- Email: [email protected]
Impact of annealing of trapping times on charge collection in irradiated silicon detectors
ARTICLE IN PRESS
Nuclear Instruments and Methods in Physics Research A 579 (2007) 762–765 www.elsevier.com/locate/nima
Impact of annealing of trapping times on charge collection in irradiated silicon detectors$ G. Krambergera,, V. Cindroa, I. Mandic´a, M. Mikuzˇa,b a
Institute Jozˇef Stefan, Jamova 39, SI-1111 Ljubljana, Slovenia Department of Physics, University of Ljubljana, SI-1000 Ljubljana, Slovenia
b
Available online 31 May 2007
Abstract The evolution of effective trapping times with time in position sensitive silicon detectors at the experiments at Large Hadron Collider (LHC) has been calculated for envisaged operation scenario. The trapping probability of holes will increase by 30% when compared to the value at the end of beneficial annealing for the same total fluence. The effective trapping probability of electrons on the other hand will decrease by around 15%. Possible operation scenarios for an upgrade of LHC (SLHC) were investigated and the differences in terms of charge trapping were compared. The simulation confirms the observations that at fluences Feq 42 1015 cm2 the long term annealing does not affect much the CCE of highly segmented nþ -p devices. r 2007 Elsevier B.V. All rights reserved. PACS: 85.30.De; 29.40.Wk; 29.40.Gx Keywords: Effective carrier trapping time; Silicon detectors; Charge collection efficiency
1. Introduction Most of the recent experiments in high energy physics use position sensitive silicon detectors for tracking of charged particles. At the Large Hadron Collider (LHC) high collision rates and track multiplicities will cause substantial radiation damage of silicon detectors [1]; even more is envisaged after a possible increase of the LHC luminosity by a factor of 10 to 1035 cm2 s1 (SLHC) [2]. Consequently, fluences received by the detectors at SLHC will range from 1014 neq cm2 at outer silicon tracker at r ¼ 100 cm to the 1016 neq cm2 for the innermost layer of the pixel detector at r ¼ 4 cm. At small radii the damage is dominated by fast charged hadrons (mostly pions), while for r420 cm the neutrons originating from the calorimeter prevail. The radiation damage reflects in an increase of: effective dopant concentration (N eff ), leakage current and effective trapping probabilities (1=teff;e;h ) of the $
Work performed in the framework of the CERN-RD50 collaboration.
Corresponding author. Tel.: + 386 1 477 3512; fax: + 386 1 425 7074.
E-mail address: [email protected] (G. Kramberger). 0168-9002/$ - see front matter r 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.nima.2007.05.294
drifting electrons and holes [3]. The latter determines the charge collection efficiency (CCE) and consequently the performance of irradiated silicon detectors. The defects in the Si-lattice which are responsible for the degradation of the detector performance change with time after the irradiation. Annealing of the defects responsible for changes of N eff were studied in details by RD48 collaboration [3]. Much scarce are on the other hand the annealing studies of 1=teff;e;h . So far, for most of the samples evolution of teff e;h ðtÞ was measured at certain temperature [4–6] and only recently systematic annealing studies were performed at different temperatures [7]. This procedure enables the scaling of the annealing time constants to close to the operation temperatures and by that the prediction of the teff;e;h evolution during the LHC lifetime. Moreover, at highest SLHC fluences the annealing of teff e;h could explain small effect of long term annealing on charge collection in highly segmented nþ -p devices observed in Ref. [8]. Due to short trapping times loss of the depletion depth does not affect CCE so much as at lower fluences.
ARTICLE IN PRESS G. Kramberger et al. / Nuclear Instruments and Methods in Physics Research A 579 (2007) 762–765
The effective trapping probability scales linearly with fluence as 1=teff e;h ¼ be;h ðt; TÞFeq
(1)
where be;h is the proportionality constant depending on time after irradiation t and operation temperature T. Most of the b measurements performed so far were done after the completion (tmin ) of beneficial annealing of N eff . The average of measurements taken from Ref. [4–7,9,10] and scaled to T ¼ 10 C are given in the Table 1. The changing of trapping times with time can be modeled by the decay of the dominant electron or hole trap into another stable one. The resulting ansatz is be;h ðtÞ ¼ b0e;h et=te;h þ b1e;h ð1 et=te;h Þ
(2)
with b0e;h and b1e;h the trapping constants at early ðt ! 0Þ and late ðt ! 1Þ annealing times, respectively. The parameters of the model are gathered in the Table 2, together with the activation energy used for scaling trapping times to lower temperatures (Arrhenius relation) [7]. For the relevant temperatures be;h ðtmin Þ b0e;h . It is worth mentioning that values for ðb0 b1 Þ=b0 are obtained as average not only from Transient Current Technique (TCT) measurements [4–7] but also from ATLAS pixel test-beam data [11]. It is assumed here that the ðb0 b1 Þ=b0 does not depend on irradiation particle type. 3. Results and discussion In terms of trapping probability the long term annealing is non-beneficial for holes and beneficial for electrons (see Table 2). This has an important consequence. Unlike in diodes the contribution of electrons ðQe Þ and holes ðQh Þ to the total induced charge Q differs and depends on geometry [12]. In segmented device the larger fraction of the signal comes from the type of carriers drifting to the sensing electrodes. For example for tracks going through the center of the ATLAS strip detector (thickness ¼ 280 mm, Table 1 Trapping time damage constants for neutron and hadron irradiated silicon detectors tmin , T ¼ 10 C
bh ð1016 cm2 ns1 Þ
be ð1016 cm2 ns1 Þ
Reactor neutrons Fast charged hadrons
5:7 1 6:6 0:9
3:7 0:6 5:4 0:4
pitch ¼ 80 mm, implant width ¼ 18 mm) the ratio is Qh =Q ¼ 0:81. Approximately the same ratio Qe =Q ¼ 0:83 is obtained for a 280 mm thick ATLAS pixel detector (pixel size 50 400 mm2 ). It is evident that even if electric field is high enough to allow to reach the saturation velocity in the entire detector volume (ideal case) the performance of detectors with pþ readout (strip detectors) will not only be significantly lower after irradiation, but it will also degrade with time. On the contrary the performance of the detectors with highly segmented nþ readout (pixel detectors) will improve. The evolution of effective trapping times was simulated using the ATLAS-SCT operation scenario: 100 days operation at 7 C, 2 days at 20 C, 14 days at 17 C and for remaining of the year at 7 C. The results are shown in Fig. 1 for detector located at r ¼ 26 cm (first layer of strip detectors in ATLAS) from the interaction point where contributions from fast charged hadrons and neutrons are approximately equal [1]. For comparison also effective trapping probabilities are shown as obtained after annealing for 4 min at 80 C which is often used to account for the maintenance period when the detectors are kept at ambient temperature. The 1=teff at the end of operation is around 30% higher for holes and 15% lower for electrons. Although the time constants in Eq. (2) are comparable with the one for reverse annealing of N eff [3], the activation energies are not. Smaller activation energies result in shorter time constants than those of the N eff annealing at relevant temperatures (10 to 20 C). As a consequence at the end of LHC lifetime N eff increases for about a 20% of the reverse annealing amplitude [1] and 1=teff changes up to more than half of its annealing amplitude. Evolution of effective trapping probabilities is governed by the first order process, hence the simulations shown for ATLAS-SCT (see Fig. 1) can be scaled to any detector (i.e. fluence) with an equivalent running scenario (time at given temperature). The scaling factor is calculated from the difference in equivalent fluence and irradiation particles composition (see Table 1). 0.14
ATLAS –SCT 1st Layer (r = 26 cm)
0.12
holes holes (min. V fd ) electrons (min. V fd) electrons
0.1 1/τ eff [1/ns]
2. Simulation procedure
763
0.08 0.06 0.04 0.02
Table 2 Parameters used to model annealing of effective trapping times
Electrons Holes
0 0
t (min at 60 C)
ðb0 b1 Þ=b0
E ta (eV)
650 250 530 250
0:35 0:15 0:4 0:2
1:06 0:1 0:98 0:1
500
1000 1500 2000 2500 3000 3500 4000 time [day]
Fig. 1. Evolution of effective trapping probabilities in ATLAS experiment at the location of the first SCT layer. The neutron contribution to the total fluence of 2 1014 neq cm2 was assumed to be 50%.
ARTICLE IN PRESS G. Kramberger et al. / Nuclear Instruments and Methods in Physics Research A 579 (2007) 762–765
764
It is also interesting to see how the 1=teff e;h would evolve at SLHC. At SLHC fluences the effective drift path (a product of saturation velocity and teff e;h ) will become as small as few tens of microns (see Eq. (1)), hence only the carriers created near readout electrodes (high weighting field, see Ref. [12]) will significantly contribute to the signal. This means that even if the depletion voltage increases beyond applied bias voltage no large drop of CCE will occur as long as the high electric field is still present at the readout electrodes. This opens up the possibility to keep the detectors with nþ readout at close to the room temperatures during non-operation periods. The beneficial long term annealing of 1=teff;e would compensate for loss of CCE due to loss of some depletion depth. Moreover the reverse annealing amplitude tends to saturate at high fluences of Feq 45 1014 cm2 for oxygen rich materials [3], hence the relative influence of reverse annealing on full
depletion voltage ðV fd Þ would be reduced. The confirmation of such reasoning are measurements of CCE using nþ p strip detectors of ATLAS geometry irradiated to 7:5 1015 24 GeV protons cm2 [13]. These measurements were simulated using simulation tools described in Ref. [14], where a constant N eff was assumed. At lower fluences the measured signal agrees well with simulation (see Fig. 2a), while at higher fluences the simulation underestimates the measurements. The reason for that could be a smaller 1=teff;e than the one predicted by Eq. (1) at high fluences; this effect has already been observed in Ref. [15]. The simulated trend of CCE during annealing agrees qualitatively with measurements as can be seen in Fig. 2b. The evolution of the effective trapping probabilities at SLHC is shown in Fig. 3 for two different operation scenarios. For the standard—LHC—operation scenario
25000 20000
Φeq =4.5 x 1015 cm –2 4000 Signal [ e ]
Signal [ e ]
Neff = – 0.0071Φ eq Vbias =–900 V
15000
Vbias = –750 V
5000
measured simulated - ATLAS geom.
T = –10 o C 10000
3000 2000 RA as for DOFZ
5000
1000
RA as for STFZ measurements
0
0 0
10
20
30
40
50
60
0
100
200
300
400
500
time [80 o C]
Φeq [1014 cm–2]
Fig. 2. (a) Comparison of measured charge and simulated charge in 280 mm thick nþ -p strip detector with pitch 80 mm for at different 24 GeV proton fluences. (b) Dependence of collected charge on annealing time at 80 C. Two cases are shown; first where reverse annealing amplitude saturates (RA - DOFZ) and second where it does not (RA - STFZ).
holes holes (min. Vfd) electrons (min. Vfd) electrons
8 1/τ eff [1/ns]
7 6 5 4
9
holes holes (min. Vfd) electrons (min. Vfd) electrons
8 7 1/τ eff [1/ns]
9
6 5 4 3
3 2
1st Layer of ATLAS–pixel (r=4 cm)
1
2
1st Layer of ATLAS–pixel (r=4 cm)
1 0
0 0
200 400 600 800 1000 1200 1400 1600 1800 2000 time [day]
0
200 400 600 800 1000 1200 1400 1600 1800 2000 time [day]
Fig. 3. Evolution of effective trapping times for: (a) for a standard LHC operation scenario and (b) if detectors are kept at 20 C during non-operation and at 20 C during 100 days of operation. The total equivalent fluence of 1016 cm2 of charged hadrons was assumed.
ARTICLE IN PRESS G. Kramberger et al. / Nuclear Instruments and Methods in Physics Research A 579 (2007) 762–765
(Fig. 3a) the effect of annealing is smaller than at the end of LHC shown in Fig. 1. On the other hand it is much larger for ‘‘warm scenario’’ where at the end of SLHC operation be;h ! b1;e;h (Fig. 3b). It is clear that detectors should be designed in such way, if possible, to have a maximum Qe =Q. Very promising candidates for the future experiments are Si detectors made on epitaxial or MCz Silicon. The V fd of such detectors can be to some extent controlled and kept sufficiently low also at very high fluences [16]. The effective acceptors formed during long term annealing compensate for the positive stable damage–donors. This leads to relatively low V fd , if an operational scenario where detectors are kept for substantial time at ambient temperature is chosen. Such a scenario would also result in significant decrease of leakage current ðI leak Þ, due to the leakage current annealing [17]. Smaller leakage pffiffiffiffiffiffiffiffiffi current leads to reduction of shot noise ðENCI leak / I leak Þ [18] and power dissipation ðPdetector / I leak Þ. 4. Conclusions The annealing of trapping times has an important effect on the optimization of the operative conditions of position sensitive silicon detector. It is beneficial for detectors with nþ side readout and non-beneficial for pþ side readout. At the end of the ATLAS experiment trapping probabilities of holes (electrons) will be around 30% larger (15% smaller) than initial probabilities at the same total fluence. At the highest SLHC fluences the reverse annealing influence to
765
CCE will be reduced and storing detectors at 20 C during beam-off period may be beneficial if detectors with nþ readout and high oxygen content are chosen. References [1] ATLAS ID Technical Design Report, CERN/LHCC/97-16, Geneve, 1997. [2] F. Gianotti, et al., hep-ph/0204087, 2002. [3] G. Lindstro¨m, et al., Nucl. Instr. and Meth. A 466 (2001) 308. [4] G. Kramberger, et al., Nucl. Instr. and Meth. A 481 (2002) 297. [5] A.G. Bates, M. Moll, Nucl. Instr. and Meth. A 555 (2005) 113. [6] O. Krasel, et al., IEEE Trans. Nucl. Sci. NS-51 (1) (2004) 3055. [7] G. Kramberger, et al., Nucl. Instr. and Meth. A 571 (2007) 608. [8] G. Casse, et al., Update of annealing measurements on heavily irradiated p-type sensors, presented at 6th RD50 Workshop on Radiation Hard Semiconductor Devices for Very High Luminosity Colliders, Helsinki, June, 2005. [9] J. Weber, et al., Measurement of the trapping time constant in neutron-irradiated silicon pad detectors, presented at IEEE NSS-MIC Symposium, San Diego, October, 2006. [10] E. Fretwurst, et al., Survey of radiation damage studies at Hamburg, presented at 3rd CERN-RD50 Workshop, CERN, Geneve, 2004. [11] T. Lari, et al., Nucl. Instr. and Meth. A 518 (2004) 349. [12] S. Ramo, Proc. IRE 27 (1939) 584. [13] P.P. Allport, et al., IEEE Trans. Nucl. Sci. NS-52 (5) (2005) 1903. [14] G. Kramberger, et al., Nucl. Instr. and Meth. A 450 (2000) 288. [15] G. Kramberger, et al., Nucl. Instr. and Meth. A 554 (2005) 212. [16] CERN-RD50 Status Report 2005, CERN-LHCC-2005-037, Geneve, 2005. [17] M. Moll, Radiation damage in silicon particle detectors-microscopic defects and macroscopic properties, Ph.D. Thesis, Hamburg, DESY-THESIS-1999-040, 1999. [18] V. Radeka, Ann. Rev. Nucl. Part. Sci. 38 (1988) 217.
Nuclear Instruments and Methods in Physics Research A 579 (2007) 762–765 www.elsevier.com/locate/nima
Impact of annealing of trapping times on charge collection in irradiated silicon detectors$ G. Krambergera,, V. Cindroa, I. Mandic´a, M. Mikuzˇa,b a
Institute Jozˇef Stefan, Jamova 39, SI-1111 Ljubljana, Slovenia Department of Physics, University of Ljubljana, SI-1000 Ljubljana, Slovenia
b
Available online 31 May 2007
Abstract The evolution of effective trapping times with time in position sensitive silicon detectors at the experiments at Large Hadron Collider (LHC) has been calculated for envisaged operation scenario. The trapping probability of holes will increase by 30% when compared to the value at the end of beneficial annealing for the same total fluence. The effective trapping probability of electrons on the other hand will decrease by around 15%. Possible operation scenarios for an upgrade of LHC (SLHC) were investigated and the differences in terms of charge trapping were compared. The simulation confirms the observations that at fluences Feq 42 1015 cm2 the long term annealing does not affect much the CCE of highly segmented nþ -p devices. r 2007 Elsevier B.V. All rights reserved. PACS: 85.30.De; 29.40.Wk; 29.40.Gx Keywords: Effective carrier trapping time; Silicon detectors; Charge collection efficiency
1. Introduction Most of the recent experiments in high energy physics use position sensitive silicon detectors for tracking of charged particles. At the Large Hadron Collider (LHC) high collision rates and track multiplicities will cause substantial radiation damage of silicon detectors [1]; even more is envisaged after a possible increase of the LHC luminosity by a factor of 10 to 1035 cm2 s1 (SLHC) [2]. Consequently, fluences received by the detectors at SLHC will range from 1014 neq cm2 at outer silicon tracker at r ¼ 100 cm to the 1016 neq cm2 for the innermost layer of the pixel detector at r ¼ 4 cm. At small radii the damage is dominated by fast charged hadrons (mostly pions), while for r420 cm the neutrons originating from the calorimeter prevail. The radiation damage reflects in an increase of: effective dopant concentration (N eff ), leakage current and effective trapping probabilities (1=teff;e;h ) of the $
Work performed in the framework of the CERN-RD50 collaboration.
Corresponding author. Tel.: + 386 1 477 3512; fax: + 386 1 425 7074.
E-mail address: [email protected] (G. Kramberger). 0168-9002/$ - see front matter r 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.nima.2007.05.294
drifting electrons and holes [3]. The latter determines the charge collection efficiency (CCE) and consequently the performance of irradiated silicon detectors. The defects in the Si-lattice which are responsible for the degradation of the detector performance change with time after the irradiation. Annealing of the defects responsible for changes of N eff were studied in details by RD48 collaboration [3]. Much scarce are on the other hand the annealing studies of 1=teff;e;h . So far, for most of the samples evolution of teff e;h ðtÞ was measured at certain temperature [4–6] and only recently systematic annealing studies were performed at different temperatures [7]. This procedure enables the scaling of the annealing time constants to close to the operation temperatures and by that the prediction of the teff;e;h evolution during the LHC lifetime. Moreover, at highest SLHC fluences the annealing of teff e;h could explain small effect of long term annealing on charge collection in highly segmented nþ -p devices observed in Ref. [8]. Due to short trapping times loss of the depletion depth does not affect CCE so much as at lower fluences.
ARTICLE IN PRESS G. Kramberger et al. / Nuclear Instruments and Methods in Physics Research A 579 (2007) 762–765
The effective trapping probability scales linearly with fluence as 1=teff e;h ¼ be;h ðt; TÞFeq
(1)
where be;h is the proportionality constant depending on time after irradiation t and operation temperature T. Most of the b measurements performed so far were done after the completion (tmin ) of beneficial annealing of N eff . The average of measurements taken from Ref. [4–7,9,10] and scaled to T ¼ 10 C are given in the Table 1. The changing of trapping times with time can be modeled by the decay of the dominant electron or hole trap into another stable one. The resulting ansatz is be;h ðtÞ ¼ b0e;h et=te;h þ b1e;h ð1 et=te;h Þ
(2)
with b0e;h and b1e;h the trapping constants at early ðt ! 0Þ and late ðt ! 1Þ annealing times, respectively. The parameters of the model are gathered in the Table 2, together with the activation energy used for scaling trapping times to lower temperatures (Arrhenius relation) [7]. For the relevant temperatures be;h ðtmin Þ b0e;h . It is worth mentioning that values for ðb0 b1 Þ=b0 are obtained as average not only from Transient Current Technique (TCT) measurements [4–7] but also from ATLAS pixel test-beam data [11]. It is assumed here that the ðb0 b1 Þ=b0 does not depend on irradiation particle type. 3. Results and discussion In terms of trapping probability the long term annealing is non-beneficial for holes and beneficial for electrons (see Table 2). This has an important consequence. Unlike in diodes the contribution of electrons ðQe Þ and holes ðQh Þ to the total induced charge Q differs and depends on geometry [12]. In segmented device the larger fraction of the signal comes from the type of carriers drifting to the sensing electrodes. For example for tracks going through the center of the ATLAS strip detector (thickness ¼ 280 mm, Table 1 Trapping time damage constants for neutron and hadron irradiated silicon detectors tmin , T ¼ 10 C
bh ð1016 cm2 ns1 Þ
be ð1016 cm2 ns1 Þ
Reactor neutrons Fast charged hadrons
5:7 1 6:6 0:9
3:7 0:6 5:4 0:4
pitch ¼ 80 mm, implant width ¼ 18 mm) the ratio is Qh =Q ¼ 0:81. Approximately the same ratio Qe =Q ¼ 0:83 is obtained for a 280 mm thick ATLAS pixel detector (pixel size 50 400 mm2 ). It is evident that even if electric field is high enough to allow to reach the saturation velocity in the entire detector volume (ideal case) the performance of detectors with pþ readout (strip detectors) will not only be significantly lower after irradiation, but it will also degrade with time. On the contrary the performance of the detectors with highly segmented nþ readout (pixel detectors) will improve. The evolution of effective trapping times was simulated using the ATLAS-SCT operation scenario: 100 days operation at 7 C, 2 days at 20 C, 14 days at 17 C and for remaining of the year at 7 C. The results are shown in Fig. 1 for detector located at r ¼ 26 cm (first layer of strip detectors in ATLAS) from the interaction point where contributions from fast charged hadrons and neutrons are approximately equal [1]. For comparison also effective trapping probabilities are shown as obtained after annealing for 4 min at 80 C which is often used to account for the maintenance period when the detectors are kept at ambient temperature. The 1=teff at the end of operation is around 30% higher for holes and 15% lower for electrons. Although the time constants in Eq. (2) are comparable with the one for reverse annealing of N eff [3], the activation energies are not. Smaller activation energies result in shorter time constants than those of the N eff annealing at relevant temperatures (10 to 20 C). As a consequence at the end of LHC lifetime N eff increases for about a 20% of the reverse annealing amplitude [1] and 1=teff changes up to more than half of its annealing amplitude. Evolution of effective trapping probabilities is governed by the first order process, hence the simulations shown for ATLAS-SCT (see Fig. 1) can be scaled to any detector (i.e. fluence) with an equivalent running scenario (time at given temperature). The scaling factor is calculated from the difference in equivalent fluence and irradiation particles composition (see Table 1). 0.14
ATLAS –SCT 1st Layer (r = 26 cm)
0.12
holes holes (min. V fd ) electrons (min. V fd) electrons
0.1 1/τ eff [1/ns]
2. Simulation procedure
763
0.08 0.06 0.04 0.02
Table 2 Parameters used to model annealing of effective trapping times
Electrons Holes
0 0
t (min at 60 C)
ðb0 b1 Þ=b0
E ta (eV)
650 250 530 250
0:35 0:15 0:4 0:2
1:06 0:1 0:98 0:1
500
1000 1500 2000 2500 3000 3500 4000 time [day]
Fig. 1. Evolution of effective trapping probabilities in ATLAS experiment at the location of the first SCT layer. The neutron contribution to the total fluence of 2 1014 neq cm2 was assumed to be 50%.
ARTICLE IN PRESS G. Kramberger et al. / Nuclear Instruments and Methods in Physics Research A 579 (2007) 762–765
764
It is also interesting to see how the 1=teff e;h would evolve at SLHC. At SLHC fluences the effective drift path (a product of saturation velocity and teff e;h ) will become as small as few tens of microns (see Eq. (1)), hence only the carriers created near readout electrodes (high weighting field, see Ref. [12]) will significantly contribute to the signal. This means that even if the depletion voltage increases beyond applied bias voltage no large drop of CCE will occur as long as the high electric field is still present at the readout electrodes. This opens up the possibility to keep the detectors with nþ readout at close to the room temperatures during non-operation periods. The beneficial long term annealing of 1=teff;e would compensate for loss of CCE due to loss of some depletion depth. Moreover the reverse annealing amplitude tends to saturate at high fluences of Feq 45 1014 cm2 for oxygen rich materials [3], hence the relative influence of reverse annealing on full
depletion voltage ðV fd Þ would be reduced. The confirmation of such reasoning are measurements of CCE using nþ p strip detectors of ATLAS geometry irradiated to 7:5 1015 24 GeV protons cm2 [13]. These measurements were simulated using simulation tools described in Ref. [14], where a constant N eff was assumed. At lower fluences the measured signal agrees well with simulation (see Fig. 2a), while at higher fluences the simulation underestimates the measurements. The reason for that could be a smaller 1=teff;e than the one predicted by Eq. (1) at high fluences; this effect has already been observed in Ref. [15]. The simulated trend of CCE during annealing agrees qualitatively with measurements as can be seen in Fig. 2b. The evolution of the effective trapping probabilities at SLHC is shown in Fig. 3 for two different operation scenarios. For the standard—LHC—operation scenario
25000 20000
Φeq =4.5 x 1015 cm –2 4000 Signal [ e ]
Signal [ e ]
Neff = – 0.0071Φ eq Vbias =–900 V
15000
Vbias = –750 V
5000
measured simulated - ATLAS geom.
T = –10 o C 10000
3000 2000 RA as for DOFZ
5000
1000
RA as for STFZ measurements
0
0 0
10
20
30
40
50
60
0
100
200
300
400
500
time [80 o C]
Φeq [1014 cm–2]
Fig. 2. (a) Comparison of measured charge and simulated charge in 280 mm thick nþ -p strip detector with pitch 80 mm for at different 24 GeV proton fluences. (b) Dependence of collected charge on annealing time at 80 C. Two cases are shown; first where reverse annealing amplitude saturates (RA - DOFZ) and second where it does not (RA - STFZ).
holes holes (min. Vfd) electrons (min. Vfd) electrons
8 1/τ eff [1/ns]
7 6 5 4
9
holes holes (min. Vfd) electrons (min. Vfd) electrons
8 7 1/τ eff [1/ns]
9
6 5 4 3
3 2
1st Layer of ATLAS–pixel (r=4 cm)
1
2
1st Layer of ATLAS–pixel (r=4 cm)
1 0
0 0
200 400 600 800 1000 1200 1400 1600 1800 2000 time [day]
0
200 400 600 800 1000 1200 1400 1600 1800 2000 time [day]
Fig. 3. Evolution of effective trapping times for: (a) for a standard LHC operation scenario and (b) if detectors are kept at 20 C during non-operation and at 20 C during 100 days of operation. The total equivalent fluence of 1016 cm2 of charged hadrons was assumed.
ARTICLE IN PRESS G. Kramberger et al. / Nuclear Instruments and Methods in Physics Research A 579 (2007) 762–765
(Fig. 3a) the effect of annealing is smaller than at the end of LHC shown in Fig. 1. On the other hand it is much larger for ‘‘warm scenario’’ where at the end of SLHC operation be;h ! b1;e;h (Fig. 3b). It is clear that detectors should be designed in such way, if possible, to have a maximum Qe =Q. Very promising candidates for the future experiments are Si detectors made on epitaxial or MCz Silicon. The V fd of such detectors can be to some extent controlled and kept sufficiently low also at very high fluences [16]. The effective acceptors formed during long term annealing compensate for the positive stable damage–donors. This leads to relatively low V fd , if an operational scenario where detectors are kept for substantial time at ambient temperature is chosen. Such a scenario would also result in significant decrease of leakage current ðI leak Þ, due to the leakage current annealing [17]. Smaller leakage pffiffiffiffiffiffiffiffiffi current leads to reduction of shot noise ðENCI leak / I leak Þ [18] and power dissipation ðPdetector / I leak Þ. 4. Conclusions The annealing of trapping times has an important effect on the optimization of the operative conditions of position sensitive silicon detector. It is beneficial for detectors with nþ side readout and non-beneficial for pþ side readout. At the end of the ATLAS experiment trapping probabilities of holes (electrons) will be around 30% larger (15% smaller) than initial probabilities at the same total fluence. At the highest SLHC fluences the reverse annealing influence to
765
CCE will be reduced and storing detectors at 20 C during beam-off period may be beneficial if detectors with nþ readout and high oxygen content are chosen. References [1] ATLAS ID Technical Design Report, CERN/LHCC/97-16, Geneve, 1997. [2] F. Gianotti, et al., hep-ph/0204087, 2002. [3] G. Lindstro¨m, et al., Nucl. Instr. and Meth. A 466 (2001) 308. [4] G. Kramberger, et al., Nucl. Instr. and Meth. A 481 (2002) 297. [5] A.G. Bates, M. Moll, Nucl. Instr. and Meth. A 555 (2005) 113. [6] O. Krasel, et al., IEEE Trans. Nucl. Sci. NS-51 (1) (2004) 3055. [7] G. Kramberger, et al., Nucl. Instr. and Meth. A 571 (2007) 608. [8] G. Casse, et al., Update of annealing measurements on heavily irradiated p-type sensors, presented at 6th RD50 Workshop on Radiation Hard Semiconductor Devices for Very High Luminosity Colliders, Helsinki, June, 2005. [9] J. Weber, et al., Measurement of the trapping time constant in neutron-irradiated silicon pad detectors, presented at IEEE NSS-MIC Symposium, San Diego, October, 2006. [10] E. Fretwurst, et al., Survey of radiation damage studies at Hamburg, presented at 3rd CERN-RD50 Workshop, CERN, Geneve, 2004. [11] T. Lari, et al., Nucl. Instr. and Meth. A 518 (2004) 349. [12] S. Ramo, Proc. IRE 27 (1939) 584. [13] P.P. Allport, et al., IEEE Trans. Nucl. Sci. NS-52 (5) (2005) 1903. [14] G. Kramberger, et al., Nucl. Instr. and Meth. A 450 (2000) 288. [15] G. Kramberger, et al., Nucl. Instr. and Meth. A 554 (2005) 212. [16] CERN-RD50 Status Report 2005, CERN-LHCC-2005-037, Geneve, 2005. [17] M. Moll, Radiation damage in silicon particle detectors-microscopic defects and macroscopic properties, Ph.D. Thesis, Hamburg, DESY-THESIS-1999-040, 1999. [18] V. Radeka, Ann. Rev. Nucl. Part. Sci. 38 (1988) 217.