SNIC Symposium, Stanford, California -- 3-6 April 2006

Precision Crystal Calorimeters in High Energy Physics Ren-Yuan Zhu California Institute of Technology, Pasadena, CA 91125, USA

Pre ision rystal alorimetry traditionally plays an important role in high energy physi s experiments. In the last two de ades, it fa es a hallenge to maintain its pre ision in a hostile radiation environment. This paper reviews the performan e of rystal alorimeters onstru ted for high energy physi s experiments and the progress a hieved in understanding rystal's radiation damage as well as in developing high quality s intillating rystals for parti le physi s. Potential appli ations of new generation s intillating rystals of high density and high light yield, su h as LSO and LYSO, in parti le physi s experiments is also dis ussed.

500 500

500

4

15000

700

2

7

65

10000 1

5000

2 1S0

2 3S1 1

2 3 4

1 3P2 1 3P1

8 5 7 6 1 1S0

8

1 3P0

100

400

200

0

500

Eγ (MeV)

1000

110

b)

2

σ = 1.06% 500

0

Endcaps σ = 0.86%

10000

0 -1 10

1

10

10

2

0

0.9

1

1.1

BGO energy / Beam energy

Energy (GeV)

Figure 2: Left: The energy resolution of the L3 BGO

alorimeter as a fun tion of ele tron energy measured in the CERN test beam. Right: The energy resolution of Bhabha ele trons observed by the L3 BGO alorimeter in situ at LEP by using the RFQ alibration.

p

120

130

E =E = 2:5%= E  0:55%  0:2=E

140

mγγ (GeV)

Figure 1: Left: An in lusive photon spe trum measured at the 0 by the NaI(Tl) rystal alorimeter at SLAC [2℄. Right: The expe ted ba kground subtra ted Higgs mass peak re onstru ted from its two photon de ays measured by the CMS PbWO4 rystal alorimeter [3℄.

Crystal alorimeters have been onstru ted, and their use has been a key fa tor in the su

essful physi s programs of many experiments. With proper ali0038

3

1

1 3S1

0 50

Events/500 MeV for 100 fb–1

/ (2.5% Bin) COUNTS

–500

3

100

Barrel

1000

Endcaps

Table I summarizes parameters of past and present

rystal alorimeters in high energy physi s. One notes that ea h of these alorimeters requires several ubi meters of high quality rystals. The most ambitious

rystal alorimeter in Table I is presumably the CMS

alorimeter whi h uses 11 m3 PbWO4 rystals. Its designed energy resolution [3℄ is

600

0

0 80

11S0-ηc

1000

2 1S0-ηc′

45.6 GeV < EBeam < 94.3 GeV

Barrel

4

Events / 0.0025

Total absorption shower ounters made of inorgani s intillating rystals have been known for de ades for their superb energy resolution and dete tion ef ien y [1℄. In high energy and nu lear physi s, large arrays of s intillating rystals have been assembled for pre ision measurements of photons and ele trons. The dis overy potential of rystal alorimetry was early demonstrated by the Crystal Ball experiment through its study of radiative transitions and de ays of the Charmonium family [2℄. Figure 1 (Left) shows nearly all the prin ipal radiative transition lines of the Charmonium system simultaneously measured by the NaI(Tl) rystal alorimeter. The designed goal of the CMS lead tungstate (PbWO4 ) rystal alorimeter [3℄ is to maximize its dis overy potential in sear hing for narrow resonan es in photon and ele tron nal states at LHC. Figure 1 (Right) shows the expe ted ba kground subtra ted Higgs peak re onstru ted with its two de ay photons by the CMS PbWO4 alorimeter. The potential of the Higgs dis overy via this de ay

hannel is dire tly related to the energy resolution of the alorimeter.

bration and monitoring, rystal alorimeters usually a hieve their designed resolution in situ [4℄. Figure 2 (Left) shows energy resolution as a fun tion of the ele tron energy obtained with the L3 BGO alorimeter in the CERN test beam, whi h is in a good agreement with the Bhabha ele tron resolutions measured in situ at LEP by using the RFQ alibration [5℄, as shown in Figure 2 (Right).

σ(E) / E (%)

1. Introduction

for the barrel, and

p

E =E = 5:7%= E  0:55%  0:25=E

(1)

(2)

for the end aps. Figure 3 (Left) shows the CMS designed energy resolution as a fun tion of energy. It an be de omposed to three ontributions from photoele tron statisti s (sto hasti ), intrinsi shower leakage (sto hasti and

onstant) and readout noise (noise). Figure 3 (Right) 1

SNIC Symposium, Stanford, California - 3-6 April, 2006

Table I Crystal Calorimeter in High Energy Physi s: Past and Present Experiment C. Ball A

elerator SPEAR 75{85 Date Crystal Type NaI(Tl) B-Field (Tesla) Inner Radius (m) 0.254 Number of Crystals 672 16 Crystal Depth (X0 ) 1 Crystal Volume (m3 ) Light Yield (p.e./MeV) 350 Photosensor PMT Gain of Photosensor Large Noise/Channel (MeV) 0.05 Dynami Range 104 a Wavelength Shifter. b Avalan he photodiode.

L3 LEP 80{00 BGO 0.5 0.55 11,400 22 1.5 1,400 Si PD 1 0.8 105

CLEO II CESR 80{00 CsI(Tl) 1.5 1.0 7,800 16 7 5,000 Si PD 1 0.5 104

shows the energy resolution as a fun tion of ele tron energy measured in the CERN test beam for two groups of 3  3 rystals, independent of their impa t position on the rystal front fa e [6℄. The measured resolution in the low and middle energy region agrees with the designed resolution. It also shows a smaller

onstant term be ause of the perfe t alibration in the test beam.

σ/E[%]

10

All

1

C. Barrel LEAR 80{00 CsI(Tl) 1.5 0.27 1,400 16 1 2,000 WSa +Si PD 1 0.2 104

KTeV Tevatron 90{10 CsI 3,300 27 2 40 PMT 4,000 Small 104

BaBar

PEP II 94{10 CsI(Tl) 1.5 1.0 6,580 16 to 17.5 5.9 5,000 Si PD 1 0.15 104

BELLE KEK 94{00 CsI(Tl) 1.0 1.25 8,800 16.2 9.5 5,000 Si PD 1 0.2 104

CMS LHC 95{20 PbWO4 4.0 1.29 76,000 25 11 2 APDb 50 30 105

2. Commonly Used Crystal Scintillators Table II lists the basi properties of ommonly used heavy rystal s intillators: NaI(Tl), CsI(Tl), undoped CsI, BaF2 , CeF3 , bismuth gemanade (Bi4 Ge3 O12 or BGO), lead tungstate (PbWO4 or PWO), Cedoped lutetium oxyorthosili ate (Lu2 (SiO4 )O or LSO(Ce)) [7℄ and Ce-doped Gadolinium Orthosili ate (Gd2 (SiO4 )O or GSO(Ce)) [8℄. Mass produ tion apabilities exist for all these rystals. All, ex ept CeF3 , LSO(Ce) and GSO(Ce), have been used in high energy physi s experiments. LSO(Ce) and GSO(Ce) are widely used in the medi al industry.

Intrinsic Photo Noise

0.1 1

10

100

1000

E[GeV]

Figure 3: Left: The designed energy resolution of the CMS PbWO4 alorimeter is shown as a fun tion of energy and orresponding ontributions [3℄. Right: The energy resolution of two groups of 9 PbWO4 rystals is shown as fun tion of ele tron energy obtained in the CMS ECAL beam test [6℄.

Over the last two de ades, however, rystal

alorimetry fa ed a new hallenge: the radiation damage aused by the in reased enter of mass energy and luminosity. To preserve rystal pre ision in a severe radiation environment rystal quality ontrol is a ru ial issue. The rest of this paper is devoted to a review of opti al and s intillation properties of heavy rystal s intillators ommonly used for the parti le physi s. The progresses a hieved in understanding rystal's radiation damage and in developing high quality rystals is also be dis ussed. 0038

Figure 4: A photo shows ten rystal s intillators with dimension of 1.5 X0 .

The opti al and s intillation properties were measured and ompared for various rystal s intillators. Figure 4 is a photo showing ten rystal samples used in this omparative study. All samples are wrapped with white Tyvek paper as re e tors. NaI and CsI based samples are sealed in quartz window of 3 mm thi k to avoid surfa e degradation aused by their hy2

SNIC Symposium, Stanford, California - 3-6 April, 2006

Table II Properties of Heavy Crystal S intillators with Mass Produ tion Capability Crystal Density (g/ m3 ) Melting Point (Æ C) Radiation Length ( m) Moliere Radius ( m) Intera tion Length ( m) Refra tive Indexa Hygros opi ity Lumines en eb (nm) (at Peak) De ay Timeb (ns)

NaI(Tl) 3.67 651 2.59 4.13 42.9 1.85 Yes 410

CsI(Tl) 4.51 621 1.86 3.57 39.3 1.79 Slight 560

230

1250

Light Yieldb;

100

165

d(LY)/dTb;d (%/Æ C)

0

0.3

CsI 4.51 621 1.86 3.57 39.3 1.95 Slight 420 310 30 6 3.6 1.1 -0.6

BaF2 4.89 1280 2.03 3.10 30.7 1.50 No 300 220 630 0.9 36 3.4 -2 0

CeF3 6.16 1.70 2.41 23.2 1.62 No 300

BGO 7.13 1050 1.12 2.23 22.8 2.15 No 480

30

300

7.3

21

0.14

-1.6

PbWO4 8.3 1123 0.89 2.00 20.7 2.20 No 425 420 30 6 0.29 0.083 -1.9

LSO(Ce) 7.40 2050 1.14 2.07 20.9 1.82 No 420

GSO(Ce) 6.71 1950 1.38 2.23 22.2 1.85 No 440

40

60

84

30

0

-0.1

a At the wavelength of the emission maximum. b Top line: slow omponent, bottom line: fast omponent.

Relative and PMT quantum eÆ ien y taken out. d At room temperature.

100

NaI(Tl)

LSO

60 em: 480 nm

Intensity (a.u.)

80

BaF2 ex: 304 nm

em: 402 nm

X-ray luminescence Peaks: 220 nm, 300 nm

ex: 358 nm

em: 410 nm

60

ex: 346 nm

40

40

20

20

0

100

PWO

LYSO

CsI(Tl)

CeF3

80

80

60 em: 424 nm

ex: 310 nm

em: 402 nm

ex: 358 nm

em: 301 nm

ex: 265 nm

em: 540 nm

60

ex: 322 nm

40

40

20

20

0

200

400

600

400

600

300

400

250

500

Transmittance (%)

BGO 80

2500 2500

Light Output (p.e./MeV)

100

while the BGO, BaF2 , NaI(Tl), PbWO4 and CsI(Tl)

rystals have their emission spe tra well within the transparent region, the UV absorption edge in the transmittan e spe tra of the LSO, LYSO and CeF3

rystals uts into the emission spe tra and thus ae ts

rystal's light output. This ee t is more seriously observed for long LSO and LYSO samples [10℄.

Light Output (p.e./MeV)

gros opi ity. To minimize sample size dependen e all samples have a ube shape, 1:5  1:5  1:5 X30 , ex ept the NaI(Tl) sample whi h is a ylinder with a length of 1.5 X0 and an area of two ends equal to 1:5  1:5 X20 to mat h the 2 in h diameter of the PMT athode.

2000

L.O = F + S ( 1 - e

1500 LSO LYSO

1000

CeF3 CsI PWO

-t/τs

)

F

S

τs

0 0 0 30 1.9

2210 2150 208 101 7.3

42 44 33 30 31

500

2000 1500 1000 500 L.O = F + S ( 1 - e

0

750

NaI(Tl) CsI(Na) CsI(Tl) BaF2 BGO

-500 0

Wavelength (nm)

0

100

200

300

Time (ns)

400

500

0

1000

F 0 0 0 98 0

S 2604 2274 2093 1051 350

2000

-t/τs

)

τs 245 693 1220 655 302

3000

4000

Time (ns)

Figure 5: The ex itation (red) and emission (blue) spe tra (left s ale) and the transmittan e (green) spe tra (right s ale) are shown as a fun tion of wavelength for eight rystal s intillators.

Figure 6: Light output measured by using the XP2254b PMT is shown as a fun tion of integration time for ve fast (Left) and ve slow (Right) rystal s intillators.

Figure 5 shows a omparison of the transmittan e (green), emission (blue) and ex itation (red) spe tra as a fun tion of wavelength for eight samples. The solid bla k dots in these plots show the theoreti al limit of the transmittan e, whi h was al ulated by using orresponding refra tive index as a fun tion of wavelength, taking into a

ount multiple boun es between the two parallel end surfa es and assuming no internal absorption [9℄. The measured transmittan e approa hes the theoreti al limits, indi ating negligible internal absorption. It is interesting to note that

Figure 6 shows light output in photoele trons per MeV energy deposition as a fun tion of integration time, measured by using a Photonis XP2254b PMT with multi-alkali photo athode, for ve fast rystal s intillators (Left): LSO, LYSO, CeF3 , undoped CsI and PbWO4 and ve slow (Right) rystal s intillators: NaI(Tl), CsI(Tl), CsI(Na), BaF2 and BGO. The undoped CsI, PbWO4 and BaF2 rystals are observed to have two de ay omponents as shown in Table II. The LSO and LYSO samples have onsistent fast de ay time (40 ns) and high photoele tron yield, whi h

0038

3

SNIC Symposium, Stanford, California - 3-6 April, 2006

is 6 and 230 times of BGO and PbWO4 respe tively. Hamamatsu PMT R2059

1



BGO: QE=8.0 ± 0.4%  LSO/LYSO: QE=13.6 ± 0.7%  CsI(Tl): QE=5.0 ± 0.2%

0.75

0.15

Hamamatsu APD 

BGO: QE = 82 ± 4%  LSO/LYSO: QE = 75 ± 4%  CsI(Tl): QE = 84 ± 4%

0.1

0.5

LSO LYSO

CsI(Tl)

0.25

0.05

APD Quantum Efficiency

PMT Quantum Efficiency

0.2

Figure 8: A photo shows four long rystal samples with dimension of 2:5  2:5  20 m3 .

BGO

500

600

700

0 800

Wavelength (nm) Figure 7: The quantum eÆ ien ies of a Hamamatsu R2059 PMT (solid dots) and a Hamamatsu S8664 APD (solid squares) are shown as a fun tion of wavelength together with the emission spe tra of the LSO/LYSO, BGO and CsI(Tl) samples, where the area under the emission urves is proportional to their orresponding absolute light output.

size. Their availability provides a new possibility for the pre ision rystal alorimetry. 1500

1500 PMT: R1306 Gate: 2000 ns E.R.: 14.7%

1000 500

500

0 CTI-LSO-L

PMT: R1306 Gate: 300 ns E.R.: 13.0%

200 100 0

CPI-LYSO-L

PMT: R1306 Gate: 300 ns E.R.: 21.2%

200 100

Table III Emission Weighted Quantum EÆ ien ies (%) Emission LSO/LYSO BGO CsI(Tl) Hamamatsu R1306 PMT 12.90.6 8.00.4 5.00.3 Hamamatsu R2059 PMT 13.60.7 8.00.4 5.00.3 Photonis XP2254b 7.20.4 4.70.2 3.50.2 Hamamatsu S2744 PD 594 754 804 Hamamatsu S8664 APD 754 824 844

Table III summarized numeri al result of the emission weighted average QE for several readout devi es. Taking into a

ount the PMT response, we on lude that the light output of the LSO and LYSO rystals is a fa tor of 4 and 200 of that of BGO and PbWO4 respe tively, as shown in Table II. Large size LSO and LYSO rystals with onsistent opti al and s intillation properties have been developed re ently for the medi al industry. Figure 8 shows four long rystal samples: SIC BGO, CTI LSO, CPI LYSO and Saint-Gobain LYSO of 2:5  2:5  20 m3 0038

0 750

CTI-LSO-L

500

ped = 43, peak = 698 L.O. = 2130 p.e./MeV

250 0 750

CPI-LYSO-L

500

ped = 43, peak = 450 L.O. = 1330 p.e./MeV

250 SG-LYSO-L

PMT R1306 Gate 300ns E.R. 11.7%

200 100 0

source: Na-22 2 × Hamamatsu S8664-55 HV = 400 V, τ = 250 ns, M = 500 ped = 43, peak = 173 L.O. = 420 p.e./MeV

1000

0

Sin e the quantum eÆ ien y (QE) of the PMT used for the light output measurement is a fun tion of wavelength, it must be taken out to ompare rystal's photon yield. Figure 7 shows typi al QE of a PMT with bi-alkali athode (Hamamatsu R2059 PMT) and an APD (Hamamatsu S8664), as well as the emission spe tra of LSO/LYSO, BGO and CsI(Tl) rystals.

SIC-BGO-L

SIC-BGO-L

Counts

400

Counts

0 300

0 750

SG-LYSO-L

500

ped = 43, peak = 504 L.O. = 1500 p.e./MeV

250 0

250

500

750

1000

0

0

250

Channel number

500

750

1000

Channel number

Figure 9: The spe tra of 0.511 MeV -rays from a 22 Na sour e measured with a oin iden e trigger using a Hamamatsu R1306 PMT (Left) and two Hamamatsu S8664-55 APDs (Right) for long BGO, LSO and LYSO samples of 2:5  2:5  20 m3 size.

Figure 9 shows spe tra of 0.51 MeV -rays from a sour e observed by these samples with oin iden e triggers. The readout devi es used are a Hamamatsu R1306 PMT (Left) and 2 Hamamatsu S866455 APDs (Right). The FWHM resolution for the 0.51 MeV -ray with the PMT readout is about 12% to 13% for the long LSO and LYSO samples, whi h

an be ompared to 15% for the BGO sample. With APD readout, the -ray peaks are well visible for the long LSO and LYSO samples, but is mu h less distinguished for the BGO sample. The energy equivalent readout noise in our laboratory set up of APD readout is below 40 keV for the LSO and LYSO samples. 22 Na

3. Crystal Radiation Damage All known large size rystal s intillators suer from radiation damage [11℄. There are three possible radiation damage ee ts in rystal s intillators. First, 4

SNIC Symposium, Stanford, California - 3-6 April, 2006

radiation would indu e internal absorption, aused by the olor enter formation, whi h would redu e the light attenuation length [9℄, and thus the light output, and may also ause a degradation of the light response uniformity. CsI(Tl)(BGRI-2)

60

BTCP-2467 80

From top to bottom: After 0, 1, 11, 31, 100 krad

20

60

Transmittance (%)

Transmittance (%)

40

From top to bottom o

200 C annealing 15 rad/h (65 h) 100 rad/h (63 h) 400 rad/h (62 h) 9000 rad/h (10 h)

40

20

0 60

CsI(Tl)(SIC-4)

40 From top to bottom: After 0, 10, 40, 100 220, 500, 1000 rad

20 0

CsI(Tl)(SIC-5)

3 2

From top to bottom: After 0, 10, 100 1k, 10k rad

1

0 300

400

500

600

700

0 300

800

350

400

450

500

550

Wavelength (nm)

600

650

700

750

800

Wavelength (nm)

Figure 10: Longitudinal transmittan e of PbWO4 (Left) and CsI(Tl) (Right) samples, showing radiation indu ed absorption bands.

Figure 10 shows the longitudinal transmittan e spe tra and their degradation under irradiation measured for full size CMS PbWO4 (23 m long, Left) and BaBar CsI(Tl) (30 m long, Right) rystal samples respe tively. The radiation indu ed absorption and

orresponding olor enter formation are learly observed in these samples. Se ond, radiation would indu e phosphores en e (afterglow), whi h would ause an in rease of the readout noise. Last, radiation may also redu e s intillation light yield. If so, both the light output and the light response uniformity would be degraded sin e the radiation dose pro le in situ is usually not uniform. 1.1

100

1

Transmittance (%)

Normalized Light Output

BTCP-2162 L.O.= 9.3 p.e./MeV (200 ns, 20.0oC)

0.9

0.8

0.7

Before Irradiation

80

100 rad (1 rad/s)

60

1 krad (11 rad/s)

40 10 krad (11 rad/s)

20 0

0.1 Mrad (24 rad/s) 1 Mrad (24 rad/s)

(a) Fast

-20 100

Before Irradiation

80

100 rad (0.03 rad/s)

60

1 krad (0.3 rad/s)

40

15 0.5

10 krad (3 rad/s)

20

0.6

0

50

dose rate (rad/h): 100 100

150

500 200

Time (hours)

0

1000

0.1 Mrad (20 rad/s) 1 Mrad (15 rad/s)

(b) Slow

-20 250

300

200

300

400

500

600

700

800

Wavelength (nm)

Figure 11: Left: Normalized light output is plotted as a fun tion of time under irradiations for a CMS PbWO4 sample, showing dose rate dependent radiation damage. Right: No dose rate dependen e was observed for the longitudinal transmittan e spe tra measured for a GEM full size BaF2 sample.

Radiation indu ed absorption may also re over under appli ation temperature through a pro ess alled

olor enter annihilation. If so, the damage would be dose rate dependent [11℄. Figures 11 (Left) shows light 0038

output normalized to that before irradiation (solid dots with error bars) as a fun tion of time under irradiation for a full size CMS PbWO4 sample. Measurements were made step by step for dierent dose rates: 15, 100, 500 and 1,000 rad/h, as shown in these gures. The degradation of the light output shows a

lear dose rate dependen e. If no re overy or the re overy speed is very slow, however, the olor enter annihilation pro ess would be less important, the olor enter density would not rea h an equilibrium under ertain dose rate rather

ontinuous in reasing until all defe t traps are lled. This means no dose rate dependen e. Figure 11 (Right) shows the transmittan e as a fun tion of wavelength for a GEM full size (25 m) BaF2 sample before and after 100, 1k, 10k, 100k and 1M rad irradiations (from top to bottom) under a fast (a) and a slow (b) dose rates. While the fast dose rate is up to a fa tor of three higher than the slow rate, the damage levels for the same integrated dose are identi al. This was expe ted, sin e no re overy at room temperature was observed for BaF2 . Table IV Radiation Damage in Crystal S intillators Item CsI(Tl) CsI BaF2 Color Centers Yes Yes Yes phosphores en e Yes Yes Yes S intillation Damage No No No Re over RT Slow Slow No Dose Rate Dependen e No No No Thermally Annealing No No Yes Opti al Blea hing No No Yes

BGO Yes Yes No Yes Yes Yes Yes

PWO Yes Yes No Yes Yes Yes Yes

Crystal radiation damage may also be ured by either thermal annealing or opti al blea hing [12℄. Table IV summarizes radiation damage observed in various rystal s intillators whi h were used in the high energy and nu lear physi s experiments. For LYSO(Ce) rystals our initial investigation on 2:5  2:5  20 m3 bar samples indi ates that they suer less radiation damage than other rystals, su h as BGO, CsI(Tl) and PbWO4 [10℄. Figure 12 (Left) shows an expanded view of the longitudinal transmittan e spe tra measured for a long LYSO samples before and after ea h step of -ray irradiations at 2, 100 and 9,000 rad/h. An immediate in rease of the longitudinal transmittan e after the rst irradiation under 2 rad/h was observed, whi h was followed by small degradation under higher dose rate. Be ause of LYSO's high light yield its natural phosphores en e and radiation indu ed phosphores en e have small effe t in readout noise. Fig. 12 (Right) shows the ray indu ed PMT anode urrent for two long LYSO samples and a linear t. The radiation indu ed phosphores en e related readout noise with 100 ns integration time is estimated to be about 0.2 MeV and 5

SNIC Symposium, Stanford, California - 3-6 April, 2006 10 19

800 PMT: R2059(BA434), bias=-900V, T=66°F

Gamma Induced Anode Current (µA)

CPI-LYSO-L

Normalized Light Output

80

400

before irradiation

78

2 rad/h (19 h) 100 rad/h (19 h) 76

74 300

200

9000 rad/h (22 h)

400

500

600

700

800

10

0.8

0.6

CsI(Tl)

0.4

Solid Line: Specification SIC-2 SIC-4 SIC-5

0.2

0

0

Wavelength (nm)

2.5

5

7.5

10

12.5

15

17.5

SIC-6 SIC-7 SIC-8

1

10 17

0.1

10 19

CsI(Tl)(SIC-2)

CsI(Tl)(Khar’kov) 10

10 18

1

10 17

20

Gamma Dose Rate (rad/h)

0.1

0 1

Figure 12: Left: The longitudinal transmittan e spe tra before and after 2, 100 and 9,000 rad/h irradiations are shown as a fun tion of wavelength for the Saint-Gobain long LYSO sample. Right: The -ray indu ed anode

urrent is shown as a fun tion of the dose rate for the CPI and Saint-Gobain long LYSO samples.

1 MeV equivalent respe tively in a radiation environment of 15rad/h and 500 rad/h for LYSO samples of 2:5  2:5  20 m3 size.

4. Crystal Development and Quality Improvement Commer ially available mass produ ed rystals usually do not meet the quality required for the pre ision

rystal alorimetry. A resear h and development program is usually needed to systemati ally study the

orrelations between rystal's radiation hardness and its impurities and point defe ts. By removing harmful impurities from the raw materials and developing an approa h to ee tively redu e the density of defe ts in the rystal during the growth and pro essing, the quality of mass produ ed rystals may be improved. This approa h has been su

essfully arried out for BGO [13℄, BaF2 [14℄, CsI(Tl) [11℄ and PbWO4 [11, 15℄. Two examples are given below in this se tion.

4.1. CsI(Tl) Development Figure 13 (Left) shows the light output as a fun tion of a

umulated dose for full size ( 30 m) CsI(Tl) samples produ ed at the Shanghai Institute of Cerami s (SIC), and ompared to the BaBar radiation hardness spe i ation (solid line) [11℄. While the late samples SIC-5, 6, 7 and 8 satisfy the BaBar spe i ation, early samples SIC-2 and 4 did not. This improved radiation hardness of CsI(Tl) rystals was also observed by BaBar and BELLE experiments [16℄. The improvement of CsI(Tl) quality was a hieved following an understanding that the radiation damage in halide rystals is aused by the oxygen or hydroxyl

ontamination. The identi ation of oxygen ontamination was a hieved by the Se ondary Ionization Mass 0038

CsI(Tl)(SIC-T3)

10 18

SG-LYSO-L

600

82

CsI(Tl)(SIC-T1)

1

Concentrations (atoms/cm3)

SG-LYSO-L

Concentrations(ppmW)

Transmittance (%)

84

10

10

2

10

3

10

4

0

2

Integrated Dosage (rad)

4

6

8

0

2

4

6

8

10

Depth (micron)

Figure 13: Left: The progress of CsI(Tl) radiation hardness is shown for full size (30 m) CsI(Tl) samples from SIC together with the rad-hard spe i ation of the BaBar experiment. Right: The depth pro le of oxygen

ontamination is shown for two rad-soft CsI(Tl) samples (SIC-T1 and SIC-2) and two rad-hard samples (SIC-T3 and Khar'kov).

Spe tros opy (SIMS) analysis arried out at Charles Evans & Asso iates. A Cs ion beam of 6 keV and 50 nA was used to bombard the CsI(Tl) sample. All samples were freshly leaved prior before being loaded into the UHV hamber. An area of 0.15  0.15 mm2 on the

leaved surfa e was analyzed. To further avoid surfa e

ontamination, the starting point of the analysis is at about 10 m deep inside the fresh leaved surfa e. Figure 13 (Right) shows the depth pro le of oxygen

ontamination for two radiation soft samples (SIC-T1 and SIC-2) and two radiation hard samples (SIC-T3 and Khar'kov). Crystals with poor radiation resistan e have oxygen ontamination of 1018 atoms/ m3 or 5.7 ppmW, whi h is 5 times higher than the ba kground ount (21017 atoms/ m3 , or 1.4 ppmw). The pra ti al solution at SIC is to use a s avenger to remove oxygen. This leads to the development shown in Figure 13 (Left).

4.2. PbWO4 Development Figure 14 shows the light output as a fun tion of time under various dose rates for CMS full size (23

m) PbWO4 samples produ ed at SIC [11℄. Samples produ ed late 2002 is mu h more radiation hard than the early samples. This improved radiation hardness of PbWO4 rystals was also on rmed by an evaluation of mass produ ed PbWO4 rystals [17℄. The improvement of PbWO4 quality was a hieved following an understanding that the radiation damage in oxide rystals is aused by the oxygen or latti e stru ture va an ies. By using Transmission Ele tron Mi ros opy (TEM) oupled to Energy Dispersion Spe trometry (EDS), a lo alized stoi hiometry analysis was used to identify oxygen va an ies. A TOPCON-002B S ope was rst used at 200 kV and 10 A. Samples were made to powders of an average 6

SNIC Symposium, Stanford, California - 3-6 April, 2006

1.1

1.1

Normalized Light Output

Sample, 05/1998

Sample, 12/1998

1

1

0.9

0.9

0.8

0.8

0.7

0.7 dose rate : rad/h 15 100 500

0.6 0

50

100

150

1.1

dose rate : rad/h 15 100 500

0.6 0

100

150

Sample, 08/2000

1

1

0.9

0.9

0.8

0.8 dose rate(rad/h): 2

0.7

50

1.1

Sample, 11/1999

15

dose rate (rad/h):

100

0 100 200 300 400 500 600

15

0.7

0

50

100

100

400

150

200

Time (hours) Figure 14: The progress of PbWO4 radiation hardness is shown for full size (23 m) CMS PbWO4 samples from SIC.

By employing a TEM with EDS system, a lo alized stoi hiometry analysis was arried out at SIC [18℄. The system is a JEOL JEM-2010 s ope and a Link ISIS EDS. The spatial resolution of this system allows a lo alized stoi hiometry analysis in a region of a diameter of 0.5 nm. An as grown sample was rst analyzed, and bla k spots were observed. Points inside and surrounding the bla k spots were analyzed as well as points far away from the bla k spots. The un ertainty of the analysis is typi ally 15%. The resultant atomi fra tions (%) at these areas are listed in Table V. A lear deviation from the atomi stoi hiometry of O:W:Pb = 66:17:17 was observed in the

enter of these bla k spots, pointing to a severe de it of the oxygen omponent. In the peripheral area, the oxygen de it was less, but still signi ant. There was no oxygen de it observed in the area far away from the bla k spots. As a omparison, the same sample after oxygen ompensation was re-analyzed. No bla k spot was found. The result of the analysis is also listed in Table V. In all randomly sele ted points no stoi hiometry deviation was observed. This analysis thus

learly identi ed oxygen va an ies in PbWO4 samples of poor radiation hardness. Table V Atomi Fra tion (%) of O, W and Pb in PbWO4 Samples Measured by TEM/EDS [18℄ As Grown Sample Element Bla k Spot Peripheral Matrix1 Matrix2 O 1.5 15.8 60.8 63.2 W 50.8 44.3 19.6 18.4 Pb 47.7 39.9 19.6 18.4

grain size of a few m, and then pla ed on a sustainÆ ing membrane. With a spatial resolution of 2 A, the latti e stru ture of PbWO4 rystals was learly visible. Figure 15 (Left) shows a TEM pi ture taken for a sample with poor radiation hardness. Bla k spots of a diameter of 5 { 10 nm were learly seen in the pi ture. On the other hand, samples with good radiation hardness show stable TEM pi ture with no bla k spots, as shown in Figure 15 (Right).

The Same Sample after Oxygen Compensation Element O W Pb

Point1 59.0 21.0 20.0

Point2 66.4 16.5 17.1

Point3 57.4 21.3 21.3

Point4 66.7 16.8 16.5

Various approa hes were tried to ompensate oxygen va an ies by annealing PbWO4 rystals in an oxygen-ri h atmosphere [11℄ and by doping [15℄. Signi ant improvement of radiation hardness was observed in both ases. The pra ti al solution at SIC is to dope PbWO4 rystals with yttrium. This leads to the development shown in Figure 14. Figure 15: TEM pi tures of a PbWO4 rystal of poor (Left) radiation hardness, showing learly the bla k spots of 5{10 nm related to oxygen va an ies, as ompared to that of a good one (Right).

0038

5. An LSO/LYSO Crystal Calorimeter As dis ussed in previous se tions LSO(Ce) and LYSO(Ce) rystals are a new type of rystal s intil7

SNIC Symposium, Stanford, California - 3-6 April, 2006

lators with light yield 4 and 200 times of BGO and PbWO4 respe tively and a de ay time about 40 ns. The LYSO rystals are also known to suer less radiation damage as ompared to other ommonly used

rystal s intillators. Mass produ tion apability of LSO/LYSO rystals has been established for the medi al industry. Crystals of size suÆ ient for building a

rystal alorimeter are routinely grown. Assuming the same readout s heme as the CMS PbWO4 alorimeter, the expe ted energy resolution of an LSO/LYSO based rystal alorimeter would be

p

E =E = 2%= E  0:5%  0:001=E;

(3)

whi h represents a fast alorimeter over large dynami range with very low noise. Su h alorimeter would provide ex ellent physi s potential for high energy physi s experiment in the International Linear Collider [19℄ or in a super B fa tory [20℄.

6. Summary Pre ision rystal alorimetry has been an important part of high energy physi s dete tor. Its energy resolution, position resolution and photon identi ation apability has been a key fa tor in many physi s dis overies. In the last two de ades, however, it fa es a hallenge: the radiation damage in s intillation rystals. Progresses have been made in understanding rystal's radiation damage and developing high quality rystals for high energy physi s experiments. Re ent availability of mass produ tion apability of large size LSO and LYSO rystals provides an opportunity to build a rystal alorimeter with unpre edent energy resolution over a large dynami range down to MeV level. This rystal alorimeter, if built, would greatly enhan e the dis overy potential for future high energy and nu lear physi s experiments.

Acknowledgments Work supported by U.S. Department of Energy Grant No. DE-FG03-92-ER40701.

[2℄ E. Bloom and C, Pe k, Ann. Rev. Nu l. Part. S i. 33 143-197 (1983). [3℄ The CMS Ele tromagneti Calorimeter Proje t, CERN/LHCC 97-33 (1997). [4℄ R.Y. Zhu, Nu l. Instr. and Meth. A537 344 (2005). [5℄ U. Chaturvedi et al., Nu l. Instr. and Meth. A461 376 (2001). [6℄ A. Zabi, in Pro eedings of the 12th International Conferen e on Calorimetry in Parti le Physi s, Chi ago, (2006). [7℄ C. Mel her, V.S. Patent 4958080 (1990) and 5025151 (1991). [8℄ K. Takagi and T. Fakazawa, Appl. Phys. Lett. 42 43 (1983). [9℄ D.A. Ma and R.Y. Zhu, Nu l. Instr. and Meth. A333 (1993) 422. [10℄ J.M. Chen et al., IEEE Trans. Nu l. S i. 52 (2005) 3133. [11℄ R.-Y. Zhu, Nu l. Instr. and Meth. A413 (1998) 297{311. [12℄ D.A. Ma and R.Y. Zhu, Nu l. Instr. and Meth. A332 113 (1993) and D.A. Ma et al., Nu l. Instr. and Meth. A356 309 (1995). [13℄ Z.Y. Wei et al., Nu l. Instr. and Meth. A297 163 (1990). [14℄ R.Y. Zhu, Nu l. Instr. and Meth. A340 442 (1994). [15℄ X.D. Qu et al., Nu l. Instr. and Meth. A480, 470 (2002), [16℄ T. Hryn'ova, in Pro eedings of the 10th International Conferen e on Calorimetry in Parti le

[17℄ [18℄ [19℄

[20℄

national Conferen e on Calorimetry in Parti le

, Ed. R.-Y. Zhu, World S ienti (2002).

Physi s

References [1℄ G. Gratta et 453 (1994).

0038

,

al.

Annu. Rev. Nu l. Part. S i.

, World S ienti , Ed. R.-Y. Zhu, 175 (2002). R.H. Mao et al., IEEE Trans. Nu l. S i. NS-51 (2004) 1777. Z.W. Yin et al., in Pro eedings of SCINT97 Int'l Conf., Ed. Zhiwen Yin et al., CAS Shanghai Bran h Press, (1997) 191. R.-Y. Zhu, An LSO/LYSO Crystal Calorimeter for the ILC, talk presented in 2005 ILC Workshop, Snowmass. See http://www.hep. alte h.edu/ zhu/talks/ryz 050818 l .pdf. W. Wisniewski, Consideration for Calorimetry at a Super B Fa tory, in Pro eedings of Tenth InterPhysi s

44

8