New trends in CdTe detectors for X and γ-ray applications Olivier Limousin CEA Saclay / DSM / DAPNIA Service d Astrophysique France New developments in photodetection, Beaune 2002 / Solid state detectors session
Summary CdTe as sensitive medium for X and γ- rays detection Bulk detectors Pixel arrays Conclusions : New trends summary
Introduction Increasing demands for new semiconductor detectors for X and γ-rays (medical, space, nuclear and physics applications) Semiconductors are well suited for compact spectro-imaging devices with a good energy resolution between scintillators and cooled Germanium Progress in technology of producing CdTe and CdZnTe (stability and reproducibility) Development of integrated front-end electronics technologies (ASIC)
CdTe : sensing medium for X and γ-rays «par excellence» High Z (Cd 48, Te 52) well suited for photelectric effect 1,0 High density (~ 6) well suited for system Photoelectric compactness 0,8 Wide band- gap and High resistivity (10 9 0,6 to 10 11 Ω cm) at room temperature Probability 0,4 Simple detector geometry 0,2 Compton High potential 0 for X and gamma rays 0 100 1000 spectroscopy Energy (kev)
CdTe versus other semiconductors Semiconductor density Z E gap ε E intrinsic [g cm -3 ] [ev] PE + Compton [ev/pair] [ev] at 100 kev (solid line) Si 2.33 14 1.12 3.6 450 Ge 5.33 PE32 only 0.67 2.9 400 (dashed) CdTe 5.85 48,52 1.44 4.43 620 CdZnTe 5.81 48, 52 1.6 4.6 700 Detection efficiency for 5100 andkev 10 gamma-ray thick CdTe photon detectors as a function as a of detector function thickness of energy in CdTe, Si and Ge Data from Takahashi and Watanabe, IEEE TNS, 2001; VOL 48; PART 4; PART 1, p 950
Two main CdTe families CdTe:Cl (THM) gap around 1.5 ev ρ 1 10 10 Ωcm p type crystals ex : 4 4 2 mm 3 10 na at 100V, 20 C µτ holes 1 10-4 cm 2 V -1 µτ elect. 1 10-3 cm 2 V -1 Uniform charge properties Up to 50mm wafer No grain boundary in wafers Cd 1-x Zn x Te (HPB) 0.08 < x < 0.15 gap around 1.6 ev ρ = 1 10 11 Ω cm n type crystals ex : 4 4 2 mm 3 1.5 na at 100V, 20 C µτ holes 2 10-5 cm 2 V -1 µτ elect. 0.5 to 5 10-3 cm 2 V -1 Very good resistivity Possible grain boundaries Bad Yield but detectors up to 1 cm 3
Signal induction principle The signal formation is described by the Schockley-Ramo theorem The signal is induced by charge carrier motion along the electric field lines This motion is seen by capacitive influence on electrodes depending on their geometry I(t) = q 0 µ E. E W E E W Applied field (stationnary regime) Weighting field (transient regime)
Let s talk about «bulk» detectors CdTe:Cl (THM) bulk detectors Ex : ISGRI (Lebrun et al.), Tokamak (Peysson et al.) CdZnTe (HPB) bulk detectors Ex : the PEGASE camera (Mestais et al.) CdTe:Cl Schottky detectors Ex : Takahashi et al. CdZnTe bulk detectors with other electrodes configurations Ex : Luke et al., Parnham et al. (ev-products) CdZnTe bulk detectors with capacitive electrodes Ex. of application : next talk (Lebrun)
Signal induction in a coplanar device γ-ray Photon -100 V 2 mm h + e - Q t U 0 Q 0 E γ
Signal induction in a coplanar device Schockley-Ramo theorem gives the instantaneous induced current I(t) If the detector is uniform, no space charge E = V 0 /L E = 1/L W The induced charge dq L at the anode is dq L = I(t) dt = q 0 dx L The induced charge is proportional to the charge carrier motion and depends on the penetration depth of the photon
Charge loss and balistic deficit 10µs Collected charge 1 µs 15% ~2µs 70% Charge loss (trapping) time time Balistic deficit (filtering)
Charge loss and balistic deficit The «collected» charge is described by the hecht relation which take into account physical trapping ie, charge transport properties (µ, τ) The hole mobility drives the rise-time, ie the balistic deficit in CdTe:Cl
Biparametric diagram 8 Pulse rise-time (µs) 6 4 2 0 0 50 100 150 Energy (kev)
Charge loss correction 8 5000 Counts P/V 3 Pulse rise-time (µs) 6 4 2 0 0 50 100 150 Energy (kev) 8000 Counts 0 0 50 100 150 200 Energy (kev) P/V 9 0 0 50 100 150 200 Energy (kev)
ISGRI :In Beaune 1999, we went with this
This time, here we are with ISGRI
with spectacular images
Calibration phase coded mask aperture shadowgram with ISGRI camera at 511 kev The eight modules spectra with 22 Na source
and spectra 8 Rise-time (µs) 6 4 2 0 0 50 100 150 Energy (kev) LT ~12 kev 7,5% 0 20 40 60 80 100 120 140 Energy (kev)
Spectral performances of ISGRI CdTe Resolution FWHM (%) 100 10 25 % (3,6 kev) 7,5 % (9 kev) 14,4 kev 1 10 100 1000 Energy (kev) 122 kev
An example of application in physics In the field of continuous thermonuclear reactions control in a Tokamak (TORE SUPRA) CdTe:Cl allows the design of compact cameras for hard X-ray tomography of the bremsstrahlung emission by electrons in tokamak fusion plamas Such electrons produce Hard X-rays between 20 and 200 kev Analyse of these electrons provides information about current density profiles Example from Peysson et al., NIMA 458, 2001, p 269
An example of application in physics Two cameras with 24 and 38 CdTe detectors (5 5 2 mm 3 ) Detectors stayed stable even under high fast neutrons flux and high magnetic field environment Example from Peysson et al., NIMA 458, 2001, p 269
PEGASE : a CZT camera for medecine Pegase is based on thick bulk CdZnTe crystals In this configuration hole signal is negligeable The associated electronics (ASIC) deals with electron pulse rise-time Example from Mestais et al., NIMA 458, 2001, p 62
PEGASE : electron loss correction All events in this window are affected to the 140 kev line of 99m Tc source 140 kev line of 99m Tc source 70% efficiency at 122 kev in a ±6.5 % window Example from Mestais et al., NIMA 458, 2001, p 62 Window selection for the line
CdTe:Cl with Schottky In contact The basic idea is to reduce the dark current noise contribution with a Schottky anode contact For thin detectors, it provides very nice spectra, NO BALISTIC DEFICIT The main problem is due to polarization effect. This can be solved by : - High bias voltage values - Negative temperature down to 40 C - Pusing the HV
CdTe:Cl close to Ge Needs a very low noise preamplifier! This often goes in the wrong direction if we must consider power consumption. FWHM 830 ev!! 2 2 0.5 mm 3 Schottky CdTe diode, 1400V, -40 C Example from Takahashi et al., NIMA 1999 & IEEE TNS 2001
Modifying weighting potentiel on CZT The idea is to reduce the influence of the penetration depth in the signal induction modifying the weighting potential Another point is to forget the holes, ie to have a single carrier collection Then, it gives the opportunity to use thick CZT detectors - electrode configuration (ex : Parnham et al., Luke et al.) - capacitive electrodes (ex : Montemont et al.)
Weighting potentiel in coplanar device Depth (mm) Depth (mm) Radius (mm) (mm) anode anode cathode cathode CAPture geometry, Parnham et al. from ev-products (USA) Scheme from Montemont, thesis université J. Fourier, Grenoble, 2000
ev-product Design : spectra Results with CAPture : -<3keV at 59.5keV, - <5 kev at 122.1 kev and - <13 kev at 662 kev -reductions in low energy tailing CAPture geometry, Parnham et al. from ev-products (USA) 5 5 5 mm 3 CZT detector Data from Parnham et al., SPIE conference, july 1999
CZT coplanar-grid array Coplanar-grid electrode pattern with edge compensation Substracting the signals from the two grids removes the hole contribution Data from Luke et al., NIMA 458, 2001, p 319 1 cm 3 coplanar-grid electrode CZT coupled to its electronics A small voltage is applied between the two grids. Electrons are collected on one grid.
Capacitive electrodes CZT Depth (mm) Radius (mm) anode cathode Dielectric film screen Capacitive electrode geometry, Montemomt et al. from CEA/LETI Data from Montemont et al., IEEE TNS, 2001; VOL 48; PART 3; PART 1, p 278
Capacitive electrodes CZT performances 4 4 6 mm 3 Schottky CdTe diode, 400V, 21 C NEW TREND! Energy (kev) Performance should not depend on the detector thicness
Bulk detectors in two words detector FWHM FWHM Thickness type [kev] at 122 kev [kev] at 662 kev [mm] CdTe 5.5 23 2 CdTe Schottky 1.5 NA 0.5 CZT bulk 6? 6 CZT Capture 5 13 5 CZT coplanar-grid 9 14 10 CZT capacitive electrode 3.6 12 6
Let s talk about «pixel» arrays Fine pixel arrays Ex : - CdTe Medipix evolution (Manach et al.) - Infocµs (Stahle et al.) Medium size pixel arrays Ex : HEFT (Ramsey, Bolotnikov, Cook et al.) Small pixel effect in CdTe arrays Thick CdZnTe pixel arrays Ex : Simbol_X
Medipix arrays characteristics European collaboration with CERN The goal is to realize an highly integrated chip (CMOS 0.25 µm) for high count rate Xand γ-rays counting imagers with semiconductor detectors First generation (Medipix 1) developped for GaAs detectors. Readout of the hole signal New generation (Medipix 2) developped for electron collection and allows the use of CdTe semiconductor Data from Manach et al., CEA/DRT/LIST and Amendiola et al., NIMA 422, 1999, p 201
Medipix arrays design Semiconductor detector Indium bump interconnexions Medipix2 readout chip (256 256 pixels) Readout cell 55µm 55µm Data from Manach et al., CEA/DRT/LIST and Amendiola et al., NIMA 422, 1999, p 201
Infocµs CZT pixel arrays In the field of hard X-rays and γ-rays astronomy Focal plane for new focussing optics in the range of 10-100 kev with grazing incidence mirrors This technic allows a very high spatial resolution The detector is made of a 26.9 26.9 mm 2 CZT crystal, 2 mm thick. It is a 64 64 pixels array. Data from Stahle et al., NIMA 436, 1999, p 138 and http://lheawww.gsfc.nasa.gov/docs/balloon/focus.html
Infocµs CZT pixel arrays Counts 2.3 kev FWHM Infocµs CZT detector assembly Energy (kev) 109 Cd source spectrum with Infocµs CZT detector Data from Stahle et al., NIMA 436, 1999, p 138 and http://lheawww.gsfc.nasa.gov/docs/balloon/focus.html
HEFT CZT pixel arrays In the field of hard X-rays and γ-rays astronomy again Focal plane for the High Energy Focussing Telescope The goal of this work is to achieve less than 1 kev at 60 kev (very low noise ASIC) The detector is made of an 8 8 pixels array (6.7 6.7 2 mm 3 ) with 680 650 µm pixel size. Data from Ramsey, Bolotnikov, Cook et al.
HEFT CZT pixel arrays HEFT CZT detector assembly Energy (kev) 241 Am spectra (a) 0.9 kev FWHM at 60 kev, 5 C (b) 1.1 kev FWHM at 60 kev, room temperature
Small pixel effect These nice results are possible because of the crystal quality, the ASIC performances and also the small pixel effect Small pixel effect is due to the weighting field distribution close the anode when the pixel size is less than a quarter of the thickness The nature works fine! Data from Eskin et al., Hage-Ali et al.
Thick CZT detectors arrays NEW TREND! 64 pixels CZT arrays, 6 mm thick
Conclusions Thanks to CdTe detectors, it is now possible to dream of high spectral performances, high spatial resolution and high efficiency simultaneously The high spectral resolution obliges to think about new geometries and high performance electronics Among these new geometries, capacitive electrodes detectors for bulk detectors and thick pixel arrays appear as major new trends