Available online at www.sciencedirect.com

www.elsevier.com/locate/nimb

Nuclear Instruments and Methods in Physics Research B 266 (2008) 1300–1306

NIMBBeam Interactions

with Materials & Atoms

Ion beam induced charge imaging of charge transportin CdTe and CdZnTe

P.J. Sellin *, A.W. Davies, S. Gkoumas, A. Lohstroh, M.E. Ozsan, J. Parkin,V. Perumal, G. Prekas, M. Veale

Department of Physics, University of Surrey, Guildford GU2 7XH, UK

Received 18 September 2007; received in revised form 23 November 2007Available online 23 December 2007

Abstract

Ion beam induced charge (IBIC) imaging is a powerful technique for quantitative mapping of the charge transport performance of widebandgap semiconductor materials. In this paper we present results from a study of electron and hole mobility–lifetime product and driftmobility in CdTe:Cl and CdZnTe, which are semiconductor materials used for radiation detector applications. IBIC imaging has beenused to produce mobility–lifetime product maps in CdTe:Cl and CdZnTe, revealing the influence of extended defects and tellurium inclu-sions and assessing the large area response uniformity of the materials. The recent extension of this method in the form of digital time-resolved IBIC is also discussed and time of flight maps are presented which give quantitative images of electron and hole drift mobility.� 2008 Elsevier B.V. All rights reserved.

PACS: 29.40.Wk; 72.20.Jv; 73.61.Ga

Keywords: IBIC; Charge transport; CdTe; CdZnTe

1. Introduction

Ion beam induced charge (IBIC) imaging is a powerfultechnique for high resolution mapping of electrical proper-ties in semiconductors. The method has been extensivelyapplied to the study of a variety of microelectronic devices,including high power transistors, solar cells and radiationdetectors. In particular, IBIC imaging of semi-insulatingmaterials for use in radiation detectors has been studiedby several groups and applied to various wide bandgapmaterials such as CdTe and CdZnTe (e.g. [1–4]), GaN [5]and diamond (e.g. [6–8]). IBIC measurements have beenperformed in both the orthogonal orientation, with thebeam incident on the top electrode, or in a lateral configu-ration with the beam scanning over the side of the device.

In this paper we present the latest results of IBIC imagingof charge transport in semi-insulating CdTe and CdZnTe

0168-583X/$ - see front matter � 2008 Elsevier B.V. All rights reserved.

doi:10.1016/j.nimb.2007.11.074

* Corresponding author.E-mail address: p.sellin@surrey.ac.uk (P.J. Sellin).

bulk layers which have applications as X-ray and c-raydetectors. In these very high resistivity materials the spacecharge region extends throughout the thickness of thedevice, which can be normally be operated under either biaspolarity. The contribution to the current pulse induced atthe device contacts is due to the drift of both electrons andholes, as discussed below, and the signal amplitude fromthe device is directly dependant on the carrier drift lengthsof both electrons and holes. In this paper we describe IBICmapping of both mobility–lifetime product, using analogueIBIC and drift mobility using digital or time-resolved IBICand the influence of extended defects and inclusions on thecharge transport properties of the material.

2. Theory of the IBIC technique

IBIC measurements of charge transport are particularlyuseful when applied to bulk materials of the type used forradiation detectors. In such devices, high field strengths arerequired over active thicknesses of several millimetres and

P.J. Sellin et al. / Nucl. Instr. and Meth. in Phys. Res. B 266 (2008) 1300–1306 1301

carrier drift length is the primary measure of charge trans-port. Due to the semi-insulating nature of these materials,the conventional discussion of majority and minority carri-ers is less applicable and the devices are considered as solid-state ionisation structures in which the drift of electronsand holes can be equally important. For a full descriptionof the application of the Shockley–Ramo theorem tocharge drift and the resulting induced current pulse, thereader is referred to the reviews by He [9] and Breese [10].

Many wide bandgap bulk semiconductors exhibit signif-icant charge trapping effects which tend to limit the elec-tron and/or hole drift lengths. In CdTe and CdZnTematerials hole trapping is a particularly strong effect, duepartly to the limited hole mobility. In general the chargecollection efficiency, CCE or e is the ratio of the chargeinduced at the sensitive electrode Q and the charge createddue to a radiation interaction Q0. For a detector of thick-ness d, an approximate expression for the CCE is given by

e ¼ QQ0

� ke þ kh

dð1Þ

where ke and kh are the mean drift lengths for electrons andholes, respectively and ke, h� d.

The charge drift length is related directly to the carrierdrift velocity and hence the mobility l by the expression

k ¼ lsE ð2Þ

where s is the effective carrier lifetime and E is the meanelectric field strength across the device. Consequently themobility–lifetime product (ls) is a primary indicator ofthe charge transport properties of the semiconductor andhence the performance of the device as a high resolutionradiation detector.

In general, the CCE is the sum of two exponentials dueto the separate electron and hole contributions to the totalinduced charge. The Hecht equation describes the CCE asa function of the interaction depth x of the radiation fromthe cathode

eðxÞ ¼QðxÞQ0

¼ kh

d1� exp

�xkh

� �� �þ ke

d1� exp

x� dke

� �� �

ð3Þ

where the electric field is assumed to be constant as a func-tion of depth (i.e. equal to V/d).

For IBIC measurements using proton and helium beamswith energies of approximately 1 MeV the interactiondepth of the charged particles is generally much less thanthe device thickness. For this limit (i.e. x� d) Eq. (3) issimplified to the ‘single carrier’ case, where the CCE isgiven by

e � lsV

d21� exp

�d2

lsV

� �� �ð4Þ

and V is the applied bias. Consequently, for a sequence ofIBIC maps acquired over a range of bias voltages, applica-tion of Eq. (4) to each pixel in the image is used to generate

a mobility–lifetime product map for either electrons orholes [11]. With irradiation of the incident particles ontothe cathode, the IBIC data is due to electron transport,whilst for anode irradiation hole transport is measured.

Time-resolved IBIC measurements are an ion beamimplementation of the more general time of flight (TOF)technique and allow measurement of carrier mobility andlifetime. In this method the complete shape of each chargepulse is digitally captured and hence the drift time of thecarriers sdr is extracted from 10% to 90% rise time of thepulse. A correction is required for fast pulses due to thefinite (�20 ns) rise time of the preamplifier, which ulti-mately limits the applicability of this method for very shortcharge drift times. For non-saturated drift velocities, theelectron and hole drift mobility is given by the ratio ofthe drift velocity to the electric field strength E. Hencefor a measured carrier drift time sdr, the mobility is givenby

l ¼ d2

V sdr

ð5Þ

Hence a plot of 1/sdr versus V is often used to extractmobility. A more complete description of IBIC drift mobil-ity mapping is given in [3,12].

3. Experimental method

IBIC measurements were carried out at the Universityof Surrey’s Ion Beam Centre using a Tandetron 2 MVaccelerator and a microbeam line supplied by OxfordMicrobeams Ltd. Further details about the layout andoperation of the beam line can be found in [13]. The SurreyIBIC facility uses a conventional analogue multi channelanalyser system [14] combined with a time-resolved digitalIBIC system. The schematic layout of the digital IBIC sys-tem is shown in Fig. 1, with full details described in [15].IBIC measurements are performed using a temperaturecontrolled sample holder, with the test sample connectedto an external charge integrating preamplifier (Ortec142A) and voltage bias supply.

For 2 MeV protons a beam diameter of less than 5 lm isroutinely obtained, which is raster-scanned across the sam-ple over an area of 2.5 � 2.5 mm at minimum magnifica-tion. Each map is acquired with a resolution of256 � 256 pixels. Collimation of the beam ensures a parti-cle flux of typically 1 kHz of protons onto the sample,allowing event by event pulse processing without pile-upproblems. The use of this very low beam current also mini-mises charge injection in the sample and any possible spacecharge effects.

The digital time-resolved IBIC system uses a 1 GS/swaveform digitiser to fully capture each pulse from thesample and simultaneously record the X,Y position ofthe scanning beam. The Surrey facility is one of the firstfull-scanning time-resolved IBIC systems and was devel-oped for TOF high resolution mapping of drift mobility

Fig. 1. Schematic of the University of Surrey digital time-resolved IBIC system. Each charge pulse is captured using a 1GS/s 8-bit Acqiris digitiser [15].

1302 P.J. Sellin et al. / Nucl. Instr. and Meth. in Phys. Res. B 266 (2008) 1300–1306

in semi-insulating bulk materials. Measurement of the risetime of the integrated charge pulse is used to extract drifttime as a function of bias voltage and hence carrier mobil-ity. Additional pulse shape analysis routines can be usedfor off-line analysis to study thermal emission of chargeand long-lived decay transients.

4. Results and discussion

A variety of IBIC studies have been carried out at theUniversity of Surrey to investigate the spatial uniformityand charge transport properties of semi-insulating CdTeand CdZnTe material. As a primary measure of chargetransport, IBIC has been used to produce maps of pro-ton-induced signal amplitude, using a conventional ana-logue IBIC system. A main objective of this work is tostudy the uniformity of response from CdTe and CdZnTematerial produced using different growth techniques andto investigate the influence of extended defects such as tel-lurium inclusions which are typically found in thesematerials.

Fig. 2 shows a sequence of three IBIC maps of signalamplitude, due to electron transport in a sample of chlo-rine-doped CdTe. The device consisted of a single rectan-gular Schottky contact on each surface, approximately4 � 4 mm in size and fabricated by thermal evaporation.The thickness of the sample was 2.15 mm. Each map inthe sequence covers an active area of 1.2 � 1.2 mm andwas acquired using a proton beam with an energy of2.0 MeV and a beam spot diameter of approximately5 lm. The three maps were acquired at bias voltages of�20, �50 and �100 V, corresponding to mean fieldstrengths of 93, 233 and 465 V/cm, respectively.

The IBIC data were acquired from regions of the CdTematerial which were known to contain large extended tellu-rium inclusions, previously observed by sub bandgap850 nm IR microscopy. The IBIC maps clearly show theinfluence of large defects on the signal response and chargetransport properties of the sample, consistent with largeinclusions being located close to the top surface of the

material. At low bias, e.g. �20 V, the influence of thesedefects is seen to extend over considerable distances, e.g.up to 300 lm from the centre of the region of poorresponse. Fig. 2 also shows vertical line profiles for eachmap, extracted along the shown line at pixel positionX = 65. As the bias voltage is increased to �100 V the lat-eral influence of the defects is strongly reduced and theoverall response uniformity of the device significantlyimproves. The region of poor response shown in the lineprofile has a FWHM which decreases from 40 pixels(�400 lm) at �20 V, to 16 pixels (�160 lm) at �100 V.At voltages beyond �100 V there is little further reductionin the lateral extent of this region of poor response.

A possible explanation for these regions of poorresponse is that the carrier lifetime in the material closeto the inclusions is reduced. For inclusions located closeto the irradiated surface (cathode), regions of reduced life-time material exist in the near-surface layer where the elec-tron–hole pairs are initially created. Charge trapping inthese regions is maximised at low voltages, since the elec-trons drift more slowly into the bulk.

A more quantitative analysis of charge transport prop-erties using IBIC is obtained by obtaining a sequence ofsignal amplitude (proportional to CCE) maps at differentbias voltages. The resulting CCE versus voltage data isthen fitted for each pixel using the single carrier Hechtequation (Eq. (4)). Fig. 3(a) shows a typical set of pulseheight spectra obtained from a single pixel of a sequenceof IBIC maps acquired with bias voltages from �30 V to�800 V, measured from a sample of CdZnTe grown by amodified Bridgman process. The device measured 6.3 � 6mm with a thickness of 2.3 mm and was fabricated usingplanar thermally-evaporated gold contacts [16]. In this datathe full energy peak is narrow and well-resolved, with asmall tail on the low energy side of the peak. Fig. 3(b)shows the resulting fit of this data to the single carrierHecht equation, which gives an electron ls value of1.4 � 10�3 cm2/V in this sample, which is within the rangeexpected for CdZnTe. By applying this algorithm to all thepixels in a map, a calibrated ls map can be produced.

Fig. 2. IBIC maps of electron signal amplitude in CdTe. The area of each map is 1.2 � 1.2 mm. The vertical line profiles are at pixel X = 66.

P.J. Sellin et al. / Nucl. Instr. and Meth. in Phys. Res. B 266 (2008) 1300–1306 1303

IBIC ‘Hecht maps’ of mobility–lifetime product in thisCdZnTe sample are shown in Figs. 4 and 5, for electronsand holes, respectively. Also shown are horizontal line pro-files of ls values obtained from the indicated line at pixelY = 53. The edge of the metal electrode is visible as thediagonal boundary in the bottom right corner of eachmap. The two sets of data are acquired under identical con-ditions, without moving the sample, except for reversal ofthe polarity of the bias voltage applied to the irradiated

contact. The data show clear differences between the elec-tron and the hole charge transport in this material. Theelectron map (Fig. 4) shows a very uniform distributionof lse value, reaching �1 � 10�3 cm2/V towards the rightedge of the sample. In contrast, the hole map (Fig. 4) showsconsiderable variation in lsh response, with differentregions of lsh extending over length scales of 500 lm–1 mm. The peak lsh value observed in this map is�4 � 10�5 cm2/V, dropping to �1 � 10�5 cm2/V in other

CCE

0.0 0.2 0.4 0.6 0.8 1.0

Am

plitu

de

0

50

100

150

200

250

300

30V100V150V200V800V

Voltage (V)0 200 400 600 800

Cha

rge

Col

lect

ion

Effi

cien

cy

0.0

0.2

0.4

0.6

0.8

1.0

μτe = (1.4 +/- 0.5) x 10-3 cm2/V

a

b

Fig. 3. (a) IBIC pulse height spectra obtained from CdZnTe with 2.0 MeVprotons, due to electron transport and (b) plot of electron mean chargecollection efficiency versus bias voltage, fitted to the single carrier Hechtequation.

Fig. 4. IBIC map of electron ls in CdZnTe showing very good uniformityand an average value of 1 � 10�3 cm2/V. The area of the map is1.5 � 1.5 mm. The horizontal line profile is at pixel Y = 53.

Fig. 5. IBIC map of hole ls in CdZnTe showing non-uniform responsewith a peak value of 4 � 10�5 cm2/V. The area of the map is 1.5 � 1.5 mm.The horizontal line profile is at pixel Y = 53.

1304 P.J. Sellin et al. / Nucl. Instr. and Meth. in Phys. Res. B 266 (2008) 1300–1306

regions. The origin of the non-uniform lsh response maybedue to spatial variation in either the hole mobility or holelifetime and further measurements such as time resolvedIBIC are required to distinguish between these twopossibilities.

Fig. 6 shows the digital IBIC map of electron drift timein CdTe acquired at a bias voltage of �20 V. This time-resolved IBIC data was acquired simultaneously with thedata set shown in Fig. 2(a). Due to the lower event acqui-sition rate of the digital IBIC system, each pixel containsfewer events than the analogue IBIC data and the drift timemap is correspondingly noisier. However, the data clearlyshows that the electron rise times are extremely uniform,with a typical value of 1.5 ls over the whole of the imagedregion. The large areas of poor response, where the pulseamplitude is below the acquisition threshold, cover thesame locations as in Fig. 2(a). The data show no decreasein drift time close to the regions of poor response, suggest-ing that both the electron mobility and the electric fieldstrength are unaffected close to these defective regions.

Fig. 7 shows a histogram of electron drift time extractedfrom a sequence of digital IBIC images similar to Fig. 6 at

Voltage (V)0 50 100 150 200 250

1/tim

e (s

-1)

0

1x106

2x106

3x106

4x106

5x106

6x106

μe = 990 +/- 50 cm2/Vs

Electrons

Holesμh = 82 +/- 2 cm2/Vs

Fig. 8. Room temperature electron and hole drift mobility in CdTe,calculated for electrons from Fig. 7 and for holes from correspondingIBIC maps of hole drift time.Fig. 6. Digital IBIC map of electron drift time in a 2.15 mm thick sample

of CdTe, at a bias voltage of �20 V. The area of the map is1.2 mm � 1.2 mm and the mean electron drift time is 1.5 ls.

-50V

Drift time (us)0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Cou

nts

0

50

100

150

200

250

-100V

-200V

-250V

Fig. 7. Histograms of electron drift time in CdTe at different bias voltages,extracted from IBIC maps similar to those shown in Fig. 6.

P.J. Sellin et al. / Nucl. Instr. and Meth. in Phys. Res. B 266 (2008) 1300–1306 1305

four different bias voltages. The expected decrease in meandrift time is observed, reducing to 180 ns at �250 V. Thereciprocal of the drift time (1/sdr) increases linearly as afunction of voltage (Fig. 8) and from this data the electronmobility is found to be 990 ± 50 cm2/V s. The correspond-ing sequence of digital IBIC maps measured for positivebias voltages confirms that the hole drift time is also uni-form across this sample, with a mean value of 1.7 ls at+600 V. Fig. 8 also shows 1/sdr values for hole drift time,which gives a hole mobility value of 82 ± 2 cm2/V s. Thesedata confirm the uniformity of electron and hole mobilityin CdTe grown by the travelling heater method, withmobility values which are very similar to those reportedelsewhere for CdTe:Cl [17]. The CdTe electron mobility isslightly lower than for CdZnTe (le � 1100–1200 cm2/V sin CdZnTe) whereas the hole mobility is slightly higher(lh � 50–80 cm2/V s in CdZnTe) [17,18].

5. Conclusion

The use of IBIC in both a conventional and time-resolved mode provides a powerful technique for quantita-tive high resolution mapping of charge transport propertiesin bulk semi-insulating compound semiconductors. Wehave presented some of the recent results of mobility–life-time product and drift mobility mapping in high qualityCdTe:Cl and CdZnTe materials which are currently beingdeveloped for radiation detector applications. The influ-ence of charge transport in CdTe due to extended telluriuminclusions has been presented and the variation of theregion of reduced mobility–lifetime product as a functionof bias voltage has been discussed. Samples of Bridgmangrown CdZnTe show excellent uniformity of electronresponse and a high electron mobility–lifetime value, whichis in contrast to the highly non-uniform distribution ofmobility–lifetime product observed for holes. Time-resolved IBIC has also been used to produce maps of elec-tron and hole drift times, confirming good uniformity ofelectron and hole mobility in CdTe:Cl. This ion beamimplementation of the more generalised time of flightmethod is a very useful technique which is currently beingextended into temperature dependent mobility measure-ments and transient spectroscopy mapping.

References

[1] G. Vizkelethy et al., Nucl. Instr. and Meth. B 158 (1–4) (1999) 437.[2] S. Rath et al., Nucl. Instr. and Meth. A 512 (2003) 427–432.[3] P.J. Sellin et al., IEEE Trans. Nucl. Sci. 52 (2) (2005) 3074.[4] N. Baier et al., Nucl. Instr. and Meth. A 576 (1) (2007) 5.[5] P.J. Sellin et al., Nucl. Instr. and Meth. A 531 (1–2) (2004) 82.[6] P.J. Sellin et al., Appl. Phys. Lett. 77 (6) (2000) 913.[7] A. Balducci et al., Diam. Relat. Mater. 14 (11–12) (2005) 1988.[8] A. Lohstroh et al., J. Appl. Phys. 101 (2007) 063711.[9] Z. He, Nucl. Instr. and Meth. A 463 (1–2) (2001) 250.

[10] M.B.H. Breese et al., Nucl. Instr. and Meth. B 264 (2007) 345.

1306 P.J. Sellin et al. / Nucl. Instr. and Meth. in Phys. Res. B 266 (2008) 1300–1306

[11] A. Lohstroh, P.J. Sellin, A. Simon, J. Phys. Condens. Matter 16 (2)(2004) S67.

[12] P.J. Sellin et al., Semiconductors 41 (2007) 395.[13] A. Simon et al., Nucl. Instr. and Meth. B 219–220 (2004) 405.[14] M.B.H. Breese, D.N. Jamieson, P.J.C. King, Materials Analysis

Using a Nuclear Microprobe, Wiley, NY, 1996.

[15] P.J. Sellin et al., Nucl. Instr. and Meth. A 521 (2–3) (2004) 600.[16] M.C. Veale et al., Nucl. Instr. and Meth. A 576 (1) (2007) 90.[17] K. Suzuki et al., IEEE Trans. Nucl. Sci. 49 (3; Part 2) (2002)

1287.[18] Y. Eisen, A. Shor, I. Mardor, Nucl. Instr. and Meth. A 428 (1) (1999)

158.