videodiagnosztika-rendszerek Budapesti Műszaki és Gazdaságtudományi Egyetem

Transcription

1 Fúziós berendezésekben használt videodiagnosztika-rendszerek sugárkárosodásának vizsgálata PhD értekezés Szerző: Náfrádi Gábor Témavezető: Dr. Pór Gábor Budapesti Műszaki és Gazdaságtudományi Egyetem Budapest, 2016

2 Investigation on the radiation damage of video diagnostic systems applied in fusion devices PhD thesis Author: Gábor Náfrádi Supervisor: Dr. Gábor Pór Budapest University of Technology and Economics 2016, Budapest

3 Contents 1 Introduction 4 2 Overview of the research field Basic semiconductor behavior p-n junction Photodiode Leakage current Basic transistor operation Radiation damage processes Displacement damage kinetics Single Event Effects Total Ionizing Dose effect (TID) Neutron and gamma radiation fields Neutron field characterization Photon field characterization NIEL scaling hypothesis Connection with DPA Parameters after irradiation Leakage current due to irradiation Semiconductor based video devices CCD based devices CMOS based devices Connecting electronics End-user radiation damage investigations Computational methods Monte Carlo calculations MCNP CAD-Monte Carlo conversion MCAM

4 CONTENTS 4 Common experimental methods Magnetically confined fusion Diagnostic systems for plasma diagnostics EDICAM system on W7-X BES system on the KSTAR tokamak BES systems on the EAST tokamak Connecting optics and their effects Radiation environment of fusion devices The neutron spectrum and fluxes The photon spectrum and dose rates Radiation of fission reactors Dose rate and flux measurements Neutron flux measurement Gamma dose rate measurement Neutron and gamma shielding Investigations of the EDICAM camera Introduction MCNP model Gamma dose and photon spectrum Irradiation Test irradiations EDICAM irradiation Evaluation of the frames I D (t, τ) normalization by the dose rate I D (t, τ) normalization by the dose Results and discussion Total Dose Effect Dose Rate Effect Conclusions Cameras of the BES system on KSTAR Introduction MCNP Calculations MCNP model Results of MCNP calculations White pixels Measurements at BME NTI Gamma-ray irradiation

5 CONTENTS Measurements at thermal equilibrium Evaluation of the recorded frames Radiation damage prevention Recent neutron shielding Results and Conclusions Shielding considerations Introduction Monte Carlo model Results Discussion Relationship with radiation weight factors Radiation shielding of the BES system of the EAST tokamak - an example for application D model with EAST relevant neutron spectrum Circular source definition Conclusions Summary 93 9 Thesis points Acknowledgements Appendix p-n junction calculations Photodiode details Other parameters after irradiation Minority carrier lifetime Change of majority carrier density Effective doping concentration Messenger-Spratt equation Further comments on MCAM

6 Chapter 1 Introduction Magnetic-confinement fusion is a great candidate to solve energy needs of mankind in the near future. This technology promises self-sustaining electric energy production without CO 2 emission or significant nuclear waste production. The fuel of the reaction is deuterium (D) and tritium (T ), two different isotopes of hydrogen. Deuterium as a natural and non-radioactive isotope is present in large quantities all around the world, however, the tritium as a radioactive isotope is to be produced due its year half-life. In magnetically confined fusion devices the recent development aims to produce the tritium from lithium by bombardment of neutrons which are born in D D and D T fusion reactions. Nowadays the D D reaction is widespread. Only in the Joint European Torus (JET) were D T campaigns executed, the last in 2003, however, a new D T campaign is planned for The presence of tritium and energetic neutrons implies that the future fusion facilities will be undoubtedly nuclear installations. The nuclear aspects of the fusion technology become more and more important as the power of the fusion installations grow to reach the power plant scales. One of major problems related to the harsh nuclear environment, is that some of the recently applied special diagnostics systems are sensitive to the neutron or energetic photon radiation. Diagnostic systems can be very diverse. Their purpose is to reveal the behavior of the magnetically confined energetic particles called plasma particles in a fine time and spatial scale. Some of the plasma diagnostics are based on very robust technology, good examples are the magnetic detectors or probes. But there are other detectors based on semiconductor technology which can be affected by the radiation. Semiconductor based diagnostics are widespread due to their high operation speed and spatial resolution. In plasma physics measurements different types of semiconductor detectors are used but most of them are some kind of cameras. The plasma diagnostic cameras detect different radiations emitted in different spectral ranges from the plasma. The examined spectrum of electromagnetic radiation starts from the infrared wavelengths and goes up to the energetic gamma ray region. Different radiations require different camera properties, camera components. In this work I will 4

7 focus on the behavior of cameras which work in the visible light spectral range. The most commonly used cameras are commercial CCD and CMOS based cameras but other exotic cameras like different photo diode arrays or Avalanche Photo Diode (APD) cameras are present, too. The commercial camera technology does not aim to develop radiation hard systems because there is a tradeoff between the time and spatial resolution of the cameras and their radiation tolerance. Another problem is that radiation hard systems are significantly more expensive. To protect the cameras around fusion devices there are few existing technology such as radiation hardening, distance protection from radiation sources or radiation shielding. This thesis will concentrate on the behavior of the commercial visible range cameras and their radiation protection including basic camera operations, radiation load calculations, irradiation experiments, shielding concepts. In the following chapters I will discuss in details the operation principles of semiconductor cameras, particularly to their radiation damage caused by neutron and gamma radiation. Detailed computational methods will be presented which are capable to estimate the neutron and gamma radiation fields around fusion devices thus the radiation loads of the cameras can be calculated. In the last chapters experimental methods will be introduced including camera operations, irradiation experiments, shielding considerations. In all concerned topics I will highlight my own work and my scientific results. 5

8 Chapter 2 Overview of the research field Both basic nuclear processes (the fission and the fusion) and other nuclear reactions are nowadays extensively researched, just think about the fourth generation reactors [1, 2], the development of ITER [3], the discovery of the Higgs boson [4, 5] or the rapidly developing space technologies [6, 7]. All of the aforementioned areas face the problems originating from the high flux and high energy particle bombardment of the applied materials. The radiation damage of the structural materials made it necessary to develop new type of steels and alloys [8], moreover the detector and diagnostic systems must be able to operate in these harsh environments. Some of these research areas put great emphasis to develop radiation tolerant detectors. Different organizations and collaborations like ROSE-RD48 1, RD50 2, RADECS 3, NASA, ESA developed new theoretical and experimental techniques to improve the radiation tolerance of electronic devices. In this chapter I will introduce the theoretical background of the operation of basic semiconductor based electronic devices, then the terminology used for the description of radiation fields will be introduced. Then the radiation damage processes in question will be presented in semiconductor based electronic components, focusing on the silicon based technologies. The reason for the choice of silicon is that silicon is the most applied semiconductor, its properties are the best known and therefore silicon is used in the most commercial technologies. Finally I will introduce the basics of CMOS and CCD camera technologies. 2.1 Basic semiconductor behavior Semiconductors are crystalline materials like metals and insulators. Their electrical behavior can be characterized by the well-known energy band structure model. This 1 Research and development On Silicon for future Experiments (ROSE) 2 Radiation hard semiconductor devices for very high luminosity colliders (RD50) 3 RADiation Effects on Components & Systems (RADECS) 6

9 2.2. P-N JUNCTION model contains a valence band and a conduction band which contain energetically allowed electron states. Valence band is the most upper band which is fully occupied by electrons when the semiconductor is in ground state. The most lower fully depleted band is the conduction band. Between these two bands an energetically forbidden band, called gap E g is present if an insulator or a semiconductor material is considered. The gap of insulator materials are larger than 3-8 ev and for semiconductors ev. The probability of the excitation of an electron from the valence band to the conduction band is proportional to p ex exp( E g /2k B T ). Comparing these gap sizes to the k B T thermal energy which is ev at room temperature and considering E g = 8 ev, the p ex But for E g = 1.11 ev the probability of a thermal excitation is p ex which can produce an electron density approximately cm 3 in the conduction band. The size of the gap essentially influences on the conduction properties of semiconductors. The conduction characteristics can be most easily changed by doping the semiconductor or by changing its temperature. More detailed description of these topics can be found in the literature [9]. Two types of doping are possible, acceptor atom doping and donor atom doping. The donor doped materials called n-type, the acceptor doped materials called p-type semiconductors. The n-type material contains dopant atoms which have more than 4 valence electrons thus more electrons will be present than in a pure intrinsic Si material. The p-type material contains dopant atoms which have less than 4 valence electrons thus less electrons will be present than in an intrinsic material. The majority charge carriers are electrons in n-type materials and holes in p-type materials. The high degree of doping used to be noted as a plus sign like p + or n + or even higher doping concentration like p ++ or n p-n junction The simplest semiconductor device is called p-n junction which is nothing more than a p-type and n-type material in contact. After the contact is established thermal equilibrium will be present due to the recombination of the mobile electrons and holes in the vicinity of the contact surface. A potential difference, called build in potential V bi arises as the recombination of electrons and holes will ionize the donor and acceptor atoms around the junction and a depleted region will be formed. If a p + -n junction is made, the depleted region will penetrate much more in n side than in the p + side. Let consider a p + n junction with a finite thickness of t diode. The concentration of the dopants are homogeneously dispersed on both side of the junction and the charge concentration is constant in the depleted region with a thickness of t depleted. The solution of the Poisson equation (2.1) will give us the electrical field strength and the potential in 7

10 2.2. P-N JUNCTION the depleted region. d 2 V (x) = eρ(x), (2.1) dx 2 ɛ where ρ(x) denotes the charge density in the depleted region, e denotes the elementary charge, and ɛ is the permittivity which is the product of the silicon relative permittivity (ɛ r ) and the vacuum permittivity (ɛ 0 ). If the applied biasing potential is zero (V (n) = 0) this leads to a parabolic potential distribution: V (x) = en d 2ɛ (x l n) 2. (2.2) If V (x = 0) = V bi V (n) (where V (n) denotes the biasing potential on the n side of the junction) than thickness of the depleted region can be easily calculated at the x = 0 position: 2ɛ t depleted = l n = (V bi + V (n)). (2.3) en d The p n junction could be forward biased or reverse biased depending on V (n). The applied voltage which ensures that the p n junction is totally depleted t depleted = t diode called V depletion. More detailed calculations can be found at the appendix (11.1) Photodiode Most of the photodiodes are based on a simple p n junction or in some cases they are based on P IN structures where I refers to intrinsic. In photodiodes during normal operation reversed biasing is used. Let consider a situation when a non-zero biasing is applied. Previously, when the Poisson equation (2.1) was solved the p n junction was considered as static. In fact this static behavior is the consequence of two currents with opposite direction. One of these currents is the diffusion current which appears due to the concentration differences of charge carriers, and the another is the drift current due to the presence of potential difference through the p n junction. The ideal diode characteristic (Figure 2.1) can be described by the following equation: I(V ) = I 0 (e ev/kbt 1), (2.4) where k B denotes the Boltzmann constant, T denotes the temperature of the diode, V the applied voltage, e the elementary charge and I 0 denotes the reverse current, dark current or leakage current. External charge injection can cause a linear shift in the current. The charge injection can be photo illumination caused electron-hole generation in the depleted region. The electron-hole pairs will be separated and drift apart due to the presence of the electric field, and a photo current will occur. Only those photons can produce electronhole pairs which energy E photon is at least equal or greater than the gap E g between the valence and conduction band (E photon E g ). This photo current is the output signal of a photodiode. More detailed description can be found in the literature [9] and in appendix (11.2). 8

11 2.2. P-N JUNCTION C u rre n t (A ) 0, , , , , , , D a rk c u rre n t Illu m in a tio n (1 x ) Illu m in a tio n (2 x ) Illu m in a tio n (3 x ) R e v e rs e b ia s F o rw a rd b ia s -0, , , In c re a s in g p h o to n flu x -0, ,0-0,8-0,6-0,4-0,2 0,0 V o lta g e (V ) Figure 2.1: Demonstration of the I-V characteristics of an ideal photo diode, at 300 K and I 0 = A in the case of dark current. 9

12 2.3. BASIC TRANSISTOR OPERATION Leakage current The leakage current of ideal diode contains only the diffusion related currents, however, in real application contaminations, impurities and fabrication related defects or surface states contribute to leakage current as well. The radiation induced leakage currents can be divided into two main components: the surface related leakage current and the bulk generation current. The dominant contributor depends on the kind of radiation and the architecture of the diode. The surface current can be dominant in charged particle irradiations when currents can flow along the ionization channel of the particles. The bulk generation current is mostly dominant in neutron irradiation however, the gamma radiation can also produce bulk and surface related leakage current. Primary charged particles (electrons, protons, ions) are not present in our experiments therefore in the following their effects will be neglected. The defects, which are located in the depleted region contribute to the bulk leakage current, but this work does not aim to reveal the physics of microscopic defects rather, the changes of macroscopic detector and camera behavior due to irradiation. The details of the radiation induced leakage current will be presented in Section As it can be seen the leakage current is a complex quantity which has many origins, here one important property of it should be highlighted, its strong temperature dependence of the bulk leakage current [10]: I(T ref ) = I(T )R(T ) where R(T ) = ( T ref T )2 exp( Eg 1 2k B [ T ref 1 ]), (2.5) T where T ref is an arbitrary reference temperature. More detailed description can be found in the literature [10, 11]. 2.3 Basic transistor operation Principally two basic types of transistors exist. The first type is called bipolar transistor, its operation is based on both majority and minority charge carriers. The second type is the Field Effect Transistor (FET), its operation is only based on the majority charge carriers. First the bipolar transistor will be introduced. Bipolar transistor consists three layers of semiconductors; two types are distinguishable, the n p n and the p n p types. All these three layers have an individual ohmic electrode contacts which names are: Emitter electrode, Base electrode and Collector electrode. To understand the operation of bipolar transistors they can be imagined like two bipolar diodes in one sandwich structure. Let consider the p n p structure shown on Figure 2.2. If forward bias is applied to the emitter-base diode the current of the majority holes will flow to the direction of the n type layer. In the n layer the holes will be the minority carriers. If reverse bias is applied to the base-collector diode a depleted region will appear but the increased number of the holes 10

13 2.3. BASIC TRANSISTOR OPERATION (minority carriers) can flow forward to the direction of the collector. The thickness of the inner n type region should be smaller than the diffusion length of the holes. According to Kirchhoff s first law it can be written in the following form: I E = I B + I C. (2.6) Another important parameter of a bipolar transistor is the base current gain which can be denoted by α: α = I C /I E (2.7) A similar parameter is called emitter current gain and denoted by β: β = I C /I B (2.8) Holes Electrons Donor atoms Aceptor atoms p Depleted region n p E Current of holes C I E U EB B I B U BC I C Figure 2.2: Schematical view of a bipolar p-n-p transistor As it was already mentioned the emitter-base diode is forward biased which means the resistance is much smaller in this region than in the base-collector region while it is reverse biased. If the I B is small it can be obtained that I E I C and considering the definition of the electric power (P ) is P = I 2 R it is obvious that the transistor can be used as an amplifier. The operation of FETs is fundamentally different. FETs are unipolar devices which are also contains three different regions or electrodes: the Source, the Drain and the Gate. The current which can flow between the source and drain is the function of the potential 11

14 2.3. BASIC TRANSISTOR OPERATION which is applied to the gate electrode. For the better understanding of the operation of FETs see Figure 2.3 which shows the simplest version of FET called Junction Field Effect Transistor (JFET). Until no bias is switched to the gate terminals the n layer will affect like a resistor, when a revers bias is switched to the gate electrodes a depleted region will appear in the n type material. The electrons as majority charge carriers can move only between the depleted regions in a narrow channel. The thickness of the depleted region is the function of the gate potential. If the thickness of the depleted region is so large that reaches the thickness of the conducting channel the source-drain current decreases to zero. The symmetric n + p n + type of FET can also be designed. The current gain can be described also with equation (2.7). More detailed description of the transistors can be found in the literature [9]. Depleated regions Holes Electrons G Donor atoms Aceptor atoms V G p + S n Current of electrons D p + V SD G Figure 2.3: Schematical view of an unipolar JFET transistor. An important type of the FETs is the Metal Oxide Semiconductor (MOS)-transistor technology or MOSFET. The structure of MOSFETs are different from JFETs, see Figure 2.4 for the details. The operation of MOSFETs are based on the minority charge carriers as it will be seen in the following. Applying a sufficient potential to the gate electrode can induce charge separation in the p-type substrate material therefore the majority charge carriers can accumulate there or a depleted region can occur or the minority carriers can create an inversion layer. Inversion layer means the minority electrons in the p-type substrate accumulate close to the insulator surface, for this a sufficiently high positive potential (threshold potential) is required on the gate. If a voltage is applied between the source and drain terminals, electron current can flow. The thickness of the 12

15 2.4. RADIATION DAMAGE PROCESSES channel depends on the distance of the source and drain terminals and on the source-drain potential. The source-drain current will increase with the source-drain voltage linearly up to the point, when the inversion channel will not reach the drain anymore, then the current saturates. More details about the operation modes are also available in the literature [11]. Two subtypes can be constructed, NMOS and PMOS transistors where N and P refers to electrons and holes in the inversion channel respectively. Source Gate Drain n n SiO 2 p type substrate n type channel Figure 2.4: Schematical view of an unipolar MOSFET. At this point the reader has arrived at one of the most important type of devices, namely the Complementary Metal Oxide Semiconductor (CMOS) structures. CMOS do not only refers to a type of a device but also to a fabrication process, but first let see details of a CMOS structure. Figure 2.5 shows a schematic view of a Negative-AND (NAND) gate. In CMOS, complementary refers to the structure of a CMOS device which uses complementary and symmetrically an NMOS and a PMOS structure but in general it refers to a structure which contains a substrate and a different type of diffused well in it which is produced by lithography techniques. 2.4 Radiation damage processes Here the possible and typical microscopic radiation damage effects will be presented which undergo during irradiations. These defects will lead to change in performance of electrical components and consequently in larger electrical circuits. 13

16 2.4. RADIATION DAMAGE PROCESSES NMOS PMOS Base Gate Source Drain Gate Source Drain Base p + n + n + p + p + n + p-substrate n-well Figure 2.5: Schematical view of an unipolar CMOS NAND gate. The Base and the Source terminals are connected as well as the two gates and the two drains Displacement damage kinetics When energetic particle radiation reaches the silicon bulk the particles can displace a so called primary knocked atom (PKA) from its former lattice position and form an interstitial and a leftover vacancy in the lattice. Both the interstitial and the vacancy can move in the lattice and can form together with impurity atoms a point defect. The PKA can displace further atoms along its path if it has sufficiently high energy, and at the end of the path a highly disordered region a so called cluster defects can be formed [12]. Point defects and cluster defects disrupt the periodic structure of the Si bulk lattice and form new electrically active states in the band structure thus changing electrical behavior of the whole material. The maximal recoil energy E R,max of a PKA which displaced by a particle with a mass m p and kinetic energy E P can be calculated in the non-relativistic elastic scattering approach as following [13]: m p m Si E R,max = 4E p (m p + m Si ). (2.9) 2 For cluster formation the minimal required recoil energy is about 5 kev [10] which means that for example a lithium atom with 0.84 MeV or a helium atom with 1.47 MeV energies can easily produce cluster defects, however, the minimal required neutron energy is also just about 35 kev. Electrons can also produce point defects if their energy is larger than 255 kev. Electrons with such energies can be produced for example via Compton scattering. For electron induced cluster defect in silicon the minimal required energy is 14

17 2.4. RADIATION DAMAGE PROCESSES more than 8 MeV. The point and cluster defects can recombine and disappear, these recombinations are highly depended on the temperature. The guided heat treatment which can heal the material from the defects is called annealing. In this work annealing is not considered because the typical fast annealing processes undergo over 60 o C, but for commercial cameras like ours, the maximum operational and storage temperatures are smaller. The properties of defects are widely studied in literature [14, 15] Single Event Effects Single Event Effects (SEE) are related to a single energetic particle impact, which was originally a charged particle or it will produce energetic charged particle, which can penetrate though different layers of an electronic device. Along the path of the particle an ionization channel will be established where highly mobile charge carriers appear. Through this ionization channel, current can flow which in unfortunate cases can cause destructive short circuit, but not every SEE is destructive or long lasting. Various types of SEEs are identified: Single Event Upsets (SEU) Multiple Bit Upset (MBU) Single Event Functional Interrupts (SEFI) Single Hard Error (SHE) Single Event Transient (SET) Single Event Latch-up (SEL) Single Event Snapback (SESB) Single Event induced Burnout (SEB) Single Event Gate Rupture (SEGR) Single event Hard Errors (SHE) Detailed description of SEEs can be found in literature [16]. 15

18 2.4. RADIATION DAMAGE PROCESSES Total Ionizing Dose effect (TID) The interaction of photons and material can be divided into three groups. In the lowest energies the dominant form of interaction is the photoelectric effect, which denotes the process when the photon interacts mainly with an inner shell electron. This electron will be knocked out of the bond if the photon has larger energy than the bonding energy of the electron. The photon will disappear in this process, and the knocked out electron will have a motional energy M = hν W bind, where h is the Planck constant, ν is the frequency of the photon and W bind denotes the binding energy of the electron. For higher energy photons the binding energy of the electron can be neglected, a more probably interaction form called Compton scattering occurs. During Compton scattering the incident photon collides with an electron which is considered to be free (in fact is relatively weakly bounded to an atom). Some part of the energy of the photon is transferred to the electron, which can be kicked out of the atom thus ionizing it. The photon losses some energy and the direction of its propagation also changes. The third interaction is called pair production. Above a threshold energy a photon can interact with the Coulomb field of the electrons or the nuclei. During pair production the photon disappears, an electron and a positron appears. The minimal energy needed for pair production is the double of the rest mast-energy equivalent of the electron E = 2mc 2 = 1022 kev. From all of the aforementioned three interactions energetic electrons can born and these electrons can easily ionize the materials, in this particular case the silicon or silicon dioxide. Ionization electronically means that electron-hole pairs will appear in the material. Four processes can undergo in the material: generation of electron-hole pairs, recombination of a fraction of the electron-hole pairs, transport of the free carriers in the oxide layers, formation of trapped charge in hole trapping sites or charge trapping in interface traps. The main ionization induced change in bulk silicon is the increase of conductivity through production of excess charge carriers. In insulator layers the trapping of charges can produce electric field and they have also chemical effects. In insulators as the electrons have a larger mobility the positive holes can trap in and produce a positive space charge region. The change of the positive charge trapped in the oxide N ot can be described with the following equation: N ot = Dκ g f y f ot t ox, (2.10) where D denotes the total ionizing dose, κ g is the electron-hole pair density per unit dose, f y is the fractional charge yield (the number of not recombined electron-hole pairs divided by the total number of generated electron-hole pairs), f ot is the hole trapping efficiency and t ox is the thickness of the oxide layer [17]. If the escaping electrons are captured in the contiguous material a negative space charge will arise. The potential difference between 16

19 2.5. NEUTRON AND GAMMA RADIATION FIELDS the space charged regions will give a rise to an electric field, a current can start to flow across the interface by satisfying Ohm s law [10, 12]. The insulator layers in Field Effect based devices like FETs play an important role as it was seen in 2.3. If such a space charge region can build up in the gate oxide layer it will affect the thickness of the channel, and the opening (threshold) voltage, while the maximal charge which can be appear in an insulator layer is a function of the thickness of the insulator (see equation 2.10). The negative shift in the threshold voltage (V t ) (both in NMOS and PMOS structures) is proportional to the square of the oxide thickness [17]: V t (N ot ) t 2 ox, (2.11) In modern devices the thickness of oxides is 2 nm or thinner. As the thickness of these oxide layers are decreasing this disturbing effect of the space charge is also decreasing. However, there is another insulator layer called Shallow Trench Isolation (STI) oxide which separates the different transistors on an integrated circuit. The thickness of this oxide layer can reach several 100 nm which means these oxides are still sensitive to TID effects. The charge trapping changes the source-drain leakage current in NMOS devices and also changes the switching threshold voltages in PMOS and NMOS transistors [18, 19]. Other effects due to interface traps in MOS devices for example 1/f noise increases or increase in the subthreshold swing can be found more detailed in the literature [17]. 2.5 Neutron and gamma radiation fields In this section those physical quantities will be presented which are the most important to characterize a radiation field which are present around nuclear devices. Every radiation can be characterized sufficiently with the distribution of the particles in time, space, velocity. Velocity is a vector quantity which can be replaced by the energy and direction of the propagation Neutron field characterization The basic parameter of the neutron field is the neutron flux. Let consider a position vector r which is surrounded by an infinitesimally small volume dv. Let n denote the density of neutrons in this dv volume in a certain time t. These neutrons are flying within the elementary solid angle d Ω around the direction vector Ω and within the energy E and E + de. The number of neutrons at a certain time in such a phase space volume is the following n( r, E, Ω, t)dv ded Ω. The neutron flux is the product of the neutron density and the velocity of the neutrons: φ( r, E, Ω, t) = n( r, E, Ω, t)v. (2.12) 17

20 2.5. NEUTRON AND GAMMA RADIATION FIELDS If one integrates overall the solid angle gets the differential flux: φ( r, E, t) = φ( r, E, Ω, t)d Ω. (2.13) 4π In typical neutron physics calculations a very important parameter is the thermal neutron flux which is the integral of the differential flux from zero energy to the thermal upper threshold energy E th : φ th ( r, t) = Eth 0 φ( r, E, t)de. (2.14) Another important parameter of the neutron field is the neutron fluence Φ which is the integral of the neutron flux in time. From any of the above mentioned flux definitions a fluence can be calculated. For example the thermal neutron fluence Φ th in a certain position is: Φ th ( r) = Tend T start φ th ( r, t)dt, (2.15) where T start and T end practically denotes the start and the end of the irradiation time. Knowing the flux in an irradiation position the reaction rate R can be calculated as follows: R( r, t) = E2 E 1 φ( r, E, t)n target σ(e)de, (2.16) where E 1 and E 2 denotes the borders of the considered energy bin, N target denotes the number of target nuclides of the sample with a certain σ(e) cross section which belongs to the investigated reaction caused by the neutron irradiation. Let introduce here the macroscopic cross section Σ(E) which is the product of N target and σ(e). The total macroscopic cross section Σ total is calculated using the total microscopic cross section σ total. The illustrative meaning of the total macroscopic cross section will be the reaction probability over unit path of the neutrons (in homogenous material). Let consider a dv volume and calculate here the reaction rate R. The reader can think about the average neutron flux φ average as the total length of the neutron paths in the dv volume. This interpretation will be useful when the flux is calculated by a Monte Carlo method based computer code for example by MCNP Photon field characterization Photon fields can be also characterized by the photon flux φ photon, but its definition is a bit different from the definition described above in equation (2.12), hence the photons travel with the speed of light. In the literature often the photon spectrum is the most important parameter. In this work the photon spectrum will refer to the differential photon flux like it is defined for particles in equation (2.13). Here we note that in radiative transfer theory the quantity which analogue to the neutron flux multiplied by the particle energy is called spectral radiance. 18

21 2.5. NEUTRON AND GAMMA RADIATION FIELDS Another important way to characterize the photon radiation is the dosimetrical approach which interested in the energy transfer from the photons to a material. In dosimetry the basic quantity is called absorbed dose D which is the total energy absorbed in unit mass of material irradiated by an ionizing radiation, like photon radiation. The unit of absorbed dose is Gray [ J =Gy]. In human dosimetry other quantities are also defined kg like equivalent dose of a tissue H T or effective dose H eff. The equivalent dose is calculated from the absorbed dose of a human tissue multiplying it with a radiation weight factor w r. The weight factor is a dimension less factor therefore the unit of equivalent dose is still [ J ] but in this case it is called Sievert [Sv]. For photons w kg r is equal to 1 at any energy. The effective dose is also a human dose term which can be calculated as follows H eff = T w T H T, (2.17) where w T is the tissue weighting factor and T w T = 1. Using dosimeters it is important to know which of dosimetrical quantities are measured and displayed by the device. For electronic component irradiation the easiest case is that when the dosimeter displays the absorbed dose. There are passive and active dosimeters, the active dosimeters measure the dose rate D(t) while the passive dosimeters measure the total dose: D = D(t)dt (2.18) during their exposure. For time dependent exposures the usage of dose rate meters is essential NIEL scaling hypothesis When a radiation reaches a material, the particles of the radiation will transfer energy to the material. This energy transfer has two cumulated effects: the Total Ionizing Dose (TID) effects and the Non Ionizing Energy Loss (NIEL) related effects. This section introduces the NIEL concept and the restrictions of its application. A strong demand was raised, as the irradiation facilities were widespread and different types of particle irradiations with different energy spectra were produced, to compare these irradiation facilities in the terms of displacement damage. This demand encouraged the development of a technique which scales different type of radiations to the same base, this is the NIEL scaling. The basic quantity of the NIEL scaling is the neutron hardness factor κ( r) in a certain irradiation position: D(E)φ( r, E)dE 0 κ( r) = D(E = 1 MeV, n) φ( r, E)dE, (2.19) 0 where D(E) is the displacement damage cross section or displacement damage function, φ( r, E) is the fluence of the radiation in the irradiation position. In the denominator 19

22 2.5. NEUTRON AND GAMMA RADIATION FIELDS D(E = 1 M ev, n) denotes the displacement function which belongs to neutrons with 1 MeV energy. According to the ASTM E (2009) standard [20], the value of D(E = 1 MeV, n) = 95 MeVmb. This way the 1 MeV equivalent neutron fluence Φ 1 MeV,eq ( r) in a certain irradiation position can be defined: Φ 1 MeV,eq ( r) = κ( r) 0 φ( r, E)dE = κ( r)φ( r). (2.20) Moreover the 1 MeV equivalent flux Φ 1 MeV,eq ( r, t) can be also introduced Φ 1 MeV,eq ( r, t) = κ( r) 0 φ( r, E, t)de = κ( r)φ( r, t). (2.21) In some applications where the bombarding particles have very large energies and their spectrum do not contain thermalized regions the integrals can derived practically from a larger energy: E min, the minimum energy in the spectrum. But in thermal reactors and application where the thermalization of the particles is significant it is advised to take the very low energy levels also into account. The definition of the D(E) functions is the following: D(E) = ν σ ν (E) ER,max 0 f ν (E, E R )P (E R )de R, (2.22) where ν denotes a reaction with σ ν cross section where an impacting particle with energy E creates a PKA with energy E R. The energy of the PKA can produce ionization and displacement, the P (E R ) function is the so called Lindhard partitioning function [21]. It describes the fraction of E R which is deposited in atomic displacements. Function f ν (E, E R ) is the probability density function which describes the probability that of a PKA wit E R will born due to an impacting particle with energy E. It is important to note that the shape of D(E) is well known in the case of silicon when the bombarding particles are neutrons which is shown at Figure 2.6 [22]. The minimum of D(E) is around 185 ev. Below that energy D(E) is increasing with decreasing energy. 185 ev is the neutron energy which is required to produce displacement via elastic scattering, below this energy the neutron capture will be the dominant process. During neutron capture a gamma photon emission can occur which will recoil the silicon atom. Figure 2.6 shows also that 3 orders of magnitude is the difference between the high energy regions (E 1 MeV) and the thermal energy regions. In Chapter 7 it will be pointed out that this large difference has important consequence on the shielding selection and planning. NIEL scaling and damage functions cover only the case of intrinsic Si material, however experiments showed with different dopant concentrations the NIEL scaling rule is valid [10], but there are some violations of the scaling [23]. The NIEL scaling hypothesis has its own restrictions but these do not cover the question of high boron doping concentrations [24] which could be important for commercial 20

23 2.5. NEUTRON AND GAMMA RADIATION FIELDS D (E )/(9 5 M e V m b ) D (E )/(9 5 M e V m b ) E th e rm a l ~ 3 o rd e rs S lo w in g o f n e u tro n s N e u tro n s k in e tic e n e rg y (M e V ) Figure 2.6: The normalized displacement function in silicon due to neutron bombardment 21

24 2.5. NEUTRON AND GAMMA RADIATION FIELDS semiconductor devices, however for a first approximation it is important to calculate the NIEL based parameters [25], especially when no other experimental data is available. Determination of the displacement function The displacement function can be calculated with different program codes for example with NJOY [26], however I have rather used displacement function from the literature. NJOY builds up from several individual modules. The HEATR module can provide radiation damage production cross sections from ENDF cross sections [27]. GROUPR module is capable to calculate the spectrum of the PKA in various materials [28] Connection with DPA DPA refers for Displacement Per Atom, this section covers the description of it. The DPA comes from the model Kinchin and Pease [29] which used to be referred as K-P model. The K-P model mostly used to describe the damage with reference to atomic displacements in metals and alloys. The K-P model has its own modifications and alternatives, like the model of Norget, Torrens and Robinson [30], the NRT model. The number of displacement per unit volume and unit time or reaction rate density R is the following: R = displacements cm 3 s = N Emax E min φ(e i )σ D (E i )de i, (2.23) where N is the atom number density, E max is the maximum energy of the incoming particle, E min is the minimum energy of the incoming particle, φ(e i ) is the energy dependent particle flux of the incoming particles and σ D (E i ) is the displacement cross section which also depends on the energy of the incoming particle. The displacement cross section can be a little more explained: σ D (E i ) = Tmax T min σ(e i, T )ν(t )dt, (2.24) where T min is the minimum energy which is transferred to a lattice atom in the collision with the incoming particle of energy E i, T max is the maximum energy which is transferred to a lattice atom in the collision with the incoming particle of energy E i, σ(e i, T ) is the cross section of the collision with T transferred energy and E i incoming particle energy, ν(t ) is the number of displacements per PKA. The last two equations can be united: R = N Emax Tmax E min From this, one can easily calculate the DPA: DP A = Rt N = t T min φ(e i )σ(e i, T )ν(t )dt de i, (2.25) Emax Tmax E min T min φ(e i )σ(e i, T )ν(t )dt de i, (2.26) 22

25 2.6. PARAMETERS AFTER IRRADIATION where t is the irradiation time assuming that the flux was constant during the irradiation. If the flux is the function of time, another integration takes place in time. Comparing equation (2.24) and (2.22) the main difference is that D(E) computes the energy which is deposited in non-ionizing processes while σ D computes the number of displacements no matter how they happened. The main parameters of the DPA calculations are the σ(e i, T ) and ν(t ). If we would like to be strict the K-P and NRT models give simple equations for the values of ν(t ). For example the NRT gives the following result: 0 if T < E d ν(t ) = 1 if E d < T < 2E d /β, (2.27) if 2E d /β < T < βt 2E d where E d denotes the binding energy of the atoms of the material and β denotes a correction factor. The DPA calculations has also their own restrictions, for example this model cannot handle the annealing processes, not applicable for compound materials etc. 2.6 Parameters after irradiation This section covers the behavior of some device parameters which are affected by irradiations. However, this section does not aim to deal with every phenomena in details because the details can be found in the literature, it only tries to present the complexity of the problem Leakage current due to irradiation To describe the change of the experienced leakage current generated in the bulk material of the examined p n junctions, a simple linear equation (2.28) was found [31, 10, 32] I = αφ eq V diode, (2.28) where I denotes the change of the leakage current, α denotes a proportionality factor called reverse current damage constant, Φ eq denotes the 1 MeV equivalent neutron fluence and V diode is the volume of the diode. An important remark is that in some of these measurements guard rings were used around the diodes and thermal annealing were also processed. Guard ring is a protection region around the active area of the diode (in the silicon bulk) which is usually biased at the same voltage than the diode, hence it collects the leakage currents generated outside the sensitive area. The annealing experiments showed that α is a function of annealing temperature and annealing time [10]. Another important note is that the leakage current does not depend on the type of the material, both n-type and p-type p n junctions satisfied the equation (2.28). This experimental result is very important because it also shows that Φ eq is independent from 23

26 2.7. SEMICONDUCTOR BASED VIDEO DEVICES the material type, and the damage function which is available only for intrinsic silicon, can be used for doped silicon as well. Here we note that in the measurements the typical boron concentration of the p-type materials was atoms/cm 3 [33]. The material independence of α could be explained by the assumption that cluster defects are mainly responsible for the leakage current generation [10] and not the point defects. The leakage current due to gamma radiation will also increase, but as the gamma radiation could hardy produce clusters it could be explained by point defects [10]. In the case of gamma irradiation the surface related leakage current contributes together with the bulk generated leakage current. The behavior of other important parameters can be found in the appendix (11.3). 2.7 Semiconductor based video devices The ancestor technology in video imaging based on vidicon tubes. The vidicon tube technology is still in use in very harsh radiation fields. The most adequate example is the video imaging based nuclear reactor and fuel inspection [34]. These devices produce an analogue signal which is displayed analogously or digitally after conversion. This technology is quite robust against the harsh nuclear environment, but other parameters of vidicon tubes do not satisfy the novel requirements of the diagnostic systems in fusion technologies. The speed, the resolution, and the light sensitivity of these cameras are not suitable for requirements of today. First, the appearance of the CCD based video techniques displaced the vidicon tube, then as the CMOS based technologies were developed CMOS imaging chips also appeared as a rival technology of CCD CCD based devices CCD stands for Charge Coupled Device which is an imaging sensor chip. High sensitivity and low noise are the main benefits of these devices. CCDs consist of pixels which are basically MOS capacitors or photo diodes. These pixels convert the incoming light into free charge carriers and these carriers are kept in a potential well until the start of the readout process. The pixels are ordered in a 2D matrix. During the readout process the pixels are read out by shifting the collected charges from a potential well to a neighboring one, across the lines or the columns of the CCD. Shifting of the charges is completed by changing the potentials of the potential wells, it is possible if both neighbors of a pixel have different contacts with different biasing options. After a charge package left the last pixel, they are collected by a shift register which transfers the charges for further processing mostly for pre-amplifying, amplifying and analog to digital conversion. To prevent the mixing the charge packages the whole process is synchronized with a clock. Most of the CCD and CMOS cameras are equipped with Pinned Photodiode, their description 24

27 2.7. SEMICONDUCTOR BASED VIDEO DEVICES can be found in the literature [35]. CCD devices have a numerous parameters which can be changed due to irradiation [36, 37]: dark current, number of defective pixels, Quantum Efficiency (QE), output amplifier gain, charge to voltage (q/v) conversion, Charge Transfer Efficiencies (CTE) etc CMOS based devices CMOS based camera chips are nowadays widespread and worthy competitors of the CCD imaging sensors. Low power consumption and faster operation are the main benefits of the technology which is based on individually addressed pixels so called active pixel architecture or Active Pixels Sensors (APS) [38]. Active Pixel means every single pixel could be individually addressed and read out. As a consequence every pixel contains in addition to the active region 3-4 or 5 additional transistors. These chips are based on 3T, 4T, 5T technologies respectively. The main three transistors are the reset transistor, the read out transistor or source follower transistor and the row select transistor. Moreover the active pixel technology makes it possible to integrate the timing or the Analog to Digital Conversion to the chip. CMOS devices have numerous parameters which can change due to irradiation [39, 40]: dark current, number of defective pixels, dark random noise, dark-signal non-uniformity, degradation of dynamic range, etc... These parameters together with the parameters of CCD devices form an upper, chip level of parameters, comparing to more microscopical parameters like the leakage current of an individual diode, the charge accumulation in an oxide layer or point defect creation due to displacement damage Connecting electronics In modern fast cameras both the CCD and CMOS chips are integrated in complex camera electronics. These electronic components can be power supplies, data processors, ADCs, memories, communication units etc. which altogether ensure the fast operation speed of the camera End-user radiation damage investigations As it was shown in the previous sections the modern video imaging chips, systems contain numerous diodes, transistors and other electrical components with different architecture and purpose. The properties of these electrical components are strongly affected by different radiations, just think about the leakage current, current gain etc. When no data available about the radiation tolerance of a camera, there are two options: It can be tried to find information about the camera components and about the radiation tolerance of the individual parts. From these informations a model can be 25

28 2.7. SEMICONDUCTOR BASED VIDEO DEVICES built up to calculate the performance change of the camera, based on the performance degradation of the components due to radiation. This comprehensive model building is usually not possible, when off the shelf cameras are chosen for plasma physics measurements and the precise work schematics are not known, partly due to industrial secrets or due to that the electrical components were not tested. Another technique is the analysis of the recorded frames - as the frames are the natural measureable output of such a camera - it is a common practice for determination of the camera degradation due to irradiation. In that case the output contains all the superimposed parameter degradations of the camera. There are various techniques for example dark signal analysis of the frames or comparison of the frames as the function of absorbed dose or particle fluence. These techniques are mainly statistics based processes and there is less chance to separate or to find the weakest electrical component in the camera. More detailed description of techniques and frame analysis of irradiated cameras can be found in the literature [41, 42, 43, 44, 45, 46, 47, 48]. The applied methods in my studies are discussed in details in Chapter

29 Chapter 3 Computational methods This section covers the estimation techniques which are used for the calculation of radiation field parameters which affect the performance of the video diagnostic systems around fusion devices. These calculations aim the prediction of the neutron flux, the gamma dose and spectrum in the camera positions. These type of calculations require precise particle transport approach. Basically two different methods are possible, the first is deterministic approach [49] and the second is the Monte Carlo method (MC) based techniques. In very complex geometries the deterministic calculation is impractical or at least not common, however the Monte Carlo based techniques could lead to success. In the following only the Monte Carlo technique will be presented. 3.1 Monte Carlo calculations The MC method is based on random number drawing as it tries to play along the physical processes like as they happen in the nature. In MC based transport calculations every parameter which has a probability distribution is drawn, like the energy, starting position, flying direction, interaction position, interaction type etc. of the source and scattered particles. For the calculations precise cross section libraries, material composition and geometry data of the problem are required MCNP The MCNP [50] which stands for Monte Carlo N-Particle transport code is one of the most popular MC based transport codes. MCNP is developed since decades by the Los Alamos National Laboratory (LANL) and distributed by Radiation Safety Information Computational Center (RSICC). The code is benchmarked and validated in many times in different research fields such as nuclear medicine, health physics, space flights, reactor physics etc. If the user is not an MCNP developer and do not want to insert his/her own code parts in MCNP then the user has to edit only an input file and interpret the 27

30 3.2. CAD-MONTE CARLO CONVERSION calculated results. The input file contains the geometry of the problem, the material composition and the references to the cross section libraries to be used. The input file should also include the precise definitions of the radiation sources (or in the case of reactor criticality calculations only positions should be defined where fission materials are present) and the quantities to be calculated, called tallies. Radiation sources can be very versatile, like point sources, surface sources, volumetric sources, moreover different energy distributions and flying direction distributions can be set. The tallies can vary also, for example volume averaged particle flux, surface averaged particle flux, energy deposition in a cell, etc. can be calculated. The precision of every MCNP calculation is based on the precision of the input parameters i.e. the precision of the geometry, the material composition and the applied cross sections. The results of the calculations are not deterministic. Only expected values, standard deviations and other statistical quantities can be calculated of tallies, however the standard deviations can be decreased by increasing the number of the started particles, or the number of calculation cycles. Moreover a set of variance reduction techniques are also implemented to MCNP. 3.2 CAD-Monte Carlo conversion As the 3D mechanical engineering Computer Aided Design (CAD) software codes were developed and the computerized design of mechanical parts became easier, a natural need for 3D MC input editor and converter softwares occurred. Nowadays, several different methods and software codes are supporting graphical input generation or conversion. In the case of MCNP the variety of available software is a bit smaller. The mainly used editor and converter software is the Visual Editor (Vised) [51] which is distributed by the RSICC like the MCNP. I have tried the 3D CAD to MCNP conversion module of it but it was not satisfactory. Another famous conversion software is McCAD [52] which is developed by Karlsruhe Institute of Technology (KIT) but it is not distributed widely. The last example of graphical MCNP editors and CAD to MCNP input converters is MCAM [53]. MCAM refers to Multi-physics Coupling Analysis Modeling Program, developed and distributed by the FDS Team, China. Both the McCAD and MCAM development is driven by the fusion research and fusion community. The name of the newest MCAM version is extended to SuperMC/MCAM which denotes that the program is compatible with the SuperMC particle transport code which was also developed by FDS Team. In my calculations I used MCAM4.8, MCAM5.2 and SuperMC/MCAM professional editions MCAM MCAM is versatile 3D MCNP input designer and CAD to MC converter code. Four types of released packages exist: the demo, the standard, the education and the profes- 28

31 3.2. CAD-MONTE CARLO CONVERSION sional edition. The main difference between them is the maximum allowed size of the convertible model. In the professional edition up to 5000 cells can be converted in one step, which is more than enough, if one keeps in mind that MCNP allows only cells in a single model. The cell and the surface cards can be arbitrary numbered and with a few disjoint numbered conversion a larger model (much larger than 5000 cells) can be converted. The separated input segments can be copied together in a final input file. MCAM provides two-way conversion between MC and CAD formats, for MCNP, SuperMC, TRIPOLI and GEANT4. Two-way means that not just the CAD to MC works, but for example: from an MCNP input file MCAM can build up the 3D visualization of the model which can be further edited. The size limit is the same, 5000 cells in the case of professional edition. There is an option to draw the geometry manually in the 3D editor or import a predefined 3D CAD design file such as STEP, CNF or INES. All the MCAM geometry files are stored in a special file format called FDS. Further information about MCAM can be found in appendix (11.4). 29

32 Chapter 4 Common experimental methods This chapter covers the basic principles of magnetically confined fusion devices in particular the three machines I have worked with. In addition typical aspects of the radiation fields and radiation measurement techniques based on literature are presented. 4.1 Magnetically confined fusion Two main types of magnetically confined fusion devices are frequently utilized: tokamaks and stellarators. The major difference between the two concepts is that in tokamaks a large plasma current flows ( 3 5 MA) to ensure the helical twist of the magnetic field lines, however in stellarators the shape of magnets are modified to achieve the helically twisted magnetic field lines. Therefore tokamaks are azimuthally symmetric machines while stellarators have discrete rotational symmetry. Helically twisted field lines are necessary to keep the drifting charged particles inside the plasma volume. The lack of plasma current indicates that steady state operation can be achieved more easily in stellarators but the calculation methods cannot be simplified to 2D problems like in tokamaks Diagnostic systems for plasma diagnostics The goal of plasma diagnostics is to determine the shape and position of the plasma, to measure the density, the isotope distribution, to measure the concentration of impurities, to measure the electron and ion temperatures, to measure the current distribution, their fluctuations etc. Such a large number of parameters requires a large number of detector types and set ups. There are a huge amount of magnetic probes, loops like Mirnow probes, Hall probes, Rogowsky coils, saddle coil, diamagnetic loop etc. Optical methods use some radiation which can be originated from the plasma or introduced from the outside and the changes of the radiation are measured. Most common techniques are: Interferometers such as micro wave interferometers or Far Infra-Red (FIR) interferometers, 30

33 4.1. MAGNETICALLY CONFINED FUSION Thomson Scattering Reflectometry Electron Cyclotron Emission (ECE) Ion Cyclotron Emission (ICE) Various X-ray measurements like X-ray spectroscopy or X-ray tomography Neutral Particle Analyzer (NPA) Charge Exchange Resonance Spectroscopy (CXRS) Bolometers and bolometer tomography Neutral Beam Injectors (NBI) connected to visible light radiation detectors like Beam Emission Spectrometers (BES) Nuclear detectors like neutron cameras or activation foils Plasma positioning and pellet detector cameras etc. In my work I have studied mainly two types of diagnostic systems: a video diagnostic system which works in the visible spectrum of light developed for the Wendelstein 7-X (W7-X) stellarator and contains a special CMOS based camera called EDICAM and BES systems which were developed for the Korea Superconducting Tokamak Advanced Research (KSTAR) and Experimental Advanced Superconducting Tokamak (EAST) tokamaks and contain various cameras like CCD, CMOS and APD based cameras EDICAM system on W7-X The W7-X is the largest fusion device of the optimized stellarator type. It consists superconducting magnets and built up from 5 almost identical modules which means W7- X has an almost five-fold discrete rotational symmetry. The major radius of the stellarator is 5.5 m while the minor radius is 0.53 m. The volume of the plasma is approximately 30 m 3. The maximal magnetic field on axis is < 3 T. The Hungarian Academy of Sciences (HAS), Wigner Research Centre for Physics (Wigner RCP) has undertaken the development of an intelligent video diagnostic system for the W7-X. The main camera of this system is the Event Detection Intelligent Camera (EDICAM) which main purpose is to detect various pre-defined events. For example when plasma reaches the first wall the EDICAM system alarms the control system [54]. Dimensions of the camera are 6 cm 20 cm. The camera contains a CMOS sensor (Lupa 31

34 4.1. MAGNETICALLY CONFINED FUSION 1300) [55] which enables to select different Region Of Interests (ROIs) and to use different frame rates in the ROIs. Pre-defined events can be real time analyzed and recognized by a Field Programmable Gate Array (FPGA) which can send alarm signals through a 10 Mbit optical transmitter. This camera system contains 10 cameras which are located at AEQ tangential ports in air atmosphere but quite close to the vacuum window. The position of the cameras indicates large radiation loads as it will be shown in section 4.2 and chapter BES system on the KSTAR tokamak KSTAR tokamak is a South Korean superconducting tokamak located in Daejon. The major radius of the vacuum vessel is 1.8 m, the minor radius of the vessel is 0.5 m, the maximum toroidal magnetic field is 3.5 T and the maximum toroidal plasma current is 2 MA. The heating system contains Ion Cyclotron Resonance Heating (ICRH), Electron Cyclotron Resonance Heating (ECRH) and Neutral Beam Injection (NBI) devices. The Beam Emission Spectroscopy (BES) system measures plasma turbulence [56], by detecting light, emitted by the neutral heating beams or from diagnostic beams. First, a trial BES system was installed to the KSTAR [57], then based on the experiences a new BES system was designed and installed in the cooperation of Wigner RCP and Budapest University of Technology and Economics (BME) on the KSTAR tokamak in Neutral beam injection systems (NBI, Li beam) inject fast atoms into the plasma, which are excited by the plasma particles and emit photons with a characteristic frequency. Variations in the intensity of light carry information on the local plasma density fluctuations. Measuring the density fluctuations can reveal information about plasma turbulences and Magneto Hydro Dynamic (MHD) modes of the plasma. The BES system was installed in the M-port of the tokamak which is situated close to the NBI box [58]. It contains a periscope which leads the light out of the plasma to two different cameras (see Figure 6.1). On one end of the light path an ADIMTECH APDCAM camera is located which contains a Hamamatsu S8550 4x8 channel Avalanche Photo Diode (APD) array, on the other end of the light path a PCO Pixelfly VGA camera is installed which contains a Pixelfly ICX414AL CCD sensor. The APDCAM is used for the fluctuation measurement and the PCO Pixelfly VGA camera is mostly used for spatial calibrations. As the cameras are located in different positions they will suffer from different radiation loads which means different neutron and gamma-ray spectra and fluences BES systems on the EAST tokamak EAST is a superconducting tokamak in Hefei, China. The major radius of the tokamak is 1.85 m while the minor radius is 0.45 m, the maximum toroidal magnetic field is 3.5 T 32

35 4.2. RADIATION ENVIRONMENT OF FUSION DEVICES and the maximum current in the toroidal plasma is 1.0 MA. The heating system contains ICRH, ECRH, Lower hybrid current drive (LHCD) and NBI heating as well. The tokamak get two new independent BES systems. A Li-BES and a H-BES system were installed in the D and P ports of the tokamak. Both BES systems contain 2 cameras an ADIMTECH APDCAM and a Photonfocus CMOS based camera Connecting optics and their effects For imaging purposes between the plasma and the camera sensors an optical system is required. In the W7-X this optical system is small because the EDICAM cameras are located in the inner end of the ports, however at the BES systems both on KSTAR and EAST optical systems contain several lenses, prisms, mirrors, wavelength filters. Since the insertion of an optical system between the plasma and cameras increases the distance from the main radiation source, thus reducing the expected radiation loads of the detectors. However, the optical system also suffers from radiation, due to neutron and gamma radiation, microscopic coloring centers [59] appears in the glass which decreases the transmission of the optical system. Moreover, practically it can be told that on every optical surface a few percent of light will be lost [60] therefore the complexity and size of the optics is limited. Another rival aspect is the necessity of shielding labyrinths for optical systems, for example in ITER. It means any optical system which leads the light through a port should contain direction changes. Every direction change requires the insertion of new optical elements where some light will be lost. There are radiation tolerant glasses at the market but their costs in the case of a complex optical system can be a large investment. Another solution can be the usage of image guides, which are arranged fiber bundles, to lead the light from the plasma to the detectors. But the performance of optical fibers is also affected by the neutron and gamma radiation [61, 60] and their price is also not negligible. 4.2 Radiation environment of fusion devices Both in stellarators and tokamaks the same nuclear reactions are desired. Nowadays, the D D reaction is in use, only in the JET tokamak performed and will perform D T plasmas in the next few years. The next large step will be the operation of the ITER tokamak whose dedicated goal is to operate with D T reaction during long plasma discharges up to 400 s for inductively driven burn and steady state operation with relatively smaller fusion power gain Q 5. Both D D and D T can produce energetic charged and neutral particles like neutrons: 33

36 4.2. RADIATION ENVIRONMENT OF FUSION DEVICES T D1 2 + D (1.01 MeV ) + p 1 1(3.02 MeV ) 50% = He 3 2(0.82 MeV ) + n 1 0(2.45 MeV ) 50% (4.1) D1 2 + T1 3 = He 4 2(3.5 MeV ) + n 1 0(14.1 MeV ). (4.2) The charged particles are confined by the present magnetic field in the fusion devices but the neutral neutrons can easily leave the plasma and reach the surrounding components like plasma facing wall, wall of the vacuum chamber, magnets, and diagnostic systems. But not only neutron radiation is which appears in the surroundings, there are naturally X-ray radiation or cyclotron radiation as the charged particles moves along the magnetic field lines, moreover gamma radiation occurs as the neutron interactions produce excited and radioactive isotopes. These three radiations can easily penetrate through the structural materials and harm the sensitive diagnostic devices. Neutron radiation is necessary to achieve sustainable fusion reaction based on equation (4.2) because the T is radioactive with a half-life a, therefore T production should be solved. A great candidate is breeding of T by neutron bombardment of Li 6 or Li 7 isotopes, which will be tested in ITER: n Li 6 3 = He T1 3 + (4.8 MeV ) (4.3) n 1 0(fast) + Li 7 3 = He T1 3 + n 1 0(slow). (4.4) The neutron spectrum and fluxes In the literature both calculations and measurements can be found which were carried out to determine the neutron spectrum and flux around different fusion devices like JET [62], W7-X [63], ITER[64] or others. These spectra can be different due to the measurement positions like: first wall, divertor position, behind biological shielding, etc. A typical first wall neutron spectrum of the ITER tokamak is seen on Figure (4.1). On JET tokamak the neutron spectrum (see Figure 4.2) is quite similar to the spectrum which is presented for ITER first wall on Figure (4.1). It is important that here the spectrum is normalized to one source neutron. To obtain the absolute value of the spectrum it should be multiplied by the source intensity. Calculations were also carried out in the case of W7-X stellarator, see Figure (4.3) for the neutron flux The photon spectrum and dose rates The gamma dose rate during operation in the ITER will be much larger than in the fusion reactors nowadays, it will reach µsv/h in the plasma facing wall regions and 10 7 µsv/h in the port regions [65]. For example 10 7 µ Sv/h is comparable with 34

37 4.2. RADIATION ENVIRONMENT OF FUSION DEVICES Figure 4.1: Calculated neutron spectrum for the ITER first wall considering D D and D T operations, and the neutron spectrum of the first wall in the Power Plant Conceptual Study model-b (PPCS-B)[64] Figure 4.2: Calculated neutron spectrum for the JET midplane vacuum vessel considering D D and D T operations together with Cf 252 spontaneous fission neutron spectrum. [62] 35

38 4.2. RADIATION ENVIRONMENT OF FUSION DEVICES Figure 4.3: Calculated neutron spectrum for the W7-X stellarator at different positions. [63] the expected monthly gamma dose in the W7-X stellarator in the EDICAM position. A typical gamma spectrum can be seen at Figure 4.4 which is a relatively hard spectrum, the upper end of the spectrum is just above 9 MeV. The length of the plasma discharges is an important parameter, because together with flux and dose rate the total irradiation time determines the fluence and the total dose absorbed by the video diagnostic systems. In the recently used tokamaks the length of the discharges are on the second scale up to approximately 60 sec. The record length of a plasma discharge (6 min 30 sec) is achieved at the Tore Supra tokamak. In stellarators a longer pulse length can be achieved for example at W7-X 30 min pulse length is planned. The pulses can be repeated daily approximately 20 times and the campaign length take a few months. The radiation of the plasma as it can be seen in section is monitored with different plasma diagnostic devices, however the radiation levels around tokamaks and stellarators are monitored mainly only due to human radiation protection purposes. The radiation levels close to the diagnostics are not measured (or very rarely), therefore only calculation methods available to predict the properties of the radiation fields. 36

39 4.2. RADIATION ENVIRONMENT OF FUSION DEVICES Figure 4.4: Comparison of the cumulated gamma spectra before 2009 and after 2012 the installation of new tandem collimators at JET. [66] 37

40 4.3. RADIATION OF FISSION REACTORS 4.3 Radiation of fission reactors Nuclear fission reactors use fissionable isotopes like U , U or P u , P u as fuel to sustain the nuclear chain reaction. During the operation of a fission reaction naturally a high amount of neutrons born. Their energy distribution is called Watt spectrum. In the case of U the average energy of the Watt spectrum is 2 MeV and the most probable neutron energy is 0.7 MeV. The mostly widespread reactors are thermal reactors which means that the neutrons inside the reactor core are slowed down and thermalized due to collisions with the atoms of the moderator material. One of the most important parameter of these reactors is the thermal neutron flux (see equation (2.14)). In dedicated high flux irradiation facilities like Institute Laune-Langevin (ILL) the thermal flux can reach cm 2 s but in a smaller reactor such as the Training Reactor (TR) of BME the maximal thermal neutron flux is cm 2 s sample irradiations, up to 110 mm, the achievable φ th = cm 2 s. At the TR for larger size ±6% and the fast neutron flux φ(e n > 1 MeV ) = ±8%. Moreover there is a large difference in the cm 2 s operation time of the fission reactors comparing to the operation times of fusion devices, thus much larger neutron fluences are achievable in fission reactors. Comparing these fluxes to the fluxes presented in section 4.2 it can be easily understood why fission reactors are used to test electronic components and study the radiation damage of electronic devices in shorter time. But the neutron spectra is slightly different, therefore the direct comparison of the irradiations is not available only after an appropriate scaling, like the NIEL scaling. During the operation of a fission reactor a significant amount of gamma radiation is also present which originates from different sources: direct emission of gamma photons from the fission events or de-excitation of nucleus during the radioactive decay of the fission products or de-excitation after neutron absorption or scattering. This also indicates that after shutdown of the reactor only the radioactive decay remains, which called as remnant radiation, which means intense gamma emission. This remnant gamma radiation can be easily used as a pure high gamma dose rate radiation field. A typical gamma spectrum of a spent fuel assembly sample can be seen on Figure 4.5, which has similar spectrum to a shutdown reactor. 4.4 Gamma dose rate and neutron flux measurements During irradiation tests it is necessary to monitor and measure the gamma and neutron radiation fields. There are active and passive techniques for both radiation. In this section a short introduction will be presented about the possible radiation measurement techniques for both fission and fusion irradiations. 38

41 4.4. DOSE RATE AND FLUX MEASUREMENTS Figure 4.5: Gamma spectrum of a spent fuel assembly sample (removed from a fission rector) after re-irradiation in PROTEUS reactor for 15 min at 800 W. The spectrum was taken from 5 min after the shutdown. [67] Neutron flux measurement For neutron flux measurement more options are available. The most common technique is the use of activation foils which is a passive method. Active techniques are mostly based on some kind of gas filled detectors like He 3, BF 3 counters or fission chambers. If not just the neutron flux but also the neutron spectrum is relevant a set of activation foils can be used or the spectrum measurement can be traced back to thermal neutron flux measurement with the use of moderator materials like in the Bonner sphere arrangement. The Bonner sphere technique is not applicable in small irradiation positions where there are not enough place for such a set up. A more point like measurement is the activation foil technique however it requires a large amount of foils if we would like to reach the sufficient energy resolution in the spectrum, moreover the unfolding technique requires too many foils to reach a fine spectral resolution. A sufficiently precise procedure can be the application of a simpler thermal flux measurement with activation foil or with an active detector together with a precise neutron transport calculation. In this case the measurement can validate the thermal flux or the total flux as well, and the calculations can give the precise shape of the neutron spectrum. The calculation of the neutron spectrum is necessary for the neutron hardness factor (κ) and the Φ 1 MeV,eq (t) (see section 2.5.3). However, there is another solution. There are calibrated solid state detectors like p i n diodes or various FETs which can be used as direct detectors of Φ 1 MeV,eq (t) [68]. These detectors also provide a direct possibility to validate these types of radiation damage calculations. After a proper check of these detectors we can easily find one that can measure up to cm 2 neutron equivalent fluences, for example BPW 34F 39

42 4.5. NEUTRON AND GAMMA SHIELDING SI-Diode can be a good choice. The detector sizes and their simple wiring makes these detectors capable of achieving local measurements with good time resolution if more of them are applied simultaneously Gamma dose rate measurement The gamma dose rate can be measured by gas filled detectors. In high gamma dose rate levels, robust and not so sensitive detectors are needed [69] like Geiger-Müller (GM) tube based detectors. Another technique is to measure the total dose during the irradiation, however in that case the development of the gamma field must be known. Passive methods [69], like the use of ThermoLuminescent Detectors (TLD) can be sufficient, if the gamma dose rate is constant or the change of it is negligible or the time dependence is known like in the case of irradiations made with isotope sources. 4.5 Neutron and gamma shielding The traditional neutron shielding covers three aspects: the slowing down of neutrons, the absorption of the thermalized neutrons and the application of additional gamma shielding to reduce gamma radiation intensity which has been born in (n, γ) reactions or in radioactive decay processes. Good neutron slowing materials contains light atoms. For the absorption of thermalized neutrons such materials are used which have large neutron absorption cross sections. The most typical absorber materials are boron, cadmium etc. Neutron absorption of B 10 and the undergoing decay process is described with the following reactions: Li 7 3(0.84 MeV ) + He 4 2(1.47 MeV ) + γ(0.478 MeV ) 94% B n 1 0 Li He MeV 6%. For example in the case of cadmium from its 8 natural isotopes the Cd 113 (4.5) has large thermal radiative capture cross section. In this reaction energetic gamma rays with energy: kev and kev can be born. The intensity of the gamma radiation can be easily decreased by applying sufficient thickness of shielding materials which contain high atomic number atoms. Does the high boron content of doped p-type semiconductor material affects negatively on the displacement damage function of the neutron irradiated semiconductor devices comparing to the damage functions of intrinsic semiconductors? The recent answer is that we do not know that exactly. The problem is - for example in the particular case of NJOY - that NJOY can calculate displacement functions only for single isotope materials. There are existing methods in the literature to handle polyatomic materials [70] and to apply these models to calculate displacement function for materials like SiC [27] but for 40

43 4.5. NEUTRON AND GAMMA SHIELDING Si the dependence of doping concentration is not clarified yet. As we saw in section for small boron concentrations both type of material behaved the same, but it is also true that in radiation hard semiconductors only B 11 enriched boron is used as dopant. 41

44 Chapter 5 Investigations of the EDICAM camera 5.1 Introduction According to the recent status of EDICAM [54], the camera head can be divided into four major sub structures. These are the CMOS sensor module, a Field Programmable Gate Array (FPGA), an Analog-to-Digital Converter (ADC) and a 10 gigabit optical link. The dominant mechanisms of radiation effects in Complementary Metal-Oxide Semiconductor (CMOS) devices are the charge-build up in the gate dielectric region, radiation induced interface levels and displacement damage in the bulk [71] [46]. For commercial FPGA-s, irradiation tests were carried out and reported that they withstand very high gamma dose values (155 Gy-384 Gy) at moderate gamma dose rate levels (1 mgy/s) without observable change in their operating properties [72]. Irradiations in high gamma dose rate fields up to very high gamma dose levels ( Gy) of commercial flash ADCs showed excellent performance [73]. High gamma dose rate irradiations (200 Gy/h) of gigabit optical links showed that such fields can worsen the performance of those devices [74]. The manufacturing conditions can be very different for off the shelf device and radiation test data for these components are rarely available, hence their behavior is hardly predictable, therefore tests needed for a final version of every assembled camera type. In the case of EDICAM which is also made up from off the shelf components, the effect of radiation on the recorded frames is the key question. Thus, the degradation of the standard output data (the frames) of EDICAM was monitored, instead of the degradation of the individual components. During operation in a fusion device, only the frames will be available. In the following sections the results of MCNP [75] calculations are presented which covers the estimation of the gamma dose in the port of W7-X where EDICAM is positioned and the results of gamma irradiation tests which are based on the 42

45 5.1. INTRODUCTION MCNP calculations. Irradiations were performed in the Training Reactor of BME. During the irradiations, the camera was operated in darkness and numerous frame series were taken with different exposure times. The behavior of the dark current frames of EDICAM during irradiation was studied MCNP model The first goal was to estimate the gamma and neutron dose and fluence values in the tangential camera port of W7-X where EDICAM will operate. In our model, some simplifications are used, which are the following: vacuum vessel was modelled as a torus with large radius of 550 cm and small radius of 50 cm. Only D + D = 3 He + n reaction was supposed, thus the neutron source was monenergistic with 2.45 MeV neutrons. The total neutron source was supposed to be n/year which means n will yield over one year operation in the whole plasma volume based on former calculations [76] [77]. The source was distributed homogenously in the vacuum vessel. The vacuum vessel was filled with vacuum, hence neutrons did not suffer any collisions and hence they did not slow down in the vacuum vessel. The tangential camera port was a tube with 10 cm radius. The port and the vacuum vessel were surrounded with the material of the magnetic coils, see Figure 5.1a. The following materials were used: 10 V/V% (concentration in volume %) of iron, 1 V/V% of niobium, 1 V/V% of vanadium and the rest of air. The wall of the vacuum vessel was modelled as 1 cm of graphite and 1.7 cm steel of type The wall of the port was steel with thickness of 0.3 cm. The port (see Figure 5.1b) had an entrance from the vacuum vessel side having a 2.5 cm thick hollow steel plug with a 5 cm diameter lead glass window. For additional radiation protection, 1.5 cm thick lead shielding was placed on the inner wall of the port. Behind the lead-glass window, the port was filled with low density silicon blocks in the whole cm port length, while the rest of it was filled with air Gamma dose and photon spectrum The most conservative case is when the 1.5 cm thick additional lead shielding is not used on the inner wall of the port. In this case, the following assumptions are valid. Gamma photons from the fusion plasma were not included in the model, only gammas generated by primary neutrons were calculated. The calculations showed that the photon spectrum would have a strong component immediately below 1 MeV and plenty of soft lines, see Figure 5.2a. In the camera port, the spatial distribution of the calculated gamma dose in silicon shows significant decrease in the direction of the port outer end (150 cm), see Figure 5.2b. The maximum estimated gamma dose is approximately 18 Gy. These results were used for planning the irradiation. 43

46 5.1. INTRODUCTION a Surrounding torus from superconducting material Vacuum vessel Optical port Vacuum Lead-glass Steel Silicon Air Lead b Graphite Figure 5.1: (a) Horizontal cross section of the vacuum vessel (black), the video port (cyan) and the surrounding torus (purple) with the material of the magnetic coils. The grey area is not part of the model. (b) Horizontal cross section of the video port. Blue is steel, yellow denotes lead-glass, olive is lead, cyan is silicon, green is graphite and grey denotes air. 44

47 5.1. INTRODUCTION 1,20E+012 1,00E+012 a Photon fluence (p/cm 2 ) 8,00E+011 6,00E+011 4,00E+011 2,00E+011 0,00E Energy (MeV) b 14 Gamma dose (Gy) Distance from lead-glass (cm) Figure 5.2: (a) Calculated gamma spectrum in silicon. (b) Calculated yearly gamma dose spatial distribution in the silicon along the port. 45

48 AMP-200 TLD detectors EDICAM Sensitive volume Sensor position Detector holder Optical cable Power supply cables 5.2. IRRADIATION 5.2 Irradiation EDICAM was irradiated in a dry vertical irradiation channel of the Training Reactor of BME (see Figure 5.3a). a b Shematic position of EDICAM during irradiation Shematic position of AMP-200 during irradiation Figure 5.3: (a) A schematic vertical cross section of the Training Reactor of BME with the vertical irradiation channel, in which a detector holder containing the AMP-200 gamma dose rate meter and the EDICAM placed. Yellow is the reactor core. (b) A schematic figure of the detector holder with the AMP-200, EDICAM and the ThermoLuminescent Detectors (TLD). The detector holder consists of two long rods which ends in bottom in an aluminum ring, in the upper part ends conically and mounted to the manipulating rods with a M6 screw. The detector holder is an open structure. The diameter of the irradiation channel is 110 mm. The EDICAM and an AMP-200 [78] gamma dose rate meter were installed on an aluminum detector holder (Figure 5.3b, 5.4). The AMP-200 was in a lower position under the EDICAM in the detector holder, because gamma dose rate depends on the vertical position and increases downwards in the channel. With this arrangement, the camera can be protected from undesirably large dose rates. A total of six TLD were also applied on the surface of the camera (Figure 5.3b) for spatial distribution measurement of the gamma dose. Irradiation was a pure gamma irradiation without neutrons. This was achieved by the residual (practically pure gamma) radiation of the core following the operation of the 46

49 5.2. IRRADIATION Figure 5.4: The AMP-200 is fixed to the aluminum detector holder, before irradiation. reactor at the nominal power level of 100 kw for 1.5 hour after which it was shut down Test irradiations Since there is always a distance between the place of the sensitive volume of AMP- 200 measuring the actual gamma dose rate and the EDICAM sensor, the dose rate ratios had to be measured between those positions. In order to determine the dose rate ratios between the selected irradiation positions and to determine the time dependence of the dose rate, long term test irradiations were carried out before the EDICAM irradiation. For these measurements, only dose rate meters, AMP-200 and FHZ-312 [79] gamma dose rate meter were used. The EDICAM was replaced by the FHZ-312 during the test on the detector holder (Figure 5.3b). We found that the gamma dose rate (DR(t)) of the detectors in each position can be described with a second order decay curve : DR(t) = DR 0 + A 1 exp( t/t 1 ) + A 2 exp( t/t 2 ), (5.1) where t denotes the time elapsed from reaching the irradiation position, DR 0 is the background dose rate, A 1 and A 2 are amplitudes and t 1 and t 2 are time constants. The obtained average of the parameters t 1 and t 2 measured by the two detectors are the following t 1 = ± h and t 2 = ± h. Parameters t 1 and t 2 were handled as well-known constant parameters in the following calculations. By dividing the measured dose rates (dose rate of AMP-200 divided by the dose rate of FHZ-312) in the final irradiation position we obtain the ratio of them G = 3.0 ± This dose rate ratio was used in the determination of the dose rate and dose of the sensor chip during the EDICAM irradiation. 47

50 5.3. EVALUATION OF THE FRAMES EDICAM irradiation For the EDICAM irradiation, we used the same detector holder positions as that were used during the test irradiations. The whole instrumentation set-up (Figure 5.3b) was lowered in the irradiation channel after the reactor was shut down. First, the detector holder reached the uppermost selected position. The set-up was kept here until the gamma dose rate decreased to 8.5 Gy/h. The absorbed dose by EDICAM in that position was 3.0 ± 0.30 Gy. Then the set-up was lowered by 0.5 m into the final irradiation position. In this place, the starting level of the gamma dose rate in the sensor chip position was 27 Gy/h. The whole set-up was kept there for 1.16 hour, at the end of irradiation the gamma dose rate was 13 Gy/h. During that the camera was connected to a measuring computer and numerous frame series were taken with 6 different exposure times (20 µs, 50 µs, 100 µs, 1 ms, 10 ms and 100 ms). The exposure time in these measurements was adjusted by only electrical way, no mechanical shutter took place. The sensor chip was dark folded to minimize the amount of visible light reaching it, thus the frames were dark current frames. Each frame series were saved in mj2 [80] file format and contained 100 frames. The gamma dose rate as a function of time (DR(t)) measured by the AMP-200 in the final position during irradiation and also a fitted second order decay curve with fixed t 1 and t 2 parameters are shown in Figure 5.5a. The dose of AMP-200 as a function of time (D(t)) (see Figure 5.5b) in the final irradiation position can be calculated by integrating the fitted second order decay curve which is shown in Figure 5.5a. The absorbed dose of the EDICAM sensor chip can be estimated by dividing the absorbed dose of AMP- 200 with the dose rate ratio measured in the test irradiations (3.0 ± 0.20). The EDICAM sensor total dose 23 Gy is the sum of the absorbed dose values in the two positions (the uppermost and the final irradiation position). Using the evaluation of TLD detectors, the normalized spatial distribution of the gamma dose on the camera surface was determined. The distribution is shown in Figure 5.5c. It can be noticed that the gamma dose is not constant on the surface of the EDICAM: there is a factor of 3 in dose between the sensor and the rear end of the camera. 5.3 Evaluation of the frames In this section the evaluation of the recorded frames will be presented. The mean p ij N value of the jth frame is m j = N i=1, where p ij denotes the dark current of the ith pixel of the jth frame, N is the total number of the pixels. The mean value of m j over a frame series is I D (t, τ), where t denotes the time, τ is the exposure time, D refers to dark current. The first five frames in every frame series were not taken into account in I D (t, τ), because their mean dark current was significantly larger than that of the others. We experienced that the camera needs some read outs to reach the nominal dark current 48

51 5.3. EVALUATION OF THE FRAMES a M o d e l: S e c o n d o rd e r d e c a y E q u a tio n : D R (t) = D R 0 + A 1 * e x p (-(t-t 0 )/t 1 ) + A 2 * e x p (-(t-t 0 )/t 2 ) W e ig h tin g : y In s tru m e n ta l C h i^ 2 /D o F = R ^ 2 = D R (t) (m G y /h ) D R t F IX A t F IX A t F IX ,8 1,0 1,2 1,4 1,6 1,8 2,0 T im e (h ) b 1,3 1,2 1,1 c D (t) (m G y ) N o rm a lis e d g a m m a d o s e 1,0 0,9 0,8 0,7 0,6 0,5 0,4 0,3 0 0, ,8 1,0 1,2 1,4 1,6 1,8 2,0 T im e (h ) 0,1-0,0 2 0,0 0 0,0 2 0,0 4 0,0 6 0,0 8 0,1 0 0,1 2 0,1 4 0,1 6 0,1 8 D is ta n c e fro m E D IC A M s e n s o r c h ip in v e rtic a l d ire c tio n (m ) Figure 5.5: (a)the measured gamma dose rate in the final irradiation position and the fitted exponential decay curve, with the confidence bands. (b) The total absorbed dose of the AMP-200 detector in the final irradiation position. (c) The normalized dose distribution on the surface of EDICAM in vertical direction, measured by TLD detectors. Position zero is the position of the sensor chip. 49

52 5.3. EVALUATION OF THE FRAMES levels. The standard deviation of I D (t, τ) were also calculated based on the propagation of uncertainty of p ij. The overall behavior of I D (t, τ) during irradiation is shown in Figure 5.6. First, it should be noticed considering Figure 5.6 that I D (t, τ) with short exposure m s 1 0 m s 1 m s µs 5 0 µs 2 0 µs I D (t, τ) (d ig it) ,8 1,0 1,2 1,4 1,6 1,8 2,0 T im e (h ) Figure 5.6: I D (t, τ) behavior during the irradiation. The time axis shows the elapsed time from the shutdown of the reactor. The EDICAM irradiation started 0.8 h after the shut down and the irradiation took approximately 1.16 h times (20 µs, 50 µs and 100 µs) is essentially the same. Only on longer exposure times (1 ms, 10 ms and 100 ms) occurred a significant shift in the offsets. However, here we should emphasize that with every exposure time a similar time evolution of I D (t, τ) can be recognized. That implies there are at least two different effects which determine the behavior of I D (t, τ). The first effect is an offset shift over a threshold exposure time (τ threshold ) which is independent from the dose rate and the dose. This threshold level is directly connected to the photodiode of pixels. A probable explanation is that only above τ threshold gamma photon excitation had time on the average to create enough electrons to increase significantly the offset with the applied dose rate levels and gamma spectrum. The second effect is the decrease of I D (t, τ) which occurs with every applied exposure time and with the evolution of time, the effect looks exposure time independent. In order to understand this effect, we should try to separate the dose and the dose rate dependent 50

53 5.3. EVALUATION OF THE FRAMES effects. For this, we suppose that I D (t, τ) can be written as a sum of three terms I D (t, τ) = I off (t, τ) + I T DE (t, τ) + I DRE (t, τ), (5.2) where I off (t, τ) denotes the offset level, I T DE (t, τ) is proportional to the dose of the sensor and I DRE (t, τ) is proportional to the dose rate in the sensor position, T DE refers to Total Dose Effect and DRE refers to Dose Rate Effect. This simple assumption gives us the possibility to normalize I D (t, τ) with the dose rate and dose, respectively, as seen in Figure 5.7a-5.7b. A separation over the τ threshold can be recognized in both Figures 5.7a-5.7b. In order to obtain a more sophisticated view of the problem, the normalizations will be deduced I D (t, τ) normalization by the dose rate is The dose rate in the sensor position is DR s (t) = DR(t)/G and the dose of the sensor D s (t) = t 0 DR s (t )dt = DR 0,s t A 1,s t 1 (exp( t t 1 ) 1) A 2,s t 2 (exp( t t 2 ) 1), (5.3) where DR 0,s, A 1,s and A 2,s are the background dose rate and amplitudes, respectively, in the sensor position. Consider equation (5.2) and explain the factors. I T DE (t, τ) = B T DE (t, τ)d s (t) and I DRE (t, τ) = B DRE (t, τ)dr s (t) where B T DE (t, τ) and B DRE (t, τ) are proportional factors of the dose and the dose rate respectively. If one do the normalization with DR s (t) the following equation can be obtained: I D (t, τ) DR s (t) = I off(t, τ) + B T DE (t, τ)d s (t) + B DRE (t, τ). (5.4) DR s (t) Substitute equation (5.3) to equation (5.4) and approximate the exponentials with Taylor series, keeping only the linear terms (exp( t/t 1 ) 1 t/t 1 and exp( t/t 2 ) 1 t/t 2 ): I D (t, τ) DR s (t) = I off(t, τ) + B T DE (t, τ)(dr 0,s + A 1,s + A 2,s )t + B DRE (t, τ). (5.5) DR s (t) Let us introduce a new notation a(t, τ) = I off (t,τ) DR s(t) +B DRE(t, τ) and b(t, τ) = B T DE (t, τ)(dr 0,s + A 1,s + A 2,s ) With these notations, the obtained equation is the following: I D (t, τ) DR s (t) = a(t, τ) + b(t, τ)t (5.6) If we assume that a(t, τ) and b(t, τ) are not functions of t the obtained from of equation (5.6) will be a simple a(τ) + b(τ)t type linear function. This assumption also means that I off (t, τ), B T DE (t, τ) and B DRE (t, τ) are also independent of t. It is very important that this derivation predicts a not necessarily zero value for a(τ). 51

54 5.3. EVALUATION OF THE FRAMES I D (t,τ)/d R s (t) (d ig it/m G y /h ) 0, , , , , ,0 1 0 a m s 1 0 m s 1 m s µs 5 0 µs 2 0 µs 0, ,8 1,0 1,2 1,4 1,6 1,8 2,0 T im e (h ) 0, , ,0 4 0 b m s 1 0 m s 1 m s µs 5 0 µs 2 0 µs I D (t,τ)/d s (t) (d ig it/m G y ) 0, , , , , , , , ,6 0,8 1,0 1,2 1,4 1,6 1,8 2,0 2,2 T im e (h ) Figure 5.7: (a)i D (t, τ)/dr(t) behavior during the irradiation. (b) I D (t, τ)/d(t) behavior during the irradiation. 52

55 5.4. RESULTS AND DISCUSSION I D (t, τ) normalization by the dose Consider equation (5.2) and normalize it with D s (t) I D (t, τ) D s (t) = I off(t, τ) + B DRE (t, τ)dr s (t) D s (t) + B T DE (t, τ). (5.7) Let us introduce a new notation: C 1 (t, τ) = I off (t,τ) D s(t) than: I D (t, τ) D s (t) + B T DE (t, τ) and C 2 (t, τ) = B DRE(t,τ) D s(t) = C 1 (t, τ) + C 2 (t, τ)dr s (t). (5.8) If C 1 (t, τ) and C 2 (t, τ) are not functions of time, we get that I D(t,τ) D s(t) follows the time evolution of DR s (t). It is very important that this derivation predicts a not necessarily zero value for C 1 (τ). 5.4 Results and discussion In this section the results obtained from the evaluation will be presented. The first result is that a decomposition of I D (t, τ) as seen in equation (5.2) works, which means the TDE and the DRE can be handled separately Total Dose Effect Total Dose Effect as a function of time First, consider the TDE as a function of time. For this, the dose rate normalized I D (t, τ) will be used as a function of time. During the irradiation, I D (t, τ) shows linear increase (see Figure 5.7a) hence the following equation can be fitted: I D (t, τ)/dr s (t) = a(τ) + b(τ)t, (5.9) where t is the time in hours. The obtained parameters of the fitted functions are in Table 5.1. The parameters of the linear curves with the smaller three exposure times (20 µs, 50 µs, 100 µs) show similar values while in the case of longer exposure times (1 ms, 10 ms, 100 ms) there is an increase in parameters a(τ) and b(τ). The values of a(τ) are around zero. Total Dose Effect as a function of dose Another representation is the dose rate normalized I D (t, τ) as a function of dose. This is shown in Figure 5.8a. It means that with growing dose the frames become brighter and brighter with increasing slope. After a dose of one year of operation in W7-X, I D (t, τ)/dr s (t) increases approximately by a factor of , depending on the applied 53

56 5.4. RESULTS AND DISCUSSION τ a(τ) (digit/mgy/h) a(τ)/a(τ) b(τ) (digit/mgy) b(τ)/b(τ) 100 ms 0, , , , ms 0, , , , ms 0, , , , µs 0, , , , µs 0, , , , µs 0, , , ,26878 Table 5.1: Values of a(τ) and b(τ) fitted parameters during irradiation, denotes the standard deviation exposure time. Accordingly, it is necessary to operate EDICAM in small dose rate level gamma fields to avoid large increase of I D (t, τ) during one year of operation. A probable explanation for TDE is the leakage current increase[71] in the photodiode of the pixels or the gate voltage shift of the pixel transistors, which handle the pixel read out sequence. This gate voltage shift of transistors is the function of the gamma dose [46]. The sensor chip is based on 6T active pixel technology hence only a resultant behavior can be observed Dose Rate Effect Dose rate Effect as a function of time To determine the dose rate connected effects of I D (t, τ), the function I D (t, τ)/d s (t) should be examined (see Figure 5.7b). On every exposure time we found an exponential decay but with some amplitude increase for the larger (100 ms, 10 ms, 1 ms) exposure times. It is interesting that it seems that I D (t, τ)/d s (t)time evolution is delimited with two curves. It seems that the curves with 10 ms and 100 ms exposure times shifted to a common curve which is the upper limiting curve and the curves with 20 µs, 50 µs and 100 µs exposure times shifted to another common curve, to the lower limiting common curve. The overall behavior is similar to the time evolution of dose rate as it was predicted in section 5.3.1, although it cannot be declared that these curves have the same time constants as we experienced at the dose rate measurements because of the small number of measured points. DRE as a function of Dose Rate Another representation is the dose normalized I D (t, τ) as a function of dose rate, which is shown in Figure 5.8b (consider that the dose rate scale is decreasing) where one can see similar behavior as seen above in Figure 5.7b. Other effects such as the Low Dose Rate 54

57 5.4. RESULTS AND DISCUSSION I D (t,τ)/d R s (t) (d ig it/m G y /h ) 0, , , , , ,0 1 0 a m s 1 0 m s 1 m s µs 5 0 µs 2 0 µs 0, D s (t) (m G y ) 0,0 5 0,0 4 b m s 1 0 m s 1 m s µs 5 0 µs 2 0 µs I D (t,τ)/d s (t) (d ig it/m G y ) 0,0 3 0,0 2 0,0 1 0, D R s (t) (m G y /h ) Figure 5.8: (a)i D (t, τ)/dr s (t) in the function of dose during the irradiation. The dose refers to the dose absorbed by EDICAM. (b)i D (t, τ)/d s (t) as a function of dose rate in the position of EDICAM during the irradiation; consider that the dose rate scale is decreasing. 55

58 5.5. CONCLUSIONS Effect (LDRE)[81] in the transistors of the pixels are probably present but we cannot separate them. 5.5 Conclusions The EDICAM device was irradiated in the pure gamma field in the Training Reactor of BME up to 23.6 ± 11% Gy dose with up to 27 Gy/h dose rates. EDICAM withstood the irradiation. The average brightness of the frame series (I D (t, τ)) were examined during the irradiation. Three main effects influence the behavior of I D (t, τ) of EDICAM during irradiation. The first effect is a nonlinear exposure time dependence of I D (t, τ) which can be tracked back to the applied dose rate levels. The current dose rate on small exposure times (20 µs, 50 µs, 100 µs) cannot generate enough electrons in the pixels to increase significantly I D (t, τ). The electron generation becomes significant after a certain threshold exposure time τ threshold in the pixel sensitive volume. The other two effects are the Total Dose Effect and the Dose Rate Effect. A simple model was presented to describe the behavior of I D (t, τ). The predictions of the applied model, in accordance with the results of the measurement showed that I D (t, τ) can be normalized by the dose rate and the dose, respectively. Accordingly, the dose and dose rate dependent effects can be handled separately. I D (t, τ)/dr s (t) grows linearly, and I D (t, τ)/d s (t) roughly follows the time evolution of the gamma dose rate. I D (t, τ)/dr s (t) grows by a factor of due to the gamma dose of a full year operation in W7-X. From the results (see table 5.1) of the test we can extrapolate and give the operational limits of gamma dose and dose rate. It can be concluded that after approximately 11 years of operation, which means 200 Gy, EDICAM will lose the half of its dynamic range on longer exposure times ( ms) which means practically inoperability. From this dose limit, the obtained prompt dose rate limit is 200 Gy/h. At this point we should emphasize that neutron irradiation could result shorter operation time limits. 56

59 Chapter 6 Cameras of the BES system on KSTAR 6.1 Introduction After a trial system [57] a new BES system was designed and installed in the cooperation of Wigner RCP and BME on the KSTAR tokamak in The BES system measures plasma turbulence [56], by detecting light, emitted by the neutral heating beams. Neutral beam injection systems (NBI, Li beam) inject fast atoms into the plasma which are excited by the plasma particles and emit photons with a characteristic frequency. Variations in the intensity of light carry information on the local plasma density fluctuations. The BES system was installed in the M-port of the tokamak which is situated close to the NBI box [58]. It contains a periscope which leads the light out of the plasma to two different cameras (see Figure 6.1). On one end of the light path an ADIMTECH APDCAM camera is located which contains a Hamamatsu S8550 4x8 channel Avalanche Photo Diode (APD) array, on the other end of the light path a PCO Pixelfly VGA camera is installed which contains a Pixelfly ICX414AL CCD sensor. The APDCAM is used for fluctuation measurements and the PCO Pixelfly VGA camera for spatial calibrations. As the cameras are located in different positions, they will suffer from different radiation loads which result in different neutron and gamma-ray spectra and fluences. Only these two radiation sources should be taken into account because the charged particles will be absorbed on the wall of the vacuum vessel and on the structural elements. The main source of the neutrons is the D + D = 3 He + n reaction which produces high energy (2.45 MeV) neutrons. Less frequently D + T = 4 He + n reactions can be observed which produces neutrons with higher energy (14.06 MeV) [82]. These reactions take place in the plasma volume. As a second source, D + D reactions will also occur in the ion dump of the NBI which is situated quite close to the BES system. The main source of the gamma-ray radiation is mostly the (n, γ) reaction as well as the activated atoms 57

60 6.2. MCNP CALCULATIONS radioactive decay emanating from the structural components of the tokamak. The neutron and gamma-ray radiation are important because they can penetrate through matter and easily reach the components of BES. Charged particles such as electrons or ions of the plasma could not leave the vacuum chamber hence they could not have contributed to the radiation damage of the BES system. The number of secondary charged particles which are born due to the interaction of primary radiation (neutrons or gamma-rays) with matter close to- or inside the BES, could be suppressed by reducing the intensity of the primary radiation. To calculate the actual neutron fluence and gamma-ray dose as well as their spectra in the camera positions, I used the Monte Carlo N-Particle extended (MCNPX) code. The MCNPX model was built by MCAM [53] developed by FDS Team, China. Unfortunately, now only calculations are available to predict the radiation fields as there are no neutron and gamma-ray detectors installed around the BES system. These calculations estimated sufficient radiation levels to damage the cameras. In the present experiment on the KSTAR we have been faced only with the white pixel generation in the CCD camera. One of the possible damages caused in the CCD cameras is the white pixel generation due to high energy neutrons [83][84][44]. This mechanism can be explained as a thermal noise effect or extreme increase of dark current in the pixels by the thermal hopping of electrons. The dark current of the pixels is the largest in regions where large numbers of deep level defects exist [83]. A typical area with a large number of mid-gap levels is the surface area but as neutron irradiation can cause displacement damage, a large number of new deep level states will also appear in the bulk material of the pixels [83]. Sometimes the white pixels refer to clusters of defects. In the APDCAM there is an increased noise level owing to radiation but this is not a permanent effect and in this work we only focus on the CCD camera radiation damage. After the campaign of the KSTAR, the CCD camera was removed from the BES system and transported back to BME NTI for further experiments. These experiments aimed at exploring the origin of the white pixel generation and at deciding whether neutron or gamma-ray radiation shielding would be necessary for the CCD camera in the future. In these experiments we irradiated the Pixelfly camera with pure gamma-ray radiation. Before, during, and after irradiation numerous frames were taken without light input with 5 s long exposure time. We measured the change of the number of white pixels in the CCD camera in thermal equilibrium up kept by a furnace. 6.2 MCNP Calculations For calculations MCNPX version 2.7a was used with ENDF 7.0 cross section libraries. For building the MCNP model, MCAM 4.8 beta standard edition was used and as an input I used the mechanical engineering design which were devised in Computer Aided 58

61 6.2. MCNP CALCULATIONS ADIMTECH APDCAM Periscope of the BES PCO Pixelfly VGA camera Figure 6.1: BES system of the KSTAR tokamak in the laboratory before installation. Clearly visible are the PCO Pixelfly VGA camera, the ADIMTECH APDCAM and the periscope part. 59

62 6.2. MCNP CALCULATIONS Three Dimensional Interactive Application (CATIA). In this section we discuss the MCNP model and the results of calculations MCNP model The MCNP model of the BES system and the KSTAR tokamak were built with some simplifications. The main parts of the model are the following: plasma volume, vacuum vessel, poloidal magnets, NBI-port, cryostat, BES system main parts, and a schematic NBI-box with an Ion dump (see Figure 6.2). The vessel and the cryostat were made out of SS 316LN steel, the poloidal field coils made of NbSn 3 and NbT i [82] the aluminum structural parts of the BES made out of AlMgSi To estimate the fluxes and doses in Plasma volume Cryostat NBI box BES system Ion Dump Poloidal magnets CCD camera NBI-port Figure 6.2: Exploded view of the MCAM model including the main parts: plasma volume, vacuum vessel, poloidal magnets, NBI-port, cryostat, BES system main parts, and a schematic NBI and Ion Dump. reality, the output of the MCNP calculations have to be multiplied by the source intensity. The maximum neutron source intensity of the plasma is n/year [82]. If we consider the ion dump as a neutron source, the situation is more complicated because no direct neutron yield measurements of NBI were available. I used the measurement results of EAST tokamak which has a similar NBI system: neutron yield is Y = n/s/a for one time unit and NBI current [85]. The maximal NBI current is: I = 50 A. I assume 60

63 6.3. WHITE PIXELS that in one year the KSTAR produces thousands of 20 s long pulses, consequently the maximum neutron source intensity is n/year Results of MCNP calculations The main results were the neutron and gamma-ray spectra, the total neutron fluence and the gamma-ray dose in the CCD camera position. The obtained neutron spectrum is shown in Figure 6.3a and the obtained gamma-ray spectrum is shown in Figure 6.3b. The calculated yearly gamma-ray dose is 3.17 mgy± 0.33 mgy and the total yearly neutron fluence is ± /cm 2 from the plasma volume. Between and MeV I used only one wide energy bin (see Figure 6.3a.). It was easier to calculate an average without unfolding the structure of the resonances in this region, which also reduced the computation time without having any influence on the final results. While comparing the spectra, the relative changes play important role and I used the same bin structure as opposed to the detailed shape which is due to resonances. Here we note that geometrical simplifications could have a more significant effect on the spectra. The calculations showed that the total neutron fluence from the ion dump of the NBI is /cm 2 ± 9, /cm 2 which is approximately one order of magnitude smaller than that of the plasma volume. The shape of the obtained neutron spectrum is similar to that of the spectrum introduced in Figure 6.3a, except that it does not contain neutrons above 2.45 MeV owing the fact that in the NBI only D-D reactions take place. The shape of the obtained gamma-ray spectrum is similar to that of the spectrum introduced in Figure 6.3b. The gamma-ray dose is Gy± Gy which is approximately one order of magnitude smaller than that of the plasma volume. 6.3 White pixels During the 2012 campaign of KSTAR, we experienced a continuously increasing number of white pixels in the PCO Pixelfly cameral. This phenomenon led us to start counting the number of white pixels. We made frames with a 10 s exposure time after the plasma shots. These frames were not dark current images as there was no mechanical shutter before the CCD camera. This signifies that direct or scattered light reached the camera during these exposures. To reduce the effect of the scattered, visible light, we counted the white pixels only in the corners (50*50 pixel regions) of these frames. The reason for this was that the aperture in front of the CCD led the light to the middle of the sensor in a circular shape and the corners mostly stayed in dark (see Figure 6.4). The white pixel definition in these measurements is the following: pixels that reach the maximum brightness (4095 digits) during the exposure are white pixels. The number of white pixels as a function of shot number is shown in Figure 6.5. In Figure 6.5 around shot numbers 61

64 WHITE PIXELS /y e a r/m e V ) E n e rg y b in n o rm e d n e u tro n flu e n c e a E n e rg y b in n o rm e d n e u tro n flu e n c e (1 /c m N e u tro n e n e rg y (M e V ) /y e a r/m e V ) E n e rg y b in n o rm e d p h o to n flu e n c e b E n e rg y b in n o rm e d p h o to n flu e n c e (1 /c m P h o to n e n e rg y (M e V ) Figure 6.3: (a) Calculated neutron flux normed by the energy bins in the CCD camera position, when the plasma is the source. We have applied normalization because the different width of the energy bins, another option could be to represent the spectra as a function of lethargy. Some energy bins have very small or zero values (because of the narrow energy bins) which cause a big vertical jump due to the logarithm scale. (b) Calculated photon flux normed by the energy bins in the CCD camera position (Si), when the plasma is the source. 62

65 6.3. WHITE PIXELS 7300 and 7700 the number of white pixels decreases. We think this can be attributed to a smaller camera head temperature. The BES system operates at room temperatures on KSTAR, but the temperature of the camera head has not been measured directly. As we can observe in the next chapter, the number of white pixels strongly depends on the camera head temperature, which is in a good agreement with the thermal hopping theory. The number of white pixels grows monotonically which confirms that number of 50 pixel 50 pixel Figure 6.4: A typical image of the CCD camera with long exposure time. The camera was not blindfolded therefore the middle region of the camera reached the maximum brightness level. For the demonstration of the white pixel generation during the campaign, only the corner regions were taken into account (see Figure 6.5). Their area was 2500 pixels each. the white pixels should be proportional to the absorbed gamma-ray dose or the neutron fluence. To determine the main source and get experimental evidence, the easiest way was to irradiate the camera in a controlled radiation environment. 63

66 6.3. WHITE PIXELS B o tto m le ft c o rn e r T o p le ft c o rn e r T o p rig h t c o rn e r B o tto m rig h t c o rn e r R a tio o f w h ite p ix e ls in th e c o rn e rs (% ) S h o t n u m b e r Figure 6.5: Ratio of white pixels in the corners of the CCD camera during the campaign on KSTAR in The area of the corners was 2500 pixels each, which was sometimes reached due to over exposure. This behavior occurs because the camera was not blindfolded, and the illumination conditions of the camera head varied or the aperture was removed. 64

67 Sensitive volume Detector holder Power supply cable AMP-200 Sensor position PCO Pixelfly VGA UTP cable 6.4. MEASUREMENTS AT BME NTI 6.4 Measurements at BME NTI Gamma-ray irradiation The PCO Pixelfly camera was irradiated in a dry, vertical irradiation channel of the Training Reactor of BME. That camera was in operation on KSTAR in the 2012 campaign. This chosen irradiation channel had already been used for radiation damage measurements [86]. The PCO Pixelfly camera and the AMP-200 detector [78] were installed Figure 6.6: A schematic figure of the detector holder with the AMP-200 detector and the PCO Pixelfly camera. The CCD chip and the sensitive volume of the AMP-200 detector were at the same distance from the reactor core. on an aluminum detector holder (Figure 6.6). The irradiation was a pure gamma-ray irradiation without neutrons, thus we could avoid the activation of the camera. This was achieved by operating the reactor at a power of 100 kw for 2 h after which the reactor was shut down. The residual gamma-ray radiation of the core produced that pure gamma-ray radiation field. For the camera irradiation the whole instrumentation set-up was lowered in the irradiation channel after the reactor shutdown. The camera was connected to a measuring computer and numerous frame series were taken during irradiation with exposure times of 5 s for each frame. The sensor chip was blindfolded to minimize the amount of visible light reaching it. Since these frames were dark current images, we could use the whole frame area and not just the corners of the frames. Each frame series were saved in a TIFF file format and contained 402 frames, each. The gamma-ray dose rate as a func- 65

68 6.4. MEASUREMENTS AT BME NTI tion of time measured by the AMP-200 during irradiation is shown in Figure 6.7a. The integrated decay curve displays the gamma-ray dose of AMP-200 (see Figure 6.7b). The total gamma-ray dose of AMP-200 was 1.7 Gy±0.17 Gy which is a good approximation of the gamma-ray dose of CCD sensor too, because the sensor chip and the sensitive volume a b G a m m a d o s e ra te (m G y /h ) G a m m a d o s e o f A M P (m G y ) ,4 1 5,6 1 5,8 1 6,0 1 6,2 1 6,4 1 6,6 1 6,8 1 7,0 1 7,2 T im e (h ) 1 5,4 1 5,6 1 5,8 1 6,0 1 6,2 1 6,4 1 6,6 1 6,8 1 7,0 1 7,2 T im e (h ) Figure 6.7: (a) Gamma-ray dose rate as a function of time measured by AMP-200 detector in the given irradiation position. (b) Gamma-ray dose of the AMP-200 detector during irradiation as a function of time. of the AMP-200 detector was at the same distance from the reactor core. Comparing the results of the MCNP calculations and the applied gamma-ray dose we can conclude that we have irradiated the camera with a gamma-ray dose that is higher than the expected yearly gamma-ray dose in a KSTAR campaign by 3 orders of magnitude Measurements at thermal equilibrium As we mentioned above, numerous dark current frames were taken before and after the irradiation at thermal equilibrium at 37±0.5 C. The goal was to detect the change in the number of white pixels due to the absorbed gamma-ray dose. The thermal equilibrium was necessary because the charge generation in white pixels is mainly a thermal effect in the thermal hopping process of the electrons. We had to compare the number of white pixels of the camera on the same temperature. To maintain the thermal equilibrium, we put the camera into a Nabertherm controller B 180 furnace and operated the camera until it reached the thermal equilibrium. During these measurements numerous dark current frames were taken with 5 s exposure times. For temperature monitoring two independent systems were used. The first system was the temperature measurement system of the furnace which measured the temperature of the air around the camera. The second system was a small thermocouple installed in a hole near the sensor chip of the camera which was connected to a hand-held Metex multimeter (see Figure 6.8). The accuracy of the 66

69 6.4. MEASUREMENTS AT BME NTI thermal measurements were 0.5 C Evaluation of the recorded frames The number of the white pixels before and after the irradiation showed a similar behavior see Figure 6.9a-6.9b. The number of white pixels follows the trend of the camera head temperature which is in a good agreement with the thermal hopping theory. Here we note that this camera is a temperature compensated camera but it seems that the camera could not handle the radiation induced thermal hopping. This result also confirms that we should make the measurement in thermal equilibrium because as the camera head temperature saturates, the number of white pixels will also saturates. During the measurement the temperature of the furnace as an environment also slowly increased by 2 3 C. It is easy to recognize a transient behavior in the number of white pixels at the beginning of each frame series both before and after irradiation. During this transient the number of white pixels exceeds the excepted trend. This overshot could be a sign of the temperature compensation. For further calculations we omitted the first 50 frames from every frame series. To quantify the behavior of the number of white pixels, saturating exponential functions were fitted. The fitted function is the following: N white (t) = N sat Aexp( (t t 0 )/t 1 ) (6.1) where N white denotes the number of white pixels, t denotes the time, t 0 denotes the starting time of the measurement, t 1 denotes the time constant of the exponential function, A denotes the amplitude of the exponential function and N sat denotes the saturation level of the white pixels. The obtained fitted curves and parameters are shown in Figure 6.10a- 6.10b. If we take into account that the temperature measurements also suffer from a measuring uncertainty of 0.5 C which translate into an additional relative error margin of 1.3% for the saturation level, we can conclude that the number of white pixels remained at the same level. This result signifies that a gamma-ray dose of this magnitude does not damage the PCO Pixelfly VGA CCD camera and does not increase the number of white pixels. The importance of this result lies in the fact that only neutron shielding is necessary to avoid the white pixel generation in the camera. This result also renders it possible to examine the number of white pixels as the function of neutron fluence (see Figure 6.11). For the fluence calculations I integrated the average flux over time when the plasma current was greater than 100 ka. This representation also shows monotone (linear) growth in the number of white pixels except those regions where we assumed lower operational temperatures (around shot numbers 7300 and 7700 and /cm 2 and /cm 2 in fluence respectively). If we assume a linear connection between the neutron fluence and the white pixel generation, we can estimate the cross section of the white pixel generation (the average fluence which is 67

70 6.4. MEASUREMENTS AT BME NTI PCO Pixelfly VGA CCD camera METEX multimeter for camera head temperature monitoring Nabertherm contrlorrel B 180 UTP and thermocouple connections Figure 6.8: PCO Pixelfly CCD camera position in the Nabertherm controller B 180 furnace before and after irradiations with thermocouple connection to Metex multimeter and UTP cable connection for measuring computer. 68

71 6.4. MEASUREMENTS AT BME NTI T e m p e ra tu re ( C ) 3 0 N u m b e r o f h o t p ix e ls ,0 8,5 9,0 9,5 1 0,0 1 0,5 T im e ( h ) 1 1, ,5 0 b N u m b e r o f h o t p ix e ls a o 3 8 T e m p e ra tu re ( C ) ,0 9,5 1 0,0 T im e ( h ) 1 0,5 1 1,0 1 1,5 Figure 6.9: (a) Number of white pixels, the camera head and furnace temperatures as a function of time before gamma-ray irradiation. Black denotes the camera head temperature, red denotes the furnace temperature and blue denotes the number of white pixels. (b) Number of white pixels and the camera and furnace temperatures as a function of time after gamma-ray irradiation. Black denotes the camera head temperature, red denotes the furnace temperature and blue denotes the number of white pixels a N u m b e r o f w h ite p ix e ls N u m b e r o f w h ite p ix e ls E q u a t i o n : N w h i t e ( t ) = N s a t - A * e x p (- (t - t 0 )/ t 1 ) W e ig h tin g : y In s tru m e n ta l C h i^ 2 /D o F = R ^ 2 = N s a t A t t0 ±2 ±2 ± F IX ,2 0,4 0,6 0,8 1,0 T im e ( h ) E q u a t i o n : N w h i t e ( t ) = N s a t - A * e x p (- (t - t 0 )/ t 1 ) W e ig h tin g : y In s tru m e n ta l C h i^ 2 /D o F = R ^ 2 = N s a t A t t ,0 b ,2 1,4 1,6 0,0 1,8 0,5 1,0 T im e ( h ) ,5 2,0 ±1 ±2 ± F IX 2,5 Figure 6.10: (a) Number of white pixels and the fitted exponential curve as a function of time before irradiation with the obtained parameters. (b) Number of white pixels and the fitted exponential curve as a function of time after irradiation with the obtained parameters. 69

72 6.5. RADIATION DAMAGE PREVENTION needed to generate one white pixel defect) from the slope of the linear. The obtained slope is cm 2 ± cm 2. Here we should note that this slope is valid only for the same neutron spectrum B o tto m le ft c o rn e r T o p le ft c o rn e r T o p rig h t c o rn e r B o tto m rig h t c o rn e r R a tio o f w h ite p ix e ls in th e c o rn e rs (% ) ,0 2,0 x ,0 x ,0 x ,0 x ,0 x ,2 x ,4 x ,6 x N e u tro n flu e n c e (1 /c m 2 ) Figure 6.11: Number of white pixels in the corners of the CCD camera as the function of neutron fluence. 6.5 Radiation damage prevention We have three obvious solutions to avoid the previously experienced radiation damage of the calibration camera of the BES system. The first solution is to design a radiation shielding around the camera position which prevents neutron radiation damage in the camera. The second solution is to replace the camera with one more tolerant to radiation. A great amount of inspiration can be derived from space missions and the high energy physics experiments for the radiation hardening of CCD and CMOS detectors. These researches make big efforts to study and improve the radiation hardness of CCD and CMOS sensors [87] [45] [88]. In modern magnetized plasma physics experiments the video diagnostics systems require high resolution and fast or ultra-fast imaging cameras and 70

73 6.5. RADIATION DAMAGE PREVENTION this demand is now fulfilled with commercial cameras. The third solution is the combination of the previous two solutions by building a shielding and replacing the camera. We have chosen the third option and have replaced the CCD camera with a Photonfocus CMOS camera and have also built a shielding around it. One way to determine the effects of the shielding is to compare the neutron spectra, another is to calculate the hardness factors (κ) with and without shielding. The obtained hardness factor without shielding is κ = when the plasma was the source and κ = when the NBI ion dump was the neutron source. The obtained 1 MeV equivalent neutron fluence without shielding from the plasma is Φ eq,plasma = /cm 2, from the NBI is Φ eq,nbi = /cm Recent neutron shielding A new multilayer neutron shielding has been built around the calibrating camera position. After having constructed the shielding, MCNP calculations were carried out to calculate the new neutron and photon spectrum as well as the gamma-ray dose and hardness factors in the camera position. The schematic view of the shielding is shown on Figure The outermost layer is made of plastic (Dehoplast PE55) with a thickness of 2.5 cm, than 0.5 cm boron containing plastic layer (Mirrobor 50-80%) stands, than 1.5 cm Dehoplast PE55, than 0.6 cm Mirrobor 50-80% and the most inner layer is 0.5 cm lead. The CMOS camera is modeled as pure Si. Plasma source The calculations showed that the shielding decreased the total neutron fluence to /cm 2 ± /cm 2 which equaling a 19% decrease. The obtained neutron and gamma-ray spectrum is shown in Figure 6.13a-b. The gamma-ray dose increased to 12.2 mgy±0.77 mgy. The hardness factor with the current shielding is κ = with the plasma as the source. The equivalent neutron fluence is Φ eq,shield,plasma = /cm 2 which means a drop of 31%. NBI source The calculations showed that the shielding reduced the total neutron fluence to /cm 2 ± /cm 2 which means an approximately 80% decrease. The obtained gamma-ray dose is 0.30 mgy±0.02 mgy. The obtained neutron and gamma-ray spectrum is shown in Figure 6.14a-b. The hardness factor with the current shielding is κ = with the NBI as the source. The equivalent neutron fluence is Φ eq,shield,nbi = /cm 2 which means a decrease of 97%. 71

74 RADIATION DAMAGE PREVENTION Dehoplast PE55 (Blue) CMOS Camera (Orange) Mirrobor 50-80% (Red) Cryostat Lead (Purple) Glasses-lenses Aluminium structural elements Figure 6.12: Schematic view of the neutron shielding around the CMOS camera. Other parts of the model stayed unchanged. /M e V ) E n e rg y b in n o rm e d n e u tro n flu e n c e (1 /c m E n e rg y b in n o rm e d n e u tro n flu e n c e a /M e V ) E n e rg y b in n o rm e d p h o to n flu e n c e (1 /c m E n e rg y b in n o rm e d p h o to n flu e n c e b E n e rg y (M e V ) E n e rg y (M e V ) Figure 6.13: (a) Calculated neutron flux normed by the energy bins in the CMOS camera position, when the plasma is the source. (b) Calculated photon flux normed by the energy bins in the CMOS camera position (Si), when the plasma is the source. 72

75 RESULTS AND CONCLUSIONS /M e V ) E n e rg y b in n o rm e d n e u tro n flu e n c e (1 /c m E n e rg y b in n o rm e d n e u tro n flu e n c e a /M e V ) E n e rg y b in n o rm e d p h o to n flu e n c e (1 /c m E n e rg y b in n o rm e d p h o to n flu e n c e b E n e rg y (M e V ) E n e rg y (M e V ) Figure 6.14: (a) Calculated neutron flux normed by the energy bins in the CMOS camera position, when the NBI is the source. (b) Calculated photon flux normed by the energy bins in the CMOS camera position (Si), when the NBI is the source. 6.6 Results and Conclusions During the campaign of KSTAR in 2012, the number of white pixels in the PCO Pixelfly VGA CCD camera of the BES system has increased which pointed to radiation damage. An MCNP model was built with MCAM to calculate the total neutron fluence, spectrum, gamma-ray dose and spectrum in the camera position. The calculated yearly gamma-ray dose from the plasma is 3.17 mgy±0.33 mgy, the dose from the NBI is one order of magnitude smaller. The total yearly neutron fluence is ± /cm 2 from the plasma volume, from the NBI the fluence is one order of magnitude smaller. In our analysis we found new evidence supporting the electron thermal hopping theory regarding the radiation damage of the CCD camera when neutron induced displacement damage creates new energy states in the bulk material of the pixels thus enabling electron thermal hopping. A gamma-ray irradiation test and temperature monitored measurements were carried out at BME NTI Training Reactor. The camera was irradiated with 1.7 ± 10% Gy, gamma-ray dose 3 orders of magnitude above the calculated KSTAR dose. These experiments showed that the number of white pixels remained constant despite of gamma-ray irradiation. We have found and given an experimental confirmation to the theory that the root cause of the white pixel generation in the PCO Pixelfly VGA CCD camera is the neutron radiation. We have determined the efficiency of the white pixel generation for neutrons at cm 2 ± cm 2 which offers the possibility to predict the failure frequency of pixels due to neutron irradiation. As a possible solution we advised to design a neutron shielding around the camera or replace it with one more resistant to radiation. The recently applied shielding reduces the 1 MeV equivalent fluence by approximately 30% which indicates 30% of longer use of the CCD camera. With the new 73

76 6.6. RESULTS AND CONCLUSIONS CMOS camera we did not experience radiation damage which indicates a lower radiation sensitiveness of the currently used CMOS camera rather than the effectiveness of the shielding. It is important to note that the gamma-ray dose increased by a factor of 4 due to (n, γ) reactions. 74

77 Chapter 7 Shielding considerations and the BES systems of EAST 7.1 Introduction Shielding is necessary to reduce the displacement damage in silicon electronic devices. In this chapter three solutions for neutron shielding are compared, applying three typical shielding materials: polyethylene, polyethylene with boron content and polyethylene with lithium content. For the comparison MCNP [89] calculations were carried out. For the calculations a multilayer structure was used. The layers can be filled up one after the other with the above mentioned shielding materials. Fusion neutrons were started as a surface neutron source on the surface of the outermost shielding layer and the surface averaged neutron spectrum was calculated after the last filled shielding layer. From the neutron spectra the 1 MeV equivalent neutron fluences were calculated and compared as a function of thickness of the shielding materials. The values of the normalized damage function D(E) (see equation (2.22)) at 2.45 MeV and 14.1 MeV where the neutrons are born in D D and D T reactions are around 1.2 and 1.8, respectively. The shape of the D(E) function (see Figure 7.1) [22] will be essential to understand the behavior of the shielding. According to the ASTM E722 (2009) standard, the value of D(E = 1 MeV, n) is 95 MeVmb in pure silicon. In the following section the detailed MCNP model and calculations are presented. The third and fourth section cover the results of the calculations and a short discussion on them, respectively. Before the conclusions a simple example is shown for application possibilities which presents the shielding of the Beam Emission Spectroscopy diagnostic (BES) [90] system developed for the EAST tokamak [91]. 75

78 7.1. INTRODUCTION D (E )/(9 5 M e V m b ) ~ 3 o rd e rs E th e rm a l D (E )/(9 5 M e V m b ) R e g io n o f s lo w n e u tro n a b s o rb tio n S lo w in g o f n e u tro n s N e u tro n s k in e tic e n e rg y (M e V ) Figure 7.1: The normalized displacement function in silicon due to neutron bombardment [22]. 76

79 7.2. MONTE CARLO MODEL 7.2 Monte Carlo model Calculations were carried out by MCNP version The geometry was built by MCAM 5.2 professional version [53, 92]. For the calculations a multilayer structure was used. The layers can be filled up one after the other with the above mentioned shielding materials. Fusion neutrons were started as a surface neutron source on the surface of the outermost shielding layer and the surface averaged neutron spectrum was calculated after the last filled shielding layer. The thickness (height of the cylinder) of the layers is 1 cm each and the radius of the cylinders is 500 cm, the cylinder was a reflecting surface. This arrangement is an ideal 1D shielding problem. A surface source was defined with 2 monenergistic peaks at 2.45 MeV and 14.1 MeV, while their fraction of intensity was 97% and 3%, respectively, which approximates the neutron spectrum of a D D plasma [93]. The momentum of the started neutrons was perpendicular to the surface of the first layer in the first case and in the second case the direction of the momentum followed isotropic distribution in the forward directions. The initial neutron hardness factor of this neutron spectrum is about 1.28 and the initial 1 MeV equivalent neutron fluence per started neutron is /cm 2 for perpendicularly started neutrons and /cm 2 for neutrons started with the isotropic distribution. After the last filled layer a surface flux tally (F2) was used to determine the neutron spectrum and also to calculate the gamma flux. The energy bin structure of the F2 tallies were the same as used in Figure 7.1. For variance reduction only the quadratic increase of importance (IMP) cards were used which means that the statistical weight factors of the neutrons and photons were doubled when they entered in a more distant layer from the neutron source. g Three types of shielding materials were compared: pure polyethylene C 1 H 2 with a density of 0.92, polyethylene with 5 m/m% boron content with a density of 0.95 cm 3 polyethylene with 7.5 m/m% lithium content with a density of 1.06 g. The ENDF/Bcm 3 VII.0 cross section library [94] was used for every calculation. The natural isotope abundances were used for every material. Every error presented in the following sections refers to one standard deviation calculated by MCNP. g cm 3 and 7.3 Results The neutron spectra were calculated using different materials and different layer thicknesses. The obtained neutron spectra for 2 cm, 6 cm and 10 cm shielding thicknesses for the three different materials are shown on Figure 7.2 as an example. These spectra contain two peaks of the source neutrons and a third peak a thermalized (Maxwell-Boltzmann) neutron distribution around 10 7 MeV. The thicker the shielding the smaller the source peaks and the larger the thermalized region are. However, the large contributors to κ and Φ 1MeV,eq are the neutrons with an energy above of 10 2 MeV. 77

80 7.3. RESULTS N e u tro n s p e c tru m (1 /c m 2 M e V /s tra te d n e u tro n ) ,1 0,0 1 1 E -3 1 E -4 1 E -5 1 E -6 1 E -7 1 E -8 1 E -9 1 E c m p o ly e th y le n e 6 c m p o ly e th y le n e 1 0 c m p o ly e th y le n e 2 c m p o ly e th y le n e w ith b o ro n 6 c m p o ly e th y le n e w ith b o ro n 1 0 c m p o ly e th y le n e w ith b o ro n 2 c m p o ly e th y le n e w ith lith iu m 6 c m p o ly e th y le n e w ith lith iu m 1 0 c m p o ly e th y le n e w ith lith iu m 1 E E E E -9 1 E -8 1 E -7 1 E -6 1 E -5 1 E -4 1 E -3 0,0 1 0, E n e rg y (M e V ) Figure 7.2: Energy bin normed neutron spectrum after different shielding thicknesses. The uncertainties of the total flux is less than 1%, bins with 100% relative error were not displayed. The neutrons were started perpendicular to the shielding layers. 78

81 7.3. RESULTS N e u tro n s p e c tru m (1 /c m 2 M e V /s ta rte d n e u tro n ) ,1 0,0 1 1 E -3 1 E -4 1 E -5 1 E -6 1 E -7 1 E -8 1 E -9 1 E c m p o ly e th y le n 6 c m p o ly e th y le n 1 0 c m p o ly e th y le n 2 c m p o ly e th y le n w ith b o ro n 6 c m p o ly e th y le n w ith b o ro n 1 0 c m p o ly e th y le n w ith b o ro n 2 c m p o ly e th y le n w ith lith iu m 6 c m p o ly e th y le n w ith lith iu m 1 0 c m p o ly e th y le n w ith lith iu m 1 E E E E -9 1 E -8 1 E -7 1 E -6 1 E -5 1 E -4 1 E -3 0,0 1 0, E n e rg y (M e V ) Figure 7.3: Energy bin normed neutron spectrum after different shielding thicknesses. The uncertainties of the total flux is less than 1%, bins with 100% relative error were not displayed. The neutrons were started following an isotropic direction distribution in the half space. 79

82 7.4. DISCUSSION Just under the D T peak the spectrum increases due to the increase in the number of small angle scatterings but these neutrons have smaller a D(E) value than the neutrons at 14.1 MeV, hence the Φ 1MeV,eq will also decrease. From the neutron spectra (see Figure ) the 1 MeV equivalent neutron fluences were calculated as shown in Table 7.1. Note that each fluence is normalized to one started neutron. Polyethylene Polyethylene with B Polyethylene with Li Φ 1MeV,eq [1/cm 2 ] Φ 1MeV,eq [1/cm 2 ] Φ 1MeV,eq [1/cm 2 ] perpendicular isotropic perpendicular isotropic perpendicular isotropic 1 cm 1.65E E E E E E-06 2 cm 1.60E E E E E E-06 3 cm 1.53E E E E E E-06 4 cm 1.44E E E E E E-06 5 cm 1.35E E E E E E-06 6 cm 1.25E E E E E E-06 7 cm 1.15E E E E E E-07 8 cm 1.05E E E E E E-07 9 cm 9.56E E E E E E cm 8.67E E E E E E-07 Table 7.1: The 1 MeV equivalent fluences as the function of the shielding thickness with the three different shielding materials. The values in the table are normalized to one started neutron. The uncertainties of the calculated values are less than 1%. The neutrons were started perpendicularly to the shielding or with isotropically distributed momentum in the forward directions. The uncertainties of the total flux values were less than 1%, however the NIEL scaling and the displacement function have their own restrictions [24] and uncertainties which could be much larger than the uncertainties of the MCNP spectra calculations. In the case of perpendicularly started neutrons the shielding is less effective. The results of the total gamma flux calculations can be seen at Figure 7.5 as the function of thickness of shielding with the two different source definitions. The typical photon spectra are shown in Figure Discussion The polyethylene provides practically and statistically the same shielding capabilities than the polyethylene with boron content. Above a thickness of 10 cm, the best shielding material is the polyethylene with lithium content. At that point the difference between the 80

83 DISCUSSION /s ta rte d n e u tro n ) N e u tro n flu e n c e (1 /c m 0, , , , , , , , , P o ly e th y le n (is o tro p ic ) P o ly e th y le n w ith b o ro n (is o tro p ic ) P o ly e th y le n w ith lith iu m (is o tro p ic ) P o ly e th y le n (p e rp e n d ic u la r) P o ly e th y le n w ith b o ro n (p e rp e n d ic u la r) P o ly e th y le n w ith lith iu m (p e rp e n d ic u la r) T h ic k n e s s o f s h ie ld in g (c m ) Figure 7.4: Total neutron fluences as a function of shielding thickness in the case of the three shielding materials with the two different source types. 81

84 DISCUSSION /s ta rte d n e u tro n ) 3,5 0 E ,0 0 E ,5 0 E ,0 0 E P o ly e th y le n (is o tro p ic ) P o ly e th y le n w ith b o ro n (is o tro p ic ) P o ly e th y le n w ith lith iu m (is o tro p ic ) P o ly e th y le n (p e rp e n d ic u la r) P o ly e th y le n w ith b o ro n (p e rp e n d ic u la r) P o ly e th y le n w ith lith iu m (p e rp e n d ic u la r) G a m m a flu e n c e (1 /c m 1,5 0 E ,0 0 E ,0 0 E ,0 0 E T h ic k n e s s o f s h ie ld in g (c m ) Figure 7.5: Total photon fluences as a function of shielding thickness in the case of the three shielding materials and with the two different source definition. 82

85 7.4. DISCUSSION P h o to n s p e c tru m (1 /c m 2 M e V /s ta rte d n e u tro n ) 1 E -3 1 E -4 1 E -5 1 E -6 1 E -7 1 E -8 1 E -9 1 E E E E E E E k e V 1 0 c m p o ly e th y le n 1 0 c m p o ly e th y le n w ith b o ro n 1 0 c m p o ly e th y le n w ith lith iu m 1 E -3 0,0 1 0, E n e rg y (M e V ) 8 M e V Figure 7.6: Photon spectra after 10 cm shielding with isotropically started neutrons in the forward directions. The minimal required energies for photon induced point and cluster defects are also denoted. Three peaks around 511 kev, 2.3 MeV and 4.5 MeV and their plateaus are clearly visible. 83

86 7.4. DISCUSSION simple polyethylene and polyethylene with lithium is less than 13%. Moreover, by using only 1 cm of additional polyethylene shielding, the lack of lithium can be compensated for. This is significant because the price of polyethylene shielding materials with boron or lithium additives is significantly higher [95, 96] than that of the simple polyethylene shielding. It is interesting to see the total neutron fluences, as it is possible to observe much larger differences (see Figure 7.4). For example there is a difference of about 20% between polyethylene and polyethylene with lithium content after 10 cm shielding with perpendicularly started neutrons and 25% for isotropically started neutrons. Moreover, a slight increase in the total fluence in the first few cm is also recognizable for perpendicularly started neutrons which occurs due to neutron production in (n, 2n) reactions. As expected the shielding slowed down the fusion born neutrons via scattering events due to the high hydrogen content. In other words, these neutrons disappeared from the high energy part of the neutron spectrum and appeared at the low energy region. At room temperature the average particle energy is around 25 mev, at this energy the value of D(E)/95 MeVmb is about 10 3 which indicates that even the number of thermal neutrons will increase due to thermalization, their contribution to the hardness factor are 3 orders of magnitude smaller than the contribution of the fusion born neutrons. Taking absorption processes into account will change the situation. The normally used neutron absorbers ( 10 B, 6 Li) are effective in the low energy regions. Their absorption cross section follows the 1/v law and reaches the maximum at lowest neutron energies. Applying these additional neutron absorbers will remove the neutrons from the lower energy region and change their contribution at D(E)/95 MeVmb from 10 3 to zero. However, the absorber atoms in the shielding material replace the effective scattering atoms, and hence decrease the slowing capabilities of the shielding [97]. In natural lithium the fraction of Li 7 is 92.5 %. The Li 7 isotope has a significant tritium production cross section in the high energy regions. The following two equations describe the ongoing neutron reactions with lithium: 6 3Li + n = 4 2 He H MeV (7.1) 7 3Li + n(fast) = 4 2 He H + n(slow) (7.2) The second process produces a slow neutron but as we saw the value of D(E)/95 MeVmb of this neutron is much smaller than of the initial one. Therefore Li is an effective neutron absorber, moderating material through the entire energy spectrum which results in the best shielding properties in this comparison. Another contributor to the results is the higher density of polyethylene with lithium content. With both source definitions the polyethylene with boron content resulted the largest photon flux, approximately three times more than the polyethylene and approximately 8 times more than the polyethylene with lithium content (see Figure 7.5). In every photon 84

87 7.5. RADIATION SHIELDING OF THE BES SYSTEM OF THE EAST TOKAMAK - AN EXAMPLE FOR APPLICATION spectrum three peaks can be seen above 255 kev (see Figure7.6) but above 8 MeV the spectrum is significantly smaller thus it can be concluded the neutron induced prompt gamma radiation will cause mainly point defects in the silicon targets after the shielding layers. Based on the above mentioned results a cheap candidate can be the simple polyethylene shielding which effectively reduces the 1 MeV equivalent neutron fluence in silicon targets such as electronic devices. Moreover, the photon fluence will be approximately only 2-3 times larger comparing to polyethylene with lithium content. If further improvement of the shielding is needed, the polyethylene can be replaced by polyethylene with lithium content Relationship with radiation weight factors In human dosimetry the analog function of D(E) is the radiation weight factor w r. This unitless factor describes the damage capabilities of different radiations. For example this factor is equal with 1 in the case of photons on all phonon energies. For neutrons according to the ICRP 2007 document the w r is a continuous function of neutron energy. The maximum of w r for neutrons is about 20 around 1 MeV and the minimum value of it is 2.5 on very small and on very large neutron energies (see Figure 7.7). It is obvious that it is not enough just to slow down the neutrons from 2.5 MeV or 14 MeV while they will have still large radiation weight factors ( 2.5 on thermal energies). In shielding problems if we would like to minimize the human doses a good strategy can be to use additional absorbers which can decrease the weight factors to zero or 1 (in case of radiative capture or activation). Based on these facts it is easier to understand why it is not necessary to apply absorber materials when the shielding should decrease the 1 MeV equivalent neutron flux especially in the case of thin shielding. 7.5 Radiation shielding of the BES system of the EAST tokamak - an example for application Two new Beam Emission Spectroscopy (BES) systems were installed at the EAST tokamak which contains 4 cameras. One of them observes visible line radiation from a diagnostic Lithium beam (Li-BES) while the other views a heating Deuterium beam (H- BES). Both of the diagnostics have an APDCAM (Avalanche PhotoDiode CAMera) and a CMOS camera from PhotonFocus. Around these cameras a development of a shielding is required to reduce the neutron induced displacement damage. The thickness of the shielding cannot exceed 6-10 cm depending on the locally available free space. Based on the above mentioned results for shielding material polyethylene Dehoplast PE

88 7.5. RADIATION SHIELDING OF THE BES SYSTEM OF THE EAST TOKAMAK - AN EXAMPLE FOR APPLICATION 2 2 E < 1 M e V 1 M e V < = E < = 5 0 M e V M e V < E 1 8 N e u tro n w e ig h t fa c to r , N e u tro n E n e rg y (M e V ) Figure 7.7: Radiation weight factor w r of neutrons. 86

89 7.5. RADIATION SHIELDING OF THE BES SYSTEM OF THE EAST TOKAMAK - AN EXAMPLE FOR APPLICATION was chosen, which does not contain any further neutron absorber materials like boron or lithium. Detailed MCNP calculations were carried out to calculate the difference between the average 1 MeV equivalent fluences in the camera positions with and without shielding. A detailed MCNP model of the two BES systems was built using MCAM including some parts of the EAST tokamak, such as the cryostat (Figures ). The model of the BES systems contain all optical elements (mirrors, lenses, and cameras), aluminum structural components as well as the shielding. APDCAM Photonfocus camera Figure 7.8: The cross section of the Li-BES system with the camera positions, the purple regions show the shielding layers, in which the cameras are located. The tokamak model itself is not sufficiently precise to directly calculate the neutron spectrum using the plasma volume as neutron source. Therefore, the source spectrum (Figure 7.11) is based on the spectrum published in the work of Qunying Huang et al. [98]. This neutron spectrum corresponds to a location 0.25 m from the outer wall of the cryostat in the midplane. Unfortunately, the angular distribution of the neutrons, which is most probably energy dependent, is not known D model with EAST relevant neutron spectrum First, the initial 1 MeV equivalent neutron fluence per started neutron was calculated from the spectrum: the obtained value was /cm 2 for isotropically started neutrons in the forward directions and /cm 2 for perpendicularly started neutrons, the hardness factor was for both case. Then, as a calculation of the 87

90 7.5. RADIATION SHIELDING OF THE BES SYSTEM OF THE EAST TOKAMAK - AN EXAMPLE FOR APPLICATION Photonfocus camera Figure 7.9: The cross section of the H-BES system with the Photonfocus camera, the purple regions are the shielding layers. APDCAM Figure 7.10: The cross section of the H-BES system with the Photonfocus camera, the purple regions are the shielding layers. 88