N A T I O N A L LABORATORY

CdZnTe room-temperature semiconductor gamma-ray detector for national-security applications

43. S. Camada, A, E. Bslotnikotr, Member? IEEE, U. Cui, Member, IEEE3 1%- Hoss~!~ , K. T. Kohan, and R B. Jmes, FeBHow, IEEE

Brookhaven National Laboratory Upton, NY 1 1793

Third Annual IEEE Long Island Systems AgpIiccstions and Technology (LdSAT) Coqference

Institute for Research & Technology Transfer, Farmingdale Stare University Farmingdale, NY

May, 4, 2007

Nonproliferation and National Security Depa Detector Development and Testing Division

Brookhaven National Laboratory P.0, Box 5000

Upton, NY 1 1973-5000 % % % $ & w a -

Motice: This muscript has been authored by employees of BmoMraven Science Associates, LLC under Contract NO. DE-AC02-98CM110886 with d ~ ~ e U.S. Dep-ent; of Energy. The publisher by accepting the manuscript for publication achowldges that the United States Government retains a non-exdusive, paid-up, imvocstb8e, world-wide license to publish or reproduce the published farm of this rnanasfipt, or allow others to do so, for United States G o v e r m t purposes. This preprink is intended for publication in a journal or proceedings. Since changes may be made before publication, it may not be cited or reproduced without the author's permission.

DISCLAIMER

This report was prepzed as an account of work sponsored by an agency of the United States Gave ent. Keither the United States Government nor any agency thereof, nor any of their empioyees, nor any of their contract(~rs: subcontra~itors, or their employees, makes any wammey, express or implied, or assumes any legal liability or responsibility for the accuracy, comflteness, or any third party's use or the results of such use of any infoma~on, appmams, produch or process disclosed, or represents that its use would not inEringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, mwufacmer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or fi%cro~ng by the United States Government or any agency thereof or its contractors or subcontractd~rs. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

> REPLACE THIS L M WITH YOUR PAPIER EDENTIFICATION NUMBER (DOUBLE-CLICK ]HERE TO EDIT) < 1

CdZnTe room-temperature semiconductor gamma-ray detector for natima -security

G, S. Cmmdh A. E. HBaEgB$nikov, ~Wember~ IEEE? Y , C U ~ , Mentber3 IEE& A. Mossah, K. T. Kohmm, and 8. B. James, Fellow, PEEE

A b s t r d 4 n e imprtant missiom of the Deparr&mepat sf Eaergy's Natieaai Nucima Security AabrnloisWa~atn is to develsp reliable gamma-ray d e t & s to meet tbe widwpailwd needs of aselrs ifBr effective Qchnigspm te detect and idewtify special. ntarcteat- sa~d rrpdioaetive-materiais Aee(~rdingiy, the Norniprialiferatin and Nationat SecariQ g7padmermt at Brwktaaven Natianal Laboratory was tasked to evaluate existing twhnotogy and to develop improved room-temperature detectors based on semiconductcrrs, such as CdZnTe (CZT). Our research covers two important areas: Improving the quality of CZT material, asd expforing new CZT-based gamma-ray detectors, Ira this paper, we report on our recent findings from the material characterlzati~n and. tats of actuai dTZg devices fabricated in oar laboratory and from materiaIs/d&ectors supplied by different commercisrl vendors, In garticuiar, we emphasize the critical roie of secondary phases in the c~rrent CZT material and issues in fabricating the CZT detectsrs, betb of which affect their performance.

I, I~TRQDUGTIQN

L ARGE volume CUnTe (CZT) semiconductor o&s several advantages as a detecting medium for x- and gamma~rays and has being considered by the Department

of Energy and other agencies for dete~titbm of special nuclear and mdioacsive materiais [I]. However, there are several obstacles that still limit the wide use sf this promising te~kitnology 12-51. Recently, evidences have grow &.&at Te hclwims (usually present in CZT mated) me the main cause alffeding the pe~omance of thick (long-M) CZT detectors, thereby h i t k g the size and efficiency of such detectors available to users [6-91, According to Rudolph [IO], Te inciusioers represent one type of several non-stoirhiometric

Mavzuscript received February 15,2007. This work was supported by U.S. f)egz;artment of Energy, Offree of NonpmlEfaaOpion Research and Engineering, NA-22. The mmuscript has been authored by Brwid38ven Science kociates, LLC wder Gontract No. DE-AC02-96143HI-885 with the U.S. Degment of Energy.

A. E. Bofohikcw, G. S. Cammda, A. Hmsain, Y. Cui, and R. B. James are with BrmWxaven National Laboratory, Upton, NY 1 11 793 USA (phone: 63 1 - 344-80 14; e-mail: b : & ~ ~ ~ i ~ & ~ ~ ~ 5 & ; & .

K.T. Kohm is with Kansas State University, Manham, KS 66506 USA.

related defects nomalIy formed during the melt growth of CZT. The other three: are intrinsic p i n t dsfscts (vacancies, intemtitials and anti-sites), which a h the bulk conductivity, dis8~cations, and Te precipitates both sf which f m during the cooling process as a result of nucleation of native defects. Hi&-resolution transmission electro~ microscopy revealed that tlme Te precipitates are 10-50 nm [lo], whife the typical diameters of Te inclusions are 1-2 ym, aitbough sizes up to 100 ura are obsesved in high-pressure and vertical Bridgman grown CZT [lo]. T'bus, the particles usually seen with IR microscopy (where optical resolution is limited to 1 pm) are attributed to tihe Te inclusions. in CZT material their conwntmtions may exceed 10' cm", but this is still several orders-of-llilmtude smaller than that of the Te precipitates P O I ,

Recently, we employed a highly collimated X-ray beam at the National Synchrotron Li&t Sowce (NSLS) at Broouaven Nationaf Lab (BNL) to measure directly the charge trapped by individual inclusions in CZT samples [6,7]. These results provided clear evidence for the long-discussed hypothesis of the cumulative effect of Te ixncitusions in thick CZT detectors. Modeling [8,9] of the electron-cloud transport through CZT material containing Te illcEusions showed that their cumulative eRect cdn explain the begm&tiom of energy resolution observed in thick CZT detectors, md Wher, that the rnagnitxde sf the effed strongly depends on their size md concen&ation. Moreover? simulati~ns predicted that hclusims of less than -3 ulnn diameter essentially bebaare as ordmary 'traps associated with point defects (native or iang~ties) in the material. Such defects trap the electrons but do not introduce dluctuations in the co8Bected charge signals, allowing for correction of the total charge loss using a depth-sensing twhique. Interestingly9 Te precipitates also may behave as psht defects, md siace their cance~tmtion in as-grown material can be hi&, tbey might control the effixtive mobility- l i f ehe pmduct for both holes and electrons. Thus, to maximize the pePgommce of CZT detectors, it is important to establish the limits of the sizes and concenm~on of the hcBusims as a fasnctisam of the device's thicknesses.

In the mwse of our project ta develop an m y of FPisch- ring CZT detectors [ 1 1,121, we fabricated and tested the spectral responses of -20 bar-shaped CZT sanzples ranging b r n 5 to 15 rn thick. Theoretically, the geometry of these samples so configured should provide a spectral response with

> REPLACE THIS LINE WITH YOUR PAPER IDENTIFlCATION WMBER (WUBLE-CLICK HERE TO EDIT) <

shapes depending on crystal's srie~fiation a d the ilib*~~hation. UsuaUy, Te inclusions have triangular 01- diamond-like shapes. In 3D images (FdixeEinte~1sity dis~bu~csms), such incBusions are represented by ~ee-dimensional sdaces, in some cases, 60mp]r3sed of several Gaussian-like functions, Small sized inclusi~~m, below 3 ,him, typically appear as bImed objects with s m m d i n g artificiaf haloes In a 3D image, suck. i~clusions me represented by sudaces with m04bnensi03~d Gaussian knctions.

We used the D L (Interactive .Data Language) pro&3ra9n;dning en~roment that is especially well suited for image mmipdatiora and processing and with p w e f i l inputtouput facilities, along with variety of e libraries for statistical ~ 1 y s i s and plotting &e pesuits. The imagi~ag setup al3owed US to acquh-e stacks of images, each focused at different depth of the cqsm~. Table 1 lists the typical pameters of the microscope. It shews hat at high mwifica~ons, larger afim x2, the depth of focus is comparable to the expected size of the precipitate -1 0 pm. In principle, with such im apparatus, the c~.grstaB's volume can be "scmedg' or "sliced" completely.

TmLE I PARAMETERS OF THE MICROSCOPE OPTICS USED FOR THE CZT ANALYSIS

The first step in image p~wesshg was to invert the original images so the featwxs representing i9lclusicrrms were regions of hi@ intensity (peaks), 7% improve image recognition backgrowid, filtering is needed to suppress the contributions ftom image noise or image defects. We used several techniques for this, from simple threshold application to bmd-pass filtering. Both techniques equally generate good- quality images wlgh uniform illumination and few noise artifacts. A rml-space, band-pass filter sappresses pixel noise md slow-scale image variations while retaining infomati,tisn: on chmckristic s h ; this is beneficial for images with now mifsm illuxrahatim or multiple surface-related defects (such as s@ratckes). Because the illmhation of the fmal plane at which images are taken em vary fkom sample to m p t e and even f o m image to image, the threshold values must be iteratively adjusted to optimize the quality oftbe daeminration af the precipitates' parameters.

After suppressing the backpun& we could begin searching for features and deteraninkg their properties, We utilized the following algorithm for peaks Hocaliatisar: First, we identified the positions of all the local maxima in the image (defined in a circular neiglhhrhood with some given d i e t e r ) , then placed a circular mask sf the same size a~oukld each marxha, a d caBcuEated the X a d Y positions ~f centroids, the total brightness of all the pixels, and the diameters of the pixels within that mask. If the initial local maximum was more than 0.5 pixels ikom the centmi4 the

mask was moved and the data reeca1cuBa$16:d. This mmipulation is useRxl for noisy data, If the image glsise was small and the featares more than about 5 pixels across, then we expected the resulting X appd Y values 8 s have errors of the order of 4.5 pixels for keasmably noise-ke images. For the band-pasass filter a d featwe search, we: used Cracker and Grim's algorithm [13f. The output of the feature-detemina~on routine is an m y which contains eXre following parmeters: XY-6;oordhat~ of the Esernd featwe, i ts integrated brigh$I.ess, ars& its diameter evaluated as ~ B ~ - ~ d ~ ~ a t - k a t i k l f - m ~ i m ~ of peaks. Suhequent analysis and identi5'ng;atim sf found features as inchsi~nas is based on these parameters. Fig. 2 gives ian

example of an identified triangular-shape inclusion, labeled with its parameters. Faint out- focus features also are clearly visible. As depicted in this picme, images may contain &Rerent Kids of features, noise, proper i~~ciusions, and faht outlines sf out-of-focus inclusiom Ehat do not "belongse to the current slice of the s m ~ l e .

Fig. 2. Example of an identified triangular-shape inclusion, fabeled with i ~ s pmameters with the faint out-of-focus features in the background.

To exclude the out-oEfocus features, we use several criteria - one of which is correlations between inclusions diameters and their brightness (Fig. 3). As ilIustrated in Fig. 3, the radius versus brightness distribaraion has two bistinguishable weas of hi& dots concentration corresponding to two different classes of objects. Large objects, with low brightness, are identified as out-sf %us irnclusioars. Objects smaller in size but with higher brightness are identified wi& true, in-ltbcaas inclusions, Visual inspection of the images confirmed this hypothesis. Visual inspection of the images confumed this hypothesis. We coneluded that integrated brightness groved to be the most disc~minaing variable, allowing us to reject a ~ t + ~ f w u s features md select in-focus inclusions. Typically, after backgoezftd subtmct~on and with a relaxed cut on brightness, the program identified up to several hundred features. After adjusting the cut on brightness, this ramber fell to less titam twaty, with the exact rmumbm depending on. the image's magnification and quality, Such visual inspections b e h md after processing the images Bed

> REPLACE THIS LINE TKiiITM YOUR PAPER IDEHTIFBCATIION NUh%BER (DOUBLE-CLICK HERE TO EDIT) < 4

us to conclude that this procedure can yieHd aaz. accwacy of detemhation of in-plane inclusions of better than 3%.

Fig. 3. Distributim of the radius ofthe Te inclusions (the objects) versus Oheix brightness. Region 3 represmts objects of small radius and high brightness: region 3 groups objects with large radius md low brightness, and finally, region. 2, bemeen regions I wd 3 cmtains objects with medium size and medim brightness.

The algorithm discards Te inclusions that are out of focus and counts only those in hcus. This requires two steps. First, a cutoAF is applied on brightness to suppress the c s n ~ b u ~ o n of objects of Bow brightness (Te inclusions out of focus) on the negative, i.e., Te inclusions in region 3 (Fig. 3) are discarded. The second step is where the algorithm selects only one of the Te illtelusions present at the same position in 3-5 difYerent layers (the number depending on the DQF). For clarity, we we layer instead of image here. The same Te inclusion can appear in sc:veral neighbring layers (is. in both regions I a d 2), but wodd be in focus only in one layer, or at least there would be a layer where that inclusion is in better focus than in the others. The algorithm selects the one in focus, and corrects for shifts of images. In this way, the algoritb ensures &at we count each Te Inclusion only once. men, by taking into account the FOV md the size of the z-scan step, the measured number of Te inclusions for each layer can be rmomafked per cm3. For each CZT sample, five different measruements were taken at five different locations over the crystal's length Erom two pzpep~dicular directions. Their avenge gives the concenmtion per cm3.

By c o r n p h g with the hi#-magnikation microscope measurements, we estimated that for the large, >3 pm, inclusions the m r in the size determination 3s -1 p. Below 3 pm, tbe measwemmh are limited by the sml! focus depth, -1 -6 p at 220x magnification. Even though fie opticat system cm locate and count the inclusions with the diameters as small as 0.5 pm, it cannot accurately determine their sizes if their actual diameter <3 prn.

1[U. WSULTS AND DISCUSSIBN

It is well established, based on the direct measwema& carried out with the highly collimated X-ray beam [6,7], &at each Te inclusion traps significant amount of charge from an electron cloud. If we consider the Te kclusions to be

nonmspaent to the electrons, one caw easily simulate their effects on the charge collection in CZT devices. The simulations very accurately reproduce the experimental ~resnlts obtained f om the x-ray scans of thin, <:! wm, planar deteclon and predicted the emulative effect of randomly distributed inclusions OD atae pedommm of the thick (long-drift) detectors like, e.g., Friseh-ring or pixel devices. Fkst of dl, the inclusions reduce the total of the collected charge pope~ioslally to the e'se~tron cloud &rift distances ist a same m m e r as ordinary traps associated with point defects do. This charge Boss can be colgected to pesewe gwd energy resoIation. Secondly, became they can trap a lmge number of electr~ns per intemction with electron clouds, hclusions cause large flwtuatiorns in the collected charge as predicted by the model [8].

Ow tgualitalrive masufements taken with - 20 FPisch-ring detectors (mom than 20 detectors have k e n tested by now) consistently demonstrate that the smples with higher concentsation or larger size Te in~lusiom showed poorer spectral responses. hrthernol.e, the comeEation &tween their size and co~mcentmtion an6 device's energy resolution becomes more promowlced for thick smples. To provide a quantitative evidence of such correlation, we selected five CZT detectors asrd acemely evaluated the size distributions and concentrations of Te inclusions. We particularly chose the samples that provided decent spectral responses and had a close to miform distribution of inclusions. The latter is important to minimize variations of the collected c h q e related to non-uniform distribution of inclusions.

Table 2 summarizes the results obtained fpom the: setected detectors, which represent different pegfomance quality. In the case of Dl, D2 and D3 detectors, the measwments were taken at 5x tnagnification. For the detectors D4 and D5, a 20x magnification was used. A quick glance at the table suggests that CZT cryseals with a high concentration of Te inclusions of -10 pm in size perform worse thm crystals with lower concenmrions of similar size incliusions. On the other hand, crystals with smal'i-size Te inclusions, of ahu t 1 pm, and a high concentration of Te inclusions, of -1 06, can give the same or better energy resolution than crystals with larger Te inclusions and lower concentrations per cm3. These results are in a good agreement with the simul&ions of the cumulative effect of Te inclusions in thick detectors //8,9].

T m L E 11 COXCENTRGT~QN AND SIZE QF XNCMJSIOXS, AND ENERGY RESOWTION

MEASURED FOR PNE a T CRYSTALS

Detector Dimmsims, Conml?s&ion, Dim- Resolution, ntpmber mm eme3 ;range, % Fvi'l-rM

pm at 662 keV 1 6x6~12 I .ox1 0" 3-20 6.2 2 6 x 6 ~ 1 2 9.6~10' 3-10 3.8 3 5~6x11 s.oxro4 3-10 1.2 4 SXSXI 4 2.7xrd <3 1.4 5 4x4~1 i 1.2xf oh 4 0.8

Figs. 4-7 show pulse-height spectra.., representative IR images and size distribution of Te inclgesions mamd for the five: samples listed in Table 2.

> REPLACE TMS LINE W-ITW YOUR PAPER fDENTIFIGATEON WMBER (llCWBLE-CLICK HERE TO EDIT) < 5

Fig. 4. Results obtairred for the smpfe Dl: (a) a pulse-height spectrum, (b) a Fig. 5. Results obtained for the sample I421: (a) a pulse-height spectrum, [b) a repr~3smbtive W image fat 5x mwification) and (c) size distribution of Te representative W image (at 5x naagnffic&on) and (c) size rfiaibutim of T6: incfusioms averaged over !ever& measurements. The total inclusions inclusions averaged over sever& measurements. The total inclusions concenWa.tion is l .OXEO' cm". Mote, the slunpfe has hi& cancentration of concentphon is 9 . 6 ~ 1 0 ~ an-'. In cornpatison to D!, D2 has rt lower large, >S pm, inclusions resulting in energy resolution of 6.2%. cmcentsation of inclusions with radii of >5 ym, resulting in a better energy

resolution of 3 -8% FFWHM at662 keV.

> WPLACE THIS LINE WITH YOUR PAPER IDENglIFICA?n[ON NUMBER WUBLE-CLICK HE= TO EDIT) < 6

Radius, urn

Fig. 6. Results obtained for &e samp1e D3: (a) a plltse-height ~pe-m, a Fig 7. Results obtained for samples I24 and DS: (a&) the pulse-height s p e w representative fR image {at 5x mi5catfonI and (c) size disgtibution of 7% and (el a representative EW image (at 2Ox pnagnificatim). These m p l s s have ZmcBusions averaged over several memwmnts. Tke totall iwcfusions only small inclusions of less tlam 3 pm wta8 concentrations of -1 .2~10~ and concm&ation is §.0x'40~ ~ n z " ~ . Note, that, I33 has a similar size distribution to 2 . 7 ~ 1 0 ~ ~ r n * ~ . Both detectom show& m bnxcellmt pesotutioxr of 1.2 and 0.8% that of detector D2 but at a l o w 'eoncentrafion of about half of the F W m & 662 keV. concentration of detectors Dl md D2, resulting jrr a significantly better resolution of 1.294,

> REPLACE ?HIS LHNE WITH YOUR PAPER XBENTIFICATION NUMEEER (DBUfSLE-CLICK HERE TO EDIT) < 7

The detector Dl has hi& co~lcen$ra~om of lmge (Xi pm) inclusions with a total concentration of i .Ox1 d anm3, resulting in its poor resolution of 6.2% FNWM at $621 keV. The detector HP2 has a compmble concentn"ation of 9 . 6 ~ 1 0 ~ cm". However, in contrast to Dl, D2 shows a lower concentration sf hcI.cxsions writ%. diameters of >5 pm, which may explain a slightly better exlergy resolution of 3.8% FWHM at 662 keV meas~aeb for this device- The detector D3 has a simsrila size d i ~ b ~ a ~ o ~ to &at of detector I32 but at a lower eollt~enm~~n of 5 . 0 ~ 1 0 ~ ane3, about haif sf the concen@aGors sf detectors DH and D2. GomespondentEy, this device shows sipificmtly better resolution of B 2%.

The detectors D4 md D5 (fabricated by a di@erent vendor) have only small inclusions of less than 3 pm but with higher concenmtions (at -~.ZXIO' cm') and 2 . 7 ~ 1 0 ~ cm", co~esgondixrgiy) than samples D 1 -D3. Both detectors showed an excellent resolution of 0.8 md 1 2 % . A slightly bigher F W of the detectat DS cm be attributed to the larger detector thickness.

These results demonstrated that art detector's responses strongly depend om comcenQation of Te incEusions greater than 5 pm. Below 3 p, inclusion sizes are very close to the resolution limit: of ow IR system, the accuracy of which was insufficient to quantitatively measure the coreltations between the device's spectral responses, precipitate concentrations, and crystal thicknesses (the etch pit technique would be more appropriate for measuring such small inclusions). Nevertheless, based on these measwmenrts, we can give an upper estimate that still is very important fur CZT crystal growers trying to minimize the impact of inclusions in their material. As we showed with IW transmission microscopy, Te inclusions with diameters of <3 pm contribute to the total energy resolution of <l% at 662 keV in CZT detectors of up to 15-wsnm thick,

IV. CONCLUSION

In this work, we deanomstrated correlations between the size a d concentmtions sf Te inclusions and the spectral response of CZT detectors. It provided the evidence &at the Te in.nclusions, often found ia3 c:upTent CZT ma&riaf, ase responsible for degrading tbc emrgy resolution of thick CZT detectors. Our experimental msufts agree well with the predictions given by the model treating Te Indmicans as rnacmscopkc meas dilled with a high concen&ation of point traps.

Furthemore, thick CZT detectors made of crystals w i h zs. small conce~tration of small Te hclusiions can potentially give the same or better energy resolution than crystals with larger Te inclusions but lower concentrations per em3. Also, we observed that CZT crystals with Te incEusEons smaller than -I pm behave as ordinary traps. Their presence in the material at: concentrations lower than does not affect significantly the detector's pedomance, e.g., a FWHM of less than 1% was obtained for 662 keV for the 1 11-ma long Frrisch-ring detector. At high concentrations or in thicker devices, the cumulative effect of small inclusions can be corrected by using a depth- sensing technique. Xn should be mentioned, however, that the

results reported i~ this article were obtained from a limited number of CZT samples. We expect to validate tbese results Mith a higher confidence level as we receive more CZT samples kom our vendors.

These results provide m insight into the critical role of the Te inclusions, whose presence in the cmernt CZT material is related to their groGh under non-stoichiomebic conditions. Hence, these data me very impartant to the CZT crystal growers because once these defects we wderstood and can be controlled, large-volume, several cm3, CZT detectors will become available to users.

The United States Govament retains, md the publisher, by accepting the article for publication, achowiedges, a world- wide license to publish or r~:produce the published f m ofthis anmwcdpc or allow others to do so: for the United States

[I] R. B. James, T. E. Schlesinger, J. C. Lmd and M. Schieber, "Cadmium Zinc Telluride Spectrometers for Gamma and X-Ray Applications", in Semiconductors for Room Temperature Nuclear Detector Applic&ions, Val. 43, edited by R. B. Jmes and T. E. Schlesinger (Academic Press, New Ymk, 13951, p. 334.

[2] C. Smles and E. E. Eissler, "Current issues of High-pressure Bridpan growth of semi-ins~irag CdZnTe", in Semiconductos for Room Temperature Radiation Detector Applications 11, Vol. 487, edited by R. B. James, T, E. Schlesinger, P. Siffert, W. Dusi, M. R. Squillante, M. O'Connell and M. Cuzin (Materids Reseaxch Society, Pittsburgh, PA, 19982, pp. 3-12.

[33 J. R. Heffelfinger* D. L. Mediin, and R. B. James, "Analysis of Grain Boundaries: Twin Boundaries and Te Inclusions in Cadmium Zinc Telluride Grown by High-Pressure Bridgman Method, in Semiconductors for Room Temperature Radiation Detector Applications 11, Vol. 487, edited by R. B. James, T. E. Schlesinger, P. SiEer:, W. Dusi, M. R. SquifEante, M. O'Connell and M. Cuzin {Matetids Research Society, Pittsburgh, PA, 1998), p. 33.

143 C . Szeles, S. E. Cameron, J.-0. Ndap, W. C. Cfiahen, "Advances in the cfyshl growth of semi-insulating CdZnTe for radiation detector app1ic&ionsm, lEEE T F C J ~ . A'ucI. Sci., NS 49, n.5, pp. 2535-2540,2002.

[53 M. Amman, J. S. Lee, and P, N. Luke, "Electron trapping ~onunifomity k hi&-pressure-Bridman-wow CdZnTe", J Appt Php., vol. 92, pp. 3 198-3206,2002.

[6j G. A. Cwiai, A. E. Bolotnikov, G , S. Camada, G. W. Wright, t. Li, and R. B. James, "Effect of Te inclusions on the p e r f o m c e of CdZnTe detectors", Appl. Phy. kt&. 88, p, 1435 1 5,2006.

[7J G. S. Cammda, A. E. Bolotnikov, G. A. Carini, anrf R. IS. Jarnes, "Effects of Teifurium Inclusions on charge collecrion in CZT Nuclear Radiation Detectors", irz Countering Nuelear and Radiologic& Terrorism, edited by S. Aprkqtan and D. Diamond, Springer, 2006, pp. 199-207.

f8f A. E. Bolotnikov, G. S. Canada, G. A. Cmini, Y. Cui, L, Li, md 8. B. James, "A. E. Bolotnikov, G. S. Camarda, G. A. Carhi, Y. Cui, L. Li, and R. B. Jmes, "Modelling the geom&ricJ effects of Te precipitates on electron @mpolz in WnTe7', Nucl. Snstr. Me&. A 571, pp. 687- 698= 2007.

[9] A. E. Bolotnibv, G. S. Camarda, G. A. CaPini, Y, Cui, K. T. Kohman, L. Li, M. B, Sdomon, aad R. B. James, "Pexfomance-limiting Defects in CdZnTe Detectors", submitted to fEEE Trans. L 4 t ~ ~ l . Scz., 2807.

El01 P. I B I u ~ o ~ J P ~ ~ , NOR-stoichimetry related d e f m at the melt growth of semiconductor compound cp~dstals - a review", Cryst. Res. Techol. Vol, 38, n. 7-8, pp. 542-554,2003.

[I I] W. J. McNeii, D. S, M&egor, A. E. Bolotplikov, G. W. Wright and R. B. Jmes, "Single-Charge-Cafiier-Type Sensing with an Insulating

> REPLACE THIS LmT W$f.% YOUR PAPER lIDEMTFICATHB16 WdMBER @OUBLE-CLECK HERE TO EDIT) < 8

Frisch Ring CdZnTe Semiconductor Radiation Dete~tor'~, Apgl. Phys. Lett. 84, pp. 1988-1990,2804.

[I25 A, E. Bototnikov, G. C , Carnsrrda, C . A. Carhi, M. Fiedale, L. Li, D. S. McGregor, W. McNeil, 6. W. Wright, md R. B. James, "Perfomatlee Cbmacteristics of Frkclh-Grid CdZnTe Detectors", IEEE Trralasactions on MucIem Science, 11's 53, n.2, pp. 607-6 14,2006.

[I31 3. C. Cxockeer md nd. G . GrIer, "Me&ods of digital video rnicr.oscopy for colloidal studies"', S. Colloid LqterFace Sci., Vol. 179, pg. 298-31 0, 1996.