t Lawrence Livermore National Laboratory PFC/JA Response of X-UV Photodiodes to kev X Rays and MeV Alpha Particles

Transcription

1 PFC/JA Response of X-UV Photodiodes to kev X Rays and MeV Alpha Particles K. W. Wenzel, C. K. Li, R. D. Petrasso, D. H. Lo, M. W. Bautz*, G. R. Picker, Jr.*, and E. Hsieht October 1992 To be published in: Review of Scientific Instruments MIT Plasma Fusion Center * MIT Center for Space Research Cambridge, MA 2139 t Lawrence Livermore National Laboratory Livermore, CA 9455 This work was supported in part by NASA Contract No. NASW-4372, U.S. DOE Grant No. DE-FG2-91ER5419, and LLNL Subcontract No. B

2 Response of X-UV Photodiodes to kev X Rays and MeV Alpha Particles Kevin W. Wenzel, Chi-Kang Li, Richard D. Petrasso, and Daniel H. Lo MIT Plasma Fusion Center, Cambridge, MA 2139 Marshall W. Bautz and George R. Ricker, Jr. MIT Center for Space Research, Cambridge, MA 2139 Ed Hsieh Lawrence Livermore National Laboratory, Livermore, CA 9455 Abstract The absolute x-ray response of three X-UV photodiodes was measured over an energy range of 1.5 to 17.5 kev so that they could be used to calibrate x-ray imaging systems for the Astro-D satellite mission. An intense electronbeam x-ray generator was used to test both the DC and AC x-ray response at 1.5, 4.5, 8., and 17.5 kev, and an 55 Fe source was used to examine one of the photodiodes at 5.9 kev. The x-ray response was determined by comparing the X-UV diode signal to that of a previously calibrated silicon surface barrier diode (SBD). The X-UV detector response was similar to the SBD response at low energies (1.5 and 4.5 kev). At 8 kev, the X-UV detectors exhibited about 7% of the SBD response, and at 17.5 kev, about 5%. This result is surprising, because the X-UV diodes actually have a greater silicon thickness than the SBD. In contrast to our findings for SBDs in the past, this implies that not the entire physical volume of these detectors comprises the active volume. The X-UV detector x-ray response was also examined as a function of the applied bias voltage. No significant bias voltage dependence of the x-ray signal was found, which indicates that the depletion layer thickness does not determine the active volume either. However, the detector noise was found to decrease substantially as the applied bias was raised from zero to a few volts. Response of these detectors, operated in pulse mode, to 22eRa a particles indicated large charge carrier recombination in the bulk silicon. This feature renders these detectors unsuitable for most chargedparticle spectroscopy applications. 1

3 Introduction Silicon photodiodes with stable, high quantum efficiency in the soft x- ray and ultraviolet wavelength regions have recently become available. 1 3 These detectors are referred to as X-UV photodiodes because of the spectral region over which they are sensitive. In addition to their high efficiency and stability, X-UV detectors have the added advantages of large dynamic range, low noise, low cost, little or no bias voltage requirements, and large sensitive areas. These advantages make X-UV detectors quite attractive for measuring radiation emitted by both space and laboratory plasmas. These detectors can be used as radiation intensity monitors if their absolute response is known as a function of wavelength. The absolute response to ultraviolet photons in the 1 ev to 15 ev region has previously been measured.- 3 The absolute x-ray response from a radioactive 55 Fe source emitting Mn K x rays at 5.9 and 6.5 kev has also been measured. 1 Most x-ray emission of importance to astrophysical and fusion plasmas lies in the.5 to 1 kev regions (especially from plasmas with electron temperatures of.5-2 kev). It was therefore the goal of this work to extend the region of absolute response measurements to x-ray energies from about 1 to 1 kev. In this paper, we describe measurements of the absolute x-ray response of three X-UV detectors at 1.5, 4.5, 5.9, 8., and 17.5 kev. These measurements were made primarily so that the X-UV diodes could be used as a standard to calibrate the x-ray imaging CCD for the Astro-D satellite mission. 4 The particular detectors examined in this work comprised two X-UV35 (.35 cm 2 area) and one X-UV1C (1. cm 2 area), all of which were manufactured by United Detector Technology 5 (see Table 1). In addition to the absolute response measurements, we also explored changes in the x-ray response and detector/preamp noise as the applied bias voltage was varied. Bias effects are important for at least three reasons. First, these X-UV detectors are claimed to have high quantum efficiency with zero applied bias. Thus it is important to determine whether the efficiency could be increased simply by applying a bias voltage. Second, we have previously seen distortions of large x-ray signals (~ 1 PA) from silicon surface barrier diodes (SBD) when their applied bias was reduced below about one volt. 7 Third, by observing the detector performance with different bias voltages, we can improve our overall understanding of these detectors. For example, it has long been believed that the depletion layer, which has 2

4 a thickness proportional to the square root of the bias voltage 8, comprises the x-ray sensitive portion of silicon detectors. 8 ' In contrast, we found for SBDs that the x-ray sensitive volume (as well as the charged-particle sensitive volume) is determined primarily by the physical silicon thickness, not the depletion layer thickness."" Thus it is of general interest to observe any bias dependence in the X-UV x-ray response. In this paper, we therefore describe the x-ray response of X-UV photodiodes as the bias voltage was varied from to -3 V. No dependence of the x-ray response was observed over this range of voltages. We conclude that the x-ray response is not determined primarily by the depletion layer. We did observe a decrease in the electronic noise as the bias was increased. For this reason, it may be desirable to operate these detectors with some low applied bias voltage (a few volts is the maximum suggested by the manufacturer). To further understand what constitutes the sensitive thicknesses of the X-UV detectors, their responses were compared to the SBD response to a particles from a 1Ra source. This isotope emits a particles with energies from 4.7 to 7.6 MeV, corresponding to ranges in silicon of 22 to 45 pm. Of greatest interest to us was to observe any degradation in the highest energy a line (7.6 MeV) which would allow us to estimate the active thickness of the X-UV detector. Although this effort was not successful, we were able to conclude that these detectors were unsuitable for most charged-particle spectroscopy applications due to larger charge recombination in the bulk silicon. However, they may be useful for the less demanding task of chargedparticle counting experiments.' This paper is organized as follows. Section I describes the electron-beam x-ray source used to characterize the X-UV detectors (and a radioactive 55 Fe source used to measure the relative X-UV response to Mn K line radiation at 5.9 and 6.5 kev). Section II describes DC x-ray measurements with zero applied bias voltage. Section III describes measurements of the detector responses to modulated x-rays using an AC-coupled current-to-voltage amplifier. The data analysis is discussed in Section IV, and the absolute response results are summarized. Section V describes the effect of varying the bias voltage on each detector. Section VI describes the a-particle response of the X-UV detectors, and Section VII contains a summary of all the results and a discussion of their implications. 3

5 I. X-Ray Sources The main x-ray source was an intense electron-beam x-ray generator (illustrated in Fig. 1).14,15 The electron beam was generated by a bare tungsten filament operated at ground potential. The beam was accelerated over a distance of less than 3 cm to an anode (target) maintained at positive high voltage. This configuration was found to be preferable to applying a negative high voltage to the cathode (and grounding the anode), because it eliminated spurious signals from electrons that were accelerated from the filament to the detector at ground potential. The anode potential could be varied between zero and +4 kv. The emission current could be varied from zero to 2 ma by changing the power input to the filament. The generator ran quite stably when the power deposited on the target was less than about 2 W. However, with higher power levels, the x-ray source would only run for a short time (typically less than two minutes) before a discharge breakdown occurred between the filament and the anode. Consequently, the DC data described below were acquired while running below the nominal 2 W threshold in order to maintain stable signal levels over several minutes. The AC data were taken over a range of operating conditions, mostly with higher emission current in order to maximize the signal level. It was possible to obtain the AC data at high power levels, because the data could be obtained in a short time period. The maximum power run was about 1 kw, during the molybdenum series, with 25 kv and 4 ma emission current. In addition to the electron-beam x-ray generator, we examined the response of one of the X-UV diodes (# 2 in Table 1) to Mn K x rays at 5.9 and 6.5 kev from a radioactive 55 Fe source. This source was well calibrated, with an activity of 2.8 mci. II. DC X-Ray Response Measurements The DC x-ray response measurements were performed by simultaneously measuring the X-UV photodiode current and the current from a standard silicon surface barrier diode (SBD) when both detectors viewed the same anode (both detectors were unbiased). For our comparison diode, we used a partially depleted SBD from EG&G ORTEC 16, serial number B, for which the x-ray response from.3 kev to 17.5 kev has been absolutely measured

6 We checked to make sure that the x-ray beam uniformly illuminated both detectors. This was done by rotating the detectors about an axis located onehalf way between them (they were separated by about 4 cm). Furthermore, with the exception of the 55 Fe case, the measurements were performed with the detectors located about 92 cm from the source anode (see Fig. 1). Both detectors were apertured with brass collimators of.32 cm diameter (except for the "Fe measurements, where a smaller aperture was used). This was done to avoid spurious signals due to x-ray absorption in the detector edge region, 7 and to make their active areas equal. The detectors were also covered, for each wavelength, with thin filters of the same composition as the anode to, first, increase the intensity of characteristic line radiation relative to the bremsstrahlung continuum, and second, to eliminate the visible light response from the x-ray source filament. We hereafter refer to this filter as the "Line Filter" (see Table 2). In addition, a separate, movable aluminum absorption filter of known thickness was used to test the spectral purity of the source by inserting the filter between the source and detectors and checking that the detector signals dropped by the predicted amount (see Fig. 1 and Table 2). Hereafter, we refer to this filter as the "Test Filter". The positive signal current from both detectors was measured with a Keithley 617 digital electrometer. 8 The background current, due to leakage current and slight visible light response, was -1 pa for the X-UV diode, and ~4 pa for the SBD. It was monitored and subtracted from the total signal current to obtain the net x-ray signal current. We found it was important to monitor the background signal over the duration of a single calibration experiment (i.e., one x-ray wavelength), because there was noticeable drift in the background leakage current. Because the leakage current increased with time, we believe this drift was due to heating of the detectors by the x-ray source filament. II.A. DC Response to 1.5 kev Al K X Rays The Al K x-ray response at 1.5 kev was measured while running the x-ray generator with an aluminum anode at potentials of 5 to 8 kv with emission currents of 6 to 14 ma. The Test Filter was 17.3 Pm Al foil (thickness measured with 55 Fe), which is predicted to transmit 15% of the 1.5 kev Al K, line (using the cross sections of Ref. 19). Both detectors also had 17 pm aluminum foils as the Line Filters for visible light and bremsstrahlung 5

7 rejection. The response currents without the aluminum Test Filter ranged from 14 pa to 44 pa for both the X-UV and SBD detectors. The response currents with the Test Filter in place ranged from 1 pa to 11 pa. 2 The Test Filter transmission varied from 17% for the 5 kv anode voltage to 3 % for the 8 kv anode voltage. The 17% result agrees well with the predicted 15% transmission, which strongly supports the conclusion that the detectors were both responding to essentially pure Al K radiation. On the other hand, the 8 kv result indicates that the spectrum may have been contaminated with some higher energy bremsstrahlung continuum. However, the same relative X-UV response was obtained with the anode at 5 kv and at 8 kv. The measured spectrum (shown in Fig. 2) also indicated quite clean line radiation at 1.5 kev. II.B. DC Response to 4.5 kev Ti K X Rays The Ti K x-ray response at 4.5 kev was measured while running the x-ray generator with a titanium anode at potentials of 7 to 1 kv with emission currents of 2 to 14 ma. The Test Filter was 17.3 pm Al foil, which is predicted to transmit 31% of the 4.5 kev Ti line (Ref. 19). Both detectors also had 25 pm titanium foils as the Line Filters for visible light and bremsstrahlung rejection. This Line Filter was predicted to transmit 3% of the Ti K" line. The response currents without the aluminum Test Filter ranged from 7 pa to 2 pa for both the X-UV and SBD detectors. The response currents with the Test Filter in place ranged from 25 pa to 7 pa. 2 The Test Filter transmission varied from 32% for the 7 kv anode voltage to 35 % for the 1 kv anode voltage. These results agree well with the predicted 31% transmission, which strongly supports the conclusion that the detectors were both responding to essentially pure Ti K radiation. The measured spectrum (shown in Fig. 3) also indicated quite clean line radiation at 4.5 kev. II.C. DC Response to 5.9 kev Mn K X Rays The Mn K x-ray response of X-UV #2 was measured using an 5 -Fe radioactive source (2.8 mci). Both detectors were collimated with.24 cm diameter brass apertures so they would view the same geometry. Furthermore, they were shielded with 17 pm aluminum foil to reject the room light. The x-ray response currents were Ix-uv=1 4 pa and ISBD= 1 8 pa, so the 6

8 relative response at this energy was IX-UV/ISBD= We also measured the spectrum with the Si(Li) spectrometer, which had an additional 12.5 pm copper foil to further reduce the x-ray signal (otherwise the geometry was very similar to the diode measurements). Fig. 4 shows the 55 Fe spectrum. To determine the uncertainty in the relative X-UV response, we estimated the uncertainty in the individual current measurements to be of the order 2 pa. Then adding the X-UV and SBD uncertainty in quadrature gives the total uncertainty to be about 18%. Then at 5.9 kev, Ix-Uv/ISBD=. 7 8±.14 (Table 1). Note that this response is lower than reported in Ref. 1. II.D. DC Response to 8. kev Cu K X Rays The Cu K x-ray response at 8. kev was measured while running the x-ray generator with a copper anode at potentials of 11 to 14 kv and with emission currents of 3 to 13 ma. The Test Filter was 86.4 pm Al foil, which is predicted to transmit 34% of the 8 kev Cu line (Ref. 19). Both detectors also had 5 pm copper foils as the Line Filters for visible light and bremsstrahlung rejection (thinner Line Filters were used for the AC measurements to obtain larger x-ray signals; compare Tables 2 and 3). This Line Filter was predicted to transmit 11% of the Cu K, line. The response currents without the aluminum Test Filter ranged from about 6 pa to about 3 pa for both the X-UV and SBD detectors. The response currents with the Test Filter in place ranged from 1 pa to 1 pa. 2 At this line energy, the X-UV diodes always exhibited lower signal currents than the SBD. The Test Filter transmission varied from 33% for the 11 kv anode voltage to 37% for the 14 kv anode voltage. These results agree well with the predicted 34% transmission, which strongly support the conclusion that the detectors were both responding to essentially pure Cu K radiation. The spectrum shown in Fig. 5 was taken with the thinner copper Line Filter used for the AC measurements (Table 3). The DC measurement used a thicker copper filter (Table 2), so the corresponding spectrum was cleaner than that shown in Fig. 5. II.E. DC Response to 17.5 kev Mo K X Rays The Mo K x-ray response at 17.5 kev was measured while running the x-ray generator with a molybdenum anode at potentials of >25 kv with emission currents of 8 to 11 ma. The Test Filter was 86.4 pm Al foil, which 7

9 is predicted to transmit 89% of the 17.5 kev Mo line (using the cross sections of Ref. 21). This Test Filter was not attenuating enough to adequately confirm the spectral purity of the Mo K line. Both detectors also had 124 pm molybdenum foils as the Line Filters for visible light and bremsstrahlung rejection. This Line Filter was predicted to transmit 9.7% of the Mo K, line. The response currents with and without the aluminum Test Filter ranged from about 1 pa to about 15 pa for the SBD and from 6 to 14 pa for the X-UV detectors. At this line energy, the X-UV diodes always exhibited lower signal currents than the SBD. The measured spectrum (shown in Fig. 6) indicated quite clean line radiation at 17.5 and 19.6 kev. III. AC X-Ray Measurements In order to test the effect of different bias voltages, we had initially planned to use the internal voltage source of the Keithley 617 digital electrometer, and measure the signal current directly with the electrometer. However, when the bias was applied, the leakage current increased dramatically (by at least an order of magnitude, depending on the specific applied bias). Moreover, this "background" current drifted over a time scale of minutes, and continued to do so over several hours. It was therefore impossible to establish a DC background current to subtract from the total x-ray current. We therefore used a simple AC-coupled transresistance current-to-voltage amplifier 22 (see Fig. 7). With this circuit, we examined the x-ray signal level and the noise level as a function of the applied bias voltage. This section describes the AC x-ray measurements. The x-ray signal was mechanically chopped by a motor-driven aluminum disk with eight rectangular apertures. This "chopper wheel" was located in the vacuum chamber, directly in front of the X-UV detector and the SBD. In this case, the Test Filter covered every other aperture in the chopper wheel, so alternate x-ray pulses were filtered. This provided a quick, in-situ determination of the x-ray spectral purity. Only those data that exhibited the expected line transmission through the Test Filter were used to determine the absolute x-ray response. Table 3 lists the Line and Test Filters used during these AC measurements. For these measurements, the detectors were moved to 71 cm from the anode, and their brass apertures were enlarged to about.52 cm diameter to maximize the signal levels. Fig. 8 shows representative signals from each anode material. 8

10 III.A. AC Response to 1.5 kev Al K X Rays The modulated Al K x-ray response at 1.5 kev was measured while running the x-ray generator with an aluminum anode at potentials of 5 to 8 kv with emission currents up to 4 ma. The Test Filter was 17.3 Pm Al foil, the same as used for the DC measurements. Both detectors also had 17 pm aluminum foils as the Line Filters for visible light and bremsstrahlung rejection. The response currents without the aluminum Test Filter ranged from 2, pa to 21, pa for both the X-UV and SBD detectors. The response currents with the Test Filter in place ranged from 1, pa to 7, pa. 2 The Test Filter transmission was 2% for both the 5 kv and the 8 kv anode voltage. These results agree reasonably well with the predicted 15% transmission, which supports the conclusion that the detectors were both responding to essentially pure Al K radiation. The measured spectrum (shown in Fig. 2) also indicated quite clean line radiation at 1.5 kev. III.B. AC Response to 4.5 kev Ti K X Rays The modulated Ti K x-ray response at 4.5 kev was measured while running the x-ray generator with a titanium anode at potentials of 8 to 16 kv with emission currents of 1 to 3 ma. The Test Filter was 17.3 pm Al foil, the same as for the DC measurement. Both detectors also had 25 Pm titanium foils as the Line Filters for visible light and bremsstrahlung rejection. The response currents without the aluminum Test Filter ranged from 5, pa to 53, pa for both the X-UV and SBD detectors. The response currents with the Test Filter in place ranged from 2, pa to 21, pa. 2 O The Test Filter transmission varied from 32% for the 8 kv anode voltage to 37% for the 16 kv anode voltage. These results agree reasonably well with the predicted 31% transmission, which supports the conclusion that the detectors were both responding to essentially pure Ti K radiation. The measured spectrum (shown in Fig. 3) also indicated quite clean line radiation at 4.5 kev. III.C. AC Response to 8. kev Cu K X Rays The modulated Cu K x-ray response at 8. kev was measured while running the x-ray generator with a copper anode at potentials of 15 to 2 kv with emission currents of 2 to 48 ma. The Test Filter was 12.5 pm Cu foil, 9

11 which was different than the Test Filter used for the DC measurements. This Test Filter was predicted to transmit 57% of the 8 kev Cu line (Ref. 19). Both detectors also had 25 pm copper foils as the Line Filters for visible light and bremsstrahlung rejection. This Line Filter was predicted to transmit 33% of the Cu K, line. The response currents without the copper Test Filter ranged from about 4,5 pa to about 5, pa for both the X-UV and SBD detectors. The response currents with the Test Filter in place ranged from 3, pa to about 28, pa. At this line energy, the X-UV diodes always exhibited lower signal currents than the SBD. The Test Filter transmission was 55% for both the 15 kv and the 2 kv anode voltage. These results agree well with the predicted 57% transmission, which strongly support the conclusion that the detectors were both responding to essentially pure Cu K radiation. The measured spectrum (shown in Fig. 5) also indicated quite clean line radiation at 8.5 and 9.5 kev. III.D. AC Response to 17.5 kev Mo K X Rays The modulated Mo K x-ray response at 17.5 kev was measured while running the x-ray generator with a molybdenum anode at potentials of Z2 5 kv with emission currents of 2 to 4 ma. The Test Filter was 62 Mm Mo foil, which is predicted to transmit 31% of the 17.5 kev Mo line (using the cross sections of Ref. 21). The response currents without the molybdenum Test Filter were in the range of 1,5 to 7,5 pa for the X-UV diode and 4, to 12, pa for the SBD. At this line energy, the X-UV diodes always exhibited lower signal currents than the SBD. The signal-to-noise ratio was too small to accurately determine the Mo Test Filter transmission, but it was about 3%, close to the predicted value. The measured spectrum (shown in Fig. 6) also indicated quite clean line radiation at 17.5 and 19. kev. IV. Data Analysis and Absolute Response Summary To determine the X-UV absolute response, we simultaneously measured the X-UV and SBD photocurrents as described above. The net x-ray response of each diode was determined by subtracting the "background" current (due to leakage current and visible light) from the total current measured when the x-ray generator was on; I = IT - IB. (1) 1

12 The uncertainty in this net current was obtained by adding in quadrature the uncertainty in IT and IB. The relative X-UV response, px, was then calculated for each measurement; PX = Op, (2) Is where Ix is the net X-UV signal current and Is is the net SBD signal current. The uncertainty in this relative response, up was obtained by propagating the uncertainties in Ix and Is, assuming they were uncorrelated 23, so that IX Xi O (s3) S Is (23 s The most likely value of the relative response was then calculated by taking a weighted average 23 of all the individual relative response values. This average is given by A = Eipx,i/Op,i Px = ~j/, (4) Eil/u,, which weights most heavily those measurements with the smallest individual uncertainty. The uncertainty in the relative response was then assumed to be the standard deviation of all the individual measurements, since the systematic variation from one measurement to another tended to be larger than the uncertainty in any single measurement of px. Finally, the absolute X-UV response was calculated from the response relative to the SBD and the previously measured SBD absolute response, Rs, taken from Refs. 17 and 7; where Rx = Pxs k ax, (5) = Sx 2 r2 ) 1/2 (6) The results from the DC measurements are sumarized in Table 4. X-UV detectors #1 and #2 have nearly the same response at 1.5 and 4.5 kev. Their responses are also very close to the "ideal" response of 4.5 x 17 C/keV (this corresponds to 3.6 ev per charge pair). The response of both diodes decreases with energy faster than the SBD response. X-UV #2 exhibited a faster roll off with energy than X-UV #1. 11

13 The results from the AC measurements are summarized in Table 5. All three diodes were again found to have similar responses at 1.5 and 4.5 kev. However, the AC measurements yielded consistently lower values (up to about 15%) for the X-UV responses than the DC measurements. On the other hand, the uncertainties in the AC and DC results overlap. At 8. kev, the AC results were nearly identical to the DC results. X-UV #2 was again found to have a faster roll off at high energy than both X-UV #1 and X-UV #3 (for which DC measurements were not performed). The measurements at 17.5 kev were consistent with these observations; X-UV #2 had the lowest response of the three detectors, and X-UV #1 and X-UV #3 were very similar to each other. The DC and AC measurements are summarized in Fig. 9. The data are plotted as a function of photon energy. The solid line in each graph represents the calculated response assuming the entire physical silicon thickness forms the active thickness (347 pm for the SBD and 375 pm for the X-UV diodes 2 ). It is clear in these data that the X-UV responses drop with increasing photon energy faster than the physical-thickness model would predict. V. Bias Voltage Effects We examined the effect of varying the bias voltage for three main reasons. First, one of the advertised advantages of silicon X-UV detectors is their ability to run without an applied bias. Second, we have previously seen signal distortions in SBDs when the bias was lowered below about 1 V and the signal levels were quite high (of order 1 pa). 7 Third, changes in the signal level with bias can indicate the influence of the depletion region on the detector response. It was previously believed that the depletion region comprised the active region of silicon detectors. 8 - However, we found in the past that the physical thickness of silicon actually determines the x-ray sensitive region. 7 It was therefore of great interest to determine if varying the bias voltage had any effect on the x-ray signal level. A representative bias scan is shown in Fig. 1. This figure illustrates the X-UV #2 and SBD responses to Cu K x rays as the bias voltage was varied from to -3 V. The net x-ray signal did not change significantly as the reverse bias voltage was changed. This fact supports our earlier contention that it is not the depletion layer thickness that determines the x-ray response of silicon diodes. However, the lack of any bias dependence is somewhat 12

14 surprising, because we have observed, in the past, distortions in SBD x-ray response when the bias was reduced below about 1 V. 7 These distortions were most severe when the x-ray signal was in the 1 to 1 pa range, and they occurred at lower bias voltage as the signal level decreased. We therefore conclude that the bias-dependent distortions were caused by a non-linear, signal-level dependent effect in the detector at high signal levels. Thus, these effects do not hinder unbiased operation at signal levels up to at least tens of na. Although it shows no effect of the bias voltage on the signal level, Fig. 1 does illustrate a different significant effect of the bias voltage. That is that the noise in the system was substantially decreased as the bias voltage was increased. Specifically, for the X-UV data shown in Fig. 1, the noise (peak-to-peak) decreased from about 6.4 na at bias down to 3.2 na at -3 V bias. Similarly, the SBD noise also decreased from about 2.4 na down to 1.6 na as the bias was increased from to -3 V. We always observed larger noise on the X-UV than on the SBD detector. This change in noise characteristics with bias voltage is probably a manifestation of the decreased diode capacitance as the bias voltage is raised. This effect points out that, while zero bias operation is feasible, it may be desirable to operate with some low level of bias to reduce the overall system noise. VI. a-particle Response Measurements We had hoped to determine the sensitive thickness of the X-UV photodiodes by examining their response to a particles, which have a short range in silicon. Specifically, we examined the a-particle response of the X-UV diodes by exposing them to a.13 pci 226la source in air and in vacuum. This source emits four main groups of a particles with energies of 4.7, 5.4, 5.9 and 7.6 MeV. These particles have ranges in silicon of about 22, 26, 3, and 44 pm respectively." The purpose of this experiment was to estimate the "active" thickness of the X-UV detector by observing degraded response to the higher energy a particles. These detectors ostensibly consist of ~375 pm thick silicon 2 5, sufficient to stop all of the a particles. For comparison, we also measured the same a particles with an unbiased SBD, #16-675B (the same detector used for x rays). The SBD is 347 pm thick, also sufficient to completely stop all the a particles. The a-particle spectrum obtained in vacuum with the SBD operated with a bias voltage of -5 V is shown in Fig. 11. This bias voltage is enough to 13

15 create a depletion layer of -1 pm in this detector, approximately twice the range of the most energetic a particle. Charge pairs created by the interaction of the a particle with the silicon are therefore swept to the detector electrodes, rather than diffusing to the electrodes. Since the charge collection time in a depleted region ( ; 1-8 s) is much shorter than a typical minority carrier lifetime (.1 to 1 ms) 6 ' 2 ", essentially all of the charge pairs are collected. The channel numbers where the particles appear in Fig. 11 therefore represent their full energies. The a-particle spectra obtained with X-UV #2 with zero and -3 V bias are shown in Figs. 12a and 12b. The corresponding spectra, obtained with the SBD operated in the same manner are shown in Figs. 13a and 13b. The a particles are barely resolved by the X-UV detector operated with no bias; in contrast, the SBD is able to resolve all of the a particles. In both detectors, however, the particles are degraded from their full energy (by about 35% for the X-UV and by about 13% for the SBD). With a bias voltage of -3 V, the a particles were still only barely resolved by the X-UV detector, and their energies were still significantly degraded. Based on these observations, we conclude that there is significantly more recombination of the charge carriers in the bulk silicon of the X-UV detectors than in the SBD. This is because at zero or low bias voltage, since there is no electric field in the device, the primary mechanism for charge collection is diffusion. 27 The primary mechanism for charge carrier loss is recombination. Since the a particles will have about the same range in the X-UV and SBD detectors, they should deposit their energy at the same depth. Subsequently, the charge carriers created by the a must diffuse the same distance in both devices before they are collected. However, the X-UV detector consistently provided smaller pulses than the SBD, which implies that a larger fraction of the charge carriers was lost to recombination. Because of this effect, we also conclude here that the X-UV photodiodes are not suitable for charged-particle spectroscopy. These diodes may be useful, however, as simple charged particle counters, because as the particles interact they should create a pulse within some range of detectable voltages. VII. Discussion The x-ray response of silicon surface barrier detectors appears to be determined by their physical thickness. 7 At first look, therefore, it is somewhat 14

16 surprising to find that the X-UV photodiodes (nominally 375 pm thick) have lower x-ray sensitivity than the comparison SBD (nominally 347 pm thick) at 8 and 17.5 kev. If the x-ray response were determined solely by the silicon x- ray absorption, as in the case of the SBD, the signal should be approximately proportional to (1 - e"pt), where p is the silicon x-ray mass attenuation coefficient, and p and t are the silicon density and thickness. The X-UV diode and the SBD would then be predicted to exhibit nearly the same x-ray response, with the X-UV response being slightly larger at higher energies (in fact, the relative X-UV response should scale as tx-uv/tsbd for high-energy x rays where ppt is small). The a-particle response of the X-UV photodiode helps to explain its smaller x-ray response at 8 and 17.5 kev. The charged particle data imply that there is enhanced charge carrier recombination in the X-UV as compared to the SBD. This effect will be especially important at low bias voltage, when the charge carriers are collected exclusively by a diffusive process. Recombination will also reduce the relative x-ray response, especially for x rays that deposit much of their energy far from the p - n junction (e.g., 17.5 kev Mo K, radiation). In conclusion, X-UV detectors exhibit x-ray responses similar to an SBD from 1.5 to about 6 kev. At energies above 8 kev, the X-UV x-ray response drops faster than the response of an SBD with similar silicon thickness. We conclude that this drop in high-energy x-ray response is due to (relatively) higher recombination in the bulk silicon of the X-UV diode. This recombination renders this detector unsuitable for charged-particle spectroscopy applications. However, these detectors may still be useful for more simple charged-particle counting. Acknowledgments We thank David Gilbert for his contribution to improving the electronbeam x-ray generator. K. W. also appreciated useful conversations with Raj Korde of International Radiation Detectors. This work was supported in part by NASA Contract No. NASW-4372 and DOE Grant No. DE-FG2-91ER

17 References 1. R. Korde and L. R. Canfield, SPIE (1989). 2. L R. Canfield, J. Kerner, and R. Korde, Applied Optics (1989). 3. L. R. Canfield, J. Kerner, and R. Korde, SPIE (199). 4. G. R. Ricker, M. W. Bautz, et al., in preparation for IEEE Trans. in Nuclear Sciences. 5. United Detector Technology, Hawthorne CA Private communication, Dean Ligocki, UDT Sensors, October (1992). 7. K. W. Wenzel and R. D. Petrasso, Rev. Sci. Instrum (1988). 8. G. F. Knoll, Radiation Detection and Measurement, Wiley, New York (1989). 9. M. Malacarne and I. H. Hutchinson, Culham Report CLM-R227, (1982). 1. J. Kiraly, M. Bitter, P. Efthimion, S. von Goeler, B. Grek, K. W. Hill, D. Johnson, K. McGuire, N. Sauthoff, S. Sesnic, F. Stauffer, G. Tait, G. Taylor, Nucl. Fusion (1987). 11. D. H. Lo, R. D. Petrasso, K. W. Wenzel, J. W. Coleman, C. K. Li, J. R. Lierzer, and T. Wei, Bull. Am. Phys. Soc (1991) and Rev. Sci. Instrum (1992). 12. In general the importance of the depletion region in determining detector response depends on the mode of operation. Radiation can interact in a partially-depleted detector either with depleted or undepleted silicon. Charge pairs created in the depletion region and accelerated to the electrodes will contribute to the signal. Charge pairs created in the undepleted region may or may not contribute to the signal, depending on the rate of diffusion, the rate of carrier recombination, and the integration time of the collection circuit (e.g., whether the system is operated in current or pulse mode). 16

18 13. D. H. Lo, K. W. Wenzel, R. D. Petrasso, J. Doty, and G. Ricker in preparation for Rev. Sci. Instrum. 14. C. K. Li, K. W. Wenzel, R. D. Petrasso, J. W. Coleman, and D. H. Lo, Bull. Am. Phys. Soc (199). 15. K. W. Wenzel, J. W. Coleman, R. D. Petrasso, C. K. Li, and D. H. Lo, Bull. Am. Phys. Soc (199). 16. EG&G ORTEC, Oak Ridge, TN R. D. Petrasso, M. Gerassimenko, F. H. Seguin, R. Krogstad, P. Gauthier, W. Hamilton, A. T. Ramsey, P. Burstein, and R. Granetz, Rev. Sci. Instrum., (198). 18. Keithley Instruments, Cleveland, OH B. L. Henke, P. Lee, T. J. Tanaka, R. L. Shimabukuro, and B. K. Fujikawa, Atomic Data and Nucl. Data Tables, 27 1 (1982). 2. In all cases, the response currents quoted with and without the Test Filter are given only to indicate typical signal levels. The high and low response currents were not necessarily obtained under the same source conditions, so the Test Filter transmission cannot be calculated from these representative values. The filter transmission results were obtained by comparing signal levels under identical source conditions. 21. W. H. McMaster, N. Kerr Del Grande, J. H. Mallet, and J. H. Hubbell, UCRL-5174-SEC-2-R1, (1969). 22. P. Horowitz and W. Hill, The Art of Electronics, Cambridge University Press (198). 23. P. R. Bevington, Data Reduction and Error Analysis for the Physical Sciences, McGraw-Hill, New York (1969). 24. The SBD response at 4.5 and 5.9 kev has not been measured. It is assumed to be close to the "ideal" response (3.6 ev per charge pair) since both the 1.5 and 8 kev measurements are close to this value (see Tables 4 and 5). 17

19 25. Private communication with Matthew Rabdau, United Detector Technology, March (1992). 26. F. S. Goulding, private communication (1988). 27. R. F. Pierret, Semiconductor Fundamentals, Addison-Wesley, Reading MA (1983). 18

20 Table 1: Three X-UV detectors were examined in this work. Detectors #1 and #2 were both Model X-UV35, and detector #3 was Model X-UV1C. All three detectors were manufactured by United Detector Technology, and some of their nominal characteristics are shown here. 5 '" Because this work examined x-ray responses for photon energies >1.5 kev, the resistivity and minority carrier lifetime values are for the bulk, p-type silicon where the majority of the x-rays are stopped. The characteristics of the SBD are also listed for comparison. Detector Model Active Capacitance Resistivity Minority Carrier I I Area (cm2) (nf) (Q-cm) Lifetime (ps) 1 X-UV X-UV X-UV1C SBD R Table 2: This table lists the materials and thicknesses used for the moveable Test Filter and the Line Filter during the DC measurements. In order to check the spectral purity of the source, the Test Filter was moved in between the source and detectors. The Line Filters, which were the same material as the anode, were attached to the front of the detectors as light and bremsstrahlung filters. Anode K, Line Test Line Material I Energy Filter Filter Al 1.5 kev 17 pm Al 17 pm Al Ti 4.5 kev 17 pm Al 25 pm Ti Cu 8. kev 86 pm Al 5 pm Cu Mo 17.5 kev 86 pm Al 124 pm Mo 19

21 Table 3: This table lists the materials and thicknesses used as the Test and Line Filters during the AC measurements. For the AC measurements, the Test Filters covered every other aperture in the eight-apertured chopper wheel, so that alternate x-ray pulses were filtered. The Line Filters func- tioned as light and bremsstrahlung filters. fanode K, Line Test Line Material Energy Filter Filter Al 1.5 kev 17 pm Al 17 pm Al Ti 4.5 kev 17 pm Al 25 pm Ti Cu 8. kev 13 pm Cu 25 pm Cu Mo 17.5 kev 62 pm Mo 124 pm Mo Table 4: This table summarizes the DC x-ray response measurements for X-UV diodes #1 and #2. Rs is the absolute SBD response (previously measured',) and R1, 2 is the absolute response of X-UV diode #1 and #2 respectively [from Eq. (5)]. All absolute response (R) values are listed in the units of 1-l 7 C/keV; 4.5x1-7 C/keV corresponds to 3.6 ev per charge pair. The uncertainties were calculated as described in Section IV. E IpV P2 Rs (kev) ± ±.1 3.8± ± ± ±.2 1.5±.5 4.5± ± ± ± ± ± ±.1.54±.6 3.7± ±.73 2.± ±.3.44± :1, 1.28it:2.88'45 2

22 Table 5: This table summarizes the AC x-ray response measurements for the three X-UV diodes listed in Table 1. Rs is the absolute SBD response (previously measured 7 ' 5 ) and R 1, 2, 3 is the absolute response of X-UV diode #1, #2, and #3 respectively [from Eq. (5)]. All absolute response (R) values are listed in the units of 1- C/keV; 4.5x1-1 7 C/keV corresponds to 3.6 ev per charge pair. The uncertainties were calculated as described in Section IV. (kev) P P2 P3 Rs R 1 R 2 R ±.9.96±.9.9±.1 3.8± ± ± ± ±.7.85±.6.89± ± ± ± ± ±.7.56±.3.73±.6 3.7±.8 2.7± ± ± ±.1.5±.1.7±.1 2.+g:g 1.2+g:l 1.-:4 1.4Ig: 21

23 k#- 1M FILAMENT POWER MorNITOR SBD HIGH VOLTAGE P1\ WER SUPPLY TEST FILTER TARGET UU txx-u FEEDTHROUGH BREAK f TFILAMEPNT PUM GATE VALVE TURBO PUMP ROUGHING PUMP 1 ROUGHING PUMP 2 Fig. 1. This schematic shows the experimental setup, including the moveable "Test Filter". Table 1 lists specific materials and thicknesses used for both the Line and the Test Filters for each calibration energy (i.e., anode material). The anode target runs at positive high voltage and the filament at ground potential. A Si(Li) spectrometer could be installed at the feedthrough position after the X-UV and SBD detectors were removed. This drawing is not to scale; the dimension is meant to indicate the overall size of the device. 22

24 C : Energy (kev) Fig. 2. The spectrum from an aluminum anode (8 kv, 1 ma emission current) was obtained with an EG&G ORTEC" 6 Si(Li) spectrometer. The spectrometer viewed the source through a 17 pm Al filter, which is the same thickness as the Line Filter used on the X-UV diode and the SBD. The emission current of the x-ray source was lower than for the diode measurements to reduce the count rate to a desirable level. 23

25 C Energy (kev) Fig. 3. The spectrum from a titanium anode (16 kv, 1 ma) was obtained through a 25 pm Ti filter, which is the same thickness as the Line Filter used on the X-UV diode and the SBD. The emission current was lower than for the diode measurements to reduce the count rate to a desirable level. 24

26 n 2 - o Energy (kev) Fig. 4. Manganese K x rays (from a 55 Fe source) were measured with the Si(Li) detector. The detector viewed the source through a.24 cm diameter aperture and 12.5 pm copper foil to reduce the count rate to desirable levels. 25

27 C Energy (kev) Fig. 5. The spectrum from a copper anode (2 kv, 1 ma) was obtained with the Si(Li) spectrometer. The spectrometer viewed the source through a 25 pm Cu filter, which is the same as the Line Filter used on the X-UV diode and the SBD during the AC measurements. For the DC measurements, we used a thicker copper Line filter, which would create a cleaner line spectrum (compare Tables 2 and 3). 26

28 4 I I I I I I I I I I 1 1 1,,,,,, I 3 C Energy (kev) Fig. 6. The spectrum from a molybdenum anode (25 kv, 1 ma) was obtained through a 124 pm Mo filter, which is the same as the Line Filter used on the X-UV diode and the SBD. The emission current was lower than for the diode measurements to reduce the count rate to a desirable level. 27

29 5 pf Bias 1 kq 1 MQ 2 jx 1 MQ F x-ray op Fig. 7. The simple AC-coupled transresistance current-to-voltage amplifier was used to measure the x-ray response while various bias voltages could be applied to the detectors. The 1 MQ feedback resistor resulted in a gain of 1 mv/na, and the 5 pf capacitor filtered out noise with frequencies above about 2 khz. 28

30 a) Al 1 11 Ib) Ti 1 I I 1 l l ) c)cuv - (~\IBD ~XfNFYUf C) CU X-Uv - SBD d) Mo - X-UV- SBD - SBD a I I I I I Fig. 8. Data from each of the anode materials is shown. (Table 3 lists the Line and Test Filters used for each measurement, and Figs. 3, 4, 6, and 7 show the corresponding spectra.) a) These aluminum data were taken with X-UV #1 while the generator operated at 8 kv and 4 ma; the scale is 1 mv and 5 ms per division. b) These titanium data were taken with X-UV #1 while the generator operated at 12 kv and 28 ma; the scale is 1 mv and 5 ms per division. c) These copper data were taken with X-UV #1 while the generator operated at 2 kv and 2 ma; the scale is 1 mv and 5 ms per division. d) These molybdenum data were taken with X-UV #1 while the generator operated at about 25 kv and 4 ma; the scale is 1 mv and 5 ms per division. 29

31 1, a) SBD b) XUV #1 4 -j-. a-.1 *C) XUV #2 TV ; 7T d) XUV #3 C Photon Energy (kev). I 1 1 Photon Energy (kev) Fig. 9. The measured responses of the SBD" (from Refs. 17 and 7) and the X-UV detector are plotted as a function of photon energy. In the X-UV data, squares with narrow error bars represent the DC measurements (Table 4), and circles with wider error bars represent the AC measurements (Table 5). The solid lines are the calculated responses assuming the entire silicon thickness forms the sensitive thickness (347 Mm for the SBD 7 and 375 pm for the X-UV 2 "). The low energy response is determined by the front surface of the devices. The SBDs have 15 A aluminum front surface, and were assumed to have an additional 2 A SiO 2 dead layer. The X-UV diodes were assumed to have only a 2 A SiO 2 dead layer. 3

32 a) I I I b) -1V 1V -SBD - SBD c) -2 V I III I li- P d) -3 V 2WWMJKI SBD a- SBD -. I, II I II I Fig. 1. The SBD and X-UV #2 detector responses to Cu K x rays are shown as a function of bias voltage. For all four sets of data, the x-ray generator was run with the anode at 2 kv and with 2 ma of emission current. The Line Filter used to obtain these data was 25 pim Cu, and the Test Filter was 12.5 pm Cu. Plots a) through d) illustrate the x-ray response obtained using the circuit of Fig. 7, as the bias was varied from V to -3 V. Neither the X-UV nor the SBD x-ray response changed significantly as the bias voltage was raised. However, the noise in the X-UV response decreased from about 6.4 na (peak-to-peak) to 3.2 na. Similarly, the noise in the SBD signal also decreased from about 2.4 na at zero bias to less than 1.6 na at 3 V bias. 31

33 35 SBD 3 - V IUS C :3 o Channel Fig. 11. The a-particle spectrum from a 22 a source was obtained with a partially-depleted SBD (#16-675B, the same SBD used for the x- ray measurements) biased at -5V bias voltage. Because this bias voltage creates a depletion layer thicker than the longest range of the a particles, the channel locations of the particles represents their full energy. 32

34 25 - a) XUV #2 V bias 2 - C = b) XUV #2-3 V bias L Channel Fig. 12. These spectra of the 22 6 Ra a emission were obtained with X-UV #2. a) With zero bias applied to the detector, the a particles with different energy could not be resolved. b) With 3 volts bias applied to the detector, the individual a particles could be barely resolved, but the collected charge from each particle was significantly degraded as compared to the SBD data shown in Fig

35 a) SBD V bias 2 - C b) SBD -3 V bias 2 - C Channel Fig. 13. These spectra of the PRa a emission were obtained with the same SBD used to obtained the spectra in Fig. 11. a) With zero bias applied to the detector, the full energy and resolution of the a particles was somewhat degraded, but the individual particles could still be resolved. b) With 3 volts bias applied to the detector, the a-particle spectrum is slightly improved over the spectrum taken with zero bias. The important feature is that even with low bias, all the a particles could be distinguished, whereas the X-UV diodes could not resolve the particles with a small or zero bias voltage appplied. 34