Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://hea-www.harvard.edu/~bradw/cv/papers/ssdSPIE.pdf Äàòà èçìåíåíèÿ: Thu Jan 2 22:59:20 2014 Äàòà èíäåêñèðîâàíèÿ: Fri Feb 28 12:18:30 2014 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï

The AXAF HXDS germanium solid state detectors
w.c. McDennottL E.M. Kellogg B.J. Warge1in, I.N. Evans', S.A. Viteka', E.Y Tsiang', D.A. Schwartz&, R. Edgar', S. Kraft',F. Schoizeb, R. Thornageib, G. jJ1b, M. Weisskopf C, S.

Odellc, A. Tennantc, J. Kolodziejczakd, and G. jpjd

Astrophysical Observatory Cambridge, Massachusetts b Physikalisch-Technjsche Bundesanstalt, Abbestr. 2-12 10587 Berlin,Germany C Marshall Space Flight Center, Alabama d USRA Huntsville, Alabama
.

a Smithsonian

ABSTRACT

The design, calibration, and performance o1 the High Purity Germanium(HPGe) solid state detectors (SSD's) used in the calibration cI the Advanced X.ray Astrophysics Facility High ResolutionMirror Assembly (URMA) is discussed. The focal plane SSD was used with various apertures to measuie the pointresponse function, as well as the effective area o the mirror. The good energy resolution of the detector allowed the effectiveenergy of the mirrors to be measured with a single exposwe using a continuum source. The energy resolution was also exploited in measuring the molecular contimiination on the mirror surfaces. The SSD's are the transfer detector standards for the HRMA calibration over the energy range from 7 eV to 10 keV. The calibration of the SSD's was performed mostly at the FIB radiometry laboratory using the electron BESSY. spectral and spacial distribution af the undispersed synchrotron radiatkm can be calculated from firststorage ringusing the The principles Schwinger Equation. With the electron storage ring being run ina reduced current mode o1 a few electrons, uncertainties in the calculated flux are below 1%. A comparison of the measuredand calculated flux made it possible to determine the detector efficiency with an uncertainty af typically 1%. Electronic effects such as pile-up, count rate linearity and deadtime have been investigated.

Keywords: AXAF, solid state detectors, X-ray detectors, absolute calibration

.

.

1.

INTRODUCTION

The HXDS solid state detectors (SSD) ai a pair cigermaniuni solidstate detector that serves as the primary transfer for the calibration of the High Resolution Mirror Assembly (HRMA) o the Advazxed X.Ray Astrophysical Facility standard in the range from 700 eV to 10 keV. One detector was absolutely calibrated and served as a flux monitor during the(AXAF) based calibration of the HRMA. The other detector was used in the focal plane o the mirror and is being calibrated ground relative to the flux mouit. The good energy resolution of the detector allowedmeasurement suh as molecular contininaion, effective energy, and effective area af the detector to be made with a sigle exposure. Also, the large array of apertures incorporated into the detector allowed the srtucture of the focused beam fromthe mirror to be explored. These characteristics and calibration descriptions are presented in the sections thatfollow.

2. DETECTOR DESCRIPTION
The HXDS SSD, purchased from Canberra Industries, is a hybrid of their LEGe detectr that has undergone modification by the Smithsonian Astrophysical Observatory (SAO). The detector consists of a high purity germanium crystal that is 6.2 mm in diameter and 5 mm thick. The active area of the detector is 30 cm2. The front angstroms of ion implanted aluminum. This contact extends around the sides surface contact (P-i.) is made of 1000 of the crystal making conducdng rather than insulating. The rear surface (N+) contact is a lithium diffused spot that does the sides of the crystal not cover surface. This unique configuration reduces the capacitance ofthe detector and thus helps to reduce noise'. the entire Visible light is blocked from reaching the Ge crystal viaan infrared shield which consists of 0.25 microns of paralyene. This brings the total amount of Al filtering to 2000 angtronis, (1000 a 1000 angstroms of Al on A on the surface of the crystal and 1000 A on the JR Shield). The detector is also ec4uipped with a 1 micron paralyene vacuum barrier, The primary purpose of this membrane was to provide the detector with its own vacuumspace in order to reduce the amount of contRmjnatjon, including water vapor, that might reach the Gecrystal oncÞ the detector was mounted in the instrument chamber that housed the JIRMA.
SPIE

Vol. 3113 · 0277-786X/97,$l 0.00

535

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 01/02/2014 Terms of Use: http://spiedl.org/terms


The Ge crystal is mounted to a 1.2 liter all.athtude portable cryostat via a cold finger. The aliattitude cryostat allows tioning the detector in any direction without loss of cryogenic fluid. With proper evacuation cif the cryostat. the cryogen would keep the detector cool without the need for a iefill for 36 hours. Since the deteCtOr would ultimately be in a vacuum itself, the cryostat was specifically designed to be able to attach cryoethc fill lines that could be used to fill the detector even when cryostat was in vacuum. A schematic of the detector is shown m Fig. 1. ·
The front of the detector is fit with a computer controlled aperture wheel that contains 20 apertures. These aperture sizes range from 5 microns to 5 mm. They also include a 132 micron Al filter and a 27 niicron Al filter which were used in the absolute calibration phase of the detectoi. The filter thicknesses were measured at the NatiOnal Synchrotron Light Source (NSLS) using an inout measurement over an energy range c 2 keV to 7 keV. AdditioiIally, a Cm source was placed in the aperture wheel to measure the ice build up on the Ge crystal that is inevitable with a cryogenically cooled detector. This source also served as a convenient way to check the gain o1 the detector system periodically during the HRMA calibration. This will be discussed further in a later section.

Figure. 1. Schematic of the HXDS Solid State detector

2.1 Signal Processing Elecfronics
The detector includes a pulsed optical resetting type preamplifier. This type o amplifier does not employ a feedback resistor which discharges the integrator portion of the pmp. This allows the charge to build up in the preamplifier. To reset the preamplifier to it's initial condition, an led is pulsed near the FEF chip and discharges it This reset rate was approximately one Hz. During this reset pulse, a veto signal is sent to the amplifier wbich disables it until thereset pulse has passed. Since this design elimirnites the need for a feedback resistor, the noise in the prnp is much lower that a typical RC feedback type preamp making it idea for low energy detection2. The pxeamp provides two outputs which allow the detector to be connected to.a shaping amplifier that is being run in a differential mode. This was necessary to limit noise pick up since the shaping amplifiers were located more than 50feet from the detector. An Ortec 671 shaping amplifier was used with an SAO built addon that allowed the settings of the shaping amplifier to be set remotely via computer controL The unipolar output was connected to an Ortec 921 MultiChsrnnel Buffer that was also remotely computer controlled. The gain of the'detector was set to give 5 eWchannel over 4096 channels. The shaping amplifier allowed shaping times selectable from 0.5 microseconds to 10 microneconds However, during the calibration only two shaping time constants, 2 ms and lOins, were utilized. The 2 ms shaping time was used when the counting rate was high (larger than 1500 Hz) while the 10 ms was used at lower counting rates.
-The bias supply for the detector was a CAEN N470 quad blab voltage power supply and was set at 485 volts. This unit was also controllable by computer commands. The bias supply included an input from the detector preamp that signalled a voltage shutdown when the detector started to warm up.

2.2 Deadtime Corrections
Deadtime corrections were planned to be made by using a pulser (Berkeley Nucleonics model BH-1) as an input signal that would suffer the same losses an x.ray event would suffer to first order. Since the pulser used was a repeating pulser, it was

536

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 01/02/2014 Terms of Use: http://spiedl.org/terms


unable to interfere with itselL This made it necessary to introduce second order statistical corrections to the data. This method was tested by using a radioactive source as a proof c principle and also at BESSY. The results ci the pulser deadtime corrections were compared with the bulit in Gedke-Hal& deadtime correction circuitry that existed m the multichannel analyzer.
The pulser amplitude was set to channel that was high enough such that the recorded pulser peak could not be influenced by any recorded xray peaks. A digital counter was used to count the number of pulses that weie injected into the preamp. A gate on the counter was connected to the MCA so that the counter would only count when the MCA was actively collecting. The radioactive source to detector distance was varied on each successive collection. The counts weie summed, corrected for deadtime, and an activity o the source as calculated for each run. The results showed that the pulser method was capable of predicting the radioactive activity very accurately, whereas the Gedke-Hale system piediction varied by sometimes more than . 8%.
.

A problem was discovered with the pulser method. When the pulsÈr amplitude was set high, and the pulser rate was also
relatively high (lOOHz+), the pulser would cause a severe undershoot of the preamp. This would cause any xray event arriving after the pulser event to be recorded lower in energy that it should be. The resulting spectrum showed peak widths much broader than the detector was capable.
.

The problem could be remedied in one of the two ways. First, reduce the pulser amplitude to a level that would cause a minimal undershoot This worked quite well, however the pulser was in a positicn that made spectroscopy at lower energies

difficult. The second solution was to decrease the pulser frequency. This minimized the number oil influences the pulser would have on the input x-ray rate.

Figure 2. Schematic repiesentation of the PHA chain with the deadtime correction counter

.

3. Absolute Calibration

The detector was calibrated using synchrotron radiation under a program developed by the Pliysikalishe-Techniche Bundesalant (PTB) at the electron storage ring BESSY4. The response function of the detector was measured at BESSY using dispersed synchrotron radiation from a monochroinator system. In the energy range from 1.1 keV to 5.9keV a double crystal monochromator was used. The upper energy was limited by mirror optics in the beapiline. The lower energy was limited by the Bragg angle of the crystals used in the monochromator. In order to get as pure a single line as possible, the second crystal of the monochromator was slightly detuned from the first crystal. This suppress much of the harmonic contamirniticm. · For energies from 600 eV to 1.1 keV, the spectral response function was measured at a grating rnonochromator beamline. Stray light and second order contaminiition effects from the monochromator were suppressed by the use of filtering. This beamline consists of an SX-700 grating monochromator with calibrated photodiodes for measuring the beam intensity downstream from the monochromator. This made it possible to simultaneously measure the detector efficiency as a function of energy. Since the monochromator was capable of reach 1.7 keV, response function data and detector efficiency data were collect upto this energy point.

The response function of the detector consists mainly of a peak and a lowÞr energy shelL The peak is obviously derived from a photon depositing all of it energy in the Ge crystal and being collect without loss. The shelf arises from the photon creating

537
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 01/02/2014 Terms of Use: http://spiedl.org/terms


photoelectrons at the surface of the crystal. These electrons must Iraverse the dead layer until they are collectas a charge. Since the mean free path for the electrons is very short in the crystal, not all of the electrons aie collected. This incomplete-charge collection results in a recorded event in an energy bin lower than the characteßstic photon energy. The absolute detector efficiency was measured at BESSY using undispersed synchrotron radiation. The absolute flux from the storage ring was calculated from first principles using the SÕhngerequation. Precise measurement of the magnetic field, ring current, source distance and aperture size allow this calculation to be made.

The storage ring is injected with a few milliamps of current and the orbit is allowed to stabilize for a couple cfhours. Once the orbit is stabilize, the source size is measured. This is done by scnnning a polarizing crystal through the x-ray beam. Once the source size is measured, the energy of the electron beam is determined by injecting an RF signal perpendicular to the electron orbit. The RF frequency is adjusted until it is resonant with the electrons in the storage ring.
After these parameters are found, the ring current is reduced by partially placing a baffle in the storage ring. This blocking plate removes some of the electrons from the orbit. Once the ring current is reduced to a few electrons, data may be taken. As the current is stepped down, it is possible to observe individual electrons being removed once the current gets down to the 1O electron mark.
.

4.

DISCUSSION OF CALIBRATION RESULTS

4.1 Response Function
A number of interesting features became apparent during the measurement of the response function at the double crystal monochromator bearnline. As the energy of the monochromator was scanned through the Ge L ifi edge, the number of counts that appeared in the shelf increased. This is shown in fig. (FIG) Since the crystal is more "absorbing" around the GeLIII edge, more the photons suffer from an incomplete charge collection as was described above. This effect was not observable in the other Ge-L edges. However, similar effects were observed when the monochromator was scanned across the Al-K edge, due to the front surface contact in the Ge crystal, and across the Ge-K edge. (see FIg. 3.)

Response Finction Above aid Below the Get Edge
1e*5

_____
B&OWG-L AboveGe-L

1e44

1S42
1.+1

ls+0

01600100800
ChalEelNumber

Figure 3. Pulse-height spectrum showing the change in the shelf as the monochromator is scanned across the GeL ifi edge.
Some other features that are observable in the response function are the presence of an Al fluorescence peak in the shelf of the spectrum when the input photon energy was larger than 1.5 key. Again, this arises from the Al infrared shield and the Al coating on the front of the crystal. Also when the photon energy is larger than the Ge-L edge, a Ge-L escape peak is present in the shelf. Both of these features are shown in fig. %%%. These two features can help set the gain of the detector

538
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 01/02/2014 Terms of Use: http://spiedl.org/terms


independent c the Cm source that was housedin the aperture wheel o the detector.
Repoie Function at 4.3 keV

10000

1000: ·1

300

600

900
ban4.er

1200

1500

Figure 4. Pulse-height spectrum showing the Al flourescence peak and the Ge-L escape peak
In spectra collected using a monochromator with an input photon energy less than the Al absorption edge, the Al fluorescence peak has been instrumental in determining the origin cif counts that are recorded at higher in energies. Originally the counts were thought to be due primarily to pile-up. However with the presence althe Al fluorescence peak, it is clear that some o the counts are actual higher energy photons (perhaps second order). The response function of the detector was relatively good. Typically the resolution was less than 170 eV at 5.9 keV using a 10 tsec shaping time constant The reason the detector did not perform better than this is because o a moderate leakage cunent that intrOdUCed noise in thepreamp and broadened the resolution of the detector. Also, the local noise environment at the synchrotron was relatively large.

4.2 Detector Efficiency

.

The detector efficiency was measured using both a calibrated photodiode as a transfer standard, and by exposing the detector to undispersed synchrotron radiation in which the flux was calculated from first principles. The calibrated photodiode was used with an SX-700 grating monochromator beaniline so that the detector efficiency could be measured as function of input energy. Detailed measurements of the fine structure of the Al infrared shield and front suiface coating could be made using this beamline. The raw data summed over the x-ray peak and shelf are shown on fig . The data were also corrected to first order for deadtime using the pulser method.
Detector E ificiency vi. Energy (Raw D at. ROl Suns)

a Is, 0
0.1

400

100

$00

1000 1200 1400 1900 1100 Eas,y (cv)

Figure 5. Detector Efficiency vs. Energy from SX700 data. The calibration data taken at the SX-700 beamline proved useful in determining the uniformity over the surface of the crystal.

539

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 01/02/2014 Terms of Use: http://spiedl.org/terms


Data was taken with the monochromator set on the three energies (Above Al-K, below AlK and at 600 eV). The data were nornialized to the photon flux measured by a calibrated PhOtOdiOde. The monochromaticbeam was collimated by an aperture to a 1 mm spot that was scanned over the surfaced the crystal in atwo dimensional raster scan. The results showed a 3%

variation in the efficiency cI the crystal. All three scans showed the same results. Since two of the scans were done above and below the Alike edge, this non.uniformity in the efficiency does not seem to be do to the JR filter and Al coating on the crystal.
.

The absolute flux from the undispersed synchrotron radiation was calculated from first principles using the Schwinger Equation. This requires a knowledge of the certain experimental parameters such at the beam current, electron energy, path length to the detector, inclination c the deteCtOr with respect to the electron orbit, and the strength of the magnetic field in the bending magnet. The uncertainty of measuring these parameters at BESSY add iii qjaclrature to be less than 1%.

The electron ring at BESSY was operated with a electron energy of 850 MeV. This gives a critical energy of &&&&. This means that most of the photons were primarily in the lower energy bins. As a results, the detector spends most of its time counting these low energy photons. In order to get reasonable statistics at the higher energies, the Al filters in the aperture wheel were use to suppress the counting of the lower energy signal. The use of these filters effectively "shifted" the energy of the input synchrotron radiation and allowed a calibration of the detector at higher energies (10 keV).
Undiaperaed Synchrotron Radiation U.ing a 5mm Open Aperture

a
U

2000 Chsnn1 Numiwir

Figure 6. Undispersed synchrotron radiation using a 5 mm open aperture.
Undiiper.ed Synchrotron Radiation Spectrum using a 25 mm Al Filter
lo+5

1·+4

0

C

:
le+2

0
le+1

1.+0
(,hannsl Numbor

Figure 7. Undispersed synchrotron radiation using a 25 mm Al filter. The filter "shifts" the photon spectrum to a higher energy.

540

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 01/02/2014 Terms of Use: http://spiedl.org/terms


Undiapersed Synchrotron Radiation with a 125 p.m Al Filter
1.+5
1e+4
a
C
C

0
a
C

a le+3
1e+2

3 0 0

1.+1 1.+0
2000 Channel Number

Figure 8. Undispersed synchotron radiation spectrum using a 125 mm Al filter. This filter allows a calibration up to 10 keV. The detector efficiency was measured in such a way that any filters used during data collection were considered to be part of the detector instead of calibrating each individual couiponenL However, the ice build up on the crystal was a variable than needed to be measured. A radioactive Cm source with and Fe taraet was placed in the aperture wheel to help measure this ice thickness. When the alpha particles form the Cm source impinged on the Fe target, itfluoresced both the Fe-K and Fe-L lines. The ratio of the Fe-L strength to Fe-K strength measured over time was used to calculate the ice thickness (see fig. ***). Another advantage of the Cm source was the ability to measure the gain of the detector at any time. The source is rich in lines resulting not only from the Fe fluorescence, but also from the emission of Pu L lines that are a result ofthe

daughter products of the Cm decay.

244cin Source Iilled in D Aperture Wheel
1000

100

a C
0 0

S

10

0

1 2000

n& .'.r

3000

4000

Figure 9. Cm spectrum used to determine the ice layer
thickness. This is allow useful for monitoring the gain.

541

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 01/02/2014 Terms of Use: http://spiedl.org/terms


SSD GainLinearity

.

I
0
1000

2000
channel Number

3000

4000

Figure 10. Gain vs. Energy. This data was obtained from
the 244Cm source.

5. USES OF THE SSD DURING THE HRMA GROUND CALIBRATION
As stated earlier, the good energy resolution of the SSD's allowed one exposure measurements during calibration of the HRMA. One such measurement is a molecular contiimintion measure. During this measurement, an electron impact source with a C target was run with a potential of 15 kV. This produced a relatively clean contimium spectrum that illuminiited the mirror surface and flouresced any contamination that might have been on th surface. One such spectrum from the molecular continmination studies is shown in Fi& 11. Simlar sources setting were used to measure the effective area of the HRMA.

Molecular Contamination Meuur.m.nt Usbig the SSD

1:

am

ma

OnndNumbcr

Figure 11. SSD Spectrum from Molecular Contamination studies during HRMA calibration.

542

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 01/02/2014 Terms of Use: http://spiedl.org/terms


ACKNOWLEDGEMENTS
We wish to thank all of the HXDS Mission Support Team. Special thanks go to Tim Norton, Roger Eng, Mark Ordway, John Moran, Nicholas Mistry, and Stephen Keleti for their design. and implemintation of the hardware and software systems of the HXDS. We also wish to thank the High Energy Astrophysics Division of the Smithsonian Astrophysical Observatory. This work was performed under the auspices of the National Aeronautics and Space Administration by the AXAF Mission Support Team under Contract No. NAS8 40224.

REFERENCES
1Canjffa Industries, "Germanium Detectors", Sec. 1.5, 1992 2G. F. Knoll, "Radiation Detection and Measurement",(New york: John Wiley and Sons), p. 592-4, 1989 3R. Jenkins, R.W. Gould, and D. Gedcke, Quantitative X-Ray Spectrometry (New York: Marcel Dekker, Inc.), 1981, pp266-267 4F. Shoize, G. U]m, Nucl. Instr and Math. A 339 p.49

543
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 01/02/2014 Terms of Use: http://spiedl.org/terms