UCRL-JC-104994
PREPRINT
l _ _'.s E'of#
'"'";'i't
, , , , ..
OCT0 4 1990
X-RAY MICROCALORIMETERS WITH
GERMANIUM RESISTANCE THERMOMETERS
S. Labor, E. Silver, T. Pfafman and Y. Wai
,
Lawrence Livermore National Laboratory
J. Beeman, F..Goulding, D. Landis,
N. Madden, and E. Hailer
Lawrence Berkeley Laboratory
SPIE's 1990 International Symposium on
Optical and Optoelectronic Applied Science and Engineering
San Diego, CA, 7-13 July i990
. August 13, 1990
.o,
ili_H!iiii_ii_iiiiiiiiiiiii::':iii!iii_ii!ii_ili!iiiii!ii!ili_iiiii!_ii!iii!!iii!iimi_iii!!!iiiimiiiiiiii!iii!iiii':ii':i_:ii_iiii::iiiii.";i_iiii::iiiiiiiiiiiiT::i_:i!iiiiii::!ii!iiiiiiiil::!il..... _" ',Xv //
iI!!INigi Niii#!ii!ili@#!iiiiiNiilNiiiiNiiiT=!!NiiigiN!Niiii_:Ni!iiiiN!iiiiiiiiiilN::IiiNNjiN i ii :, ii!iiiiNilgiii!iii#iiiNiiii!Niii!ii!iNTiiiiiiiiiiiiiii!!iiii'=ii{
...."_iiiiiiiiiiiiiiiiiiiii_,:. x_ _'_ .:,::!i
1
This is a preprint of a paper intended for publication in a journal or proceedings, Since
changes may be made before publication, this preprint is made available with the
understanding that it will not be cited or reproduced without the permission of the
author.
#
;.;.:.:;.:,:.:.::...-:...-.;.:,• :.:..:
_P
r
DI','_,CI, _.I M _ R
This doctlmenl _'as prepared as an account of _'ork Sl_)nsored h)' an agency of lhc
United StalL,s Government. Neither the li=died Stales Government nor the [_niver._lt)'
of California nor an)' of their employees, makes an)' _arran¢)', express or implied, or
as._umes ant legal liability ,Drresponsibility for the accuracy, _'ompletenes_, or u._eful-
ne._sof a;l_ information, apparalu._, product, or process disclosed, or repre_enl._ thai
its u_c _ould not infringe prlvatelt o_m,d rights. Reference herein to ant' specific
commercial products, prcK'es_,or serHce bf' trade nam(,, trademark, manufacturer, or
_!ther_is¢,, d,_esnot n¢.c('_sarilt' con._tilute or Implt' it_ endorsement, recommend_ztlon,
or I'avorin_ hy the LLlntlcd States (;ovcrnmenl or the Llntver_itt' of ('altl'ornia. The
He_'s' and oplnions of author._ expressed herein do not neces._arily state or reflect
those of the United States Government or the Universttt' ¢)f ('allfornla, and shall nut
be used for ad_'ertisinl_ or product endorsement purpo._es.
¢
X-ray microcalorimeters with germanium resistance thermometers
S. Labov, E. Silver, T, Pfafman and Y. Wai
Laboratory for Experimental Astrophysics
' Lawrence Livermore National Laboratory
' U CRL- Jt-- 104994
J. Beeman, F. Goulding, D. Landis, N. Madden, and E. Hailer
; Lawrence Berkeley Laboratory D E9 1 00 0 6 5 8
ABSTRACT
We report on the current status of our work on x-ray microcalorimeters for Use as high resolution x-ray spectrometers. To
maximize the x-ray collecting area and the signal to noise ratio, the total heat capacity of the device must be minimized. This
is best achieved if the calorimeter is divided into two components, athermal sensor and an x-ray absorber. The thermal sensor
is a neutron transmutation doped (NI'D) germanium resistor made as small as possible to minimize the heat capacity of the
calorimeter. The thermistor can be attached to a thin x-ray absorber with large area and low heat capacity fabricated from
superconducting materials such as niobium. We discuss results from our most recent studies of such superconducting
absorbers and present the x-ray spectra obtained with these composite microcalodmeters at a temperature of 0.1 K. An energy
resolution of 19 eV FWHM has been measured.
I. INTRODUCTION
We are developing microcalorimeters for high resolution x-ray spectroscopy. When an x-ray photon is absorbed
in one of these detectors, a rise in temperature results which can be measured with great accuracy. In principle, an x-ray
microcalorimeter can detect x-ray photons between 0.1 and 10 keV with nearly 100% efficiency and resolution of a few eV. 1
'Fhis high resolution and high efficiency over a broad range of energies is not available in other x-ray spectrometers. The
potential applications of these detectors include atomic physics, plasma physics and x-ray astrophysics.
The x-ray induced temperature pulse of a calorimeter is measured by a thermal sensor. While capacitive2, 3 and
inductive4,5,6 thermal sensors are under development, most high resolution measurements to date have been obtained with
resistive sensors. Detector arrays with ion-implanted silicon thermistors have been produced at Goddard Space Flight Center
which yield f': eV FWHM resolution 7 when operated at 0.065 K. We report here on developments utilizing neutl'on
transmutatio, doped (NTD) germanium thermistors. Although NTD germanium has a larger specific heat than ion-implanted
silicon, the loping in NTD germanium is more reproducible. Once the material for a given neutron dose has been
character _ d, one can routinely fabricate devices with the desired resistance at a chosen operating temperature.
This paper discusses the operation of these NTD germanium thermistors and their response to x-rays absorbed in
the germanium, in the gold electrical contacts, and in a superconducting film deposited directly on the sample. The paper also
describes our plzms for improving the re_lution and increasing the area and efficiency of such detectors.
2. NTD GERMANIUM CALORIMETER PERFORMANCE
2.1 Calorimeter geometry
, The results presented in this section were obtained by direct irradiation of an NTD germanium calorimeter 100
_tm x 100 _n x 250 gm long, operating at a temperature of 0.1 K. The two ends of the sample have been implanted with
boron ions and coated with 200 nm of gold for electrical contact. Graphite fiberswhich are 7 pna in diameter and have been
, coated with 4 I-truof indium, are silver epoxied to the samples. These indium/graphite wires are the electrice_ bads and also
serve as the mechanical support for tt_edevice. The length of each wire is typically -150 _xrn. The detector is biased at a
voltage which is held constant tor the duration of the x-ray pulse. The absorbed x-ray heats the thermistor and reduces its
resistance. The resulting current pulse is amplified by a JFET negative feedback amplifier.8, 9
2.2 Non.ohmic behavior
A thermistor is used to convert the x-ray induced temperature change into a current pulse. The _'csistance of the ,
thermistor, however, doesnot depend only on the temperature. As the voltage across a semiconductor resistor at cryogenic
temperatures is increased, the resistance falls even if the temperature of the device is fixed. We have observed this nonlinear '
relation between voltage and current in NTD germanium at temperatures between 0.08 K and 1.2 K.9 It has also been
observed by others in ion-implanted silicon ./as well as NTD germanium between 4 K and 1 KtO and at 0.02 K.11
This nonlinear behavior directly effects the performance of the calorimeter. As the bias voltage is increased, not
only does the resistance fall, but the gradient of the resistance as a function of temperature (dR/dT) flattens severely. The
detector's signal to Johnson noise ratio is therefore reduced, and the resulting energy resolution is much worse than the
predictions based on the low power measurements of resistance as a function of temperature.
The nonlinear behavior also makes it difficult to characterize the basic parameters of the system. Under normal
operating conditions, the calorimeter is weakly coupled to the cold sink. In this case, the resistance will fall as the bias
voltage is increased due to a combination of the voltage effect and ohmic heating of the device. The usual technique for
calculating the thermal conductance between the calorimeter and the cold sink, is to apply power to the device and measta'e the
resulting change in resistance. One then assumes that ali the change in resistance is due to ohmic heating, and then calculates
the change in temperature using the low power resistance-temperature relation of the device. At typical operating powers and
applied voltages, however, the electrical properties of the device do not follow the low power behavior, and not ali oi"the
resistance change is due to ohmic heating. Since the temperature of the device is uncertain, special test device,;, are required to
accurately measure dR/dT at the operating power. Preferably two thermistors thermally connected to each other should be,
used, one monitoring the temperature of the other. Since the non-ohmic behavior varies with the volume and geometry of the
resistor, numerous test devices are required to fully characterize the nonlinear effect. We have not yet undertaken such an
effort, but it will be necessary to do so in order to properly optimize the thermistor and achieve the best possible signal to
noise ratio.
To obtain an estimate of theresistance,temperature slope (dRJdT) encountered at operating powers, we measure
dR/dT of thedevice with a moderate amount of power, in between the low power limit and the higher powers required for
calorimeter operation. In this measurement, we apply enough bias voltage to change the resistance, but not enough to
significandy heat the device. Rough estimates of the thermal conductance are adequate to determine that the device is not
being heated with these moderate power levels. The resistance is measured in this way at several cold sink temperatures to
determine dR/dT. The resulting dR/dT is much closer to the actual dR/dT encountered under operating conditions than the
OR/dT found with measurements in the low power limit. With this improved estimate of dR/dT, we cart calculate the thermal
conductance, and using the decay time constant of the pulse, calculate the heat capacity.9
2.3 Estimation of operating parameters
The detector described above was cooled in an adiabatic demagnetization refrigerator which has a t,:mperature
stability of 2 _tK rms 9,12 Operating at a base temperature of 0.1 K, we measure a decay time constant of 0.7 msec, a
thermal conductm_ce of 3 x 10-1o W/K and a heat capacity 2 x 10 d3 J/K. The fractional change in resistance per degree K
with 3.5 mV across the 2 Ml2 thermistor is 1/R dR/dT = 28 K-1.
From these numbers, 5.89 keV of energy absorbed in the detector should produce a voltage signal of 0.18 V at
the output of the preamplifier. The peak voltage measured from the gold absorption of a 5.89 keV x-ray is 0.175 V, in close
agreement with the prediction. This conf'trms the approximate values of the heat capacity and other parameters given above,
but the analysis is not sufficient to accurately preztict the absolute pulse height. A much more thorough analysis of the
,¢
voltage-resistance-temperature, characteristics of the thermistor would be necess,'u'y to determine the exact amount of energy
required to produce a given pulse height.
I,
2.4 Pulse height analysis
The 0.1 K detector was irradiated with an 55Fe source producing Mn Ktx (5.89 keV) and K[3 (6.49 keV) x-rays,
and the resulting spectrum is shown in Figure 1. An electronic pulser was used to simultaneously measure the electronic
noise, and the resulting line profile appears in Figure 1. As described previously,9 these NTD germanium detectors produce
two different pulse heights for each x-ray energy. The Kct, KI_ pair with lower pulse heights (and large number of counts)
arise from x-rays absorbed in the germanium, When x-rays are absorbed in the gold contacts, a larger pulse height results.
For x-rays absorbed in the germanium, some of the energy does not appear as thermal energy within the el,_ctronic processing
' time of about 1 msec.
I
........ ,....... ,
1 2500 ' ' " l ' ' ' I _ , ' I ' ' ' I ' ' '
- Ko_ _
- electronicpu[ser -
2000 --
- 12eV -
m _
_-
1500 -- Ge absorption _
o_ 57eV
¢j ,-
_
1000
500 _ _ Au absorption _
__2
t li --_1_--- P"
,XJ,L_A .... ....
0 400 800 1200 1600 2000
pulse height channel
Figure 1. An x-ray spectrum obtained with an bfFD germanium calorimeter 100 _tm x 100 I.tm x 250 t.tm long operating
at 0.1 K. The detector was illuminated with Mn Kc_ and KI3x-rays giving rise to the four peaks in the center of
the figure. The peak on the far right is from an electronic pulser which is used to measure the electronic noise.
The 25 eV FWf2.1 resolution indicated by the Au absorption peak includes contributions from the Mn Ko_1and
KR:z lines. Accounting for these two lines within the peak shown yields an actual energy resolution of 19 eV.
In solid state ionization detectors where x,rays are absorbed in a semiconductor, some of the energy goes into
creating electron-hole pairs. At 77 K, 2.96 eV of x-ray energy is required, on average, to produce one electron-hole pair in
germanium. Since the band gap of germanium is 0.67 eV, only 22.6% of;the x-ray energy goes into electron-hole pairs, The
remaining 77.4% of the x-ray energy is converted to heat during the cascade of elecu'on energies which is initiated by the
photoelectric absorption. At 0.3 K, the ratio of the germanium absorption pulse height, to the gold absorption pulse height
is 0.778. 9 This strongly suggests that the electron-hole pairs in the germanium are not recombining within the duration of
the signal pulse processing time, whereas, when x-rays are absorbed in the gold the energy rapidly appears as thermal energy.
At 0.3 K, we have found that the fraction of the x-ray energy appearing as a thermal signal in the germanium is always 77%,
, regardless of detector voltage, and pulse shape filtering. This same fraction was observed at 0.3 K with two different detector
sizes and germanium doped at two different levels.
In contrast, at 0.1 K the fraction of the x-ray energy which appears as a thermal signal in the germanium varies
, from 79% to 85% depending on detector voltage and pulse shape filtering time ,scale. Possible explanations for this varmtion
include partial electron-hole recombination, a change in the band gap or in the average energy required to create an electron
hole pair. The lower operating temperature and/or higher germanium doping concenmltion could be responsible for some of