Properties of LYSO and Recent LSO Scintillators for ... - IEEE Xplore

IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 51, NO. 3, JUNE 2004

789

Properties of LYSO and Recent LSO Scintillators for Phoswich PET Detectors Catherine Michelle Pepin, Student Member, IEEE, Philippe Bérard, Student Member, IEEE, Anne-Laure Perrot, Member, IEEE, Claude Pépin, Daniel Houde, Roger Lecomte, Member, IEEE, Charles L. Melcher, Member, IEEE, and Henri Dautet, Member, IEEE

Abstract—The luminescence and nuclear spectroscopic properties of the new cerium-doped rare-earth scintillator lutetium-yttrium oxyorthosilicate (Lu0 6 Y1 4 Si0 5 :Ce, LYSO) were investigated and compared to those of both recent and older LSO crystals. UV-excited luminescent spectra outline important similarities between LYSO and LSO scintillators. The two distinct Ce1 and Ce2 luminescence mechanisms previously identified in LSO are also present in LYSO scintillators. The energy and timing resolutions were measured using avalanche photodiode (APD) and photomultiplier tube (PMT) readouts. The dependence of energy resolution on gamma-ray energy was also assessed to unveil the crystal intrinsic resolution parameters. In spite of significant progress in light output and luminescence properties, the energy resolution of these scintillators appears to still suffer from an excess variance in the number of scintillation photons. Pulse-shape discrimination between LYSO and LSO scintillators has been successfully achieved in phoswich assemblies, confirming LYSO, with a sufficient amount of yttrium to modify the decay time, to be a potential candidate for depth-of-interaction determination in multicrystal PET detectors. Index Terms—Intrinsic resolution, LSO, LYSO, scintillation detector, scintillation processes.

I. INTRODUCTION

T

HE last decade has seen the introduction of several new high luminosity scintillators that are promising candidates for applications in medical imaging. The availability of a variety of crystals with a range of different scintillation properties has triggered a renewed interest for phoswich detectors, which allow the depth of interaction (DOI) information within multicrystal assemblies to be determined simply by pulse shape discrimination (PSD). However, the number of scintillators with suitable characteristics for PSD identification and coincidence detection in PET remains fairly limited. The first choice is evidently LSO [1], because it has about the same stopping power as BGO, but is seven times faster with a light output three (APD)

Manuscript received January 28, 2003; revised April 5, 2004. This work was supported in part by the Natural Sciences and Engineering Council of Canada under Collaborative R&D Grant CRD 217968-98, in collaboration with PerkinElmer Optoelectronics and CTI, Inc. C. M. Pepin, P. Bérard, C. Pépin, D. Houde, and R. Lecomte are with the Department of Nuclear Medicine and Radiobiology, Université de Sherbrooke, Sherbrooke, QC J1H 5N4, Canada (e-mail: [email protected]). A.-L. Perrot is with the European Organization of Nuclear Research (CERN), CH-12111 Geneva 23, Switzerland (e-mail: [email protected]). C. L. Melcher is with CTI, Inc., Knoxville, TN 37932 USA (e-mail: [email protected]). H. Dautet is with PerkinElmer Optoelectronics, Vaudreuil, QC J7V 8P7, Canada (e-mail: [email protected]). Digital Object Identifier 10.1109/TNS.2004.829781

to five (PMT)times higher. LSO is now a mature crystal that can be produced in quantity with stable scintillation properties [2], [3]. A variant of LSO in which some of the lutetium is replaced by yttrium atoms has recently been developed at CTI, Inc. (Knoxville, TN). Cerium-doped lutetium-yttrium oxyortho-silicate (Lu Y SiO :Ce, LYSO) has a comparable light yield to LSO with a slightly longer decay time of 53 ns, making it an attractive candidate for PSD identification in phoswich detectors. In this work, the scintillation performance of the new LYSO scintillator was investigated with PMT and APD readouts, and compared to the most recent LSO production. Older LSO crystals were also measured concurrently as a reference. II. THEORY The scintillation properties of Ce-activated rare-earth oxyorthosilicates are controlled by two distinct luminescence centers, known as Ce1 and Ce2 [4], [5], which are attributed to Ce substituted lattice sites and interstitial Ce centers, respectively [6]. The presence of two concurrent scintillation processes is believed to be detrimental to the energy resolution of Ce-activated crystals [7]. The energy resolution of scintillators coupled to APDs can be , expressed as the sum of contributions from electronic noise , and scintillator resolution [8], [9]: multiplication noise . The scintillator resolution can be further broken down into a statistical term , which is dependent on the photon yield , and the crystal intrinsic resolution . The excess noise factor , describing the can be used as variance in excess of Poisson statistics, and parameters to describe scintillator intrinsic characteristics. III. MATERIALS AND METHODS A. Description of Crystals The properties of the scintillators LYSO, LSO, GSO and BGO are summarized in Table I. For nuclear spectroscopy of measurements, crystals were wrapped in several layers Teflon tape and optically coupled with optical grade silicone. All crystals were cleaned successively with ethanol, acetone and methanol prior to wrapping. As a reference, three older LSO crystals previously used in scintillation studies with APDs ([3], [7], [8], LSOA and LSOB [3]) were also tested. The dimensions and surface treatment of the crystals are summarized in Table II.

0018-9499/04$20.00 © 2004 IEEE

790


Fig. 1. UV-excitation and emission spectra for LSO scintillator.

TABLE I SCINTILLATOR PROPERTIES

luminescence spectra were obtained by irradiation with a Cs source (662 keV). Light emitted by the crystals was concentrated onto the aperture of an ISA monochromator (HR 320) with a BK7 lens of 100 mm focal length. The -ray excited spectra were then collected by an intensified photodiode array equipped with an S-20 photocathode (IPDA 1024 from Princeton Instruments). All measurements were performed at room temperature and corrected for the spectral efficiencies of the spectral-dependent monochromator and photodetector. C. Nuclear Spectroscopy Measurements

TABLE II DIMENSIONS AND SURFACE TREATMENT OF SCINTILLATORS

A standard procedure described elsewhere [10]–[13] was used to chemically etch the crystals. Measurements of light output and energy resolution were performed with the crystals optically coupled to a PMT before and after polishing to assess the efficiency and reproducibility of the etching process. B. UV- and -Excited Luminescence Measurements UV-excited luminescence spectra were obtained using a Hitachi F2000 spectrofluorometer with the 2 2 10 mm LSO and LYSO scintillators. Using the same crystals, -excited

The second set of measurements involved standard nuclear spectroscopy techniques to determine the relative light output, energy resolution at several gamma-ray energies, coincidence timing resolution, as well as the pulse-shape discrimination performance of LYSO and LSO crystals. All measurements were performed using PerkinElmer APDs connected to a low-noise phointegrated CMOS charge-sensitive preamplifier and a tomultiplier tube (PMT). An ORTEC 452 spectroscopy amplifier was employed for all energy resolution measurements with a shaping time of 0.25 s. Scintillation decay time measurements were carried out for the LSO#9g and LYSO#9g scintillators using standard time correlated single photon technique originally developed by Bollinger and Thomas [14]. The crystals were excited by a Na source. The data were fitted with a single exponential term to extract the decay time. Coincidence timing measurements as well as PSD measurements of LYSO/LSO phoswich were performed using a zero-cross time circuit based on two CF discriminators with 10 ns (Ortec 579) and 100 ns (Ortec 474) shaping times [15]. IV. RESULTS A. UV-Excited Luminescence Spectra Typical LSO and LYSO UV spectra are shown in Figs. 1 and 2. At first glance, the luminescence spectra of the new generation LSO and LYSO crystals are quite similar. However, when compared to the emission spectra of the older LSOref scintillator [3], [7], one observes that the excitation bands at and nm, associated with Ce2 emission, are significantly

PEPIN et al.: PROPERTIES OF LYSO AND RECENT LSO SCINTILLATORS FOR PHOSWICH PET DETECTORS

791

Fig. 2. UV-excitation and emission spectra for LYSO scintillator.

Fig. 3. -excited luminescence spectra for LSOref and new generation LSO and LYSO. TABLE III RELATIVE LIGHT OUTPUT AND ENERGY RESOLUTION OF SCINTILLATORS AT 662 keV (

Cs) BEFORE AND AFTER ETCHING MEASURED WITH A PMT READOUT

enhanced. This Ce2 luminescence mode is revealed by the maximum of the emission spectra, which is clearly shifted toward 440–445 nm for both LSO and LYSO when the crystals are exnm. cited at

UV-excited spectra. For the newer LSO and LYSO, the maxnm with a prominent imum emission is shifted toward shoulder at longer wavelengths (see arrow in Fig. 3), which can be correlated with the more efficient Ce2 luminescence.

B.

C. Effect of Surface Treatment

-Excited Luminescence Spectra

The -excited luminescence spectra for LSOref and the new generation LSO and LYSO are reported in Fig. 3. The maximum emission of LSOref is centered at 425 nm, in agreement with the

The light output and energy resolution of unpolished and chemically etched LSO and LYSO scintillators have been measured. The results are summarized in Table III. The etching

792


Fig. 4. Typical energy spectrum from with APD readout.

Cs obtained for an LYSO scintillator

Fig. 5. Typical energy spectrum from with APD readout.

Cs obtained for an LSO scintillator

2 2

Fig. 6. Energy resolution of 4 4 4 mm LSO and LYSO scintillators as a function of APD bias at 662 keV ( Cs, = 250 ns).

Fig. 7. Components of energy resolution in LSO#9g. LIGHT

TABLE IV OUTPUT AND ENERGY RESOLUTION SCINTILLATORS AT 662 keV (

2 2 10 MM

4 4 Cs)

OF

light output was more important in LYSO than LSO. Similarly, the improvement in energy resolution after etching is impresto % at sive with the 2 2 10 mm , dropping from 662 keV. For the larger crystals, the gain in energy resolution is still significant and the resulting resolution is comparable to the one measured for the mechanically polished LSOref scintillator. D. Nuclear Spectroscopy

process slightly reduced the weight of the crystals ( % for % for LYSO), the smaller crystals being more LSO and affected because of their larger surface to weight ratio. The LYSO was also more sensitive to etching. Before etching, the light yield of the recent LSO and LYSO scintillators with a 4 4 mm cross section was already superior to that of LSOref. For 2 2 10 mm crystals, the light yield could be dramatically lower, by up to a factor of two in % LYSO. While the gain in light output is only marginal with the cubic 4 4 4 mm crystals after etching, it is signif% with 4 4 10 mm crystals and striking with icant the 2 2 10 mm crystals (20%–40%). In all cases, the gain in

The averaged light output and energy resolution of 4 4 10 mm LYSO and recent LSO crystals irradiated Cs -ray source (662 keV) are reported in Table IV by a and compared to data obtained for GSO, BGO and older LSO scintillators. Typical LSO and LYSO spectra are shown in Figs. 4 and 5. In spite of higher light outputs (from 11 to 36%) than the older LSOref, the newer crystals show little gain in energy resolution. Fig. 6 shows the energy resolution of scintillators LSO#9g and LYSO#9g as a function of the APD bias at 662 keV. The behavior of these two crystals is very similar, both achieving an energy resolution of about 10%. Further investigation of the was performed by measuring the elecscintillator resolution tron yield and energy resolution as a function of -ray energy. Figs. 7 and 8 display the components of energy resolution in typical LSO and LYSO scintillators, respectively, as a function


Fig. 8.

793

Components of energy resolution in LYSO#9g. Fig. 10. energy.

Nonproportionality of LSO light output as a function of incident -ray

Fig. 9. Components of energy resolution in LSOref.

TABLE V COMPARISON OF MEASURED SCINTILLATION CHARACTERISTICS

Fig. 11. Nonproportionality of LYSO light output as a function of incident

-ray energy.

but a deteriorated of . Significant nonproportionalities, as shown in Figs. 10 and 11, were noticed in the scintillation response of the recent LSO and LYSO crystals, in agreement with previously reported data [16]–[19]. of primary electrons generated in the APD. These results can be compared to those obtained for the older LSOref crystal (Fig. 9). The scintillation parameters extracted from similar measurements performed on several recent LSO and LYSO samples are reported and compared to previously reported data in Table V. The electron yield of newer LSO and LYSO crystals is almost % higher than the electron yield of the older identical, and LSOref. However, there is no progress in energy resolution, as this is reflected in the values obtained for the excess variance and the intrinsic enin the number of scintillation photons for recent ergy resolution . The fitted value LSO is similar to the one of 6.5 reported in a previous work [8] and to the one of 5.2 measured in the present study for LSOref. However, the intrinsic energy resolution of newer LSO is even % versus 3.5%). The worse than the one of LSOref ( LYSO crystals achieve a better intrinsic resolution than LSOref

E. Coincidence Timing and Pulse Shape Discrimination Figs. 12 and 13 present the scintillation decay time measurement for LSO#9g and LYSO#9g scintillators. The decay time constants are 41.9 ns for LSO#9g and 50.8 ns for LYSO#9g. This difference in decay time allowed PSD crystal identification to be successfully achieved with APD readout in an LYSO/LSO phoswich assembly, provided that photoelectric energy gating was implemented (Fig. 14). PSD time resolutions of 1.94 ns and 2.34 ns were obtained for LSO and LYSO, respectively, with a separation of 4.1 ns. Coincidence time resolution measurements in reference to a fast plastic/PMT detector as a function of the APD bias are presented in Figs. 15 and 16 for LSO#9g and LYSO#9g, respectively. These measurements yielded timing resolutions of 1.3 ns FWHM for LSO#9g and 1.5 ns FWHM for LYSO#9g when read out by the APD operated at a gain of 57 (bias of 560V).

794


Fig. 15. Fig. 12.

Coincidence timing resolution as a function of APD bias for LSO#9g.

Scintillation decay time measurement for LSO#9g.

Fig. 16. Coincidence timing resolution as a function of APD bias for LYSO#9g. Fig. 13.

Scintillation decay time measurement for LYSO#9g.

surface treatment. Chemical etching would therefore be the preferred procedure with these small crystals. The gain in performance was observed to be slightly superior for LYSO, probably as a result of the more aggressive etching process with this lower density material. B. Scintillator Performance Characteristics

Fig. 14. Energy-gated PSD time spectrum of a LYSO/LSO phoswich detector optically coupled to an APD.

V. DISCUSSION A. Effect of Surface Treatment The etching process used to polish the crystals had a beneficial effect on light yield, but variable outcome for energy resolution. The small crystals with a large aspect ratio (e.g., 2 2 10 mm ) clearly benefited the most from the

The light yield of the most recent LSO scintillators has improved significantly in comparison to older LSO. The new scintillator LYSO also exhibits a high light output, though slightly lower than LSO, in spite of the presence of yttrium atoms, which would be expected to enhance scintillation efficiency. Unfortunately, the higher light output does not translate into any significant improvement of the energy resolution. Even with highly optimized detector setups enhancing light collection and avoiding systematic position dependence, the energy resolution can not be improved. As this has been pointed out previously [7], [16]–[19], some fundamental process other than photon statistics obviously must be responsible for the limited energy resolution of these scintillators. Our results suggest that both the intrinsic crystal resolution and the excess variance in the number of scintillation photons contribute to degrade the overall energy resolution. Since the material is clear and exempt of bubbles or impurities on


visual inspection, the nonzero intrinsic resolution in the small cubic 4 4 4 mm crystals used in this study is likely due to inhomogeneities of the scintillation efficiency across the crystal. Further investigation could be planned to verify if such spatially dependent luminous efficiency is present on a macroscopic or microscopic scale by irradiating selected regions of the crystals with a highly collimated source. cannot be excluded as another sigThe excess variance nificant source of resolution degradation in LSO and LYSO. The excess variance is commonly attributed to the important nonproportionality of the photon yield observed in rare-earth oxyorthosilicate crystals [16]–[19]. The origin of this nonlinear response is not well understood. However, the energy resolution was related to the existence of two competing scintillation processes in cerium-activated crystals [7]. The higher light output in recent LSO and LYSO was obtained by enhancing Ce2 emission, while preserving high Ce1 efficiency. As previously postulated, the coexistence of the two efficient luminescence processes, which appears to be necessary to achieve a high light yield in rare-earth Ce-activated oxyorthosilicate crystals, may well be the cause of the limited energy resolution of current LSO and LYSO. C. Pulse Shape Discrimination The discrimination of LSO and LYSO in a phoswich assembly was shown to be possible using the PSD technique, proving LYSO to be a potential candidate for use in conjunction with LSO in multicrystal PET detectors. However, PSD is achieved at the expense of a significant loss of detection efficiency because photoelectric energy gating must be applied. Progress in light output, APD performance (quantum efficiency, gain) and pulse shape discrimination techniques will be required for this feature to be easily implemented in PET scanners. ACKNOWLEDGMENT The authors would like to thank D. Rouleau for his technical support. REFERENCES [1] C. L. Melcher and J. S. Schweitzer, “Cerium-doped lutetium oxyorthosilicate: A fast, efficient new scintillator,” IEEE Trans. Nucl. Sci., vol. 39, pp. 502–505, 1992.

795

[2] C. L. Melcher, M. Schmand, M. Eriksson, L. Eriksson, M. Casey, R. Nutt, J. Lefaucheur, and B. Chai, “Scintillation properties of LSO: Ce boules,” IEEE Trans. Nucl. Sci., vol. 47, pp. 965–968, 2000. [3] C. M. Pepin, P. Bérard, and R. Lecomte, “Comparison of LSO LGSO and MLS scintillators,” in Proc. 2001 IEEE Nuclear Science Symp./Medical Imaging Conf. Rec., vol. 1, San Diego, CA, Nov. 2001, pp. 124–128. [4] H. Suzuki, “UV and gamma-ray excited luminescence of cerium-doped rare-earth oxyorthosilicates,” Nucl. Instrum. Methods Phys. Res. A, vol. 320, pp. 263–272, 1992. [5] H. Suzuki, T. A. Tombrello, C. L. Melcher, and J. S. Schweitzer, “Light emission mechanism of Lu (SiO )O:Ce,” IEEE Trans. Nucl. Sci., vol. 40, pp. 380–383, 1993. [6] J. D. Naud, T. A. Tombrello, C. L. Melcher, and J. S. Schweitzer, “The role of cerium sites in the scintillation mechanism of lso,” IEEE Trans. Nucl. Sci., vol. 43, pp. 1324–1328, 1996. [7] A. Saoudi, C. Pepin, C. Pépin, D. Houde, and R. Lecomte, “Scintillation light emission studies of LSO scintillators,” IEEE Trans. Nucl. Sci., vol. 46, pp. 1925–1928, 1999. [8] R. Lecomte, C. Pepin, D. Rouleau, H. Dautet, R. J. McIntyre, D. Sween, and P. Webb, “Investigation of GSO, LSO and YSO scintillators using reverse avalanche photodiodes,” IEEE Trans. Nucl. Sci., vol. 45, pp. 478–482, 1998. [9] R. Lecomte, C. Pepin, D. Rouleau, H. Dautet, R. J. McIntyre, D. McSween, and P. Webb, “Radiation detection measurements with new ‘buried junction’ silicon avalanche photodiode,” Nucl. Instrum. Methods Phys. Res. A, pp. 92–102, 1999. [10] J. S. Huber, W. W. Moses, M. S. Andreaco, M. Loope, C. L. Melcher, and R. Nutt, “Geometry and surface treatment dependence of the light collection from LSO crystals,” Nucl. Instrum. Methods Phys. Res. A, vol. 437, pp. 374–380, 1999. [11] K. Kurashige, Y. Kurata, H. Ishibashi, and K. Susa, “Surface polishing of GSO scintillator using chemical process,” IEEE Trans. Nucl. Sci., vol. 45, pp. 522–524, June 1998. [12] R. Slates, A. Chatziioannou, B. Fehlberg, T. Lee, and S. Cherry, “Chemical polishing of LSO crystals to increase light output,” IEEE Trans. Nucl. Sci., vol. 47, pp. 1018–1023, June 2000. [13] D. Strul, J. Sutcliffe-Goulden, P. Halstead, and P. K. Marsden, “Optimization of fiber-optic readout of LSO scintillation crystals with acid etching,” IEEE Trans. Nucl. Sci., vol. 49, pp. 619–623, June 2002. [14] L. M. Bollinger and G. E. Thomas, “Measurement of the time dependence of scintillation intensity by a delayed-coincidence method,” IEEE Trans. Nucl. Sci., vol. NS-32, pp. 1044–1050, 1961. [15] A. Saoudi, C. M. Pepin, F. Dion, M. Bentourkia, R. Lecomte, M. Andreaco, M. Casey, R. Nutt, and H. Dautet, “Investigation of depth-of-interaction by pulse shape discrimination in multicrystal detectors read out by avalanche photodiodes,” IEEE Trans. Nucl. Sci., vol. 46, pp. 462–467, 1999. [16] P. Dorenbos, J. T. M. de Haas, C. W. E. van Eijk, C. L. Melcher, and J. S. Schweizer, “Non-linear response in the scintillation yield of Lu SiO :Ce ,” IEEE Trans. Nucl. Sci., vol. 41, pp. 735–737, 1994. [17] P. Dorenbos, J. T. M. de Haas, and C. W. E. van Eijk, “Nonproportionality in the scintillation response and the energy resolution obtainable with scintillation crystals,” IEEE Trans. Nucl. Sci., vol. 42, pp. 2190–2202, 1995. [18] T. D. Taulbee, B. D. Rooney, W. Mengesha, and J. D. Valentine, “The measured electron response nonproportionalities of CaF , BGO and LSO,” IEEE Trans. Nucl. Sci., vol. 44, pp. 489–493, 1997. [19] B. D. Rooney and J. D. Valentine, “Scintillator light yield nonproportionality: Calculating photon response using measured electron response,” IEEE Trans. Nucl. Sci., vol. 44, pp. 509–516, 1997.