Nuclear Instruments and Methods in Physics Research A 504 (2003) 149–153

Technology development of 3D detectors for medical imaging

Giulio Pellegrinia,b,*, P. Royb, A. Al-Ajilib, R. Batesb, L. Haddadb, M. Hornb,K. Mathiesonb, J. Meloneb, V. O’Sheab, K.M. Smithb, I. Thaynea, M. Rahmanb

aDepartment of Electronics and Electrical Engineering, Glasgow University, Glasgow G128QQ, UKbDepartment of Physics and Astronomy, Glasgow University, Glasgow G128QQ, UK

Abstract

Fabrication routes to realising ‘3D’ detectors in gallium arsenide have been investigated and their electrical

characteristics measured. The geometry of the detector is hexagonal with a central anode surrounded by six cathode

contacts. This geometry gives a uniform electric field with the maximum drift and depletion distance set by electrode

spacings rather than detector thickness. The advantages of this structure include short collection distances, fast

collection times and low depletion voltages depending on the electrode diameter and pitch chosen. These characteristics

are fundamental for the application of 3D detectors in, for example, medical imaging and protein crystallography.

r 2003 Elsevier Science B.V. All rights reserved.

Keywords: Medical imaging; GaAs; Laser drilling

1. introduction

The study and quest for new semiconductordetectors have become very active in recent years.Silicon is the most widely used material in particledetection but, due to its low absorption efficiencyat photon energies above 10 keV, it has limitedapplication in high energy X-ray detection. Mate-rials with greater absorption coefficient, such asGaAs and CdZnTe, have been investigated [1] forthis application. Detectors produced in GaAssuffer from incomplete charge collection resultingfrom imperfections in the crystal structure. Thisreduces the mean free drift path of the chargecarriers in these materials and limits the thicknessof the sensitive region. Additionally, problemswith the growth of GaAs have led to high effective

carrier concentrations and so increased operat-ing voltages. This implies thin detection layersand, in some cases, the presence of an undepletedlayer.To avoid these type of limitations, a new

detector architecture has been proposed [2]. Athree-dimensional array of electrodes that pene-trate into the detector bulk makes up the so-called3D detectors, as shown in Fig. 1. The aim is to setthe maximum drift (x) and depletion distances (W )by the electrode spacing rather than by thedetector thickness as in the more conventionalplanar technology. The advantages of this struc-ture include collection distances and collectiontimes reduced by over one order of magnitudecompared to standard planar pixel and microstripdetectors, and depletion voltages up to two ordersof magnitude lower. The main applications ofthese detectors are medical imaging and precisiontracking measurements.

*Corresponding author.

E-mail address: giuliop@a5.ph.gla.ac.uk (G. Pellegrini).

0168-9002/03/$ - see front matter r 2003 Elsevier Science B.V. All rights reserved.

doi:10.1016/S0168-9002(03)00811-8

2. Detector fabrication

To create a 3D detector an array of holes mustbe formed through the substrate. The techniqueused to create high aspect ratio holes (depth todiameter ratio) in semiconductor material here waslaser drilling. The main advantages of using a laseris that it is independent of the material drilled (e.g.silicon, gallium arsenide (GaAs), silicon carbideand CdZnTe) and it is the best technique availablecurrently for GaAs.

The drilling operation was carried out atStrathclyde University using a Ti:Sapphire femto-second laser [3]. This system can provide 3mJ laserpulses of 40 fs duration at a pulse repetition rate of1 kHz. The ultrashort pulses ablate material viathe rapid creation of a plasma that absorbs theincident energy, resulting in direct vaporisationfrom the target surface. This produces negligiblecollateral heating or shock-wave damage.Holes obtained in GaAs using this technique [4]

are shown in Fig. 2. A layer of photoresist was

h+

e-

+ve -ve +veSiO2

x3D

W3D

E

Bulk

-ve -ve-ve

h+

e-

+ve

x2D

W2DE

p+

n

n+

Ionising Radiation

Fig. 1. Cross-section of a 3D detector (left) and a planar detector (right).

Fig. 2. (a) Top surface of a 10mm hole; (b) hexagonal geometry (pitch 85mm); (c) cross-section of laser drilled holes (sample thickness500mm) and (d) metal evaporated on the sidewall of a silicon 10 mm hole.

G. Pellegrini et al. / Nuclear Instruments and Methods in Physics Research A 504 (2003) 149–153150

spun on the surface of the samples which were thenplaced in a vacuum chamber to prevent the laserfrom ionising the air around the area it hits. Testshave shown that the photoresist protects thesurface of the sample from debris ejected duringlaser ablation. The samples were mounted on a x–ystage controlled by a PC in order to create thearray of holes. A sacrificial layer, about 200 mmthick, was placed on top of the samples. This layerprovides further protection of the sample surfaceand, due to tapering, decreases the spot size of theholes in the sample. The spot size of the holes inthe sacrificial layer is about 30 mm while the size inthe sample is 10 mm. The holes obtained in GaAssamples have diameters of about 8–10 mm (Figs. 2aand b) and go through the whole thickness of thematerial. The samples are about 500 mm thick. Theholes follow a straight line for the first 300 mm andthen start to diverge as the energy densitydecreases beyond a threshold point as shown inFig. 2c. For the applications of 3D detectors inX-ray detection, 300 mm is a suitable depth. Mostof the debris on the surface was cleaned up byremoving the layer of photoresist. However, acidcleaning for 10min in a solution of 1:1:50HCl:H2O2:H2O can be used to further removethe debris left on the surface.Alternative methods for creating the holes in

GaAs and silicon, such as dry etching andelectrochemical etching, are under investigation[5] but not reported here.

3. Electrode formation

After creating the array of holes in the sub-strate, electrodes must be formed within the holes.This was done by evaporating a layer of metalinside the holes in order to form Schottkycontacts.Before metal deposition, the silicon sample was

de-oxidised for 30 s in a 1:4 solution of hydro-fluoric acid (HF) and water. This removed anyoxide that had grown on the internal walls sincethe etching process. The metal deposition wascarried out using a Plassys MEB 450 ElectronBeam Evaporator. A layer of 33 nm of Ti, 33 nm ofPd and 150 nm of Au [6] was evaporated inside the

hole forming Schottky contacts. The uniformdistribution of the three metals within the holewas verified by using a scanning electron micro-scope (SEM). In Fig. 2d is shown the layer ofmetal (brighter colour) on the sidewall of a hole.Previous tests done in silicon showed that the fullamount of metal was deposited at the bottom of a130 mm hole while only about one-third of themetal was deposited on the sidewall. In order toconfirm that the evaporated gold was deposited allaround the wall, these holes were filled byelectroplating gold inside them (a previous layerof gold being necessary to start the growth). Thesample was then cut and a cross-section wassuccessfully checked with a SEM. To connect theelectrodes to the electronics, a circuit was printedon the surface of the sample by evaporating a layerof 150 nm of aluminium on the oxide layer (about400 nm) present on the surface. This is shown inFig. 3.

4. Electrical measurements

Electrical characteristics of a single cell of a 3Ddetector structure were measured. Experimentalcurrent–voltage and X-ray spectra were studied.The results presented are for GaAs devices withholes generated using laser drilling and Schottkycontacts made as shown in Fig. 3.

Fig. 3. Hexagonal geometry of a 3D detector with a pitch of

85 mm. The central electrode is wire bonded to the read out

electronics while the surrounding holes are grounded.

G. Pellegrini et al. / Nuclear Instruments and Methods in Physics Research A 504 (2003) 149–153 151

4.1. Current–voltage measurements

Current–voltage measurements were performedusing a Keithley 237 source. The Keithley wascontrolled by a LabView program. The sampleswere mounted on a support and connected to theelectronics. The samples were then placed in ametal box to provide electrical and light shielding.As the current is temperature dependent, themeasurements were carried out in a Hareausenvironmental chamber, in order to keep thetemperature constant at 20�C. A voltage stepwas applied and the current measured after a 10 sdelay.The characteristic obtained is not symmetric for

reverse and forward bias, as shown in Fig. 4a. Thiscould be due to the asymmetric geometry of thecell structure. When the central electrode isforward biased, a reverse biased Schottky barrierwill be created on the surrounding electrodes,whereas only the central electrode will support aSchottky barrier when it is reverse biased.The leakage current is relatively low (about

30 nA per unit cell at �35V when the detector isfully depleted). The barrier height value of theSchottky contacts created was measured applyinga forward bias in the range 0.1–1V. Then, a linearfit was used to extrapolate the value of the bar-rier height [7]. This was found to be (0.970.1) eV,in agreement with the data reported in theliterature [8].

4.2. X-ray measurements

Some X-ray pulse height spectra were measuredwith the fabricated detectors. For light andelectrical screening the sample was placed in ametal box which had a hole on one side. A variableX-ray source was placed at this hole to irradiatethe device. The signal from the detector was readby an Ortec pre-amplifier connected to a post-amplifier with a 500 ns shaping time. The signalwas then processed by a PC-based multichannelanalyser to obtain the spectrum. All the measure-ments were carried out at room temperature. Theresults were repeated for several cells created onthe same piece of material.Fig. 4b shows the pulse height spectra of

fluorescent X-rays of Am, Mo, Ag, Ba, and Tb.The energies of X-ray emitted by this source are inthe range 17 keV (Mo) to 60 keV (Am). The energyresolution obtained from this measurement wascalculated to be about 40% while the detectionintrinsic efficiency was of the order of 80%. Bothwere measured at a bias of �35V.The limiting factor of the resolution measure-

ments may be due to the fact that only one cell wasbiased while the surrounding cells were leftfloating. Future work will investigate the beha-viour of a full array of hexagonal cells using bumpbonding technology to connect each electrode tothe read-out electronics.

-50 -40 -30 -20 -10 0 10 20-80

-60

-40

-20

0

20

40

60

80

100

Current-Voltage curve

Cu

rren

t (n

A)

Bias (V)

0

200

400

600

800

1000

0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0

Energy (keV)

Sig

nal

(co

un

ts)

Mo (17 keV)

Ag (22 keV)

Ba (32 keV)

Tb (45 keV)

Am (60 keV)

(a) (b)

Fig 4. (a) Leakage current measured at room temperature and (b) spectra obtained illuminating the 3D detector by X-rays with

different energies (17–60keV).

G. Pellegrini et al. / Nuclear Instruments and Methods in Physics Research A 504 (2003) 149–153152

5. Conclusions

So far, 3D detectors in GaAs using Schottkycontacts have been successfully created and tested.Process steps for the fabrication of 3D radiationdetectors with Schottky contacts have been devel-oped and complete detectors made. A laser drillingtechnique has been used to create holes in GaAsmaterial obtaining holes with an aspect ratio betterthan 30:1. Measurements relating to X-ray spec-troscopy have been successfully carried out on thefabricated samples.

Acknowledgements

I would like to acknowledge the TOPS colla-boration for providing the laser facility and all themembers of the Detector Development group atGlasgow University for assistance in developing

the 3D technology. This research was supportedby PPARC (UK) and by the Fifth FrameworkProgramme (EU).

References

[1] S. Manolopouulos, et al., IEEE Trans. Nucl. Sci. NS-45 (3)

(1998) 394.

[2] S. Parker, Nucl. Instr. and Meth. A 395 (1997) 328.

[3] TOPS, Strathclyde terahertz to optical pulse source, http://

dutch.phys.strath.ac.uk/TOPS/.

[4] D.R. Jones, et al., Femtosecond laser micromachining of

sub 10 micron diameter holes at up to 100:1 aspect ratio.

ICALEO 2001, Jacksonville, FL, October 15–18, 2001.

[5] G. Pellegrini, et al., Nucl. Instr. and Meth. A 487 (2002)

19.

[6] R. Bates, et al., Nucl. Instr. and Meth. A 392 (1997) 269.

[7] S.M. Sze, Semiconductor Devices Physics and Technology,

Wiley, New York, 1985.

[8] E.H. Rhoderick, Metal-Semiconductor Contacts, Oxford

Science Publication, UK, 1998.

G. Pellegrini et al. / Nuclear Instruments and Methods in Physics Research A 504 (2003) 149–153 153