Leroy C., Rancoita P.-G. Principles Of Radiation Interaction In Matter And Detection

Подождите немного. Документ загружается.

January 9, 2009 10:21 World Scientiﬁc Book - 9.75in x 6.5in ws-bo ok975x65˙n˙2nd˙Ed

520 Principles of Radiation Interaction in Matter and Detection

0 2 4 6 8 10 12 14

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

Detection efficiency [%]

Neutron energy [MeV]

Fig. 6.38 Fast neutron detection eﬃciency (%) in CH

as a function of the incident neutron energy

(MeV) [Gutierrez (2007)]. The detection eﬃciency increases with the incident neutron energy.

The diﬀerential elastic cross section is smaller for an incident neutron of 14 MeV

compared to 1 MeV as increasing neutron energy opens more inelastic channels,

then, decreasing the elastic cross section. The probability that the proton is scat-

tered at a given angle for an elastic scattering can be calculated from the diﬀerential

cross section. That probability depends on the silicon detector area and the sum of

probabilities is normalized to 1. The maximum thickness of the CH

converter to

be used depends on the proton range, R(θ). This thickness will change with respect

to the recoil energy, E

(θ), and is given by

max

= R(θ) cos θ. (6.158)

The incident neutrons may follow a direction either along the normal or at angle

with the normal to the converter area (Fig. 6.35) [Gutierrez (2007)] In all cases

one has to evaluate i) the probability that the proton can come out the conver-

ter into the silicon detector, depending on its range in the converter and ii) the

probability as a function of the scattering angle. One has also to account for the

increase of the collision probability with the converter thickness. Figure 6.36 shows

the detection probability of the proton as a function of the converter thickness for

neutrons incident along the normal to the converter area [Gutierrez (2007)]. The

optimal thickness of the CH

converter as a function of the incident neutron energy

is shown in Fig. 6.37 [Gutierrez (2007)]. The optimal thickness increases with the

incident neutron energy. For incident neutrons along the normal to the converter

area, one ﬁnds the detection eﬃciency as a function of the incident neutron energy

(Fig 6.38) [Gutierrez (2007)]. The results found in the ﬁgures above can be checked

with a simple calculation similar to those done for the thermal neutrons (see for

January 9, 2009 10:21 World Scientiﬁc Book - 9.75in x 6.5in ws-bo ok975x65˙n˙2nd˙Ed

Solid State Detectors 521

n n n

Silicon detector

Converter

Fig. 6.39 Detection of thermal neutron detection in a detector structure with pores.

instance [Wielunski et al. (2004)]). The signal in silicon covered by a CH

conver-

ter is essentially given by the recoiling hydrogen, i.e., the proton. Considering the

example of 5 MeV incident neutrons, the maximum energy transfer to hydrogen

∼ M

) and carbon are 5 MeV and 1.42 MeV, respectively. These values cor-

respond to a range of 347.8 µm for recoil hydrogen and 3.1 µm for recoil carbon in

polyethylene. At the same time, the elastic scattering cross section for 5 MeV inci-

dent neutrons on hydrogen and carbon are 1.63 b and 1.12 b, respectively (the cross

section values can be found in [NNDC (2008a)], for instance). The concentrations of

hydrogen and carbon in polyethylene are 7.986 ×10

and 3.993 ×10

atoms/cm

respectively. Therefore, for maximum transfer energy from fast neutrons to hydro-

gen and carbon in polyethylene, one expects:

= 0.45 × 10

−2

+ 0.14 × 10

−4

. (6.159)

The signal expected from carbon recoil is negligible compared the signal from hy-

drogen.

6.7.2 3-D Neutron Detectors

The way to improve the limited detection eﬃciency for neutrons of planar detectors

is to enlarge the surface between the neutron converter and the silicon detector [Uher

et al. (2007b)]. This increases the probability that the heavy charged particles will

reach the detector sensitive volume. Current semiconductor fabrication technologies

allow creation of pores into the silicon detector. The width of these pores can be

from a few micrometers up to hundreds of micrometers. The depth can be hundreds

January 9, 2009 10:21 World Scientiﬁc Book - 9.75in x 6.5in ws-bo ok975x65˙n˙2nd˙Ed

522 Principles of Radiation Interaction in Matter and Detection

of micrometers (depending on the pore width, technology and material used). These

pore dimensions are well suited for thermal neutron detectors. The principle of neu-

tron detection in a detector structure with pores is shown in Fig. 6.39. Pore depths of

200 µm or more can be fabricated by Deep Reactive Ion Etching (DRIE) [Wikipedia

(2008b)] or Electrochemical Etching (EE) [Badel (2003)] technologies. DRIE oﬀers

a wider choice of pores sizes and shapes than EE [Uher et al. (2007b)]. On the other

hand DRIE requires expensive equipment.

6.8 Radiation Eﬀects on Silicon Semiconductor Detectors

In this section are reviewed the general detection properties of irradiated silicon

detectors based on the standard planar (SP) and MESA technologies. The later,

less familiar nowadays, is a rather old technology introduced in 1956 [Tanenbaum

and Thomas (1956)] and has been recently revived as an alternative to standard

planar technology with the possibility to produce detectors at lower costs [Sopko,

Hazdra, Kohout, Mr´azek and Posp´ıˇsil (1997)]. The section contains a review of the

spectroscopic features of SP and MESA detectors and their behavior under parti-

cle irradiation. For the comparison, the scanning of non-irradiated and irradiated

detectors is done with heavy charged particles (e.g., α-particles and protons). Pulse-

height spectra are obtained by measuring the signal of the diodes with a standard

spectroscopy system. The study of the peaks evolution with bias voltage can provide

precise information on the active medium structure and charge collection perfor-

mances of the detector. In the case of irradiated detectors, the study of the peak

evolution with bias voltage and irradiation ﬂuence allows the measurement of the

charge collection eﬃciency degradation in various regions of the diodes and structure

alteration. This section also includes a discussion of the violation of non-ionizing

energy-loss (NIEL) scaling. This scaling factor, which expresses the proportional-

ity between NIEL value and resulting damage eﬀects, is modiﬁed for low energy

protons.

6.8.1 MESA Radiation Detectors

The MESA process for building silicon diodes has been known for a long

time [Tanenbaum and Thomas (1956)]. It has been recently revived as an alter-

native to the planar process of silicon detector production that is relatively simpler

and cheaper [Sopko, Hazdra, Kohout, Mr´azek and Posp´ıˇsil (1997)]. This relatively

inexpensive high production yield of MESA detectors combined with satisfactory

charge collection capabilities and overall performance under irradiation could favor

their choice for applications where a large number of detectors is needed such as ac-

tive medium of trackers (pixels and microstrips) or calorimeters-preshowers (pads)

for particle physics experiments. Pad and strip MESA detectors can be produced

according either to standard MESA (SM) technology or to planar MESA (PM)

January 9, 2009 10:21 World Scientiﬁc Book - 9.75in x 6.5in ws-bo ok975x65˙n˙2nd˙Ed

Solid State Detectors 523

-2

-1

1 10 10

V (volts)

Leakage Current (µA)

Standard MESA

Planar MESA

Fig. 6.40 The bulk leakage current (in µA) as a function of the applied bias voltage (in volts)

for standard MESA (SM) and planar MESA(PM) detectors, detector area of 25 mm

(from [Leroy

and Rancoita (2007)]).

technology. The possibility to use SM and PM silicon diodes as radiation detec-

tors has been investigated [Casse et al. (1999b); Chren et al. (2001); Houdayer et

al. (2002)]. The SM process has the disadvantage of yielding diodes that have high

surface current due to improper passivation of cut edges. The standard MESA tech-

nology has been modiﬁed in order to decrease the surface current. More generally,

one also observes much lower bulk leakage currents for planar MESA (PM) detec-

tors compared to standard MESA (SM) detectors (Fig. 6.40). After PM detectors

became available, SM detectors were no longer or rarely used. The leakage current

is larger for the SM detector (about 10 µA for a bias voltage of 100 V, detector area

of 25 mm

) than for the PM detector (around 50 nA for a bias voltage of 100 V,

detector area of 25 mm

This improvement observed from SM to PM results from the application of

chemical etching for producing a MESA structure with subsequent oxidation of

detector edges. As demonstrated next, the PM detectors show electrical features

comparable to those of SP detectors.

The leakage currents of the PM and SP detectors used as reference are very low

and remain low (a few nA cm

−2

for SP detectors, a few tens of nA cm

−2

for PM

detectors) with increasing applied bias voltage, even for bias voltage beyond the full

January 9, 2009 10:21 World Scientiﬁc Book - 9.75in x 6.5in ws-bo ok975x65˙n˙2nd˙Ed

524 Principles of Radiation Interaction in Matter and Detection

n+ p+n

Back Front

00’ w’

dead

w-X

dead

Fig. 6.41 Schematic representation of a MESA detector with the dormant or dead layers on each

side; r represents the range of the incident particle (adapted from Nucl. Instr. and Meth. in

Phys. Res. A 460, Chren, D., Juneau, M., Kohout, Z., Lebel, C., Leroy, C., Linhart, V., Pospisil,

S., Roy, P., Saintonge, A. and Sopko, B., Study of the characteristics of silicon MESA radiation

detectors, 146–158, Copyright (2001), with p ermission from Elsevier).

depletion value.

6.8.1.1 Electrical features of Planar MESA Detectors

Several studies were done to compare the performances of planar MESA (PM) detec-

tors to standard planar (SP) detectors (these studies also involved standard MESA

(SM) detectors) [Chren et al. (2001); Houdayer et al. (2002)]. The detectors used for

these studies were typically PM detectors of average resistivity of 2 kΩ cm, 25 mm

area and 280 µm thickness. The SP detectors had a resistivity from 2.5 kΩ cm to

6.0 kΩ cm, an area from 25 mm

to 100 mm

, and a thickness of 300 µm. A charge

transport model has demonstrated the existence of dormant or dead layers on each

side of the MESA junctions (Fig. 6.41) [Leroy, Roy, Casse, Glaser, Grigoriev and

Lemeilleur (1999b)]. A layer of about 5 µm on each side of a PM diode is acting as a

dead or dormant layer (this layer was 14 µm deep on each side of a SM diode [Leroy,

Roy, Casse, Glaser, Grigoriev and Lemeilleur (1999b)]), consistent with charge col-

lection data. The active thickness of the MESA detector, of which a thickness X

dead

is considered dormant on each side as shown in Fig. 6.41, can be deﬁned as

= w − 2 X

dead

The capacitance-versus-voltage (C −V ) curves are shown in Fig. 6.42 for a PM

detector and a SP detector of same area (25 mm

). The C − V curves are typical

and indicate a full depletion voltage, V

, in the range of (120–150) V for both

types of detectors (PM and SP). These rather large values of V

reﬂect the low

resistivity (2–2.5) kΩ cm of the PM and SP detectors used for the example. The

capacitance measured is larger for the PM detector compared to the SP detector

since the thickness of the PM detector, 270 µ m (280 µm corrected for the 10 µm

dead layer), is smaller than the thickness of the SP detector, of 300 µm.

The average charge collection (in fC) from the signal induced by relativistic elec-

January 9, 2009 10:21 World Scientiﬁc Book - 9.75in x 6.5in ws-bo ok975x65˙n˙2nd˙Ed

Solid State Detectors 525

100

1 10 10

V(volts)

Capacitance (pF)

Fig. 6.42 The C − V curve features of a PM and a SP diode are compared. The full depletion

voltage, V

, is found to be in the range of (120–150) V for both type of diodes (reprinted from Nucl.

Instr. and Meth. in Phys. Res. A 476, Houdayer, A., Lebel, C., Leroy, C., Roy, P., Linhart, V.,

Pospisil, S., Sopko, B., Silicon planar MESA diodes as radiation detectors, 588–595, Copyright

(2002), with permission from Elsevier).

trons from a

106

Ru source is shown in Fig. 6.43 as a function of the applied bias

voltage for a PM detector compared to that of a SP detector. Saturation is achieved

at and beyond V

(value consistent with C − V measurements) for the detectors

shown in Fig. 6.43. The average charge value of ∼ 3.2 fC found at full depletion and

beyond for the PM detector is in agreement with an eﬀective thickness of 274 µm

which is (3–4) % less than the thickness of ∼ 283 µm measured with a micrometer,

reﬂecting the existence of dormant or dead layers which could correspond to the

extent of the B or P diﬀusion zone in the PM process. The total dormant or dead

zone is larger for SM detectors (28 µm) than for PM detectors (10 µm). The average

charge value of ∼ 3.6 fC measured at full depletion and beyond for a SP detector

taken as reference for comparison is consistent with its measured (with a microm-

eter) thickness of 306 µm which is about 10% more than the thickness of the PM

detector. Note that the ratio of the average charge measured for the PM and SP de-

tectors is equal to the ratio of their thicknesses, as it should be. The resistivity of the

particular SP detector used as reference for the measurement was 5.9 kΩ cm, which

explains the smaller value of V

(∼ 50 V) observed for this detector in Fig. 6.43.

January 9, 2009 10:21 World Scientiﬁc Book - 9.75in x 6.5in ws-bo ok975x65˙n˙2nd˙Ed

526 Principles of Radiation Interaction in Matter and Detection

0.5

1.5

2.5

3.5

100 200

V (volts)

Charge (fC)

Fig. 6.43 Comparison of the charge collected (in fC) as a function of the applied bias voltage (in

volts) for PM and SP detectors illuminated by electrons (mips, electron energy larger than 2 MeV

from a

106

Ru source), (reprinted from Nucl. Instr. and Meth. in Phys. Res. A 476, Houdayer,

A., Lebel, C., Leroy, C., Roy, P., Linhart, V., Pospisil, S., Sopko, B., Silicon planar MESA diodes

as radiation detectors, 588–595, Copyright (2002), with permission from Elsevier).

6.8.1.2 Spectroscopic Characteristics of MESA Detectors

The spectroscopic technique has been applied to the study of PM detectors [Chren et

al. (2001); Houdayer et al. (2002)]. Alpha sources (

239

Pu,

241

Am,

244

Cm) were used

to perform spectroscopic measurements with α-particles in vacuum. The spectra

obtained with

239

Pu (α-particle energy of 5.157 MeV),

241

Am (α-particle energy of

5.486 MeV), and

244

Cm (α-particle energy of 5.805 MeV) are shown in Fig. 6.44 for

α-particles incident on the front and back sides of a PM detector for two shaping

times of 0.25 µs [Fig. 6.44(a)–(c)] and 1.0 µs [Fig. 6.44(d)–(f)]. The comparison of

the peaks evolution with voltage for the PM and a silicon surface barrier detector

(SSBD) of reference allows the determination of the CCE.

Shifts among peaks and their splitting are caused by diﬀerent energy losses of

particles entering through non-sensitive layers close to or on the detector surface. As

already emphasized in the section on standard planar detectors, the capability to

measure shifts at the level of a keV is equivalent to measuring layer thicknesses

of about ten nanometers. Therefore, an investigation of the detector ﬁne-structure

becomes possible [Chren et al. (2001)] as illustrated in Fig. 6.45.

January 9, 2009 10:21 World Scientiﬁc Book - 9.75in x 6.5in ws-bo ok975x65˙n˙2nd˙Ed

Solid State Detectors 527

Fig. 6.44 Spectra representing the response of a PM detector to α-particles from Pu, Am, and Cm

sources incident on the front side [(b) and (e) for shaping time

of 0.25 and 1.0

s, respectively] and

back side [(c) and (f) for shaping time τ of 0.25 and 1.0 µs, respectively] of the detector, compared

to the response [(a) and (d) for shaping time τ of 0.25 and 1.0 µs, respectively] of a silicon surface

barrier detector (SSBD) used as reference. The applied reverse bias voltage was ∼ 48 and 80 V for

the SSBD and PM detectors, respectively. The ﬁne structure of alpha spectra is clearly resolved

by the SSBD. The inﬂuence of PM detector electrode structure is explained in Fig. 6.45. The peak

positions are determined with a precision better than 5 keV (reprinted from Nucl. Instr. and

Meth. in Phys. Res. A 460, Chren, D., Juneau, M., Kohout, Z., Lebel, C., Leroy, C., Linhart,

V., Pospisil, S., Roy, P., Saintonge, A. and Sopko, B, Study of the characteristics of silicon MESA

6.8.2 Results of Irradiation Tests of Planar MESA Detectors

The characteristics of MESA detectors as a function of ﬂuence were investigated

in view of their possible use in high particle ﬂuence environment. The irradiation

studies of PM detectors have really been limited up to now to one experiment where

January 9, 2009 10:21 World Scientiﬁc Book - 9.75in x 6.5in ws-bo ok975x65˙n˙2nd˙Ed

528 Principles of Radiation Interaction in Matter and Detection

Fig. 6.45 Illustration of the sensitivity of the spectroscopic method applied to PM detectors. The

response of a SSBD used as reference (a) is compared to the response of a PM detector illuminated

by α-particles (b) on the front side and (c) on the back side. Shifts among p eaks and their splitting

are caused by diﬀerent energy-losses of particles entering through non-sensitive layers close to or

on the detector surface. d

is the diﬀerence of front side electrode structure thickness between the

reference detector (Au + p − n junction) and the PM detector (Al + p − n junction) expressed

in energy losses; d

is the diﬀerence of thickness expressed in energy losses in front side reference

detector structure (Au + p − n) and back side PM detector (Al + n

n). d

is the diﬀerence

expressed in energy losses in front side PM detector structure (Al + p − n) thickness and back

side PM detector structure (Al + n

− n) thickness. Also deﬁned are ∆: the energy loss in front

side Al of PM detector, and ∆

: the energy loss in front side SiO

layer in MESA valley of PM

detector (reprinted from Nucl. Instr. and Meth. in Phys. Res. A 460, Chren, D., Juneau, M.,

Kohout, Z., Lebel, C., Leroy, C., Linhart, V., Pospisil, S., Roy, P., Saintonge, A. and Sopko, B,

Study of the characteristics of silicon MESA radiation detectors, 146–158, Copyright (2001), with

permission from Elsevier).

a set of PM diodes (25 mm

area, ∼ 280 µm thickness and ∼ 2 kΩ cm resistivity)

were exposed to low energy (10 MeV) protons (hardness factor

∗

: theoretical is κ ∼

4.5, measured is ∼ 2.2) [Bechevet, Glaser, Houdayer, Lebel, Leroy, Moll and Roy

(2002)]. The eﬀective doping concentration (N

eﬀ

) of a semiconductor diode can be

extracted from measurement of the capacitance (C) as a function of the applied

∗

The reader can see Sect. 4.1.3 (and references therein) for an general treatment of the hardness

factor.

January 9, 2009 10:21 World Scientiﬁc Book - 9.75in x 6.5in ws-bo ok975x65˙n˙2nd˙Ed

Solid State Detectors 529

Table 6.4 Range of protons in silicon as a function

of energy (from [Leroy and Rancoita (2007)]). The

ranges were calculated as a function of the proton

energy using the SRIM code, which employs Ziegler’s

stopping tables [Ziegler, Biersack and Littmark

(1985a)]

Energy (MeV) 7 8 9 10

Range (µm) 380 476 584 700

bias (V ) (see Sect. 6.1.3).

The eﬀective doping concentration (N

eﬀ

) as a function of ﬂuence (Φ) is given

by Eq. (6.75) [Pitzl et al. (1992)]. The values of the parameters of the irradiated

PM and SP diodes extracted by ﬁtting Eq. (6.75) to the experimental data (N

eﬀ

as a function of proton ﬂuence) show that the parameter b is not aﬀected by the

MESA process [Bechevet, Glaser, Houdayer, Lebel, Leroy, Moll and Roy (2002)],

while the parameter c could be aﬀected since its value should be smaller [Dezillie et

al. (1999)], reﬂecting the higher initial donor concentration (N

) of the PM diodes

compared to the SP diodes.

It is a standard procedure to heat the detectors at 80 C for up to 17 hours to

simulate their ageing during the 10 years of operation of the LHC experiments. The

irradiation induced change in the eﬀective doping concentration (∆N

eﬀ

) as a func-

tion of the heating time (t) is given by [Lindstroem, Moll and Fretwurst (1999)]:

∆N

eﬀ

(t) = N

exp(−t/τ

) + N

+ N

1 −

1 + t/τ

, (6.160)

where N

and τ

are the short-term annealing constants; N

represents the so-called

stable damage constant, which consists of an “incomplete donor removal” and the

introduction of negative space-charge proportional to ﬂuence; ﬁnally, N

and τ

are the long-term reverse annealing constants. From results reported in [Bechevet,

Glaser, Houdayer, Lebel, Leroy, Moll and Roy (2002)] it is observed that before

“inversion”

†

(Φ < 10

p cm

−2

) the short term annealing (N

) is about 3 times

more in amplitude for SP than for the PM detectors, while after inversion (Φ >

−2

), the diﬀerence is about a factor 2. The anti-annealing parameter, N

for PM and SP detectors is comparable before inversion but N

is about 25 % larger

for the PM detectors after inversion.

The leakage current was recorded (guard ring connected) during the measure-

ments of the capacitance (C) as a function of the applied voltage (V ). These

measurements were performed at room temperature with the leakage current

†

The reader can also see the discussion in Sects. 4.3.5 and 4.3.6 about the experimental obser-

vation that, at large particle ﬂuences, the Hall coeﬃcient changes sign in n-type substrate with

high-resistivity (before irradiation).