm·cm (Figure 4.4). Regions II and III of the reverse bias branch also become deformed (Figure 4.3). Let us analyze the processes in the π–layer. There is a large increase in the resistance of the π–layer, as shown in Figure 4.4. A classical conductivity modulation method was used to establish the lifetime of the minority carriers (electrons) in the π–layer. A fixed-length forward bias rectangular pulse was applied to the structure, resulting in the injection of electrons into the π–region from the displaced π-ν junction. In a fixed time interval the rectangular test pulse was applied to the 42 Chapter 4. Neutron Irradiation of GaAs Structures

Figure 4.4: Resistivity of the π–layer vs. neutron irradiation (left). Photovoltaic e.m.f. vs. neutron irradiation (right). Fluence scale from 11 16 2 4 9 −3 10 to 10 n/cm , ρ-scale from 10 to 10 Ohm·cm and Uo from 10 to 1 V.

structure again. Because there is no field in the π–layer (voltage not applied to detector) the shape of the output pulse shows the recovery of the π-ν region after the injecting pulse. The minority carriers were washed out due to recombination only. When the fluence of neutrons increased, the lifetime decreased from 10−4 to 10−10 s. At a neutron fluence higher than 1014 − 1015 n/cm2 it was impossible to create an injection of electrons into the π–region. Evidently the reason for this phenomenon is the very high resistance of this region and the weak charge carrier injection into the π-ν junction. All these results are in good agreement with the model, where we can consider our structure as a chain of two connected resistors the π-ν junction and the π–region. Due to the influence of the absorbed neutrons, the additional compensation of the π and ν regions can be observed. The resistance of the π–region increases, (as seen in Figure 4.4), but the π-ν junction is washed out. As a result, at a limiting dose of 1015 n/cm2 a typical p-i-n structure develops. The same conclusion can be obtained from C-V variations, (cf. the region of the forward step 4.2. Irradiation induced defects and test structures properties 43

Figure 4.5: C–V variations as a function of bias for diﬀerent neutron irradiation. The lines are drawn to guide the eye.

in Figure 4.5), and from the large decrease of the photovoltaic e.m.f. with increased radiation dose (Figure 4.4). This is because charge carriers created by absorption of light cannot be effectively separated by the field of the space-charge region. Such a model easily explains the high radiation resistance of the π-ν-n structures doped with iron because of the higher initial level of doping and much lower resistance of the π–layer. GaAs samples of n+-π-ν-n structures have also been irradiated using the U-70 accelerator, with a fluence of neutrons around 1017 n/cm2. After irradiation the structure became like a pure resistor type. This can be confirmed by the changing of the forward I–V characteristic. The I-V characteristic not presented here, it is absolutely equal to the I-V characteristic of the resistor. The growth of the resistance in the forward region can be seen in these I–V characteristics which are typical of p-i-n structures. At the same time, the response of the structures to beta particles from a 106Ru source completely disappears even at near-breakdown bias voltages. The specific resistance of the i– region reached 109 Ohm·cm. This result confirms the earlier conclusion that a peculiarity of highly irradiated samples of GaAs will have i– 44 Chapter 4. Neutron Irradiation of GaAs Structures

Figure 4.6: Noise (RMS) vs. neutron irradiation. The lines are drawn to guide the eye.

type conductivity independently of the initial material’s conductivity type [40]. This behavior is in agreement with the change of detector capacitance due to the creation of compensation impurities (centers) during irradiation (Figure 4.5).

4.3 Sensitivity of GaAs structures to minimum ionizing beta particles

Figures 4.6, 4.7, 4.8 show the experimental data for irradiated samples which were biased up to 300 V during irradiation. The behavior of the samples (not shown here) which had no bias voltage applied during the irradiation is similar. It was not possible to estimate the inﬂuence of the bias voltage on the rate of degradation of the structures because of the small number of structures investigated during the experiment, but in each case GaAs structures which were chromium compensated degraded much earlier: neutron ﬂuences of 4 · 1014 n/cm2 led to results similar to those for 1.0−1.2·1015 n/cm2 for samples doped with iron. The deterioration in 4.3. Sensitivity of GaAs structures to minimum ionizing beta particles 45

Figure 4.7: Charge collection (spectrum peak position for most probable energy loss) vs. neutron irradiation. The lines are drawn to guide the eye.

Figure 4.8: Signal to noise ratio vs. neutron irradiation. The lines are drawn to guide the eye. 46 Chapter 4. Neutron Irradiation of GaAs Structures

S/N ratio due to the irradiation was not more than 20% for the latter. The much lower radiation resistance of chromium doped samples can be explained by the following. To decrease the specific capacity of these structures (which leads to an increase in the S/N ratio for samples before irradiation (Figure 4.8)), the sample processing used a chromium concentration which was ten times less than that of the iron in the case of doping by the latter. It is well known that radiation-induced defects in GaAs give deep trapping centers which have compensating properties like iron and chromium [46]. Consequently at a higher concentration of the chromium or iron dopants the relative influence of created traps at higher neutron fluence is smaller. One of the properties of GaAs after irradiation is the decrease in the intensity of radiative recombination, Refs. [40–44], pointing to the predominant creation of non- radiative recombination centers. Significant distortion of the lattice near the radiation-induced defects leads to an increase of the probability of multi-phonon non-radiative recombination of charge carriers and distortion of the lattice also accounts for the recombination-stimulated migration of defects [45]. All of this leads to a decrease of the lifetime of a non-equilibrium charge carriers concentration in π– and ν–layers caused by ionization by beta particles. For neutron fluences > 1015 n/cm2 full charge collection requires an increase of the electric field in the structure so that the charge collection time is not longer than the lifetime of non-equilibrium charge carriers (Figure 4.7). As discussed in Section 4.2, during the neutron beam irradiation of GaAs, additional compensation of the semiconductor takes place, leading to an increase in the specific resistance of the π– and ν–layers of the structure. The specific resistance of the π–region can be approximately ten times that of other layers. It is supposed that the extension of the space-charge region with increasing reverse bias voltage will take place mainly in the ν–region but with the increase of the specific resistance of the π–region, the voltage decrease does not occur over the whole sensitive thickness of the sample. To create the required electric field inside the whole sensitive region of the sample, it is necessary to raise the bias voltage [19]. This change of the voltage distribution between π– and ν–layers together with the decreased lifetime of non-equilibrium charge carriers leads to the observed degradation of the sample response to beta particles. 4.4. Conclusion 47

4.4 Conclusion

Fe doped detectors based on laboratory grown material (Bridgeman) shows better radiation resistance compared to GaAs detectors based on commercial (Czochralski) low resistivity Cr doped materials. The main reason is assumed due to to the lower dopant concentration. GaAs structures based on Czochralski low resistivity materials with increased chromium doping concentration can be in principle created. It can be expected that the radiation resistance of such structures will be similar to that of similar structures compensated by iron. The present results conﬁrm the possibility of the use of such structures as a basis for the fabrication of radiation-resistant coordinate detectors. The diﬀusion technology for the formation of compensated high- resistivity GaAs layers on industrial substrates is now better under- stood. This ensures the reproducibility of the electrophysical characteristics of the devices and furthermore provides the necessary surface areas to permit the application of planar technology in the industrial fabrication of strip detectors. Chapter 5

Irradiation in Mixed Beam at CERN

In this chapter results from irradiation of GaAs detectors at intense particles beam are presented. The irradiation was made just after the beryllium target in the special beam-line for the neutrino experiments. The target was bombarded with proton beams at 540 GeV at CERN. The study of the radiation resistance of the GaAs samples has shown that their main characteristics (charge collection eﬃciency, signal/noise ratio) degrade less than factor of two at the integral ﬂuence of 2·1014 particles·cm−2.

5.1 Introduction

Proceeding from the data on GaAs detectors irradiation by various types of particles, a hypothesis about the correlation between degradation of the properties of the detectors and the rate of irradiation is set forth. The results of the preceding investigations are summarized in Table 5.1. Therefore a test was made in an environment with a diﬀerent irradiation rate at CERN. The 540 GeV proton beam from CERN SPS accelerator was sent to the target with the following time structure: ten repeated cycles of two extractions, 1.5 · 1015 protons each, separated by 2.76 s. Clearly, the degradation of detectors is in proportion to the rate of irradiation in the range (1.5–3)·1012 — 5 · 1014 part.×cm−2 per hour. The estimates

48 5.2. Investigated structures 49

Rate of irradiation Final ﬂuence Beam type Accelerator particles×cm−2 at 50% less per hour signal degradation

20 MeV neutrons Protvino I-100 (4 − 5) · 1012 1.2 · 1015

1 GeV protons Protvino Booster 5 · 1013 1.5 · 1014

1 GeV protons Protvino Booster 5 · 1014 1.1 · 1013

Secondary beam CERN SPS (2 − 4) · 1012 1.3 · 1015 of neutrino facility

Table 5.1: Summary results of hadron beam irradiation. The lines before the last is preceding measurements and was done on other facilities. of the degradation were done using the signal-to-noise ratio. The results obtained using the facility of the CHORUS experiment (CERN) are based on the possibility of on-line measurements of the signal from detectors exposed to a mixed beam after target.

5.2 Investigated structures

For the test p-π-ν-n structures, the GaAs used was the industrial material produced by the Czochralski method. The samples were plates of low-resistivity n−type GaAs with surface area 2 × 2 mm2 and 200 − 300 µm thick. In the industrial substrates a high-resistivity layer of 70 µm thickness was prepared by in-diﬀusion of chromium as a dopant. Inside the high-resistivity layer, by means of controlled doping, a π-ν junction was formed at the depth of 20 − 30 µm. 50 Chapter 5. Irradiation in Mixed Beam at CERN

5.3 Experimental details

The 4-channel data acquisition system was developed and manufac- tured in NIM standard to ﬁt the special experimental area, which was a remote pit. Using this system, the following measurements were done:

1. Identify of the beam spill number dumped from SPS onto target. This was required since the two spills per cycle diﬀer.

2. Read-out the BCT (Beam-Current Transformer) counter data that is proportional to the number of protons hitting the target.

3. Measure the signal response to the secondary particle beam.

The module control and data acquisition was made using a PC that was connected to the module through a serial port. The duration of a single measurement cycle, with the bias voltage polarity ﬁxed, was around 2.5 hours. First, before placing the detectors along the secondary beam in

Figure 5.1: Signal response vs. muon fluence for different bias voltage and two different samples N1 and N5. Pulse height is normalized to BCT counter. the neutrino cave, calibration in a muon beam was done. Also, the 5.4. Exposure in mixed beam, basic results 51 module control and data acquisition system was tested. The GaAs detectors were placed close to the silicon ones used to measure the muon flux profile after the decay channel. The signals from the GaAs detectors were in correspondence with those from the silicon detectors and showed a linear dependence of the response to the muon flux up to 6.3 · 107µ/cm2·spill. At a bias of 70 V, the amplitude of the response reached a plateau. The obtained data are presented in Figures 5.1 and 5.2, respectively. It was expected that detectors would show a

Figure 5.2: Signal response vs. bias voltage. Lines for two diﬀerent samples N1 and N5 are drawn to guide the eye. linear dependence of response with increasing beam intensity, making it possible to estimate the ﬂux.

5.4 Exposure in mixed beam, basic results

Four GaAs detectors were located at three control points, co-axially within a gas ionizing chamber to control the ﬂux. Figure 5.3 shows the layout of the detectors and the equipment. 52 Chapter 5. Irradiation in Mixed Beam at CERN

Figure 5.3: Experimental layout. In the lower part the beam enters perpendicularly to the drawing in the center of the 2 m gas ionizing chamber. The GaAs detectors were placed at points 1, 2 and 3.

Two detectors were placed oppositely in the beam center (p.1), the 3rd and 4th detectors were put 30 and 80 cm (p.2 and 3 in Fig- ure 5.3) from the secondary beam axis. The ratio of the particle intensity at these points was 1 : 0.32 : 0.16 respectively. For the operating mode in the neutrino cave, the secondary flux in the center was 2.5 · 109 particles/cm2·spill. Figure. 5.4 shows the signal from one of the detectors, as well as the BCT counter data. The difference between the RMS distribution amplitudes was less than 0.5%. The experiment revealed that sometimes the number of protons on the target did not correlate with the amplitude of the signal coming from the detector. The explanation might be either variation in the impact parameter value in beam ejection, or the fine structure of the beam. This effect shows itself noticeable in Figure 5.5 where the BCT counter data and the signal responses of the detectors are given for only the first spill each cycle. 5.4. Exposure in mixed beam, basic results 53

Figure 5.4: Raw data from one of the detectors and the BCT counter.

Figure 5.5: GaAs (signal amplitude) response vs. mixed beam ﬂuence. 54 Chapter 5. Irradiation in Mixed Beam at CERN

Figures 5.6, 5.7 and 5.8 display responses of the detectors versus the irradiation dose for three values of bias voltage, respectively. The last point for the 4th (p.3 in Figure 5.3) detector was obtained at the minimum rate of irradiation under the exposure to the ﬂux 2 · 1014 after 7 days. The degradation factor was equal to 1.6. The lowest curve corresponds to detector placed in a maximum irradiation ﬂux (center).

Figure 5.6: The degradation of signal response for bias value 23 V. The three lines correspond to the diﬀerent positions for the GaAs detectors. 5.4. Exposure in mixed beam, basic results 55

Figure 5.7: The degradation of signal response for bias value 46 V. The three lines correspond to the diﬀerent positions for the GaAs detectors.

Figure 5.8: The degradation of signal response for bias value 76 V. The three lines correspond to the diﬀerent positions for the GaAs detectors. 56 Chapter 5. Irradiation in Mixed Beam at CERN

5.5 Signal estimation

To measure the ﬂux, only the data of the gas ionizing chambers was used because of the nonlinear response of the GaAs detectors to ionization. The signal amplitude of the detector can be written as:

h³ dE ´ i dx Φ U = · qe /ts Rin (5.1) ∆Ecr where,

dE dx – ionizing energy loss; Φ – ﬂux;

∆Ecr – electron-hole pair creation energy;

qe – charge of electron;

Rin – input impedance of ampliﬁer;

ts – spill time.

A simple calculation using(5.1) reveals that at full charge collection, the signal amplitude would be around 100 V with Rin = 1 MOhm. The signals ∼ 20 V that are observed at the bias voltage 70 V testify to the fact that the basic signal production mechanism is of drift nature. The value of the signal relaxation time after spill transition, provided one takes into account that the cable length is 60 m and the total capacitance of the detector plus cable is 2500 pF, also speaks in favor of the absence of either breakdown, or current avalanche in the detectors. The data shows that the degradation of the detectors, as expected, is a function of the irradiation rate. Chapter 6

Microstrip Detectors Test

In this chapter the results of the first steps of studying pioneering GaAs prototypes of radiation resistant coordinate detectors using charge particle beams are presented. The structure design of two detector prototypes with 150 µm and 50 µm pitch and their main characteristics are described. The results of radiation tests show good operating conditions for protons and neutrons with fluxes up to 2 · 1014 p/cm2, 1 · 1015 n/cm2 and γ-rays absorbed dose up to 110 kGy in CH(polyethylene)- equivalent. The response of the detectors to the charge particles and the detection efficiency have been measured.

57 58 Chapter 6. Microstrip Detectors Test

6.1 Main parameters of the microstrip detectors

Figure 6.1: Intrinsic structure (Au)n+-π-ν-n type.

1. Ohmic contact Au-n+. 2. Microstrip n+ – type. 3. π – layer. 4. ν – layer. 5. Substrate n – type. 6. Ohmic contact Au-n.

The ﬁrst GaAs detector prototypes employed one type of intrinsic structure: (Au)n+-π-ν-n, developed at the Siberian Institute for Physics and Technology, (SIPT), Tomsk. The GaAs industry-standard Czochralski material substrate was taken as the basis. The detectors, with surface area 20 × 30 mm2, were made on 300 µm thick wafers of low-resistivity doped with Sn (5·1016 −1.5·1017 cm−3) n-type GaAs in which a high resistivity 125 ± 25 µm thick layer was formed by means of chromium diﬀusion. Inside this layer we have a π-ν junction at a depth of 70 ± 20 µm (Figure 6.1). The π and ν layers may be consider as i-type of GaAs. The thickness of the π and ν layers is a function of the chromium concentration ratio. 6.1. Main parameters of the microstrip detectors 59

Figure 6.2: The forward branch of I–V characteristics. The voltage scale is from 10−2 to 103 V and current scale is from 10−11 to 10−4 A. The lines are drawn to guide the eye.

Figure 6.3: The reverse branch of I–V characteristics. The voltage scale is from 10−2 to 103 V and current scale is from 10−11 to 10−4 A. The lines are drawn to guide the eye.

The readout strips, with pitch 150 µm and strip width 50 µm, were 60 Chapter 6. Microstrip Detectors Test formed from n+-type GaAs with Au on top. The I–V characteristics for the (Au)n+-π-ν-n detector, measured for a single strip, are presented in Figure 6.2 and Figure 6.3. The (Au)n+-π-ν-n detector have two back-to-back junctions: n+-π and π-ν. Here ”forward” and ”reverse” bias voltages refer to the π-ν junction.

6.2 Experimental set-up

Two versions of the (Au)n+-π-ν-n microstrip detectors were produced, with pitch 150 and 50 µm and strip width 50 and 10 µm respectively. The schematic layout of the experimental set-up is shown in Fig-

Figure 6.4: Experimental set-up. ure 6.4. The strips were DC coupled to fast current sensitive ”Garan- tiya” preampliﬁers with about 10 ns peaking time [47]. The signals arriving at the ADC LeCROY 2249A were integrated with a gate of 20 ns. The ADC trigger used a coincidence of the four scintillator counters installed on the beam axis. 6.3. The beam test results 61

6.3 The beam test results

The detection efficiency of the 150 µm pitch detector was studied in a 40 GeV/c π− beam at IHEP (Protvino). To generate the trigger from beam particles we used up-stream beam counters S1,S2, a scintillator fibre counter placed close to the microstrip detector, with the fibre diameter 300 µm oriented along the GaAs detector strips, and a 0.7 × 1 cm2 finger counter perpendicular to the fibre. Figure 6.5 shows the spectrum of signals from a single strip at 300 V bias voltage. The most probable signal corresponds to 12 · 103 e− and the signal-to-noise ratio is approximately 6. The ”shadow” of the fibre on the detector can be seen in Figure 6.6. The detection efficiency was estimated to be only 50% at 4σ threshold. The reason for this low value was not further investigated.

Figure 6.5: Spectrum of the signals for a single strip (pitch 150 µm). Dashed line is Gaussian noise ﬁt and dotted is Gaussian-Exp ﬁt of signal.

Similar measurements were made with the 50 µm pitch detector in the 70 GeV/c proton beam at IHEP. The spectrum and the ﬁbre ”shadow” are presented in Figure 6.7 and Figure 6.8. 62 Chapter 6. Microstrip Detectors Test

Figure 6.6: ”Shadow” of the ﬁbre (pitch 150 µm). The two lines represent diﬀerent thresholds 3σ and 4σ of noise in the data analysis.

Figure 6.7: Spectrum of the signals for a single strip (pitch 50 µm). The most probable signal for a minimum ionizing particle (MIP) is indicated. 6.3. The beam test results 63

Figure 6.8: ”Shadow” of the ﬁbre (pitch 50 µm). The two lines represent diﬀerent signal thresholds in the data analysis.

Figure 6.9: Spectrum of the signals for a single strip (pitch 50 µm) for 5.9 GeV/c π−.

The most probable signal is equal to 10·103 electrons and the signal- to-noise ratio is near to 7 at the bias voltage of 95 V. Taking into account events with charge division between adjacent strips and accidental events we estimate the detection eﬃciency of the 64 Chapter 6. Microstrip Detectors Test

50 µm pitch detector to be 76% for signals greater than 4 · σnoise and 98% for signals grater than 3 · σnoise. The response of the detector in a 5.9 GeV/c π− beam at CERN was also measured. The same detector shows a better signal-to-noise ratio ∼9.6 at the bias voltage of 75 V (Figure 6.9). Diﬀerent readout electronics was used for this test and the noise was less due to better ground quality.

CLUSTER PROBABILITY

Detectors type 1 strip 2 strips 3 strips 4 strips 5 strips Threshold in % in % in % in % in %

pitch 50 µm 66 26 3.6 3.0 0.6 1.5 σnoise pitch 150 µm 63 29 5.8 1.6 0.7

pitch 50 µm 73 23 0.6 2.4 0.6 2 σnoise pitch 150 µm 78 19.7 1.8 0.3 0.09

pitch 50 µm 80 16 1.8 1.8 0.4 3 σnoise pitch 150 µm 88 11.5 0.5 0.07 0.01

pitch 50 µm 84 14 0.6 1.4 0.0 3.5 σnoise pitch 150 µm 92.7 7.0 0.17 0.05 0.02

Table 6.1: Cluster probability for 50 µm and 150 µm pitch detectors.

In Table 6.1. the charge size cluster probability for 50 µm and 150 µm pitch detectors is presented. To measure the signal cluster size events with the largest analogue signals in a certain strip were accumulated. The cluster size was then deﬁned as the number of neighboring strips with a signal above the threshold.

One can see from Table 6.1 and Figure 6.10 with threshold=3·σnoise and S/N=10 that the average cluster√ is only one strip. As a result the position resolution is equal to pitch/ 12 in these cases. 6.4. Investigation of radiation hardness 65

Figure 6.10: Mean cluster of events (pitch 50 µm).

6.4 Investigation of radiation hardness

The GaAs samples were made on the same wafer as the strip detectors have been exposed to beams of protons and neutrons to confirm their radiation hardness. The neutron irradiation was carried out at Rutherford Appleton Laboratory with a neutron energy ∼ 1 MeV. The results of this experiment are shown in Table 6.2. The study of proton irradiation was realized at the Booster Accel- erator (IHEP, Russia) with proton energy 1 GeV. There was also a neutron flux with energy 1-6 MeV and the same fluence as protons. The irradiation was carried out in stages, with the response of the samples to β-particles from a 106Ru source being checked. Fluences up to 2 · 1014 p/cm2 and approximately 2 · 1014 n/cm2 were obtained (Figure 6.11). 66 Chapter 6. Microstrip Detectors Test

Sample N110-10-25 (n+-π-ν-n) before irr. ∼ 5 · 1014 n/cm2 ∼ 1015 n/cm2 Q[e−] 12000 12500 10400

SNR 3.9 4.0 3.4

Sample N110-0-1 (n+-π-ν-n) before irr. ∼ 5 · 1014 n/cm2 ∼ 1015 n/cm2 Q[e−] 7800 12000 10500

SNR 2.5 3.9 3.4

Table 6.2: Most probable collected charge from detector and SNR for GaAs detectors after neutron irradiation at Rutherford Appleton Laboratory.

Figure 6.11: The collected charge versus proton irradiation. The ﬂu- ence scale is from 1012 to 2·1014 protons·cm−2. The lines are drawn to guide the eye. 6.4. Investigation of radiation hardness 67

The initial response to β-particles was taken as 100%,on the Y-axis. The difference between the two samples is the initial concentration of compensation impurity. The main radiation resistance parameters are defined by the donor concentration of the initial substrate and the value of the n-type material compensation, which depend on the acceptor concentration distribution. To obtain the high radiation hardness the admixture concentration of the initial substrate must be large. However, the chromium solubility is limited in GaAs. For the present samples the high resistivity layer with up to 120 µm thickness the initial Sn concentration is ∼ 2 · 1017 cm−3. The limited solubility explains a large gradient of Cr concentration and unevenness of the electric field through the structure. The signal rise with irradiation up to ∼ 2 · 1013 p/cm2 (Figure 6.11 protons) and ∼ 5 · 1014 n/cm2 (Table 6.2 neutrons), is caused by an increase of the high resistivity layer thickness from additional radiation compensation of GaAs material. Some of a radiation damages have acceptor’s properties and ionization energy like Cr. Chapter 7

GaAs detectors for α-particles and heavy ions spectrometry

7.1 Introduction

One main requirement for the fabrication of detectors for α-particles and heavy ions spectrometry is the uniformity of the detector thickness and its electrophysical characteristics. Only in this case will fluctuations of the particles energy losses be less or comparable to the statis- tical fluctuations of the created charge, which is formed by the particle in the detector structure. Detectors based on GaAs, due to its wide 2 band gap (Eg=1.43 eV) and high electron mobility (µ=8600 cm /Vs), offers the prospect of high speed particles detection and signal processing. The GaAs wafers are generally prepared by means of liquid-phase epitaxy and the results of studies of such detectors can be found in references [48–53]. The aim of the present work is to study charged particle GaAs detectors fabricated by means of vapour-phase epitaxy (VPE) in Siberian Institute for Physics and Technology, Tomsk, Rus- sia.

68 7.2. Detector manufacture technology 69

7.2 Detector manufacture technology

The VPE process of growing GaAs wafers on GaAs substrates doped with Te up to the concentration n+(Te)=(0.5−1)·1018cm−3 and 400 µm thickness is performed in a Ga-AsCl3-H2 gas transport system. The crystal orientation is 2o relative to the (100) direction. The process results n-type crystal in concentrations in the range of 1012 −1014cm−3. The thickness of the n-epitaxial layer is 30±3 µm. To form a p+-contact 17 19 −3 with the hole concentration 10 − 10 cm Zn is added to the AsCl3- + + H2 gas mixture. The p -layer thickness is ∼ 0.7µm. On the p - and n+-surfaces an 0.1 µm thick Au-Ge layer is evaporated to provide the ohmic contact. For all samples of 3×3 mm2 size the detector structure was of p+-n-n+ type with p+-n junction close to the surface. The detector structure is presented in Figure 7.1.

Figure 7.1: Inner structure of the GaAs detectors studied here.

7.3 Electrical characteristics

For the detector structure the I-V and C-V characteristics were measured. The frequency independence thus suggest that there not multiple structure thus indicating the absence of deep level centers. Chapter 7. GaAs detectors for α-particles and heavy ions

70 spectrometry

¥£

¥¢

¥ #

¥¡

¥ ¤ £ ¢ ¡

¦ ¦ ¦ ¦ ¦

§ ¨ ©

Figure 7.2: C-V characteristic.

In Figure 7.2 typical C-V curves are shown, as measured for 1 MHz and 1 kHz. The dependence is linear in the 1/C2 versus (−V ) co- ordinates. Thus, the doping concentration can be obtained from the experimental value of the slope parameter as

∆(1/C2) 2 tan(φ) = = , (7.1) ∆V qεε0NB

where q is the electron charge, εε0 the dielectric constant for GaAs. Using the experimental value of tan(φ) the n-region doping concentra- 13 −3 tion can be estimated to be NB = 4.2 · 10 cm . This corresponds a W0 = 3.2 µm wide space charge region (SCR).

7.3. Electrical characteristics 71

¡¢ ¢ £¤ ¢

! "

¡¢ ¢ £¤ ¢ ©

# $ &

! "

¡¢ ¢ £¤ ¢¨

¡¢ ¢ £¤ ¢ §

¡¢ ¢ £¤ ¢ ¦

¡¢ ¢ £¤ ¢ ¥

¡¢ ¢ £¤ ¢

¡¢ ¢ £¤

¢ ¡¢ ¢ ¡ ¢ ¢ ¢ ¢ ¢ ¢

Figure 7.3: I-V characteristics.

The I-V curve is presented in Figure 7.3 and is typical for a p- n junction. For the forward direction there is a linear ln(I) vs. V ∆(ln(I)) dependence, with a slope tan(θ) = ∆V = 23. According to the Sah- Noyce-Shockley generation recombination model [54] in the wide band gap semiconductors at low electric ﬁeld the recombination in the SCR p-n region is of main importance and thus the recombination current is written as:

qniWo qV Irec = S exp , (7.2) 2τ0 βkT 6 −3 where ni = 1.8 · 10 cm – the charge carrier concentration in GaAs, −8 τ0 ∼ 10 s – the charge carrier life time [55], β = 2, q – charge of electron, S – detector area, Wo – wide of space charge region (SCR). The experimental quantity β = q/(tan(θ)kT ) can be derived from the slope of the linear ln(I) vs. V dependence and the obtained value of 1.7 is in agreement with the theoretical prediction. Using the experimental β- value the recombination current at V = 0.6 V gives −6 −6 Irec.exper. = 8 · 10 A to be compared to Irec.theor. = 4 · 10 A. In the backward I-V region two current components are of importance. The Chapter 7. GaAs detectors for α-particles and heavy ions 72 spectrometry

ﬁrst one shows a I ∼ V 1/n dependence with n = 2 and is interpreted as a generation current in SCR.

qniW0 Ig = S. (7.3) 2τ0 1/n This current is strongly dependent on W0 ∼ V with n = 2 for −10 the p-n region. At V = 1 V the theoretical value Ig = 9.2 · 10 A is in a good agreement with the experimental Ig. The second current component exhibits an exponential behavior, I ∼ exp(µV ) where µ = 0.025 V−1 and corresponds to a microplasma current. At V = 100 V a rapid current increase is observed. This is because of the W0 ' 30 µm, i.e. the n-region is fully depleted and thus avalanche processes are of importance. An optimum detector bias of V = 70 V was chosen, since it corresponds to the maximum signal to noise ratio.

7.4 Detector performance for α-particles

2 32 4

2 9:

6 7

2 3 8

2 3 5 6 7

2 3 3 4

¡¢ £ £¤ ¥¦ § ¨ ©¨ ¤ £¤ ¤ ©

"#" $%& ' (") *

Figure 7.4: Energy calibration using α-sources.

The detector performance was studied using a standard set-up for alpha spectrometry. The energy calibration for alpha energies in the 7.4. Detector performance for α-particles 73 interval 4824 keV−8784.5 keV is presented in Figure 7.4 and described

by a linear relation.

¥£

>9 : :

> > : :

R STU V W XY S V \]

¥¢

>8 : :

> A : :

>@ : : I

¥¡

> ?::

> : : :

8 = : :

8 < : :

8 ; : :

,- 6 7 ,

! "#$ %& ' ( ) * + *. / 01 ! ! 23 4 5 + .

8 9 : :

>@ : >8 : >9 : > < : 9 : : 9 @ : 9 8 :

BCDEEFGH

¢ ¢ ¡ ¦ ¢ ¦ ¢ § ¦ ¦ ¦¡ ¦ ¦ ¦ ¦ § ¦ £ £ ¡ ¦ £ ¦

¨ ©

Figure 7.5: Energy spectrum measured with the Si detector.

In Figure 7.5 the energy spectrum for a triple α-source (233U, 239Pu, 238Pu) measured with a Si detector is displayed, to be compared with spectrum in Figure 7.6 taken with the GaAs detector. The GaAs detector spectra were taken at two different temperatures, namely at room temperature and at 77 K (LN2 temperature). The observed energy shift by eight channels is most likely due to slightly decreased average energy for electron-hole pair creation at higher temperature. The energy resolution (FWHM) measured for the 4824 keV α-line, the charge collection efficiency (ε) and the slope parameter from line fit for spectra in Figure 7.5 and 7.6 are given in Table 7.1. The different slope values for Si and GaAs can be explained by the difference in the ionization energy. In fact, the ionization energy ratio for the two materials and the slope ratio are roughly the same within a 10% accuracy. Chapter 7. GaAs detectors for α-particles and heavy ions

74 spectrometry

£ ¡

¥ ¥

25 ) )

$ % & ' ' () * + , - (* 2 '3

! " # ./ 0 1

2 2 ) )

26 ) )

BCDE F G H IFJ K FL MM NO

2 * ) )

BCDE F G H IFJ K FL PQ Q NO

23 ) ) ?

2 '))

;

2 ) ) )

¢¡

6 8 ) )

6 7 ) )

$ % & ' ' (3 ) 2 + , - ' 3 (5 '

! " # ./ 4 1

6 ))

6 3 ) 6 6 ) 6 5 ) 6 7 ) 2 ) ) 23 )

§ ¨©

¤ ¤£ ¡ ¤ ¡ ¤ ¥ ¡ ¦ ¦ £ ¡ ¦ ¡ ¦ ¥ ¡ ¡ ¡£ ¡ ¡ ¡

§ ¨©

Figure 7.6: Energy spectrum measured with the GaAs detector.

Table 7.1: Results for the Si and GaAs detectors obtained from the α-spectrum measurements.

FWHM [keV] ε [%] slope T [K]

Si 25.75 60.8 8.5984 300

GaAs 67.23 57.6 11.205 300

GaAs 43.02 57.6 11.03 77 7.4. Detector performance for α-particles 75

Conclusion

The present studies of a GaAs detector based on an epitaxial material showed a linear signal response in a wide alpha energy range. The good results for the energy resolution and the charge collection eﬃ- ciency indicates abilities for α- spectrometry applications. This may be well valid even for heavy ion detection, which can be a subject for further studies. The Si detectors give us a better resolution, but the main advantage of GaAs detectors is the radiation hardness which is an important for heavy ion spectrometry. To obtain radiation hardness for ion spectrometry applications, investigations should be continued with compensated GaAs. Chapter 8

X-ray Irradiation of Silicon Detectors

Introduction

Silicon detectors are used extensively in X-ray imaging, either as pri- mary detector for the X-rays or in combination with a scintillator as a detector for secondary light. Long term stability is usually desired for any application in medical imaging or non-destructive testing, and any radiation related measurable impact on the detector performance must be low or corrected for. The radiation hardness thus plays an important role for the choice of the detector material. Applications range from detectors for portal imaging to detectors for computed to- mography (CT) and mammography. The impact of ionizing radiation on Silicon detectors has been described in the literature [56, 57], however all studies have been performed at higher energies, e.g. using a 60Co gamma-ray source or hadrons. The damage depends on the radiation energy. X-rays at high enough energy, 145-173 keV accordingly to [58,59], create defects in the Silicon lattice at the Si/SiO2 interface. Data for these higher energies cannot be extrapolated to lower energies, less than 50 keV in our case. Z. Marka et al. [60] have studied the long-term time dependence of the observed leakage current in samples irradiated with low energy X-ray at 10 keV. They investigated the damage enhanced electron transport across thin oxides in Si/SiO2. The behavior of a silicon pad detector produced by Detection Tech- nology [61] has been studied in terms of current versus voltage (I-V)

76 8.1. Experimental layout 77 and capacitance versus voltage (C-V) for different amounts of delivered radiation dose. The investigation was partly initiated due to a non-standard I-V behavior of freshly delivered device. The aim was to study how this device was affected by radiation and how the pre- irradiation behavior can be restored. The process of applying reverse bias voltage over an extended period of time and thus reducing the detector current is referred to as voltage training or restoring. The X-ray source used corresponds to what is typically used in medical imaging, i.e. 10-100 keV energy. Furthermore, it was also studied how fast the detectors recovered its pre-irradiated behavior under reverse biasing after terminating the irradiation. In the investigation the read-out electronics is shielded from direct exposure to the X-rays and it can be assumed that the measured change is only due to a change in detector parameters. Nevertheless, the overall dependence on the impact of radiation on a detector system will to a large extent depend on which front-end electronics is used to read out the detector, e.g. a change in input capacitance or dark current will change the noise level and possibilities to partly compensate this with preamplifier adjustment will be crucial. A detailed description of electronics typically used in medical applications can be found in [62]. During the present measurement no amplifiers were used.

8.1 Experimental layout

Figure 8.1: Inner structure and layout of pad detector.

The pad detector layout is shown in Figure 8.1. The detector is 3.5×3.5 mm2 and is surrounded by a 50 µm thick guard-ring. The 78 Chapter 8. X-ray Irradiation of Silicon Detectors distance from the guard-ring to the detector edge is 400 µm. The size of the active area is 2.5×2.5 mm2, and its thickness is 500 µm. The estimated capacitance is 15 pF and the resistivity for the fully depleted detector is 6–10 kOhm·cm. The aluminum thickness of the conducting layers are 10 µm and the thickness of thermal silicon dioxide layer and the passivation layer are 1 µm and 0.5 µm respectively. The pad detector breakdown voltage is in the range 350±20 V.

Figure 8.2: Schematics for the pad detector I-V and C-V measurements. The detector is drawn like a diode.

The electric scheme for the I-V and C-V measurements is given in Figure 8.2. A Keithley electrometer 610C was used for the current measurement and a TTi LCR400 Precision LCR Bridge operated at 1 kHz frequency was used for the capacitance measurement. To avoid inﬂuence from the inner capacitance of the voltage source on the C bridge-meter, two resistors were connected in series with the detector. The guard-ring was grounded separately from the active area contact (the guard-ring diode is not present in electric scheme).

The detector was positioned with the X-rays incident perpendicularly to the front side surface at a distance of 40 cm from the X-ray source. The X-ray source used for the measurements, a Philips X-ray tube [63] with a W anode and 0.5 mm thick Al-ﬁlter combination, is standard equipment for crystallography. The X-ray energy spectrum was measured, according to Figure 8.3, using a commercial XR-100T- CZT gamma ray detector system from AMPTEK [64]. 8.2. Results and discussions 79

3500

3000

2500

2000

1500

Number of photons 1000

500

0 5 10 15 20 25 30 35 Energy [keV]

Figure 8.3: The X-ray spectrum obtained with a XR-100-CZT detector. The X-tube parameters were U=30 kVp and I=10 mA. The distance from the source to the detector was 10 m.

8.2 Results and discussions

Before any radiation test was performed, the initial I-V and C-V characteristics was measured. As can be seen in Figure 8.4, the ”As received” device has a higher leakage current at 200V and above compared to after biasing at 150 for 24 h. After the 24 h voltage training at 150V the sharp rise of the current between 170 and 200V is smothened out. This non-standard behavior initiated the study. 80 Chapter 8. X-ray Irradiation of Silicon Detectors

10 -3

10 -4 As received After biasing at 150 V for 24 h

10 -5

10 -6

10 -7 Current [A] Current

10 -8

10 -9

10 -10 0 50 100 150 200 250 300 350 Voltage [V]

Figure 8.4: The I-V curves before irradiation. Errors are within the size of the markers.

10 -3 1h Irr. I=10mA 10 -4 150V 24h 150V 49h 150V 72h 10 -5 150V 96h 150V 168h

10 -6

300 -7 10 250 Current [A] Current 200 150 10 -8 100

Infl.point [V] 50 Infl = 129Ln(t) - 341 -9 10 0 0 50 100 Time [h] 10 -10 0 50 100 150 200 250 300 350 Voltage [V]

Figure 8.5: I-V curves obtained after the X-ray irradiation (200 Gy) and indicated hours of voltage training. The subplot shows voltage where the sharp leakage current increase occurs as function of time after the irradiation. Errors are within the size of the point markers. 8.2. Results and discussions 81

The detector was then irradiated during 1 hour with 10 mA X- ray tube current and 30 kV voltage. The estimated absorbed dose given to the Si/SiO2-interface is 200 Gy, assuming the detectors are irradiated perpendicularly to the wafer and that the photons enter through the front-side. Immediately after the irradiation I-V and C-V measurements were performed. Subsequently a bias voltage of 150 V was applied to the detector for several days. Measurements of I-V and C-V were made during about 1 hour once a day.

0.0032 150V 168h

270V 0.0027

0.0032 150V 96h

250V 0.0027

0.0032 150V 72h

230V 0.0027

0.0032 150V 49h

220V 0.0027

0.0032 150V 24h 170V 0.0027

0.0032 1h Irr. I=10mA

0.0027

] 300 -2 250 200

[pF 0.0022

2 150 U U [V] 100 U(t) = 51ln(t) + 13.4

1/C 50 0.0017 0 0 50 100 150 200 Time [h] 0.0012 0 50 100 150 200 250 300 350 Voltage [V]

Figure 8.6: C-V curves obtained after the X-ray irradiation (200 Gy). The upper plots show the plateau region for C-V curves in the same scale for diﬀerent restoring time. The triangle on every plot indicates the start position for the capacitance instability region (see text). 82 Chapter 8. X-ray Irradiation of Silicon Detectors

In Figure 8.5 the results of the I-V measurements are presented. Immediately after irradiation (”1h Irr. I=10 mA” curve), the sharp current rise was now found between 70 V and 100 V. During the voltage training, the voltage where the of current rise occurs is displaced to higher voltages, depending on time of restoration under applied bias. The subplot in Figure 8.5 shows the voltage at which the sharp leakage current increase occurs as a function of time after irradiation. After one week of voltage training, the pre-irradiation behavior was recovered. In a similar test without voltage training i.e. applying reverse bias, the I-V characteristics from just after irradiation did not change with time during more then a week. In Figure 8.6 the detector capacitance vs. voltage is presented. The capacitance measurement becomes unstable giving values ﬂuctuating by about 10% above voltage values indicated by triangles in the ﬁgure. These voltages coincide with the voltages where the leakage current increases were found in the I-V measurements.

10 -4

10 -5

I = 4x10 -5 e-0.22t

10 -6 Current [A] Current

10 -7

R= τττ/C=0.22/1.4x10 -11 =15.7 [GOhm]

10 -8 0 10 20 30 40 Time [sec]

Figure 8.7: Current stabilization time. The lines are drawn as ﬁt result.

Four cycles of irradiation were carried out with ”low rate” param- 8.2. Results and discussions 83 eters, 10 mA and 30 kV. Identical restoring process was observed in all four cases as presented on Figure 8.5 and Figure 8.6. The voltage stimulated improving eﬀect is well demonstrated. It is important to note, that the I-V and C-V characteristics were completely restored for all four irradiation cycles. Assuming that an increase in bias voltage over the detector will result in a charge up current that will stabilize after some time at the leakage current, the time constant of the current decrease and the measured capacitance can be used to estimate the inner serial resistance of the detector. Figure 8.7 shows the current as function of time applying 150 V bias. It can be ﬁtted between 0 and 30 s by I = 4 · 10−5e−0.22t with χ2/ndf=8.77/2, where the current is in Ampere and time is in sec- onds. The estimated serial resistance equals 16 GOhm, which for this detector typically corresponds to the resistance for a 0.1 µm thickness of the silicon dioxide layer. It is noteworthy that for voltages above the sharp rise in leakage current, the current does not decrease with time.

2.5*10 -6 -8 -8 Idet = 3x10 Itube + 5x10

2*10 -6

1.5*10 -6

1*10 -6 Detector current [A] current Detector

5*10 -7

0 0 20 40 60 80 Tube current [mA]

Figure 8.8: Response of detector. The lines are drawn as ﬁt result.

The same detector sample was subsequently irradiated at higher dose rates. The detector response to the X-ray tube current was 84 Chapter 8. X-ray Irradiation of Silicon Detectors checked to ensure that the irradiation was performed in the linear region of the detector response. The result is presented in Figure 8.8. The resulting equation of the linear ﬁt with χ2/ndf=4.03/4 for detector response vs. X-ray tube current is shown in the same plot.

10 -2 2h Irr. I=50mA 150V 26h 10 -3 150V 48h 150V 120h 150V 210h -4 10 150V 302h 150V 400h 10 -5

10 -6 210

Current [A] 160 10 -7 110 -8 10 60 Infl= 51Ln(t) - 121 Infl. point [V] 10 -9 10 0 200 400 Time [h] 10 -10 0 50 100 150 200 250 300 350 Voltage [V]

Figure 8.9: I-V curves obtained after the X-ray irradiation (high rate, 2kGy) and indicated hours of voltage training. The subplot shows voltage where the sharp leakage current increase occurs as function of time after the irradiation. The errors are within the size of the point markers.

In Figure 8.9 the results of I-V measurements are presented for the detector after 2 hours of X-ray irradiation with tube current 50 mA and tube voltage 30 kVp. The estimated dose given to the Si/SiO2- interface, assuming the detector is irradiated perpendicularly to the wafer and that the photons enter through the front-side is 2 kGy for this ”high rate” case. The subplot in Figure 8.9 shows the voltage at which the sharp leakage current increase occurs as a function of time after irradiation. As can be seen from the plot, the restoring process after high irradiation rate is slower compared to the previous radiations and not completely ﬁnished after more than two weeks. The detector capacitance vs. voltage is presented in Figure 8.10 for the ”high rate” case irradiation. 8.2. Results and discussions 85

0.0032

201V 0.0027 150V 400h 0.0032

197V 0.0027 150V 302h 0.0031

171V 0.0026 150V 210h 0.0031

154V 0.0026 150V 120h 0.003

116V 0.0025 150V 48h 0.003 70V

0.0025 150V 26h 0.003 2h Irr. I=50mA

0.0025 ] 250 -2 200 150

[pF 0.002 2

U U [V] 100 50 U(t) = 46.6ln(t) - 73.6 1/C 0.0015 0 0 100 200 300 400 Time [h] 0.001 0 50 100 150 200 250 300 350 Voltage [V]

Figure 8.10: C-V curve obtained after the X-ray irradiation (high rate). The upper plots show the plateau regions for C-V curves in the same scale for diﬀerent restoring time. The triangle on every plot indicates the start position for capacitance instability region. The start position for unstable region vs. restoring time is shown in the subplot. 86 Chapter 8. X-ray Irradiation of Silicon Detectors

8.3 Summary

Applying post-irradiation bias voltage to the studied silicon pad detector, produced by Detection Technology [61], has been demonstrated to effectively restore the basic parameters, thus providing long-term stability in X-ray field. The conclusion from the two different irradiation rate and dose cases is that the degradation effect require more restoring time under bias for the ”high rate” case. A detector receiving up to 200 Gy/week should be possible to operate provided that it is constantly kept under reverse bias. Chapter 9

Summary and Outlook

9.1 Comparison with prediction from Lo- cal Charge Neutrality model

The Local Charge Neutrality (LCN) model permits the estimation of the changes in parameters for GaAs compensated by deep level centers. The model [65,66] of hadron induced displacement damage in GaAs was developed in the Siberian Physical Technical Institute (SPTI) based on data from hadron irradiation at IHEP and CERN. Local charge neutrality predicts that acceptor- and donor-like states that appear as a result of displacement damage due to radiation, should respect charge conservation resulting in a total charge equal to zero. There should thus exist an ”irradiation stable” (LCN) position of the Fermi level. From any position of Fermi level in initial material under irradiation, the Fermi level shifts to the local charge neutrality level with energy ELCN . It is also correct for the intrinsic semiconductor material. In the intrinsic semiconductors the Fermi level is located in the middle of the band gap Eg. The maximum shift of the Fermi level is equal to Eg/2 − ELCN . This shift equals to 0.17 eV for silicon and 0.08 eV for GaAs according to the values of ELCN in these materials given in Table 9.1. The tendency of increasing resistivity of GaAs under irradiation is demonstrated in Figure 9.1. The changes in resistivity of the compensated GaAs is presented by the line with ﬁlled circle data points.

87 88 Chapter 9. Summary and Outlook

C Si GaAs CdTe

300K ELCN [eV] 2.0 0.39 0.63 1.38

300K EF [eV] 2.75 0.56 0.71 0.72

0K EF [eV] 2.8 0.58 0.76 0.76

Table 9.1: Local charge neutrality (LCN) and Fermi levels in intrinsic semiconductor materials at diﬀerent temperatures [67].

Figure 9.1: Resistivity of GaAs material as a function of radiation ﬂux by fast protons, neutrons and electrons. (no is an initial density of electrons in GaAs.) Open symbols are earlier measurements by Tomsk group. Solid circles are results from one of the neutron irradiated samples. The same ﬁgure appeared in Chapter 4.)

In the resulting equation from LCN theory according to [66] the −1/3 critical value of the neutron ﬂuence Φc ' N for a given level of 9.2. Rate of Irradiation 89

doping with compensating impurity Nd is determined by:

³ ´1/2 −1/3 εkT N = 2 2 (9.1) πe Nd

For neutron fluences up to Φ < Φc the charge collection efficiency of the detector remains acceptable. For higher neutron fluence, before significant destruction of the detector, the requirement of the larger concentration of the impurity Nd is important. The Fermi level in the damaged area correspond to the ELCN and is in balance with EF in the surrounding, intact region of the crystal. At the boundary of the damaged region there appears a potential barrier of height ∆E = Ef − ELCN . Because of this barrier, a space-charge region, as described in [65] is established around the damaged region. The important parameter of this region is its spatial extent.

2 2 K0 ≈ πe Nd/εkT

Where 1/K0 is the radius of effective damage in GaAs (spatial extent of the region around the cascade in which the space charge exists), e is electron charge, k is Boltzmann constant and Nd is concentration of deep level centers. The average distance between the radiation induced energy levels (cascades) is N −1/3. The space-charge fields between the cascades do −1/3 −1 not overlap, when N > 2K0 . Calculation of 1/K0 gives an effective damage radius equal to 1.6 · 10−6cm for donor concentration 15 −3 −7 17 −3 Nd = 10 cm and 1.6 · 10 cm for Nd = 10 cm . According to LCN theory GaAs detectors should be able to with- stand hadron irradiation up to 1017 particles/cm2 for initial doping 17 −3 concentration of Nd = 10 cm [65,66]. But in experimental tests we lost functionality of the detectors at neutron flux 1.2·1015 n/cm2 and proton flux 1.5 · 1014 p/cm2. The reason is most likely due to the rate of irradiation.

9.2 Rate of Irradiation

It is supposed, that the rate of irradiation imposes huge inﬂuence on the degradation of detector parameters. With an increase in the rate of irradiation, the degradation of detector parameters occurs faster. 90 Chapter 9. Summary and Outlook

10 16 Secondary beam of neutrino facility CERN SPS Ff = 7 x10 25 xRate -0.86

20MeV neutrons 10 15 Protvino I-100 ] 2 1MeV neutrons 1GeV protons RAL ISIS Protvino Booster

10 14 Final fluenceat50% Final S/N degradation [par./cm degradation S/N

10 13 1GeV protons Protvino Booster

10 12 10 12 10 13 10 14 10 15 Rate of irradiation [par./cm 2 per hour]

Figure 9.2: Half SNR degradation dependance for diﬀerent rates of irradiation. Line is drawn as power ﬁt result.

Figure 9.2 presents the half SNR degradation dependence for diﬀerent rates of irradiation. The ﬂux of hadrons can be up to 1016 part.×cm−2 within less then 50% of SNR degradation for a rate of irradiation less then 1011 part.×cm−2 per hour.

9.3 Conclusion

In this thesis I have summarized the current understanding of the effects of radiation on GaAs detectors fabricated using the technology developed at SPTI (Tomsk, Russia). The position resolution and charge collection efficiency for ionizing particles at perpendicular incidence have been measured. The signal to noise level was measured and clustering effects investigated. The conclusion was that that the GaAs detectors would fulfill the technical requirements for the studied parameters. Another goal was to measure and optimize the radiation hardness of GaAs detectors and to estimate aging effects during the foreseen op- 9.3. Conclusion 91 eration in a hadron collider. The radiation hardness was estimated by exposing the detectors to gamma, neutron, proton and mixed beams of short-lived particles leading to a decrease of 20% in the charge- collection efficiency. There is a small increase in the noise due to the irradiation, but the S/N ratio remains above 10 in the worst case expected. No conductivity type inversion has been observed and the detectors can therefore be operated at room temperature (20oC). The data suggest changes in the production of the GaAs compared to industrial production and also in the concentration of compensation impurities and the thickness of the active layer. The data suggest that further improvement in performance could be expected if the sensitive thickness could be made almost equal to the physical thickness of the wafer material and if the leakage current density could be reduced. I believe that GaAs detectors in the future will be a versatile tool in several fields such as medical imaging, non-destructive testing and in physics experiments with high charge multiplicity. Chapter 10

Acknowledgments

This work could not have been done without the support from several people. I am very grateful for the opportunity given to me by my supervisors Prof. Per Carlson who took care of my Licentiate. Many thanks go to him for giving support and assistance. Especially thanks going to Prof. Bengt Lund-Jensen, who was the final and a very efficient supervisor, going so deep into all details and aspects of this work, thanks a lot. I would like to thank Prof. Ken- way M. Smith from Physics department of Glasgow University for his encouragement and guidance in Glasgow and CERN. Many thanks to Prof. Stanislav Khludkov and Dr. Oleg Tolbanov for providing re- sources and working environment. I would like to thank Stefan Rydströmfor very helpful discussion of experimental results for semiconductor irradiation and for proof read- ing of thesis. Finally, I am grateful to all my colleagues and friends from Protvino, Glasgow and Stockholm groups for providing such great and friendly work-places.

92 Bibliography

[1] Czochralski crystal-growth system. NASA Tech Briefs, 21:p10, 1997.

[3] Z. Naturforsch H. Welker. Vol.A, No.7:p.774, 1952.

[4] J. Blakemore. Semiconducting and Other Major Properties of Gallium Arsenide. J. Appl. Phys., Vol.53, No.10, 1982.

[5] Sammy Kayali. GaAs Material Properties. JPL Publication 96-25, December 15, 1996.

[6] S. M. Sze. Semiconductor Devices Physics and Technology. J. Wiley, New York, 1985.

[7] R. F. Pierret. Semiconductor Fundamentals. Addison-Wesley Pub- lishing, New York, 1989.

[8] R. Williams. Modern GaAs Processing Methods. Artech House, Dedham, MA, 1990.

[9] Y. Nanishi. J. Appl. Phys., Vol.22, 1983.

[10] S. Miyazawa. Appl. Phys. Lett., Vol.44, 1984.

[11] T. Honda. J. Appl. Phys., Vol.22, 1983.

[12] H. C. Casey. Diﬀusion in III-V Compound Semiconductors. Atomic Diﬀusion in Semiconductors, D. Shaw, Editor, Plenum, New York, 1973.

93 94 Bibliography

[13] S. Y. Chiang. Properties of Vacancy Defects in GaAs Single Crys- tals. J. Appl. Phys., Vol.45:p.2986, 1975.

[14] S. Markram-Ebied. Nature of EL2: The Main Native Midgap Electron Trap in VPE and Bulk GaAs. Semi-insulating III-V Ma- terials, D. Look, Editor, Shiva Publishing Ltd., England, 1984.

[15] Tolbanov O.P. et al. I-V characteristics of gallium-arsenide π–ν–n structures, doped with Fe and Cr, in strong electrical ﬁelds range. Semiconductor physics and technology., Vol.20, No.11:p.2072, 1985.

[16] Khludkov S.S. GaAs shower S-diodes. Izvestiya vuzov SSSR, Fizika, Vol. 26, No. 10:p.67 (in Russian), 1983.

[17] Gaman V.I. C-V characteristics of diode structures based on the GaAs, compensated by Mn and Fe. Izvestiya vuzov SSSR, Fizika., v.26, No.10:p.79 (in Russian), 1983.

[18] Chmill V.B. et. al. An exploration of GaAs structures for solid- state detectors. Nucl. Instr. and Meth., A326:p.310, 1993.

[19] Chmill V.B. et. al. Exploration of GaAs structures with π-ν junction for coordinate sensitive detectors. Nucl. Instr. and Meth., A340:pp.328–340, 1994.

[20] Chmill V.B. et al. Radiation hard microstrip detectors based on gallium arsenide. Nucl. Instr. and Meth., A379:pp.406–408, 1996.

[21] Chmill V.B. et al. Radiation resistanse of GaAs structures. ATLAS Inner detector note no.143, 1995.

[22] Chmill V.B. et al. Particle detector based on GaAs. Radia- tion hardness and spatial resolution. Nucl. Instr. and Meth.,A 409:pp.247–250, 1998.

[23] Chmill V.B. et al. The Result of GaAs Irradiation at Mixed Beam Neutrino Cave CERN. Fifth International Workshop on Gallium Arsenide Detectors and Related Compounds (Cividale del Friuli, Italy June17-20 1997) Nucl. Instr. and Meth., A 410:pp.54–60, 1998. Bibliography 95

[24] V.Chmill et al. Investigation of epitaxial GaAs charged particle detectors. Nucl. Instr. and Meth., A438, Number 2:pp.362–367, December 1999.

[25] V.Chmill. X-ray Irradiation of Silicon Detectors. Nucl. Instr. and Meth., A:submitted, 2006.

[26] R.E. Kremer et al. Low dislocation density GaAs grown by the bridgman technique. Journal of Material Research, Vol.5, No.7.:1468, 1996.

[27] Golovin V.M. et al. Hybrid low-noise ampliﬁer with bipolar tran- sistors. Preprint IHEP 89-231. Protvino, Russia,, 1989.

[28] Vorobiev A.P. et.al. Searching gallium arsenide structures for coordinate sensitive detectors. Preprint IHEP 91- 192. Protvino, Russia, 1991.

[29] Demeester G.B. Ramanamurthy P.V. Nucl. Instr. and Meth., A56:p.93, 1967.

[30] Akimov Yu.K. et al. Semiconductor detectors in experimental physics. Energoatomizdat, Moscow, 1989.

[31] Tove P.A. Seibt W., Sundstrom K.E. Nucl. Instr. and Meth., A113:pp.317–324, 1973.

[32] Chmill V.B. et al. Exploration of GaAs structures for solid-state detectors. Preprint IHEP 92-59, Protvino, Russia,, (Submitted to NIM). NIM A326:pp.310–312, 1993.

[33] Schoklye. Van Roosbroeck W. Problems of semiconductor physics. Moscow, (in Russian), 1958.

[34] Van Roosbroeck W. Physycs Revew, V.91:p.282, 1953.

[35] Keldysh L.V. Physics of Semiconductors. ZhTEF (in russian), Vol.34:p.1138, 1958.

[36] Bertin R. et al. Nucl. Instr. and Meth., A294:P.211, 1990.

[37] Beaumont S. et al. CERN/DRDC/90-58,, page DRDC/P13, 1990. 96 Bibliography

[38] Beaumont S. et al. Gallium arsenide microstrip detectors for charged particles. CERN–PRE/92-51, 1992.

[39] Beaumont S. et al. Gallium arsenide microstrip detectors. Nuclear Physycs, B23A:pp.417–418, 1991.

[40] Jeong M.U. et. al. Radiation Eﬀects in Semiconductors. Gordon and Breac Publish. N.Y., page p.287, 1971.

[41] Aukerman L.W. Semiconductors and Semimetals. N.Y.Acodemic Press ed. R.K. Willardson and A.C. Beer., V.4, 1968.

[42] Coates R. and Mitchell E.W.J. Adv. Physics, V.24:p.593, 1975.

[43] Lang D.V. Inst. Phys. Conf. Ser., n31:p.70, 1977.

[44] Brudnyi V.N. et. al. Solid State Comm., V.34:pp.117–119 Perga- mon Press Ltd., 1980.

[45] Krivov M.A. Brudnyi V.N. Radiations Damage in Gallium Ar- senide. Izvestia VUZ, N1(212):(in Russian), 1980.

[46] Logan R.A. Lang D.V. Radiation eﬀects in semiconductors. J. Appl. Phys., V47:p.3587, 1987.

[47] V.M. Golovin et. al. Hybrid low-noise ampliﬁer with bipolar tran- sistors. Preprint IHEP 89-231, Protvino:Russia, 1989.

[48] M.Bruzzi. GaAs Detectors and Related Compaunds. Proceed- ing of the International Workshop(1997) Nucl. Instr. and Meth., A395:pp.1–151, 1997.

[49] J.E. Eberhardt et al. . Nucl. Instr. and Meth., A395:pp.463–476, 1971.

[50] P.E.Gibbons and J.H.Howes. . IEEE Trans. Nucl. Sci., NS- 19(3):p.353, 1972.

[51] K.Hesse et al. . Nucl. Instr. and Meth., 101:p.39, 1972.

[52] Kobayashi et al. . IEEE Trans. Nucl. Sci., NS-19(3):p.324, 1972.

[53] Kobayashi et al. . IEEE Trans. Nucl. Sci., NS-20(1):p.310, 1973. Bibliography 97

[54] Shockley W. Sah C.T., Noyce R.N. . Proc. IEEE, 45:p.1228, 1957.

[55] S.M. Sze. Physics of Semiconductor Devices. Wiley, New York, 1983.

[56] M.Bruzzi. Material engineering for the development of ultra- radiation hard semiconductor detectors. Nucl. Instr. and Meth., A518:pp.336–337, 2004.

[57] R. Rando et al. A. Candelori, D. Bisello. Radiation hardness of silicon detectors for high-energy physics applications. IEEE Trans. Nucl. Sci., vol.50:pp.1121 –1128, 2003.

[58] P. Rappoport J. Loferski. . Phys. Rev. III, page pp. 32, 1958.

[59] R. Bauerlein. . Zs. Phys., 176 (4):pp. 498, 1963.

[60] Z. Marka et al. . J. Appl. Phys., Vol.93:No.4, 15 February 2003.

[61] Detection Technology web site. http://www.deetee.com.

[62] John Yorkston. Digital Radiographic Technology. Advances in Digital Radiography:, RSNA Categorical Course in Diagnostic Ra- diology Physics:pp 2336, 2003.

[63] Philips X ray system. PW1830/00/25/26/40. HT Generator In- struction Manual.

[64] Amptek Inc. company web site. http://www.amptek.com.

[65] S.S. Khludkov et. al. Proton-induced displacement damage in GaAs and radiation hardness of semiconductor detectors for high energy physics. Nucl. Instr. and Meth., A395:pp.60–64, 1997.

[66] S.S. Khludkov et. al. Radiation hardness of semiconductor detectors for high-energy physics. J. Phys. D:Appl. Phys., 29:pp.1559– 1563, 1996.

[67] C. Kittel. Introduction to solid state physics. 6th Ed., New York: John Wiley:p.185, 1986. List of Figures

1.1 Unit cube of GaAs crystal lattice...... 5 1.2 Energy band diagram for GaAs...... 6 1.3 Energy band structure of GaAs...... 7 1.4 Energy band diagram of GaAs with impurities. . . . . 10

2.1 Distribution of shallow donors, acceptors, deep acceptors. Energy diagram ...... 15 2.2 Energy zone diagrams...... 17

3.1 Forward and reverse I-V branch of the samples. . . . . 20 3.2 Spectra of the signals from p-π-ν-n structure registered in the charge sensitive channel for different bias voltage. 22 3.3 C–V of the p-π-ν-n sample (at the frequency 1 MHz). . 23 3.4 Spectrum of the signals from the n+-π-ν-n structure, exposed to 2 MeV β–particles ...... 23 3.5 Spectrum of the signals from the n+-π-ν-n sample exposed to radioactive 241Am α–particles, registered in the charge sensitive channel...... 24 3.6 Estimated dependence of the charge collection time taking into account the plasma effect for the release of 50000 electrons...... 27 3.7 Estimated dependence of the charge collection time taking into account the plasma effect at the integrated potential difference 50 mV as function of the numbers of injected electron...... 27 3.8 The charge spectra of the signals from the p-π-ν-n sample exposed to monoenergetic 0.6 MeV β–particles at various bias voltages...... 28

98 List of Figures 99

3.9 The charge spectra of the signals from the p-π-ν-n sample irradiated by 2 MeV monoenergetic β–particles at different bias voltages...... 29 3.10 The charge spectrum of the signals from p-π-ν-n samples irradiated with 2 MeV by β–particles, registered in the current sensitive channel at 100 V bias...... 30 3.11 The temperature dependence of the noise in GaAs samples. 32 3.12 Temperature dependence of the signal peak position in GaAs samples...... 33 3.13 The signal-to-noise ratio vs. temperature in GaAs samples. 33 3.14 The noise of GaAs samples vs. absorbed dose of 137Cs γ-rays for two different samples...... 34 3.15 The signal peak position vs. absorbed dose of 137Cs γ- rays for two different samples...... 35 3.16 The SNR vs. absorbed dose of 137Cs γ-rays for two different samples...... 35

4.1 Resistivity of GaAs material as a function of radiation ﬂux by fast protons, neutrons and electrons...... 39 4.2 Forward I–V characteristic as a function of neutron irradiation...... 40 4.3 Reverse I–V characteristic as a function of neutron irradiation...... 41 4.4 Resistivity of the π–layer vs. neutron irradiation (left). Photovoltaic e.m.f. vs. neutron irradiation (right). Flu- ence scale from 1011 to 1016 n/cm2, ρ-scale from 104 to 9 −3 10 Ohm·cm and Uo from 10 to 1 V...... 42 4.5 C–V variations as a function of bias for diﬀerent neutron irradiation...... 43 4.6 Noise (RMS) vs. neutron irradiation...... 44 4.7 Charge collection (spectrum peak position for most probable energy loss) vs. neutron irradiation...... 45 4.8 Signal to noise ratio vs. neutron irradiation...... 45

5.1 Signal response vs. muon ﬂuence...... 50 5.2 Signal response vs. bias voltage...... 51 5.3 Experimental layout...... 52 5.4 Raw data from one of the detectors and the BCT counter. 53 100 List of Figures

5.5 GaAs (signal amplitude) response vs. mixed beam ﬂuence. 53 5.6 The degradation of signal response for bias value 23 V. 54 5.7 The degradation of signal response for bias value 46 V. 55 5.8 The degradation of signal response for bias value 76 V. 55

6.1 Intrinsic structure (Au)n+-π-ν-n type...... 58 6.2 The forward branch of I–V characteristics...... 59 6.3 The reverse branch of I–V characteristics...... 59 6.4 Experimental set-up...... 60 6.5 Spectrum of the signals for a single strip (pitch 150 µm). 61 6.6 ”Shadow” of the ﬁbre (pitch 150 µm)...... 62 6.7 Spectrum of the signals for a single strip (pitch 50 µm). 62 6.8 ”Shadow” of the ﬁbre (pitch 50 µm)...... 63 6.9 Spectrum of the signals for a single strip (pitch 50 µm) for 5.9 GeV/c π−...... 63 6.10 Mean cluster of events (pitch 50 µm)...... 65 6.11 The collected charge versus proton irradiation...... 66

7.1 Inner structure of the GaAs detectors studied here. . . 69 7.2 C-V characteristic...... 70 7.3 I-V characteristics...... 71 7.4 Energy calibration using α-sources...... 72 7.5 Energy spectrum measured with the Si detector. . . . . 73 7.6 Energy spectrum measured with the GaAs detector. . . 74

8.1 Inner structure and layout of pad detector...... 77 8.2 Schematics for the pad detector I-V and C-V measurements ...... 78 8.3 The X-ray spectrum obtained with a XR-100-CZT detector...... 79 8.4 The I-V curves before irradiation...... 80 8.5 I-V curves obtained after the X-ray irradiation (200 Gy) 80 8.6 C-V curves obtained after the X-ray irradiation (200 Gy) 81 8.7 Current stabilization time ...... 82 8.8 Response of detector ...... 83 8.9 I-V curves obtained after the X-ray irradiation (high rate, 2kGy) ...... 84 8.10 C-V curve obtained after the X-ray irradiation (high rate) 85 List of Figures 101

9.1 Resistivity GaAs samples as a function of radiation ﬂux by fast protons, neutrons and electrons...... 88 9.2 Half SNR degradation dependance for diﬀerent rates of irradiation...... 90 List of Tables

1.1 Room-temperature properties of GaAs and Si...... 6

2.1 Type of tested GaAs detectors ...... 18

4.1 Parameters of the I-100 proton beam...... 38 4.2 Main characteristics of the irradiation volume...... 38

5.1 Summary results of hadron beam irradiation. The lines before the last is preceding measurements and was done on other facilities...... 49

6.1 Cluster probability for 50 µm and 150 µm pitch detectors. 64 6.2 Most probable collected charge from detector and SNR for GaAs detectors after neutron irradiation at RAL. . 66

7.1 Results for the Si and GaAs detectors obtained from the α-spectrum measurements...... 74

9.1 Local charge neutrality (LCN) and Fermi levels in intrinsic semiconductor materials...... 88

102