Abstract--The work presents analysis of the correlation existing

between the spectrometric performance of commercially produced hemispherical detectors of volume 500 mm3 and the quality of starting CdZnTe material. More then 100 detectors made from CdZnTe crystals grown by eV Products, Saint-Gobain and Yinnel Tech. were studied. For the incoming inspection of the material an infrared (IR) transmission microscope was used. The presence of structural defects such as grain boundaries, twins, inclusions and cavities was checked and their shapes, sizes and spatial distributions were examined; the values of electron mobility-lifetime product (µτ)e were measured. The hemispherical detectors were made by a technology elaborated at the RITEC. Spectrometric performance of the detectors was verified by measuring the energy resolution and the peak-to-Compton ratio at the 662 keV line. It was found that all tested detectors had numerous structure defects, whose shapes, sizes, numbers and spatial distributions varied from sample to sample. Were found that hemispherical detectors made from samples containing inclusions of irregular shape had a poor spectrometric performance, while the detectors from samples containing large inclusions (>50 µm) of regular shape (triangular, hexagonal) as a rule performed well. It was also found that a highly non-uniform spatial distribution of structural defects usually leads to a poor spectrometric performance.

Index Terms--CdZnTe, hemispherical, gamma-ray detector, structural defects.

I. INTRODUCTION CdZnTe material is successfully used for making

commercial room-temperature gamma-ray hemispherical detectors [1]-[3] with an energy resolution <18 keV at 662 keV line and volume of 0.5 cm3. The progress achieved recently in the field of crystal growing [4], [5] makes it possible to improve of the quality of the starting material and to raise the good-to-bad ratio of the material for making commercial detectors.

However, despite the mentioned progress the yield of commercially available detectors possessing high spectrometric performance is, as of yet, rather low. This is so both for hemispherical detectors and the coplanar grid detectors [6]. Therefore it is now important to determine the factors influencing – and limiting - the characteristics of the detectors.

This work was supported in part by the International Atomic Energy Agency under research contract number No.10401/R2

The experience accumulated in production of spectrometric quasi-hemispherical detectors of the CZT/500(S) type shows that quite often it is insufficient to have only a small set of parameters characterizing the material in order to estimate its fitness for making high-quality detectors of gamma-radiation. Well-known parameters - specific resistance, electron mobility-lifetime product (µτ)e, visually controlled absence of structural defects in the form of large-angle grain boundaries and twins, energy resolution in the 59.6 keV line – not always can fully determine the material’s quality. It often happens that detectors made from materials with identical characteristics and by the same technology possess essentially differing characteristics. The difference can be determined by the presence of diversified and non-uniformly distributed in a detector’s spatial structural defects.

We have tried to generalize the experience gained in production of hemispherical detectors from materials of different producers.

Hemispherical detectors belong to the class of so-called detectors with a single-charge collection [7]. In such detectors the charge collection is done at the cost of the electronic component. The efficiency and uniformity of the electron collection mainly determine detectors performance. Presence of structural defects inside of detectors in many respects influences on charge collection uniformity.

In our work we have tried to reveal the most specific features of materials of different producers and to determine the extent of their influence on the parameters of the hemispherical detectors. Study of structural defects nature, reasons and conditions of theirs formation is out of scope of the work presented.

II. CONTROL PROCEDURES For the examination, more then 100 commercially-produced

CZT/500S detectors with a sensitive volume of 0.5 cm3 were chosen. These quasi-hemispherical 1.0х1.0х0.5 cm3 detectors were mainly produced from the material of eV Products [8] and also from the material of Yinnel Tech. [9] and Saint-Gobain [10]. All the detectors were made by the same technology (worked out at the RITEC Ltd [1]) comprising a procedure of incoming inspection.

The procedure of incoming inspection consists in the following: visual examination of the mechanically polished

Correlation Between Quality of CZT Crystals and Spectrometric Performance

of Hemispherical Radiation Detectors V. Ivanov, L. Alekseeva, P. Dorogov, A. Loutchanski

RITEC Ltd., Riga, Latvia

0-7803-8701-5/04/$20.00 (C) 2004 IEEE

surface of the samples, IR transmission microscopic control, making of a planar detector, measurement of alpha-spectra, and estimation of the (µτ)e value. The incoming inspection results are used to reject the flawed samples.

Visual examination. Visual examination of the mechanically polished samples allows revealing the presence of large-angle grain boundaries and twins, which are clearly seen on the polished surface in the reflected light.

IR transmission microscope control. To reveal defects inside the crystals an IR transmission microscope was applied. The microscope’s field of vision at the minimum magnification was about 1x1 mm2. Therefore to control the entire surface the scanning was performed by moving the sample along with taking the photos of specific defects. Scanning across the item’s thickness was also done.

Measurement of alpha-spectra and calculation of the (µτ)e value. The alpha-spectra were measured on plane planar detectors. These detectors were made by standard technology, comprising the mechanical polishing, chemical polishing in a bromine-methanol etching agent and deposition of electrodes from a solution of Au-hydrochloride acid.

To register the alpha-spectra a 238Pu alpha source was applied, with two close alpha-lines of average energy of 5.5 MeV.

Measurements with using alpha-source were also taken for estimation of the electron mobility-lifetime product (µτ)e value by a time-of-flight method with using of the familiar single carrier Hecht equation [11].

Making of hemispherical detectors. In order to obtain hemispherical detectors the electrodes of the plane planar detectors were removed by mechanical and chemical polishing. Prior to deposition of the electrodes typical of the hemispherical geometry of detectors [12] a visual examination of the etched surface was performed to reveal possible exits of structural defects into the surface. As a rule this phenomenon manifests itself as etching pits at the places of defects exits.

The spectrometric characteristics of hemispherical detectors: the energy resolution (FWHM) and the peak-to-Compton ratio were measured at the 662 keV line of 137Cs isotope. All measurements were taken at room temperature.

III. THE RESULTS

A. Visual examination The samples from all suppliers as a rule had no visible large-

angle grain boundaries, while the presence of twins was revealed in many samples. On the etched surface of some samples pits and sometimes hillocks were observed, in the quantities varying from sample to sample. As a rule, the etching pits were seen along the twinning plane or grain boundaries.

B. IR transmission microscopic control The control using the IR transmission microscope has

revealed the presence of variously shaped structural defects in the samples of all producers. The shapes of defects observed by us and their spatial distributions changed considerably from sample to sample. Fig.1 shows the mapping of defects distributions in area and the photos of their forms for some samples. We have tried to classify the most typical forms of the revealed defects and the variants of their distributions. The results are given in Fig. 2

Most typical defects were small inclusions <20 µm uniformly distributed in volume. Such inclusions are present practically in all samples tested. The medium-sized inclusions (about 20÷40 µm) are also met quite often. These inclusions may have both regular geometric shape (triangular and/or hexagonal) and irregular one. Some samples exhibit large inclusions (>50 µm). Such inclusions can also be both regularly and irregularly shaped. As a rule, samples with large regularly shaped inclusions possess considerably fewer small inclusions.

The revealed inclusions might be either Te enriched inclusions or Te precipitates. As was noted in the paper [13], usually Te enriched inclusions are much larger then Te precipitates with an average diameter in the 1÷50 µm range.

In some samples of all the producers known [13] linear tubular defects (“tubes” or sticks) were revealed.

In several samples cavities 30÷50 µm in size were observed. The spatial distribution of inclusions can strongly vary from

sample to sample. They can be uniformly distributed in volume or can create aggregations in the form of clouds in a definite part of a sample.

Fig. 1. Examples of IR mappings of about uniform (a) and non-uniform (b), (c) structural defects spatial distribution.

0-7803-8701-5/04/$20.00 (C) 2004 IEEE

We have classified extended aggregations of defects – both

linear and plane – as a separate group. In many samples there were met aggregations of inclusions as extended linear branched defects that can occupy the entire space or its major part. For many samples it is typical to have rectilinear aggregations (cellurar or segregated) defects stretching through the entire sample. Such defects, as a rule, are located along the planes of twinning or grain boundaries [14]

Many samples have simultaneously a few types of structural defects.

A. Alpha-spectra measurements The alpha-spectra obtained for plane planar detectors had as

a rule no marked distortions. This evidences the absence of pronounced defects.

But some detectors had distorted shapes of alpha-peaks (see Fig.3). The detectors with false and distorted peaks were rejected and were not used for production of hemispherical detectors. However these results must be interpreted with caution. The shape of the registered alpha-spectrum is sensitive to the quality of a detector’s surface treatment. The generation of charge carriers at absorption of alpha particles proceeds in a thin contact layer and therefore any distortions, mechanical included, can lead to worsening of the energy resolution in the alpha-line and even to appearance of false peaks. These, for

example, can be mechanical damages arising at cutting the crystals.

B. Measurements of electron mobility-lifetime product (µτ)e The average measured value of the (µτ)e parameter for the

samples from the Yinnel Tech. was about 3.2∗10-3 cm2/V, for the samples from the Saint-Gobain about 3.3∗10-3 cm2/V and for the samples from the eV Products – about 5.5∗10-3 cm2/V.

Selective estimation of the (µτ)e parameter for the holes in the materials of all suppliers has given the value less than 5∗10-5 cm2/V.

The calculation of the (µτ)e parameter was performed under the assumption of uniform electric field distribution in the detector, which is not always true.

The shape of the induced charge signal edge for some detectors was far from ideal, i.e. linearly-rising (Fig. 4). The induced charge signal edge for these detectors had a complicated shape, which can evidence that there are some transport inhomogeneities [13].

The specific resistance of all the samples from different producers was in the range of (0.7÷12.0)∗1010 Ω∗cm.

Fig. 2. IR micrograph of main types of observed structural defects.

(a) Uniform distributed small (<20 µm) inclusions of regular and irregular shapes. (b) Average (20÷40 µm) inclusions of regular and irregular shapes. (c) Large (>50 µm) inclusions of regular shapes. (d) Cavities or pores. (e) Large (>50 µm) inclusions of irregular shapes. (f) Aggregations of inclusions. (g) Branching linear clusters of inclusions. (h) Rectilinear or plane cellurar or segregated inclusions. (i) Tubular linear rod-like defects.

0-7803-8701-5/04/$20.00 (C) 2004 IEEE

Fig. 3. Alpha spectra recorded by uniform and non-uniform planar detectors.

Fig. 4. Planar detectors output induced charge signals for uniform (a) and non-uniform detector (b). Detectors illuminated with alpha particles from the side of negative contact.

A. Measurements of spectrometric characteristics of hemispherical detectors All detectors were made by similar technology.

Spectrometric characteristics of the detectors were measured under the same conditions for the 662 keV line of the 137Cs isotope at room temperature. The energy resolution (FWHM) of the ready-made detectors was in the range of 10.5-40 keV, except for several detectors with poorer performance. The peak-Compton ratio was in this case in the limits of 2 - 8.

Fig. 5 demonstrates the correlation between the energy resolution and the peak-Compton ratio. The best energy resolution obtained for the detectors made from Yinnel Tech. material is 14.5 keV with the peak-Compton ratio being equal to 5.2. The best energy resolution obtained for the detectors made from Saint-Gobain material is 11 keV with the peak-Compton ratio being equal to 7.0. The best detector from eV Products material had approximately the same characteristics 10.5 keV, with the peak-Compton ratio equal to 7.5.

One of the goals of the work was to find a possible correlation between the spectrometric characteristics of hemispherical detectors and the (µτ)e value. Fig. 6 presents the

Fig. 5. Measured hemispherical gamma-ray detectors energy resolution (FWHM) at 662 keV vs. peak-to-Compton ratio.

Fig. 6. Measured hemispherical gamma-ray detectors energy resolution (FWHM) at 662 keV vs. electron mobility-lifetime product.

dependence of the energy resolution (FWHM) in the 662 keV line on the (µτ)e value.

It is seen that the energy resolution in a wide range of (µτ)e variations does not depend on the value of this parameter, and only at its low values (less than 2∗10-3 cm2/V ) a deterioration of the detector performance is observed. A definite number of the detectors with poorer performance that were made from the material with (µτ)e value in a range of (2.5÷4.5)∗10-3 cm2/V can be associated with the presence of defects in the material.

Our measurements have shown the existence of a certain correlation between the performance of detectors and the type of defects, namely: - the use of a material with large inclusions of irregular shape does not allow for making detectors with good spectrometric characteristics; - the use of a material with extended tubular defects of the “tube” or stick type does not, as a rule, allow for making the detectors with good spectrometric characteristics; - the use of a material with aggregations of inclusions in the form of branched defects occupying the major part of a crystal considerably reduces the possibility of making high-quality detectors;

0-7803-8701-5/04/$20.00 (C) 2004 IEEE

- non-uniform distribution of inclusions worsens the performance of detectors; - the presence in the central positive electrode region of exits to a detector’s surface of defects in the form of pits worsens the performance of a detector and decreases the value of its possible working voltage; - no marked correlation has been observed between the concentration of uniformly distributed inclusions with a small sizes and the quality of a detector; - the use of a material with individual large inclusions of regular geometric shape does allow, as a rule, to make detectors with good spectrometric performance.

B. Correlation between performances of hemispherical detectors and plane planar detectors We have carried out a search for the correlation between the

energy resolution of hemispherical detectors at the 662 keV line and the energy resolution of the plane planar detectors made from one and the same sample for the 59.6 keV and 122 keV lines and the peak-valley ratio at the 122 keV line. This correlation was determined only for some detectors made from the material of eV Products. We have not revealed any noticeable correlation between the energy resolution at 59.6 keV lines and the energy resolution at the 662 keV line. A noticeable correlation is observed between the peak-valley ratio at the 122 keV line and the energy resolution at the 662 keV line (Fig. 7). Practically all the hemispherical detectors possessing the worst energy resolution at the 662 keV line were made from the planar detectors with a poor energy resolution (worse than 20%) and a low peak-valley ratio (<2.0) at the 122 keV line. At the same time, some planar detectors with poor energy resolution and low peak-valley ratio at the 122 keV line were used for making hemispherical detectors with good parameters.

IV. SUMMARY Our investigations have shown that materials of all the three

producers possess similar structural defects. They essentially vary in their forms and spatial distributions from sample to sample. In practice, no identical samples exist, even if taken from one and the same producer. The classification of the revealed defects carried out by us allowed for an analysis of the correlation between the presence of various defects and their spatial distribution from the one hand and the performance of the hemispherical detectors on the other.

The accomplished investigations have also shown the absence of a noticeable correlation between the value of (µτ)e parameter and the spectrometric characteristics of the ready-made hemispherical detectors.

The quality of the material of all the three producers and the quality of the detectors made from this material were approximately the same.

It has been established that the structural uniformity of material is one of the main factors influencing the quality and yield of hemispherical detectors.

Fig. 7. Measured hemispherical gamma-ray detectors energy resolution (FWHM) at 662 keV vs. planar detector peak-to-valley ratio at 122 keV.

V. ACKNOWLEDGMENTS The authors are deeply indebted to R. Art (IAEA) for his

permanent interest and support of our work, to L. Li (Yinnel Tech.) for providing some samples that we used in our studies and to eV Products and to Saint-Gobain as suppliers of CdZnTe material.

VI. REFERENCES [1] RITEC Ltd. Riga LV-1006, Latvia, http://www.ritec/mt.lv [2] V. Ivanov, P. Dorogov, "Development of large volume hemispheric

CdZnTe detectors for use in safeguards applications", Proc. ESARDA 19th Annual Symposium on Safeguards and Nuclear Material Management, pp. 447-453, Montpellier, France, 1997.

[3] V. Ivanov, P. Dorogov, A. Loutchansky, L. Aleksejeva, E. Mozaev, " Further development of hemispherical CdZnTe detectors for safeguards applications", Proc. ESARDA Symposium on Safeguards and Nuclear Material Management, pp. 470-484, Sevilla, Spain, 1999.

[4] L. Li, F. Lu, K. Shah, M. Squillante, L. Cirignano, W. Yao, R. W. Olson, P. Luke, Y. Nemirovsky, A. Burger, G. Wright, R. B. James, “A new method for growing detector-grade cadmium zinc telluride crystals”, Proc. IEEE Nuclear Science Symp. Conf. Rec., vol. 4, 2002, pp. 2396-2400.

[5] C. Szeleles, “Advances in the crystal growth and device fabrication technology of CdZnTe room temperature radiation detectors”, IEEE Trans. Nucl. Sci., vol. 51, pp. 1242-1249, June 2004.

[6] P. N. Luke, M. Amman, J. S. Lee, "Factor affecting energy resolution of coplanar-grid CdZnTe detectors", IEEE Trans. Nucl. Sci., vol. 51, pp. 1199-1203, June 2004.

[7] H. L. Malm, C. Canali, J. W Mayer, M-A Nicolet, K. R. Zanio, W. Akutagawa, " Gamma-ray spectroscopy with single-carrier collection in high-resistivity semiconductors", Appl. Phys. Lett., pp. 344-346, 1975.

[8] eV Products a division of II-VI, Inc., Saxonburg, PA 16056, US, http://www.evproducts.com

[9] Yinnel Tech. Inc., South Bend, IN 46619, US, http://www.yinnel.com [10] Saint Gobain Crystals & Detectors, Newbury, Ohio 44065, US,

http://www.detectors.saint-gobain.com [11] K. Hecht, "Zum Mechanismus des lichtelektrischen Primärstromes in

isolierenden Kristallen", Z. Physik, 77, p. 235, 1932. [12] L. A. Alekseeva, P. G. Dorogov, V. I. Ivanov, A. K. Khusainov,

"Cadmium telluride gamma radiation detectors with a high energy resolution", Pribory i Teknika Eksperimentya, no. 1, pp. 54-58, 1985.

[13] C. Szeles, E. E. Eisler, Current issues of high-pressure Bridgman growth of semi-insulating CdZnTe", MRS Symp Proc., vol. 487, Warrendale, pp. 3-12, 1998.

[14] J. R. Heffelfinger, D. L. Medlin, R. B. James, "Analysis of grain boundaries, twin boundaries and Te precipitates in Cd1-xZnxTe grown by high-pressure Bridgman method", MRS Symp Proc., vol. 487, Warrendale, pp. 33-38, 1998.

0-7803-8701-5/04/$20.00 (C) 2004 IEEE