Hatano Y., Katsumura Y., Mozumder A. (Eds.) Charged Particle and Photon Interactions with Matter - Recent Advances, Applications, and Interfaces

Подождите немного. Документ загружается.

850 Charged Particle and Photon Interactions with Matter

The unique degradation behavior is also observed in I

for 3J solar cells by proton irradiation

with various energies. Figure 30.10 shows the value of I

for InGaP/GaAs/Ge 3J solar cells as func-

tions of irradiated proton energy and uence (closed squares). The values of I

for InGaP (open

triangles) and GaAs (open circles) sub cells, estimated from spectral sensitivity measurements, are

plotted in Figure 30.10. At a uence of 10

−2

, the value of I

for the 3J solar cells does not

depend on the value of proton energy, and is almost the same as the value of I

for InGaP solar cells.

This indicates that the InGaP cell is the current-limiting cell in this uence region. The degradation

of I

for GaAs solar cells irradiated with protons at hundreds of keV is larger than that of GaAs solar

0.1

0.01

Proton energy (MeV)

Remaining factor of I

(%)

Fluence (cm

–2

)

Figure 30.10 I

for InGaP/GaAs/Ge 3J solar cells as functions of irradiated proton energy and uence

(closed squares). For comparison, the values of I

for InGaP (open triangles) and GaAs (open circles) sub cells

estimated

from spectral sensitivity measurements are also plotted in the gure.

100

0.01 0.1 1 10

Proton energy (MeV)

Relative damage coefficient

Figure 30.9 Relative damage coefcient of V

for InGaP/GaAs/Ge 3J solar cells as a function of irradi-

ated

proton energy.

Radiation Effects on Semiconductors and Polymers for Space Applications 851

cells irradiated at another energy. The value for GaAs solar cells irradiated with protons at hundreds

of keV becomes almost the same value for InGaP solar cells irradiated at a uence of 10

−2

. The

for GaAs solar cells dramatically decreases and becomes the lowest value at 10

−2

. Thus,

since the current-limiting cell changes to the GaAs cell from the InGaP cell, the value of I

for 3J

solar cell becomes the same as that of the GaAs cell. This faster decrease in I

for the GaAs sub

cell by irradiation of protons with hundreds of keV is explained as follows. The protons with given

energy stop around the pn junction in the GaAs sub cell. This indicates that heavy damage is created

in this region by the irradiation. Since the radiation resistance of GaAs is not higher than that of

InGaP, the degradation of the GaAs sub cell is much faster than the InGaP sub cell by hundreds keV

range proton irradiation. From this result, it can be concluded that the improvement of the radiation

hardness of the GaAs sub cell is important to improve the overall radiation hardness of 3J solar cells.

30.2.5 prediction Methodology of triple-junction Solar cell degradation

Protons and electrons with a very wide range of energies exist in space. However, available energies

for ground-based irradiation experiments are limited by the capability of accelerators. Therefore,

it is important to develop accurate prediction methods on the radiation degradation of solar cells

in real space, using results obtained from ground tests. Presently, the damage-equivalent uence

method, which was proposed by the U.S. Jet Propulsion Laboratory (JPL) more than 20 years ago,

is the most popular prediction method (Tada et al., 1982). According to the JPL method, a result

obtained by 1 MeV electron irradiation testing is standard, and the relative damage coefcient is

estimated from the comparison of results using electrons/protons with other energies to the result

obtained by 1MeV electron/proton irradiation. In the case of single junction solar cells, the JPL

method is thought to be a reasonable way for the lifetime prediction of solar cells, since a single

peak is usually observed in the relative damage coefcient as a function of proton/electron energy.

But, for triple-junction solar cells, as shown in Figure 30.9, three peaks appear in the proton/electron

energy dependence of the relative damage coefcient. This indicates that proton/electron irradia-

tion experiments using various energies are necessary to clarify the relationship between proton/

electron energy and the relative damage coefcient for 3J junction solar cells. From the standpoint

of saving time and money, increasing the number of irradiation experiments is not a good idea.

Instead of uence, displacement damage dose (D

), based on the concept of non ionizing energy

loss (NIEL), has been proposed by the U.S. Naval Research Laboratory (NRL) (Summers et al.,

1987; Messenger et al., 2001). The NIEL is a concept for the energy reduction of incident particles in

materials,

which is for displacement of atoms at lattice sites and is estimated as D

and is dened as

D E

ref

NIEL fluence

NIEL

( )

= × ×













−

(30.3)

where

NIEL(E

ref

) is NIEL for a standard energy such as 1 MeV electrons

is a correction number usually between 1 and 2 (Messenger et al., 2001)

the D

is applied as the value of damage creation, since the D

does not depend on the energy of

incident protons/electrons, the degradation curves for solar cells irradiated with protons/electrons at

various energies should be single line. Figure 30.11 shows the degradation curve for V

of InGaP/

GaAs/Ge 3J solar cells as a function of D

. Two degradation curves are shown in the gure. Since the

projection range for protons with lower than 100keV is within the InGaP sub cell, the obtained result

can be explained in terms of the curve obtained by low energy proton irradiation for the InGaP sub

cell, and the curve obtained by high-energy proton irradiation for GaAs and Ge sub cells. From this

result, it is apparent that the degradation of V

can be described by D

on the basis of the NIEL con-

cept. However, the degradation of I

for 3J solar cells is not simple, even if D

is applied, as shown in

852 Charged Particle and Photon Interactions with Matter

Figure 30.12. The results at D

around 10

MeV/g obtained by 250 and 380keV proton irradiations are

not plotted on the line. Since proton irradiation at these energies creates damage into GaAs sub cells,

the obtained result is attributed to the remarkable degradation of I

by the current-limiting cell chang-

ing from the GaInP subcell to the GaAs subcell. This result suggests that it is difcult to describe the

degradation of I

using D

, if the change occurs in the current-limiting cell. Thus, although the NRL

method on the basis of NIEL concept is very convenient to reduce irradiation tests, we need to know

the D

or uence value for the change in current-limiting cell by some other method.

Modeling of degradation behavior of InGaP/GaAs/Ge 3J solar cells using spectra response (quan-

tum efciency) was proposed by JAEA and Japan Aero eXploration Agency (JAXA) (Sato et al.,

2009a,b). In this modeling, the change in quantum efciency for 3J solar cells due to proton/electron

1.0

0.8

0.6

0.4

30 keV

50 keV

70 keV

100 keV

250 keV

380 keV

1 MeV

2 MeV

5 MeV

Displacement damage dose (MeV/g)

Remaining factor of V

Figure 30.11 Degradation curve of V

for InGaP/GaAs/Ge 3J solar cells irradiated with protons with

various

energies as a function of D

1.0

0.8

0.6

0.4

0.2

0.0

30 keV

50 keV

70 keV

100 keV

250 keV

380 keV

1 MeV

2 MeV

5 MeV

Displacement damage dose (MeV/g)

Remaining factor of I

Figure 30.12 Degradation curve of I

for InGaP/GaAs/Ge 3J solar cells irradiated with protons with

various energies as a function of D

Radiation Effects on Semiconductors and Polymers for Space Applications 853

irradiation is measured, and the simulation of the measurement result is carried out to obtain L and

p of each sub cell. Then, the K

and R

for each sub cell are estimated from the uence dependence

of L and p. Since I

and V

for 3J solar cells can be estimated from the simulation using K

and

, the degradation curves for I

and V

are obtained. Figure 30.13 shows the degradation curves

of I

for 3J solar cells irradiated with either protons or electrons. The simulation results mentioned

above are also depicted as solid lines. Excellent agreement between experimental and calculated

results is observed as shown in Figure 30.13. This indicates that the degradation curve for 3J solar

cells can be described by this model, even if the current-limiting cell changes from the InGaP cell

to the GaAs cell. Furthermore, it was found that the value of K

and R

can be described as func-

tions of NIEL (Sato et al., 2009b). This indicates that the degradation curve for 3J solar cells can be

predicted using the K

and R

as a function of NIEL, which can be obtained from minimal irradia-

tion testing. Since a wide variety of materials are used in solar cells and the structure becomes more

complicated, the degradation modeling is more important from the viewpoint of lifetime prediction

and development of new space solar cells. To obtain highly accurate modeling of radiation degra-

dation of solar cells, further investigation is necessary, especially for unique effects such as room

temperature annealing effect and minority carrier injection effect.

30.3 eleCtroniC deviCes

30.3.1 overview of Single-event effectS in Space

The electronic devices in the articial satellites are exposed to high levels of space radiations. The

high-energy heavy ions in cosmic rays especially cause the malfunction of electronic devices known

as SEEs. It is well known that the energy of heavy ions in space exceeds GeV (Xapsos, 2006). As an

example, the energy spectrum of iron ions at a geostationary orbit is shown in Figure30.14 (Dodd

et al., 1998). The energy at the peak ux for iron ion is around 300 MeV/amu. The atomic mass unit

(amu) for iron is 56; therefore, the energy of iron at the peak ux is 16.8GeV. The experiments, using

the high-energy heavy ions provided by accelerators, are carried out on the ground to understand

the mechanism of SEEs and to predict the probability of SEEs during space missions. Figure 30.14

shows typical ion irradiation facilities for lower and higher energies. In the case of the test facility

1.0

0.8

0.6

0.4

0.2

0.0

0.00

Protons

10 MeV

3 MeV

150 keV

30 keV

Electrons

Symbols: Experimental

Lines: Calculation

2 MeV

1 MeV

Fluence (cm

–2

)

Remaining factor of I

Figure 30.13 Degradation curves of I

for 3J solar cells irradiated with either protons or electrons.

Symbols

and lines represent experimental and calculated results, respectively.

854 Charged Particle and Photon Interactions with Matter

called Takasaki Ion Accelerators for advanced Radiation Application (TIARA) of JAEA, ion energy,

accelerated by the AVF cyclotron, ranges from 2.5 to 27MeV/amu (Onoda et al., 2006; Kurashima

etal., 2007). In this gure, the LET of iron ion along the curve is indicated with graycoding.

A SEE is categorized as having either nondestructive or destructive effects. The former is often

referred to as “soft error,” while the latter is referred to as “hard error” (Petersen, 2008). A soft error

occurs when the amount of collected charge is large enough to reverse or ip the data state of a

memory cell, register, latch, or ip-op. The error is “soft” because the circuit/device itself is not per-

manently damaged by the radiation. If new data is written to the bit, the device will store it correctly.

The single-event upset (SEU) and the multiple-bit upset (MBU) in a static random access memory

(SRAM) and a dynamic random access memory (DRAM) are notable examples. Other good example

is the single-event functional interrupt (SEFI) in eld programmable gate array (FPGA) or DRAM

control circuitry. The single-event transient (SET) is also a serious problem in analog electronics

and digital logic cells. In general, SETs in analog electronics are referred to as ASETs while those in

digital combinatorial logic are referred to as DSETs. In contrast, a “hard” error is manifested when

the device is physically damaged such that improper operation occurs, data is lost, and the damaged

state is permanent. The single-event latch-up (SEL), the single-event burnout (SEB), and the single-

event gate rupture (SEGR) in power electronic devices provide examples of the hard errors.

The SEEs in devices, circuits, and applications used in radiation environments are too complicated

to be discussed here in detail. In this section, therefore, the DSET will be discussed with a focus on

the kind of mechanism in operation. With this aim, three factors seem to be helpful in attempting to

sketch its mechanism: the rst is the generation of charge and ion track when an ion passes through a

semiconductor, the second is the transient currents resulting from the charge transportation in semi-

conductor, and the third is the propagation of transient voltage pulse in logic circuits.

30.3.2 generation of charge and ion trackS in SeMiconductorS

When the incident ion passes through a semiconductor, it loses kinetic energy predominantly through

interactions with the orbital electrons of the semiconductor crystal. As a result, an ion leaves dense

electron-hole (e-h) pairs along its trajectory. Of course, the ion can also interact directly with crystal

–2

–3

–4

–5

–6

–1

“High”

Energy

Test

Range

Ion LET (MeV cm

/mg)

0.2 5 20

Low

Energy

Test

Range

(GANIL

& MSU)

(BNL)

Energy (MeV/amu)

Ion fluence (#/cm

/day/MeV/amu)

Figure 30.14 Energy spectrum of iron ion in a geostationary orbit at solar minimum conditions behind

100mil Al shielding. Ion LET is indicated by the gray of the symbols. Typical heavy-ion testing is performed

at energies of 1–10MeV/amu. (Reprinted from Dodd, P.E. et al., IEEE Trans. Nucl. Sci., 45, 2483, 1998. With

permission.)

Radiation Effects on Semiconductors and Polymers for Space Applications 855

atoms but the probability of this reaction is orders of magnitude less than the electronic interaction.

If the generated charges (dense electrons and holes) are collected at an electrode and exceed a thresh-

old, then SEE is initiated. The charge collection threshold for the SEE is called the critical charge.

The probability of SEE depends on how much energy is deposited (thus, how many charges are

created) in the sensitive region of electronic devices. The deposited energy can be expressed as a

function of LET. The LET is dened as the average energy loss per unit path length, normalized

by the density of the target material, ρ, and its unit is typically MeV cm

/mg. The LET can be cal-

culated by using the advanced computer codes that employ Monte Carlo techniques and detailed

physical models. One of the most popular simulators in the SEE community is SRIM code (Ziegler

al., 2008). The deposited energy, E, over a path length, l, is expressed as follows:

E l x dx= ⋅

∫

ρ LET( ) .

(30.4)

The LET of an ion can be assumed to be a constant value when the ion has a long range compared to

sensitive volume geometry and when multiple scattering is negligible. The above equation reduces to

E l x

= ⋅ ⋅ρ

LET( ).

(30.5)

It has long been considered that the LET is the most important key factor to dominate the mechan-

ism of SEE. At the same time, ion track distributed along the trajectory of the ion in three dimen-

sions has been thought to be a second-order effect. In principle, different track structures lead to a

different charge collection and SEE response in spite of the same LET. At a given LET, the higher

energy ion produces a longer, wider, and less dense track than the lower energy ion. Various ion

track models have been suggested and one of these is by Kobetich and Katz (1968). Since this model

overestimates the e-h density at the core of the ion track, an improved model using empirical equa-

tions was presented by Fageeha et al. (1994). These models use the delta-electron production under

MeV ion penetration and the keV delta-electron transmission theory in the matter to calculate the

initial radial track distribution. Figure 30.15 shows the predictions using a theoretical model for two

ions with the same LET (11MeV cm

/mg) (Dodd et al., 1998). As shown, the ion track radius of

Kr at

5.04

GeV

is wider than that of Cr at 210

MeV.

The tracks reach radii of over μm.

0.001 0.01 0.1

Radial distance (μm)

Charge density (cm

–3

)

210 MeV Chlorine

5.04 GeV Krypton

Gaussian

approximation

1 10 100

Figure 30.15 Radial track structure of low and high-energy ions with the same incident LET. Also

shown is a Gaussian approximation to the track center region. All three curves have the same integral charge.

(Reprinted

from Dodd, P.E. et al., IEEE Trans. Nucl. Sci., 45, 2483, 1998. With permission.)

856 Charged Particle and Photon Interactions with Matter

If the radial extent of the ion track is larger than the sensitive volume geometry, then one could

imagine that less charge would be collected. On the other hand, the charge outside the sensitive vol-

ume is transported laterally by diffusion toward other electrodes. Figure 30.16a shows the calculated

results of potential represented in counter by using two-dimensional numerical codes in cylindrical

coordinate (Zoutendyk et al., 1988). The direction of carrier motion by drift is indicated by means of

arrows. In contrast, the counter map of electron density is represented in Figure 30.16b. As shown,

the electrons move laterally by diffusion. As a result of lateral charge transport, the MBU occurs

near the position where the ion hits. Figure 30.17 shows the typical MBU cluster induced in 256k

DRAM caused by a normal incidence Kr at 295MeV (Zoutendyk et al., 1989). It follows from what

has been said, that the ion track plays an important role in SEE. However, in some cases, no signi-

cant difference was observed between high and low energy ions exhibiting the same LET (Dodd

al., 1998). Dodd et al. (1998) concluded that the track structure is not important. They concluded

that the central cores of the track structures were nearly identical for low and high-energy ions.

t > t

Diffusion

Electron

density

(b)

–3

t > t

–5 V

Equipotentials

(a)

Electron

drift-field

funnel

Figure 30.16 Counter maps of potential (a) and electron density (b) being calculated by using two-

dimensional numerical codes in cylindrical coordinate after ion track was formed. The direction of carrier

motion by drift and diffusion is indicated by means of arrows. (Reprinted from Zoutendyk, J.A. et al., IEEE

Trans. Nucl. Sci., 35, 1644, 1988. With permission.)

Radiation Effects on Semiconductors and Polymers for Space Applications 857

Three-dimensional simulations conrmed that charge collection was similar in the two cases. This

is partially because the sensitive volumes do not have an abrupt boundary, and the initial ion track

expands very rapidly by radial diffusion. In any case, the inuence of ion track on the charge col-

lection and SEEs is still one of the most interesting issues because of these contrasting observations.

30.3.3 tranSient currentS occurring SubSeQuent to ion track generation

The dense carriers (electrons and holes) created by an ion are transported to an electrode by drift

and diffusion processes in semiconductors. As a result, transient current ows in electronic devices.

Since the behavior of the transient current strongly depends on the impact point of ions in semicon-

ductor devices, the size of the ion beam as well as the information on the impact point of ions are

necessary

to clarify the detailed generation mechanism of SEEs.

With

this aim, Wagner et al. (1987) have developed the measurement system using collimated

beam and wide bandwidth electronics. The beam provided by the accelerator was radially scattered

by a Pt thin lm, and the scattered beam penetrated a Pt micro-collimator at a farther, angled loca-

tion. Using this method, though the parallel pencil beam can be obtained easily, the beams contain the

spattered atoms with low energies as mixtures, and its energy spectrum becomes slightly broad and

shifts toward the left. Since the LET generally decreases with increasing ion energy after reaching

the maximum at several MeV, the low energy ions including the spattered ions induce a larger tran-

sient current than scattered ions, which leads to the lower signal-to-noise ratio. In the 1990s a MeV

range-focused microbeam at TIARA facility was used for transient current measurement instead of

a collimated beam (Nashiyama et al., 1993; Laird et al., 2001). In this system, ion beams with a spot

size of 1μm at energies from 6 to 18MeV (less than about 1MeV/amu) can be irradiated into samples.

Combining the focused microbeam with wide bandwidth electronics, the transient current with a

spatial imaging can be measured. This system is referred to as transient ion beam induced current

(TIBIC) system. A similar system called time-resolved ion beam induced current (TRIBIC) has been

also developed at Sandia National Laboratories (SNL) (Schone et al., 1998). Both methods allow one

to map the spatiotemporal response of a semiconductor device and examine ultrafast charge collec-

tion dynamics (Breese, 1993). These systems have a key advantage in being able to probe a specic

region in a device at high resolution; however, its energy is much lower than that of the cosmic ray

in which we are interested, as shown in Figure 30.14. To overcome this problem, the high-energy

focused microbeam connected to the AVF Cyclotron at JAEA TIARA facility has been successfully

Figure 30.17 Typical MBU cluster induced in 256k DRAM caused by a normal incidence Kr at 295MeV.

(Reprinted

from Zoutendyk, J.A. et al., IEEE Trans. Nucl. Sci., 36, 2267, 1989. With permission.)

858 Charged Particle and Photon Interactions with Matter

developed in the 2000s (Oikawa et al., 2003, 2007). The 260MeV

and the 520MeV

14+

ions

are available for microbeam. The details of the high-energy TIBIC system are given below.

Figure 30.18 shows the schematic diagram of the high-energy TIBIC system (Hirao et al., 2009;

Vizkelethy et al., 2009). The TIBIC system contains a beam chopping system; a beam focusing system

including a beam shifter, micro-slits, a scanner, and a quadruplet quadrupole magnets; an irradiation

chamber; and electronics for TIBIC measurements. In this system, the beam chopping system is used

for controlling the beam current. In addition, the beam current can be reduced by the beam attenuators

located next to the chopper. The beams accelerated by the AVF Cyclotron are optimized by the focus-

ing system. The focused microbeam is transported into the irradiation chamber. The scintillator lm

(ZnS) and the #1000 Cu mesh lie in the same plane of the sample holder in the irradiation chamber.

These are used for detecting the microbeam position and for evaluating the full width at half maximum

(FWHM) of the microbeam spot size, respectively. Software control sequence for TIBIC imaging pro-

ceeds as follows. First, a constant reverse bias is applied to the sample. During this period, the beam is

removed using a chopping system controlled by a transistor–transistor logic (TTL) signal provided by a

device acquisition (DAQ) board. Next, the microbeam is electrostatically scanned by the raster method

with a speed of about 330s/frame. The beam ux is controlled to remain at several tens of ions per

second.

When a transient current is generated by an ion strike, the digital storage oscilloscope (DSO)

is triggered and the beam is again stopped by the TTL signal from DSO. The DAQ board also receives

the TTL signal from DSO and simultaneously records the x–y positions. After the transient current and

positions are stored in PC, the process is repeated over the complete scan area. Finally, we perform

the pulse shape analysis (PSA) on each transient current and generate images based on extracted PSA

parameters, such as the charge, the peak current, the fall time, and the rise time.

Figure 30.19 shows the typical TIBIC peak image observed from a Si pin photodiode by using

high-energy TIBIC at JAEA (Vizkelethy et al., 2009). This image indicates that that the diode

region covered with the circular electrode is SEE sensitive. The peak current at the edge of elec-

trode is lower than that at the center region. Although not shown here, it is known that no remark-

able difference between TIBIC charge images at the edge and the center is observed. In contrast

with the edge, both the peak current and the charge at the bonding pad are much higher than those

at the center. The reason why the charges at the bonding pad is larger than that at the center is

simply interpreted in terms that a penetrated ion loses its energy in the bonding wire and the ion

having lower energy generates the charge in the depletion layer. In other words, the obtained result

indicates that the charge generated in Si under the bonding pad at the thickness of 20 μm can be

evaluated by using ions with several hundreds of MeV. Thus, this result suggests that integrated

Divergence

defining slits

Beam shifter

PC for

microbeam

Cyclotron

Micro

slits

XY scanner

DAQ

PC for

TIBIC

GPIB

DUT

Oscilloscope

Trigger output (TTL)

Power supply

Amp.

BIAS-T

Chamber

Chopper

Ion source

Quadruplet

quadrupole

magnet

Figure 30.18 Schematic diagram of the high-energy TIBIC system connected with the AVF Cyclotron.

(Reprinted

from Hirao, T. et al., Nucl. Instrum. Methods. B, 267, 2216, 2009. With permission.)

Radiation Effects on Semiconductors and Polymers for Space Applications 859

circuits with multiple wiring can be evaluated by the developed TIBIC system. Figure 30.20a

and b show the typical transient current at a reverse bias of 20V when Ne at 260MeV and Ar at

520 MeV strike at the edge, the center, and the bonding pad of electrode, respectively (Vizkelethy

et al., 2009). The oscillation around 3ns is probably caused by the impedance miss-matching at the

feedthrough terminal on the irradiation chamber. Figure 30.20c represents the transient currents at

the center on a log scale. The carrier dynamics in Si pin photodiodes after an ion strike has been

studied by a numerical device simulator, the so-called technology computer aided design (TCAD)

(Weatherford, 2002; Laird etal., 2005; Onoda et al., 2006). As a result, the carrier dynamics can

be simplied into the ambipolar, the bipolar, and the diffusion phases. These phases are distin-

guished in Figure 30.20c. As shown, the plasma induced by Ar at 520MeV makes the ambipolar

phase longer due to dense plasma. Because of the oscillation, it is difcult to evaluate the boundary

between the bipolar phase and the diffusion phase. However it is considered that the overall current

is less affected by the diffusion current. Figure 30.21a shows another example of transient currents

measured at the source and drain electrodes of a 0.25μm bulk n-MOSFETs by using TRIBIC

system at SNL (Ferlet-Cavrois et al., 2004). The Cl ion at 35MeV strikes the drain region with the

transistor biased either in the OFF- or TG-state. In the OFF-state, the drain is biased and the other

electrodes are grounded. The OFF-state corresponds to a common logic sensitive conguration. In

the TG-state, both the source and drain are biased, while the other electrodes are grounded. In both

cases, the drain current has the typical shape of charge collection in a reverse biased junction as

shown in Figure 30.20. The charge collection on the source and drain electrodes can be calculated

by integrating the transient currents and these are plotted in Figure 30.21b. For both the OFF- and

TG-states, a positive charge is collected on the drain. The drain charge increases monotonically

and saturates after 5ns.

30.3.4 Modeling of tranSient voltage pulSe in logic circuitS

The transient voltage pulse, which is referred to as DSET, is likely to become the dominant

source of soft errors for advanced CMOS logic very-large-scale-integrations (VLSIs) (Buchner

300 μm

7.0E–4

0.0E+0

Edge

Bonding pad

Peak current level

Figure 30.19 Typical TIBIC peak image observed from Si pin photodiodes at a reverse bias of 20V by

using 520MeV Ar microbeam. This image indicates that that the diode region covered with the circular elec-

trode is SEE sensitive. (Reprinted from Hirao, T. et al., Nucl. Instrum. Methods. B, 267, 2216, 2009. With

permission.)