CdTe Detectors for Quantitative X-Ray Fluorescence 6 of 25 Apparatus Thermoelectrically Cooled Solid...

2007 Denver X-ray Conference Page 1 of 25

CdTe Detectors for Quantitative X-Ray Fluorescence

R. Redus, J. Pantazis, T. Pantazis, A. Huber

Amptek, Inc14 DeAngelo Dr, Bedford MA 01730

B. CrossCrossRoads Scientific

414 Av. Portola, El Granada, CA


Why use CdTe?Much higher sensitivity > 20 keV

RoHS/WEE demands accurate measurement of metals

With CdTe, one can measure K X-rays (with few interferences) with much higher sensitivity than Si diodes Energy (keV)

1 10 100 1000

Inte

ract

ion

Prob

abili

ty

0.01

0.1

1

O F NeNaMgAl Si P S Ar Ca Ti CrFeNiZnGeBr Sr Mo AgSbBa TaAuPb U Fm

500 μm Si-PIN1 mil Be window

200 μm (active) Si-PIN0.5 mil Be window

750 μm CdTe4 mil Be window

1 mm CdTe

Photoelectric Interactions

1000 μm Si-PIN


57Co spectra measured with 25 mm2 Si-PIN and CdTe

– 6.4 keV Fe Kα X-ray• Equal photopeak area

– 14.4 keV γ-ray• CdTe has 25% more

photopeak area

– 122 keV γ-ray• CdTe has 140x photopeak

area0.0E+00

1.0E+04

2.0E+04

3.0E+04

4.0E+04

0 20 40 60 80 100 120 140

Energy (keV)

Cou

nts

122 keV γ-ray CdTe Si-PIN

14.4 keV γ-ray

6.4 keV Fe Kα X-ray

0 5 10 15 20Energy (keV)

Cou

nts


• Resolution– Electronic Noise– Hole tailing

• Escape peaks• Continuum

– Compton– Dead layer

• Other– Stability– Linearity

Key Spectral Characteristics of CdTe

1.0E+00

1.0E+01

1.0E+02

1.0E+03

1.0E+04

1.0E+05

0 20 40 60 80 100 120 140

Energy (keV)

Cou

nts

ElectronicNoise

Hole Tailing

EscapePeaks

ComptonContinuum


ApparatusDetectors

– CdTe• Compound semiconductor with wide bandgap (4.4 eV), high density (6.2

g/cm3), and high atomic number (48,52)• Charge transport better than most alternatives μτh=2x10-4 cm2/V• Studied and used for γ-ray spectroscopy since late 1960s

– Amptek detectors• Schottky (blocking) contacts to reduce leakage current• Idark ≈ 5 nA/cm2 at 500V and 300K • Mπn structure from Acrorad, Inc

– Good yield, reproducible properties

– Amptek diodes are 0.5 to 1 mm thick from 3x3 to 7x7 mm2

– Results here are for 0.75 x 5 x 5 mm3 unless otherwise stated

Page 6 of 25

Apparatus

Thermoelectrically Cooled Solid State Detector

• Reasons for thermoelectric cooling– Reduces shot noise and thermal noise– Cooling invisible to user

• Two stage cooler for >80˚C differential– 215K for lab use– 240K for field use (at ambient of 45˚C)

• FET and feedback components on cooler– Leakage currents as low as 5 fA– Low stray capacitance, reduced EMI pickup

• Continuous feedback preamp using current divider

+VL

Rd

FETDetector

Cf

OutputThermoelectric Cooler

RfCurrentDivider

iR

α iR

(1-α) iR


ApparatusX123, PX4, DP4

– All are complete spectroscopy systems– All share core technologies

• Digital pulse processor for pulse shaping, selection logic, and multichannel analyzer

• Power supplies, including 1.5 kV bias supply and closed loop temperature control

• USB interface, +5V power input

– Targeted at different applications

PX4 and XR100 for benchtop & laboratory

DP4 and PA210 for embedding in instruments

X123 for compact, packaged system


Electronic Noise

CdTe Noise Components– Typical results with a 25 mm2 x0.75 mm detector– Noise corner <600 eV FWHM near 6.4 μsec peaking time– Noise dominates below 30-50 keV, Fano broadening above

1.0E-01

1.0E+00

1.0E+01

1.0E-01 1.0E+00 1.0E+01 1.0E+02Peaking Time (msec)

Res

olut

ion

(keV

FW

HM

)

122 keV

59.5 keV

14.4 keV

Electronic Noise

CdTe 25mm2 x 0.75 mm500V bias, 230K


Hole Tailing: OriginsLifetimes

τe = 3 μsecτh = 1 μsec

Transit timeTe (max) = 10 nsecTh (max) = 110 nsec e-

h+

IncidentX-rays

+

Cathode (-)

Anode (+)

e-

+

e-+

h+ h+

1

2

3

-5.0E-08 5.0E-08 1.5E-07 2.5E-07

Time (sec)

Prea

mp

Out

1: At Cathode (all electrons)2: Middle (both)3: At Anode (all holes)

0.900

0.925

0.950

0.975

1.000

0 0.025 0.05 0.075

Distance (cm)

Cha

rge

Num

ber

Charge: Q(x)Number: N(x)

0.0E+00

5.0E+02

1.0E+03

1.5E+03

2.0E+03

2.5E+03

0 20 40 60 80 100 120 140

Energy (keV)

Nor

mal

ized

Cou

nts

Cathode interactions

Anode interactions


Hole Tailing: Effects

“Physics” based model– Charge collection ⇒ Q(x)– Photoelectric absorption ⇒ N(x)– Combination ⇒ N(Q) – Convolve with Gaussian

ENC and Fano

– Fairly accurate representation– Estimate performance– Not in analytical form, so difficult

to use for spectra fitting

110 115 120 125 130Energy (keV)

Cou

nts

Ideal Gaussian

Ideal Hole Tailing Exponential attenuation Hecht equation

Gaussian convolved with

ideal tailing

110 115 120 125 130Energy (keV)

Mod

el C

ount

s

Mea

sure

d C

ount

s

Model

Measured


Hole Tailing: Effects

Photopeak shape<20 keV: All interactions near cathode, no tailing, Gaussian≈50 keV: Small asymmetric correction to Gaussian≈100 keV: Interactions uniform, tail important, shape complex

Can model using Van Espen type tail but truncate at Qanode

0 5 10 15 20Energy (keV)

Cou

nts

54 56 58 60 62Energy (keV)

Cou

nts

110 115 120 125 130Energy (keV)

Cou

nts


Escape Peaks

Much more important in CdTe than in Si– Cd & Te ωK ≈ 85%, 5% in Si– K X-ray attenuation lengths 0.1 to 0.2 mm

1.0E+00

1.0E+01

1.0E+02

1.0E+03

1.0E+04

1.0E+05

0 10 20 30 40 50 60 70

Energy (keV)

Nor

mal

ized

Cou

nts

Compton backscatter from source (48.3 keV)

Photopeak59.5 keV

Log

Linear

Te Kβ Te Kα Cd Kβ Cd Kα

X-ray energy (keV)

Te Kβ 31.0

Te Kα 27.5

Cd Kβ 26.1

Cd Kα 23.2


Escape Peaks: Correction

Intensity vs energy– Computed using EGS4

Monte Carlo softwareCarried out by Paul Bennett of RMD, Inc.

– Data using filtered isotopic sources.

Algorithm– 4 analytical equations, one

for each of the weighted centroids for Kα and Kβ of Cdand Te

– Subtraction starts at high energy, looking for all events above K edge

0%

5%

10%

15%

20%

0 50 100 150

Incident energy (keV)

Esca

pe p

eak

/ pho

tope

ak r

atio

Modeled Cd Kα

Measured Cd Kα

Cd Kβ

Te Kα

Te Kβ

Equations of the general form:

Esc = a0 + a1•e + a2•e2 + a3•e3 + a4•e4 + a5•e5

where Esc=escape events, e=parent energy, and the a terms are coefficients.



Results of Correction: Filtered Ag Tube Spectrum at 30 kVp– Filter should remove everything below 14 keV.– Raw spectrum (blue) shows broad peak around 5 keV, due to escapes– Correction algorithm moves to the gray, removing almost entirely.– Had to adjust Cd escape edge from 26.7 to 26.0 keV – not clear why.



Results of Correction: W Tube Spectrum at 80 kVp– Raw spectrum (blue) has large “steps” at K edges– Theoretical correction (yellow) helped but left some structure at K edges– Adjusting the Cd-K edge from 26.7 to 26.0 keV improves greatly, leaves a little

structure at the K edges.



Result of Correction: Pure Lead– Raw spectrum (blue) has clear escape peaks from 40 to 55 keV– Four primary peaks, plus continuum, each with four escape peaks– Gray represents the “reassigned” photons– Yellow processed spectrum shows change to continuum and peaks


Compton Background

Continuum Removal– CdTe spectra, at higher energy, have more scattering into detector– Plot shows result of applying Si parameters to CdTe– Yields residual false peaks

Continuummodels

Residuals


Compton Background

Continuum Removal– Adjust parameters to give high curvature background continuum

• First, Cd-Te escape peaks are removed (partially at least)• Second, automatic background function applied to spectrum

– Very little residual continuum


Low Energy Background

Dead Layer Effects– CdTe dead layer much more

significant than Si• Metal contact (200nm Pt)• CdTe higher Z and density

– Secondary electrons deposit more energy while escaping

– At low energies, peak to background ratio lower for CdTe


Stability

Does CdTe polarize?– At room temperature and low electric field strength, CdTe

Schottky diodes polarize– Polarization slows rapidly with cooling and high bias voltage– As operated in XR100-CdTe, negligible on time scale of days– Recovers within minutes at zero bias

Cathode (-)

Anode (+)

e-

+

h+

Cathode (-)

Anode (+)

e-

+

h+

++ +

+++

TrappedChargesinitialE

Reduced

E

0

400

800

1200

45 50 55 60 65 70

Energy (keV)

Cou

nts

Start2 hours4 hours6 hours8 hours

16 hours18 hours20 hours22 hours24 hours

Data taken at 300K, 200V


Stability

– Left: photopeak centroid and counts over 5 days• Gain fluctuations consistent with 30 ppm/˚C temperature coefficient • Count rate follows radioactive decay of 57Co

– Right: Spectra measured 60 hours apart– Stable over period of days. Expect drift at some longer time scale

7.80E+04

8.05E+04

8.30E+04

8.55E+04

8.80E+04

10/6 12:00 10/7 12:00 10/8 12:00 10/9 12:00 10/10 12:00

Date and Time

Phot

opea

k C

ount

s

121.25

121.50

121.75

122.00

122.25

Cen

troi

d (k

eV)

Centroid15 ev rms, tracks at 30 ppm/°C

Photopeak Counts

57Co decay


Linearity

Highly Linear at X-ray Energies

– 14 to 136 keV– R2 = 0.0.999997

– Use peak channel, not centroid– At higher energies, use Qmax

from photopeak fit 0

1024

2048

3072

4096

0 25 50 75 100 125 150

Energy (keV)

Cha

nnel

-2

-1

0

1

2

Diff

eren

ce (k

eV)

R2=0.999997


Reproducibility

– Plots show data from production lot of 20 detectors– One detector exhibited higher noise and worse tailing– Other nineteen consistent

1.0E+00

1.0E+01

1.0E+02

1.0E+03

1.0E+04

0 10 20 30 40 50 60 70

Energy (keV)

Cou

nts

1.0E+00

1.0E+01

1.0E+02

1.0E+03

1.0E+04

0 10 20 30 40 50 60 70

Energy (keV)C

ount

s


Conclusion

• Sample analysis of Pb-Sn Solders

• Spectrum Processing– Cd-Te escape– Background– Gaussian peak fits

• FP Calibration– Pure Sn & Pb stds.

• FP Analysis– 68% Sn for Sn-Pb

thickness = 1 mm– 63% Sn for Sn-Pb

thickness = 0.5 mm– (Nominal Sn = 63%)

•Raw Spectrum•Processed Spectrum•Gaussian Peak Fit Spectrum

Even though Pb-Kα peaks have significant low-energy tails, quantitative analysis is possible.


Conclusions– CdTe is a powerful tool for measuring metals via XRF

• It has high sensitivity for K lines, with fewer interferences• One can carry out quantitative analysis

– Spectral characteristics require changes to algorithms• Hole tailing shape is different• Escape peaks more significant• Continuum more significant and shape different• Amptek’s XRS-FP will implement these corrections

– For more information, go to www.amptek.com

Date post:	15-Apr-2018
Category:	Documents
Upload:	duongnguyet
View:	218 times
Download:	3 times

CdTe Detectors for Quantitative X-Ray Fluorescence 6 of 25 Apparatus Thermoelectrically Cooled Solid...

Documents