khaled bekhouche : measurement of charge transfer efficiency of a silicon particle detector based on...

Upload: salah-eddine-bekhouche

Post on 07-Apr-2018

222 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    1/88

    Examination committee:

    Full Name Title Quality University

    Dehimi Lakhdar Pr Chairman Biskra

    Sengouga Nouredine Pr Supervisor Biskra

    Saidane Abdelkader Pr Examiner Oran

    Debilou Abderrazak M.C.A Examiner Biskra

    Melaab Djamel M.C.A Examiner Batna

    by : Khaled BEKHOUCHE

    Thesis

    Submitted in fulfilment of the requirements of the degree of Doctor of Science in Electronics

    Entitled:

    Measurement of Charge Transfer Efficiency of a Silicon Particle

    Detector Based on a Charge-Coupled Device

    Democratic and Popular Republic of Algeria

    Ministry of Higher Education and Scientific Research

    Mohammed Khider University

    Faculty of Sciences and TechnologyDepartment of electrical engineering

    November 2010

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    2/88

    i

    Abstract

    The X-ray technique is used to measure the CTI in a CCD. The source is a 55Fe and the CCD

    is a T-type Column Parallel (CPC-T) which has 4 channels, 500 pixels each. The CTI is

    measured for one channel since they are identical. A LABVIEW software is used to acquire

    data via an ADC converter while a MATLAB code is used to analyse the measured data and

    extract the CTI. The CTI is calculated for two number of frames, 1000 and 10000. This

    allows the study of the effect of the statistical errors. Smaller statistical errors are obtained

    with 10000 frames. For an unirradiated CPC-T, the CTI is only a few 10 -5. An analytic model

    is used to fit the experimental results and found to be in good agreement when a low density

    of two electron traps located at 0.37 eV and 0.44 eV below the conduction band is considered.

    The measured noise has a parabolic-like shape which indicates that the trap distribution has

    the same shape too. Using a simple analytic model, the estimated occupancy is in good

    agreement with the measured one except at the edges of the CCD. It was found that for each

    clock voltage value there is an operating window where the CTI is low. Beyond this window,

    the CTI increases rapidly and the electrons can not reach the sense node. The lower is the

    clock voltage the lower is the energy consumption but the narrower is the operating window.

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    3/88

    ii

    Acknowledgements

    All my thanks to Allah my lord

    This work was carried out in the Laboratory Particle Physics of the University of Oxford.

    Many people have contributed to the work presented in this thesis. I would like to thank them

    for their help and support. First, I would like to thank my thesis advisor, Professor Nouredine

    Sengouga and my co-advisor Professor Andr Sopczak from Lancaster University for the

    guidance and encouragement. I would like to thank doctors Andrei Nomerotski and Rui Gao

    for the great help in carrying out the measurements. I would like to thank Professor LakhdarDehimi and Aoulmit Salim for their invaluable help throughout the study. I would like also to

    thank all LCFI members for their critical remarks and suggestions. Lastly but not least I

    would like express my sincere appreciations for the Ministry of Higher Education and

    Scientific Research and the University of Biskra for providing long term grant to Lancaster

    and Oxford University.

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    4/88

    iii

    CONTENTS

    Introduction .....................................................................................................................1

    Chapter I Particle Detectors..........................................................................................3

    I. 1 Radiations sources ......................................................................................................3

    I. 2 Radiation detectors......................................................................................................4

    I. 2. 1 Detector operation modes....................................................................................4

    I.2.1.1 Current operation mode..................................................................................5

    I.2.1.2 Mean Square Voltage operation mode ...........................................................5

    I.2.1.3 Pulse operation mode .....................................................................................5

    I.2.2 Pulse height spectra and energy resolution............................................................7

    I.2.4 Dead time.............................................................................................................10

    I.3 Radiation damage in particle detectors......................................................................11

    I.3.1 Surface damage....................................................................................................11

    I.3.2 Bulk damage........................................................................................................12

    I. 4Trapping and generation-recombination at deep levels ............................................15

    Chapter II Charge Coupled Devices..........................................................................19

    II. 1 CCD operation.........................................................................................................19

    II.1.1 Four-phase CCD ................................................................................................19

    II.1.2 Three-phase CCD ..............................................................................................20

    II.1.3 Two-phase CCD ................................................................................................21

    II. 2 CCD array architecture...........................................................................................21

    II.2.1 Linear arrays ......................................................................................................22

    II.2.2 Full frame array .................................................................................................22

    II.2.3 Frame transfer....................................................................................................24

    II.3 Readout time requirement for the vertex detector at the Future International Linear

    Collider (ILC)..................................................................................................................25

    II. 4 Metal-Oxide-Semiconductor capacitor theory ........................................................26

    II.4.1 P-MOS capacitor................................................................................................27

    II.4.2 N-P MOS capacitor............................................................................................29

    II.5 Charge generation.....................................................................................................32

    II.6 Charge collection......................................................................................................34II.7 Charge transfer .........................................................................................................35

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    5/88

    iv

    II.7.1 Self-induced drift ...............................................................................................37

    II.7.2 Fringing field drift .............................................................................................37

    II.7.3 Thermal diffusion ..............................................................................................38

    II.8 Charge measurement ................................................................................................38

    II.8.1 Charge conversion .............................................................................................38

    II.8.2 Correlated Double Sampling .............................................................................39

    II.9 Noise sources............................................................................................................40

    Chapter III Experimental setup and CTI calculation .............................................42

    III. 1 Experimental setup.................................................................................................42

    III.1.1 Hardware part ...................................................................................................45

    III.1.1.1 CPC-T motherboard...................................................................................46

    III.1.1.2 Temperature controller ..............................................................................47

    III.1.1.3 ICS-554 Analog-To-Digital Converter......................................................48

    III.1.2 Software part ....................................................................................................49

    III.1.2.1 Bias tab ......................................................................................................50

    III.1.2.2 Sequencer tab.............................................................................................51

    III.1.2.3 Amplifier tab..............................................................................................52

    III.1.2.4 DAQ system tab.........................................................................................52

    III.2 Charge Transfer Inefficiency Calculation............................................................55

    Chapter IV Charge Transfer Inefficiency results and discussion ..........................58

    IV. 1 Introduction............................................................................................................58

    IV. 2 Charge transfer inefficiency measurement ............................................................58

    IV.2.1 Low statistics....................................................................................................58

    IV.2.2 High statistics...................................................................................................61

    III.3 Charge Transfer Inefficiency analysis.....................................................................64

    III.4 Distribution of X-ray events....................................................................................66

    III.5 Noise effect .............................................................................................................68

    III. 6 OFFSET of the pedestal voltage and clock voltage effects ...................................69

    Conclusion and Outlook ..............................................................................................70

    Bibliography ...................................................................................................................72

    Appendix A .....................................................................................................................75

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    6/88

    1

    IntroductionOver the years, the technologies of optical imaging (still largely based on photographic films)

    and particle tracking (increasingly using electronic detectors such as spark chambers and

    multi-wire gaseous chambers) drifted apart. However, the invention of the Charge-Coupled

    Device (CCD) in 1970 started a revolution which is still having profound effects in the fields

    of optical imaging, particle tracking, X-ray detection, analog storage devices, etc [1-3].

    Charge coupled devices have been successfully used in several high-energy physics

    experiments over the last 20 years. Their small pixel size, excellent precision and high

    quantum efficiency (QE) over a wide wavelength range going from the X-rays to the far

    infrared [4] provide a superb tool for studying short-lived particles and understanding their

    nature at a fundamental level. Over the last few years the Linear Collider FlavourIdentification (LCFI) collaboration has developed Column-Parallel CCDs (CPCCD) and

    CMOS readout chips, to be used for the vertex detector at the International Linear Collider

    (ILC). The CPCCDs are very fast devices capable of satisfying the challenging requirements

    imposed by the beam structure of the superconducting accelerator [5]. Another type of a

    CCD-based device is In-Situ Storage Image Sensor (ISIS). In ISIS, each pixel has an internal

    memory implemented as a small CCD register. The charge is collected under a photogate and

    is transferred to a few-pixel storage CCD inside the same pixel [6].Physics studies continuously show that extremely precise vertexing will be needed to uncover

    the interactions at a future TeV-scale e+e- linear collider. A very good vertex detector is

    crucial for the high quality b and c tagging, required to further explore the physics at high

    energies. CCD-based vertex detectors have proven their potential for such studies, which was

    shown in the excellent results of the vertex detector VXD3 at the Stanford Large Detector

    (SLD). The excellent gain uniformity, high precision and small layer thickness of the CCD

    are still difficult to achieve with other semiconductor sensors [7, 8]. The transfer

    characteristics of charge-coupled devices have been investigated theoretically and simulated

    experimentally since the invention of these devices [9-12].

    Using analytic models, the charge transfer inefficiency is determined then compared to TCAD

    simulations and/or to experimental results [13, 14].

    In this work, the results of the CTI measured at the test stand at Oxford University are

    presented. It is an unirradiated Column Parallel CCD (CPCCD) called CPC-T. The setup is

    controlled by a Laboratory Virtual Instrumentation Engineering Workbench (LABVIEW)

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    7/88

    2

    software [15]. The binary file, containing data, generated by the latter is then analysed by a

    MATrix LABoratory (MATLAB) code [16] to calculate the CTI.

    This thesis is organized as follows:

    Chapter I: The general definitions and aspects in particle detectors domain are presented in

    some details. This includes radiation sources which have a great effect on the detector

    performances depending on their nature. Common operation modes of a particle detector are

    explained in brief. Then, the characteristics and performances of the detectors are emphasised

    such as the detection efficiency. Finally, the effect of radiation on creating defects in the

    detector is studied.

    Chapter II: This chapter concentrates on charge-coupled devices. First the three usual

    operations of a CCD are explained: two, three and four-phase operation systems. Then, the

    different architectures of the CCD are mentioned. As CCDs will be used in the inner vertex

    detector of the future ILC, time requirement for this project is given in brief. The theory of P-

    MOS and NP-MOS capacitors is given in some details because they are the element cells in

    the CCDs. The three processes (charge generation, charge collection, charge measurement)

    involved to properly convert the photo-generated electrons to an electric signal are presented.

    Finally, we present the theory of the generation-recombination processes resulting from the

    presence of the defects.

    Chapter III: We present the setup used in this work to measure the CTI. It is composed of two

    parts: hardware and software. Then, the method used to analyse data and calculate CTI is

    explained.

    Chapter IV: This chapter and the previous one are the heart of this work. In this chapter we

    present the results obtained after measuring the CTI of an unirradiated CCD. An analytic

    model is used to fit the CTI-T characteristics. Some other parameters are estimated such as:

    noise and occupancy. The effect of temperature and pixel number on the CTI is also studied.

    This thesis terminates by a conclusion in which the results obtained are summarized and a

    possible future work is suggested. An appendix for the MATLAB code developed is also

    included.

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    8/88

    3

    Chapter I

    Particle Detectors

    I.1 Radiations sourcesRadiations can be categorized into four general types. Fast electrons and heavy charged

    particles are considered as charged particle radiation. Electromagnetic radiation and neutrons

    are considered as uncharged radiation. Fast electrons are energetic electrons emitted in

    nuclear decays or produced by any other process. Alpha particles, protons, fission products or

    the products of many nuclear reactions are parts of heavy charged particles. The

    electromagnetic radiation includes X-rays emitted in the rearrangement of electrons shells of

    atoms, and gamma rays that originate from transitions within the nucleus itself. Neutrons

    generated in various nuclear processes are divided in slow neutrons and fast neutrons

    subcategories. Radiations differ in their ability to penetrate into a material. This property is of

    considerable concern in determining the physical form of radiation sources or the structure

    and the physical parameters of the detector. Soft radiations, such as alpha particles or low-

    energy X-rays, penetrate only a few micrometers of depth. Beta particles are generally more

    penetrating; up to a few tenths of a millimeter of depth can usually be achieved. Harder

    radiations, such as gamma rays or neutrons, can penetrate to a depth in millimetres orcentimetres without seriously affecting its properties [17]. A radiation source is characterized

    by its activity which is defined as the sources rate of decay and is given by:

    Ndt

    dNdecay = (I.1)

    whereN is the number of radioactive nuclei and is defined as the decay constant. The SI

    unit of activity is the Becquerel (Bq) defined as one disintegration per second. The historical

    unit of activity has been the Curie (Ci), defined as 10107.3 disintegrations per second. Thus

    Ci10703.2Bq1 11= (I.2)

    The specific activity of a radioisotope source is defined as the activity per unit mass of

    radioisotope sample and it is given by

    M

    A

    ANM

    N v

    v

    ===

    /mass

    activityactivitySpecific (I.3)

    where Mis the molecular weight of the sample,Av is Avogadros number and is given by

    1/2T/)2ln(=

    (I.4)where T1/2 is the half-life of the radiation source.

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    9/88

    4

    I.2 Radiation detectors

    The net result of the radiation interaction in a wide category of detectors is the appearance of

    a given amount of electric charge within the detector active volume. A simplified detector

    model assumes that a charge Q appears within the detector at time t= 0 resulting from the

    interaction of a single particle or quantum of radiation. Next, this charge must be collected to

    form the basic electrical signal. Typically, collection of the charge is accomplished through

    the imposition of an electric field within the detector, which causes the positive and negative

    charges created by the radiation to flow in opposite directions. The time required to fully

    collect the charge varies greatly from one detector to another. The sketch in Fig.I.1 illustrates

    one example for the time dependence the detector current might assume, where tc represents

    the charge collection time.

    Fig.I.1: An illustration of the time dependent of the detector current during charge collection time tc.

    The total charge generated in that specific interaction is equal to the time integral over the

    duration of the current, thus

    =ct

    dttiQ0

    )( (I.5)

    I.2.1 Detector operation modes

    There are three general modes of operation of radiation detectors: the current mode, the pulse

    mode and the mean square voltage mode (abbreviated MSV mode, or sometimes called

    Campbelling mode). Pulse mode is the most commonly applied of these, but current mode

    also finds many applications. MSV mode is limited to some specialized applications that

    make use of its unique characteristics.

    Time

    i(t)

    tc

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    10/88

    5

    I.2.1.1 Current operation mode

    In the current mode, a current-measuring device (practically a picoammeter) is connected

    across the output terminals of a radiation detector as shown in Fig.I.2. The current mode

    operation is used with many detectors when event rates are very high.

    Fig.I.2: Radiation detector in current mode. (a) Current-measuring device (picoammeter) connected across the

    output terminals of a radiation detector. (b) Recorded signal from a sequence of events.

    I.2.1.2 Mean Square Voltage operation mode

    The MSV mode of operation is most useful when making measurements in mixed radiation

    environments when the charge produced by one type of radiation is much different than that

    from the second type. In this mode, the average current I0 is blocked. Then, by providing

    additional signal-processing elements, the time average of the squared amplitude of the

    fluctuations is computed. The processing steps are illustrated in Fig.I.3.

    Fig.I.3: The processing steps using in a detector operating in the mean square voltage (MSV) mode.

    I.2.1.3 Pulse operation mode

    Most applications are better served by preserving information on the amplitude and timing of

    individual events that only pulse mode can provide. The nature of the signal event depends on

    the input characteristics of the circuit to which the detector is connected (usually a

    preamplifier). The equivalent circuit can often be represented as shown in Fig.I.4.

    Detector pA

    Time

    i(t)

    Ion chamber Squaring circuitAveraging

    Output

    Variance computing circuit

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    11/88

    6

    Fig.I.4: The equivalent circuit of both, detector output and the input of the circuit connected to it.

    Figure I.5.a shows the signal pulse produced from a single event in a detector operated in

    pulse mode. Two separate extremes of operation can be identified that depend on the relative

    value of the time constant of the measuring circuit. In the case where the time constant, given

    by =RC, is kept small compared with the charge collection time, so the signal voltage V(t)

    produced under these conditions has a shape nearly identical to the time dependence of the

    current produced within the detector as illustrated in Fig.I.5b. Radiation detectors are

    sometimes operated under these conditions when high event rates or timing information ismore important than accurate energy information. But it is more common to operate detectors

    in the opposite extreme in which the time constant of the external circuit is much larger than

    the detector charge collection time as illustrated in Fig.I.5c.

    Fig.I.5: (a) The signal pulse produced from a single event in a detector operated in pulse mode. (b) The case of a

    small time constant load circuit. (c) The case of a large time constant load circuit.

    Detector C R V(t)

    i(t)

    Time

    V(t)

    Time

    V(t)

    Time

    = dttiQ )(

    (a)

    (b)

    (c)Case 2 :RC >> tc

    Vmax = Q/C

    Case 1 :RC

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    12/88

    7

    I.2.2 Pulse height spectra and energy resolution

    The pulse amplitude distribution is a fundamental property of the detector output that is

    routinely used to deduce information about the incident radiation or the operation of the

    detector itself. The most common way of displaying pulse amplitude information is through

    the differential pulse height distribution. Figure I.6 shows an example of such distribution.

    The abscissa is a linear pulse amplitude scale in volts or analog to digital converter (ADC)

    counts. The ordinate is the differential number dN of pulses observed with an amplitude

    within the differential amplitude increment dH, divided by that increment.

    Fig.I.6: An example of differential pulse height spectra.

    The number of pulses whose amplitude lies between two specific values, H1 andH2, can be

    obtained by integrating the area under the distribution between those two limits, as shown by

    the cross-hatched area in Fig.I.6. Peaks in the distribution, such as H4, indicate pulse

    amplitudes about which a large number of pulses may be found. On the other hand, valleys or

    low points in the spectrum, such asH3, indicate pulse amplitudes around which relatively few

    pulses occur. The physical interpretation of differential pulse height spectra always involves

    area under the spectrum between two given limits of the pulse height.

    In many applications of radiation detectors, the objective is to measure the energy distribution

    of the incident radiation. These efforts are classified under the general term radiation

    spectroscopy. One important property of a detector in radiation spectroscopy can be examined

    by noting its response to a monoenergetic source of that radiation. Under these conditions, the

    differential pulse height distribution is called the response function of the detector for the

    energy used in the determination as illustrated in Fig.I.7. The energy resolution of the detector

    H2H1 H4H3 Pulse height

    dN/dH

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    13/88

    8

    is conventionally defined as the full width at half maximum (FWHM) divided by the location

    of the peak centroidH0, thus

    0

    FWHM

    HR = (I.6)

    For peaks whose shape is Gaussian with standard deviation , the FWHM is given by

    35.2 . The energy resolutionR is thus a dimensionless fraction conventionally expressed as

    a percentage. It should be clear that the smaller the figure for the energy resolution, the better

    the detector will be able to distinguish between two radiations whose energies lie near each

    other.

    Fig.I.7: An example of response function and definition of detector resolution.

    I.2.3 Detection efficiency

    The detection efficiency depends on the type of radiation. For charged radiation such as alpha

    and beta particles, they will form enough ion pairs along its path to ensure that the resulting

    pulse is large enough to be recorded. The detector is said to have a high counting efficiency.

    On the other hand, uncharged particle such as gamma rays and neutrons can travel large

    distances between interactions. The detector in this case has less counting efficiency. It then

    becomes necessary to have a precise figure for the detector efficiency.

    It is convenient to subdivide counting efficiencies into two classes: absolute and intrinsic. The

    absolute efficiency is defined as

    sourceby theemittedquantaradiationofnumber

    recordedpulsesofnumber=

    abs (I.7)

    dN/dH

    HH0

    Y/2

    Y

    FWHM

    Resolution :R = FWHM/H0

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    14/88

    9

    The intrinsic efficiency is defined as

    detectoron theincidentquantaradiationofnumber

    recordedpulsesofnumberint = (I.8)

    The two efficiencies are simply related for isotropic sources by

    =

    4int abs (I.9)

    where is the solid angle of the detector seen from the actual source position. A commonly

    encountered circumstance is shown in Fig.I.8. It involves a uniform circular disk source

    emitting isotropic radiation aligned with a circular disk detector, both positioned

    perpendicular to a common axis passing through their centers. The solid angle is given by the

    following approximate equation calculated numerically [17]

    ( ) ( )

    +

    +

    + 2

    31

    22/52/1 18

    3

    1

    112 FF

    (I.10)

    where

    ( ) ( ) 2/92

    2/71164

    35

    116

    5

    +

    +=F ;

    ( ) ( ) ( ) 2/133

    2/11

    2

    2/92 11024

    1155

    1256

    315

    1128

    35

    ++

    +

    +=F

    2

    =

    d

    s ;

    2

    =

    d

    a

    This approximation becomes inaccurate when the source or detector diameters become too

    large compared with their spacing.

    Fig.I.8: A uniform circular disk source emitting isotropic radiation aligned with a circular disk detector, both

    positioned perpendicular to a common axis passing through their centers.

    d

    a

    s

    Source Detector

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    15/88

    10

    I.2.4 Dead time

    Dead time is defined as the finite time after the registration of an event, before the detector is

    able to accurately register another incident event [18]. Dead time is the minimum amount of

    time that must separate two events in order that they be recorded as two separate pulses. Two

    models of dead time behaviour of counting systems have come into common usage:

    paralyzable and nonparalyzable response. The fundamental assumptions of the models are

    illustrated in Fig.I.9. At the center of the figure, a time scale is shown on which six randomly

    spaced events in the detector are indicated. At the bottom of the figure is the corresponding

    dead time behaviour of a detector assumed to be nonparalyzable. A fixed time is assumed to

    follow each true event that occurs during the live period of the detector. True events that

    occur during the dead period are lost and assumed to have no effect whatsoever on the

    behaviour of the detector. In the case of non-paralyzable detectors and electronics, the

    expression giving the recorded events rate m as a function of delivered events rate or true

    interaction rate n can be calculated for random time distributions of the events for a given

    dead time as

    n

    nm

    +=

    1(I.11)

    For paralyzable systems (dead time adds up for subsequent events) the rate of recorded events

    readsnenm = (I.12)

    Fig.I.9: An illustration of two assumed models of dead time behaviour for radiation detector. (a) Paralyzable

    model. (b) True events in detector. (c) Nonparalyzable model.

    Time

    Time

    Time

    Dead

    Live

    (a)

    (b)

    (c)

    Dead

    Live

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    16/88

    11

    I.3 Radiation damage in particle detectors

    Semiconductor devices are sensitive to radiation, both ionizing and non-ionizing. It was

    found, that the degradation of the parameters of bipolar and MOS devices is caused by

    radiation-induced surface effects at the Si-SiO2 interface, as well as by defects in the bulk

    silicon [19].

    I.3.1 Surface damage

    The passage of an ionising radiation in the depletion region in silicon detectors creates

    electron-hole (e-h) pairs that are collected by the electric field at the electrodes and form the

    signal. In the undepleted bulk of the semiconductor, where there is no electric field, the high

    carrier density allows the deposited charge carriers to recombine. Therefore, the

    semiconductor does not show permanent traces of the passage of a charged particle that loses

    energy by ionisation.

    On the contrary, the passage of an ionising radiation in the oxide causes the built up of

    trapped charge in the oxide layers of the detector [20]. If an MOS structure (Fig.I.10) is

    exposed to a short radiation pulse while under an applied bias voltage, electron-hole pairs are

    generated in the oxide. Under the applied bias, the electrons that escape early recombination

    with the holes are rapidly (within picoseconds) swept out of the oxide and leave behind an

    instantaneous, essentially uniform distribution of relatively immobile holes. This positive

    charge distribution causes an initial negative voltage shift, V(O+), in the capacitance voltage

    (C-V) characteristic of an MOS capacitor or the current- voltage (I-V) characteristic of the

    equivalent MOSFET [20]. In terms of the continuous time random walk (CTRW) model [21],

    holes then begin to move under the influence of the applied bias to either the gate or the Si02-

    Si interface by a slow polaron-hopping process that requires many decades in time. As the

    transporting holes reach the interface most of them are removed, causing the voltage shift,

    V(t), to decrease with time. Figure I.10 shows schematically the expected behaviour of the

    voltage shift as a function of time for a sample irradiated under positive bias (a) and negative

    bias (b).

    In addition to the trapped charge, the ionising radiation also produces new energy levels in the

    band gap at the SiO2-Si interface. These levels can be occupied by electrons or holes,

    depending on the position of the Fermi level at the interface and the corresponding charge can

    be added or subtracted to the oxide charge.

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    17/88

    12

    Fig.I.10: Contributions of holes to voltage shift in MOS structures. Dotted curves correspond to the transporting

    holes. Dashed curves correspond to the trapped holes at interface. Continue curves correspond to the total. (a)

    Positive bias. (b) Negative bias.

    I.3.2 Bulk damage

    While in the silicon crystal itself ionization is a reversible process, and therefore does not

    cause any damage, the energy transfer to crystal atoms which is the non-ionizing energy loss

    (NIEL) of the incident particle causes displacement damage [22]. The bulk damage is caused

    by the NIEL interactions of a primary particle with mass mp and energy Ep with a lattice

    silicon atom with mass MSi. The energy transferred in the interaction is, in the non-relativistic

    case:

    ( ) pSipSip

    EMm

    MmE

    +=

    2sin4 2

    2

    (I.13)

    where is the scattering angle. Displacement damage occurs when the energy transferred to

    the silicon atom is sufficient to remove it from the crystal lattice. The atom is then called

    primary knock-on atom (PKA). The minimum threshold energy for the displacement is

    eV15 in silicon. The vacancy-interstitial (V-I) silicon created is called a Frenkel pair. The

    minimum energy for particles like neutrons or protons, with mass u1 , required to create a

    Frankel pair is eV110 . The same threshold energy for electrons is keV260 . The energy

    of the particle used for the irradiation studies reported in the present work is several orders of

    magnitude higher than the threshold level, as it will be in the background radiation in theinner detectors of the Large Hadron Collider (LHC) [19].

    Gate (+) Gate (-)

    SiO2 Si SiO2 Si

    1.0

    0V

    V

    0

    Log(t)Log(t)

    0V

    V

    1.0

    0

    (a) (b)

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    18/88

    13

    The recoil energy of the PKAs can be up to 130 keV and therefore they can remove other

    atoms from the crystal lattice, giving rise to a PKA cascade. It has been estimated that about

    50% of the energy of the recoil atom is deposited via ionisation and the displacement

    dominates when the recoil atom loses its final 5-10 keV. The fraction of energy going to the

    non-ionising interaction is described by the Lindhard partition function. The cascade results in

    the formation of two or three terminal clusters of 50 linear dimension with high

    concentration of Frankel defects. About 90% of the vacancies recombine with interstitials,

    leaving no net damage in the crystal. Some vacancies can form stable divacancy (or multi-

    vacancy) defect complexes and the remaining vacancies and interstitials diffuse through the

    crystal and react with other defects or impurity atoms always present in the silicon crystal (O,

    C, P, B) to form stable complexes. The introduction rate of vacancies (V) and divacancies

    (V2) have been determined to be1cm5.01.2 =V and

    1cm4.07.4 =V for 1 MeV

    neutron irradiation. Figure I.11 shows an example of the final damage due to aggregation of

    point defects (V2, VP).

    Fig.I.11: A diagram of some defects in the n-type silicon crystal lattice due to point defect complexes.

    Closely situated multiple displacements, which can interact electrically is called defect

    cluster. Light particles, such as electrons generate mostly point defects, and need more than 5

    MeV to produce clusters. The energy threshold for electrons to displace a Si atom has been

    estimated to 260 keV, whereas only 190 eV is required for a neutron to do the same effect.

    Heavy particles easily create clusters because of the high energy of the primary recoil Siatom. Neutrons, protons and pions need only about 15 keV to create clusters and generate

    DivacancyVacancyInterstitiel silicon(self interstitiel)

    Interstitielimpurity

    Phosphorus Substitutional impurity Vacancy-phosphorus

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    19/88

    14

    point defects as well. The final radiation damage is due to the thermally stable defects formed

    by the reaction of primary defects (V, interstitial) with other defects or atomic impurities.

    Some of the possible reaction have been identified and are reported in Table I.1 [19]. Some of

    the stable complexes are electrically active, therefore the changes of the electrical properties

    of detectors are correlated with the electrical activity of the stable damage. Several defects

    have been identified by means of different measurement techniques, such as electron

    paramagnetic resonance (EPR), photoluminescence, current or capacitance deep level

    transient spectroscopy (I-DLTS, C-DLTS), thermally stimulated current (TSC). Table I.2 lists

    some of the identified defects, their charge states and their associated energy levels in the

    band gap [19]. Most of these energy levels situate in the deep region of the silicon band gap,

    close to the middle.

    I reaction V reaction Ci reaction

    I + Cs Ci V + V V2 Ci + Cs CC

    I + CC CCI V + V2 V3 Ci + O CO

    I + CCI CCII V + O VO

    I + CO COI V + VO V2O

    I + COI COII V + P VP

    I + VO O

    I + CV2V

    I + VP P

    Table I.1: A few defect reactions in silicon. The subscript i stands for interstitial, s for substitutional, I for Siinterstitial, V for vacancy, C for carbon, O for Oxygen and P for phosphorus.

    Defect Energy level Defect type

    VO EC-0.17 acceptor

    V2O EC-0.50 acceptor

    V2

    EC-0.23

    EC-0.42

    EV+0.25

    acceptor

    acceptor

    donor

    VP EC-0.45 acceptor

    CC EC-0.17 acceptor

    CO EV+0.36 donor

    Table I.2: Identified defect states with their energy levels in eV.

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    20/88

    15

    I.4. Trapping and generation-recombination at deep levels

    If the energy levels introduced in the forbidden gap of a semiconductor lie greater than 0.1 eV

    from the valence or the conduction band-edges, the level is commonly referred to as a deep

    level. Deep levels can be formed by either introduction of impurities or be the result of

    inherent crystal defects. They may be, for example, large foreign atoms positioned

    substitutionally or interstitially, vacancies, host atoms on the wrong site in compound

    semiconductors (anti-site defect), defect due to dislocations or damages induced by irradiation

    or ion-implantation.

    Deep levels may behave as carrier traps or as generation recombination centres if they are

    near to mid-band-gap. As traps they can capture the free carriers supplied by the dopant

    atoms, thus compensating the shallow levels, reducing the effective doping density and

    increasing the resistivity of the material. Deep levels, behaving as recombination centres,

    provide a path for the generation and recombination of electron-hole pairs across the band-

    gap. Deep levels may be characterized by three parameters: the activation energy (Et) which is

    related to the position of the level in the band-gap, its concentration (Nt) and its capture cross-

    section () for carriers which provides a measure of the ability of the deep level to trap

    carriers.

    A theory describing the generation and recombination (g-r) processes has been established by

    Shockley, Read and Hall [23, 24]. More details on g-r processes can be found in the literature

    [25, 26]. Therefore, the effect is throughout the literature referenced as Shockley-Read-Hall

    (SRH) generation/recombination. Four sub-processes are possible:

    a) Electron capture. An electron from the conduction band is captured by an empty trap

    in the band-gap of the semiconductor. The excess energy ofEc Et is transferred to

    the crystal lattice (phonon emission).

    b) Hole capture. The trapped electron moves to the valence band and neutralizes a hole

    (the hole is captured by the occupied trap). A phonon with the energy Et Ev is

    generated.

    c) Electron emission. A trapped electron moves from the trap energy level to the

    conduction band. For this process additional energy of the magnitude Ec Ethas to be

    supplied.

    d) Hole emission. An electron from the valence band is trapped leaving a hole in the

    valence band (the hole is emitted from the empty trap to the valence band). The energy

    necessary for this process isEtEv.These four processes are illustrated in Fig.I.12.

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    21/88

    16

    Fig.I.12: The four sub-processes in the Shockley-Read-Hall generation/recombination process. (a) Electron

    capture. (b) Hole capture. (c) Electron emission. (d) Hole emission.

    The process (a) i.e. capture of electrons by the deep centre from the conduction band has the

    following equation

    npnpR TnTthnna =>

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    22/88

    17

    npnedt

    dnTnTn += (I.18)

    Similarly the net rate of holes leaving the valence band is given by

    pnpedt

    dpTpTp += (I.19)

    Thus the net rate of increase of density of filled deep levels is given by

    pnpenpne

    dt

    dp

    dt

    dn

    dt

    dn

    TpTpTnTn

    T

    +++=

    =

    (I.20)

    Using the total density of traps TTT pnN += , we get

    ( ) ( ) TppnnTpnT nececNecdt

    dn++++= (I.21)

    where cn and cp are the capture rates of electron and hole, respectively.

    According to the detailed balance principle, in thermal equilibrium, the rate of any physical

    process and its reverse must balance each other. Thus, in this case, the rates for holes

    emission and capture must be equal. Similarly the rate of emission of electrons and the

    corresponding capture rate must also exactly cancel. One can obtain the electron and hole

    thermal emission rates from a deep level using the detailed balance principle and the Fermi

    Dirac distribution function. The probability of an electron occupying an energy level Et isgiven by

    +

    =

    kT

    EEEf

    Ft

    t

    exp1

    1)( (I.22)

    The emission and capture rates for electrons and holes are equal according to the balance

    principle, thus

    ca RR = (I.23.a)

    db RR = (I.23.b)

    The number of filled traps is given by

    TtT NEfn )(= (I.24)

    The density of electrons in the conduction band is given by

    =

    kT

    EENn CFCexp (I.25)

    Using equations (I.14), (I.16), (I.22), (I.23.a), (I.24) and (I.25), we get the thermal emissionrate of electrons

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    23/88

    18

    >=l1 && xdataC2(i)maxy

    maxy=ydataC2(i);

    imaxy=i;

    end

    end

    end

    ipeak1=imaxy;

    limit1nC2=xdataC2(ipeak1)-noisewidth/0.6*GV/Factor; %%%%Noise limit1

    limit2nC2=xdataC2(ipeak1)+noisewidth/0.6*GV/Factor; %%%%Noise limit2

    else

    limit1nC2=fresult2C2.b1-3*fresult2C2.c1;

    limit2nC2=fresult2C2.b1+3*fresult2C2.c1;

    end

    xdata2C2=[limit1nC2:Mbin:limit2nC2];

    ydata2C2=(hist(data_col(:,M),xdata2C2));

    fresult2C2=fit((xdata2C2(2:length(xdata2C2)-1))',(ydata2C2(2:length(ydata2C2)-1))','gauss1');

    limit1xC2=fresult2C2.b1+xraynoise/0.6*GV/Factor-nsigma*fresult2C2.c1; %%%%X-ray limit1

    limit2xC2=fresult2C2.b1+xraynoise/0.6*GV/Factor+nsigma*fresult2C2.c1; %%%%X-ray limit2

    xdata3C2=[limit1xC2:Mbin:limit2xC2];ydata3C2=(hist(data_col(:,M),xdata3C2));

    fresult3C2=fit((xdata3C2(2:length(xdata3C2)-1))',(ydata3C2(2:length(ydata3C2)-1))','gauss1');

    if (PP>(Pend-MPX)) || (PP==(Pstart+Pend)/2) || (PP==(Pstart+MPX))

    figure;

    semilogy(xdataC2(2:length(xdataC2)-1),abs(ydataC2(2:length(ydataC2)-1)),'k-','LineWidth',1);

    hold on;

    semilogy(xdata2C2(2:length(xdata2C2)-1),fresult2C2(xdata2C2(2:length(xdata2C2)-1))','r-

    ','LineWidth',3);

    hold on;

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    85/88

    80

    semilogy(xdata3C2(2:length(xdata3C2)-1),fresult3C2(xdata3C2(2:length(xdata3C2)-1))','b-

    ','LineWidth',3);

    axis([fresult2C2.b1-3*fresult2C2.c1 fresult3C2.b1+3*fresult3C2.c1 1 2*fresult2C2.a1]);

    grid on;

    xlabel('ADC codes');ylabel('Counts');

    legend('Distribution of ADC codes','Fit noise','Fit X-ray');

    end

    limit1noise(2)=fresult2C2.b1-ms*fresult2C2.c1;

    limit2noise(2)=fresult2C2.b1+ms*fresult2C2.c1;

    limit1xray(2)=fresult3C2.b1-ms*fresult3C2.c1;

    limit2xray(2)=fresult3C2.b1+ms*fresult3C2.c1;

    for j=P1:P2

    xrayC2(KK2)=0;

    nxrayC2(KK2)=0;

    noiseC2(KK2)=0;

    nnoiseC2(KK2)=0;

    for k=1:iframe

    if ((datatot(k,j,M)>=limit1xray(2))&&(datatot(k,j,M)=limit1noise(2))&&(datatot(k,j,M)

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    86/88

    81

    pixel_number

    end

    end

    end

    %%% Start: CTI determination %%%figure;

    LLC2=length(XrayADC2);

    plot(XpixelC2,XrayADC2,'b-','LineWidth',0.5);

    fresult4C2=fit(XpixelC2',XrayADC2','poly1');

    hold on;

    plot(XpixelC2,fresult4C2(XpixelC2),'r-','LineWidth',3);

    grid on;

    axis([Pstart Pend fresult3C2.b1-fresult3C2.c1 fresult3C2.b1+fresult3C2.c1]);

    xlabel('Pixel number (COL 2)');

    ylabel('X-ray peak');

    legend('X-ray peak','Linear fit');

    CTI2=-fresult4C2.p1/fresult4C2.p2

    errorsC2=confint(fresult4C2);

    dp1C2=(errorsC2(2,1)-errorsC2(1,1))/2;

    dp2C2=(errorsC2(2,2)-errorsC2(1,2))/2;

    dCTIC2=abs(CTI2)*(dp1C2/abs(fresult4C2.p1)+dp2C2/abs(fresult4C2.p2))

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    87/88

    Publications related to this work

    A. Sopczak et al., Measurements of Charge Transfer Inefficiency in a CCD With High-

    Speed Column Parallel Readout,IEEE Trans. Nucl. Sci., vol. 56, no. 5, pp. 29252930,2009.

    A. Sopczak et al., Comparison of Measurements of Charge Transfer Inefficiencies of a High

    Speed Column Parallel CCD,IEEE Trans. Nucl. Sci., vol 57, no. 2, pp. 854859, 2010.

  • 8/6/2019 Khaled BEKHOUCHE : Measurement of Charge Transfer Efficiency of a Silicon Particle Detector Based on a Charge-

    88/88

    55Fe .CCD))CTI

    CTI.500 )CPC-T)T-type CPCCD

    LABVIEW.

    .CTI MATLAB ))ADC

    CTI 100010000)Frame.( .

    , CPC-T .10000

    CTI10-5.

    . 0.44eV0.37eV

    Parabolic .

    .CCD

    .CTI )Clock voltage(

    CTI )Sense node.(

    . ,

    Rsum

    La technique des rayons X est utilise pour mesurer la CTI dans un CCD. La source est une55Fe et le CCD est un CPC-T (T-type column parallel CCD), qui dispose de 4 canaux, 500

    pixels chacun. La CTI est mesure pour un seul canal, car ils sont identiques. LabVIEW est

    utilis pour acqurir des donnes via un convertisseur ADC. Un code MATLAB est utilis

    pour analyser les donnes mesures et extraire la CTI. La CTI est calcule pour deux nombres

    de frames, 1000 et 10000. Cela permet l'tude de l'effet des erreurs statistiques. Les petites

    erreurs statistiques sont obtenues avec 10000 images. Pour un CPC-T non irradi, la CTI est

    de lordre de 10-5. Un modle analytique est utilis pour ajuster les rsultats exprimentaux et

    sest trouvs en bon accord dans le cas dune faible densit de deux piges lectrons situ

    0,37 eV et 0,44 eV en dessous de la bande de conduction. Le niveau de bruit mesur est de