- Email: [email protected]
A direct measurement of plasma delays in surface barrier detectors
NUCLEAR
INSTRUMENTS
AND METHODS
151 ( 1 9 7 8 )
529-536
; Q
NORTH-HOLLAND
P U B L I S H I N G CO,
A DIRECT MEASUREMENT OF PLASMA DELAYS IN SURFACE BARRIER DETECTORS* H. HENSCHEL and R. SCHMIDT
Abt. Angew. Kernphysik, Strahlenzentrum der Universitiit Giessen, Giessen, W. Germany Received 26 July 1977 A method for the measurement of the difference between the plasma delays caused by fission fragments and alpha particles in surface barrier detectors (SBDs) is presented. With SBDs fabricated from n-type silicon of resistivities ranging from 300 I2 cm to 1000 g2 cm and applied voltages from 30 V to 80 V the resultant plasma delay differences are about 2 ns. This is approximately the plasma delay from fission fragments, if one assumes that the plasma delay caused by alpha particles is several times smaller.
1. Introduction Surface barrier detectors (SBDs) show even for very heavy ions (e.g. fission fragments) a good relative energy resolution (about 1%). Using them as stop detectors in time-of-flight spectrometers however becomes uncertain if velocities of ions with very different masses and energies have to be determined with the same relative accuracy as their energies. Heavy ions create along their ionization tracks a plasma of high charge carrier density (average densities are about 1016 charges/cm 3 for 5 MeV alphas and about 1018 charges/cm 3 for fission fragments). The applied electric field cannot penetrate this plasma at once. Pulse formation first begins when the plasma becomes diluted by charge carrier diffusion and, in the main, by the iLnfluence of the electric field. Further the pulse rise times are longer compared with light particles, because the charge carriers only gradually come under the influence of the electric field. The duration of these plasma effects is of the order of 1-100 ns, depending on the electric field strength and mass and energy of the detected ions. This is the reason why timing with SBDs in many cases gives an insufficient time resolution. The first hint of the existence of "plasma times" was given by Miller et al. t) in 1960. They ,obtained (with 252Cf fission fragments) pulse rise times of 10 ns instead of the expected 2 ns. In the following years numerous experimental determinations of "plasma times" were carried out2-8). Mostly the plasma time was calculated by quadratic subtraction of the whole instrumental rise time from the measured rise time of the voltage pulses (t~ = /2meas_ tinstr). 2 In refs. 6-8 "plasma times" were * The work is supported by the "Deutsche Forschungsgemeinschaft".
derived from the width of current pulses by linear 6'7) or quadratic s) subtraction of the residual parts. In all these investigations of the current or voltage pulses an extension of the charge collection time by the plasma is determined, that means the time from the beginning of the pulse formation to the moment, when all charge carriers are under the influence of the electric field. Alberigi Quaranta et al. 9) showed for the first time (as far as we know) that there exists another plasma effect, a "plasma delay". They found a retardation between particle impact and pulse formation in the detector which can be measured directly, contrary to the above defined "plasma time". For its determination one needs a start pulse (for the time-to-pulse-height converter) and, besides the particles whose "plasma delay" is to be measured (in ref. 9 deuterons and alphas), others which cause a several times smaller delay, to fix the time-zero mark. In ref. 9 the start pulse is derived from the pulsed accelerator. The time-zero mark is delivered by protons in cooled SBDs. In ref. 10, where the "plasma delays" caused by natural alphas are determined, the start pulse is produced in a scintillation detector by the coincident conversion electrons. In ref. 11 plasma delays of 252Cf fission fragments were determined. The start pulse was made by the second fragment in another SBD. In the last two references no comparison with particles causing relatively small plasma delays was possible and hence no fixation of a time-zero mark. Therefore direct determinations of plasma delays were not practicable. For this reason it is interesting to carry out direct measurements of the "plasma delay" for heavier particles than deuterons and alphas too. Today determinations of the plasma delay are of practical interest, because after the wide-spread
530
H. HENSCHEL AND R. SCHMIDT
use of constant fraction discriminators (in the amplitude-and-rise-time-compensated triggering mode, see for example ref. 12), it should be mainly the "plasma delay" which injures timing with SBDs, whereas changes of the rise time of voltage pulses (as a result of different plasma times) should hardly affect the time resolution.
get a more exact value for tpj.d As done by the authors of refs. 10 and 11 we start from the equation (1) tpa = kd/F
2. Principle of measurements, treatment of data We are measuring the difference of the plasma delays from fission fragments (t~f) and alphas (tdp~) of a 252Cf-source (about 36 times more alpha decays than fissions," in future we write tp'J for "plasma delay", tp for "plasma time"). The moments of alpha decays and fissions are indicated by a scintillation detector which detects the prompt low energetic ~,'s and X-rays following the alpha emission ( 2 5 2 C f ~ 2 4 8 C m ) and the prompt neutrons and y's emitted from the fission fragments respectively. The photomultiplier pulses start the time-topulse-height converter (TPHC), whereas the pulses of the SBD to be examined stop it. As done by Alberigi Quaranta et al. 9) we make use of the fact that the plasma delay of a light particle (here alphas, there protons) is small against that of heavy ones (here fission fragments, there alphas). For "plasma times" it was found tpj-~ 4 tp~ (ref. 2). Similar relations are expected for the "plasma delays", because the ratio of the constant k~ found in ref. 11 to the constant k~ found in ref. 10 is about 4.8:1 [k is the particle typedependent constant of the equation tp = k/F" ( F = electric field strength, n ~ 1) which seems to be valid for plasma times 2,4-s) as well as for plasma delaysX°)]. In ref. 13, too, it is expected that the plasma delays caused by alphas should be several times smaller than those caused by fission fragments. Therefore the time difference measured in our experiment should be approximately equal to the "plasma delay" of fission fragments*. Nevertheless one should try to determine t ~ to
F(x) =
*
As an alternative method the comparison of the plasma delays of fission fragments and alpha particles being emitted in a ternary fission event (LR alphas) was taken into account: Coincident detection of alphas and fragments in different SBDs to start and stop (resp. stop and start) a TPHC. The LR alphas, however, have kinetic energies of about 25 MeV and, as a consequence, ranges of about 300/lm in silicon (compared with about 15 ~m for fission fragments). Because of these very different conditions electric field strengths in partially depleted SBDs) we gave up this (simpler) method.
The constant k d can be determined (following refs. l0 and 11)from the variation of tap with the electric field strength (resp. the applied bias voltage UD). If one uses the relations
d-x d-x , eeo/2p a
(2)
(e.g. ref. 14) with d the depth of the depletion layer, x the distance from the entrance window, t and g0 the relative and absolute permittivity, p the resistivity and/~ the charge carrier mobility of the detector material, a for abbreviation, and d = x/(2 eeo#P Up) = X/(2 a UD),
(3)
one gets a tp
ak a x/(2 a Uo) - x'
_
t3td -dUD
½akd x/(2a)
(4) (5)
4 U D [ X / ( 2 a U D ) -- X] 2"
With different values for the centroids of the ionization tracks of alphas and fission fragments (x~,xj) it follows from eq. (4)
tpaf = kdf ( d- x~,']2 tpa,
k~a \ d - x J
'
(6)
and from eq. (5)
A tapf _ kdf ( d - x,'~ 2 '
\d-x~ resp.
k~ = Atap~k d - x J
(7)
With eq. (5) it is now possible to calculate the constants k~ and kid, using the measured shifts of t~ with Uo (zlt~/AUo). These constants allow thereafter [with eq. (1)] the calculation of t ~ and t~f as a function of the electric field strength (method I). Better results for t~f one should get, however using the direct measured difference tpl-tp~.d d Substituting eq. (7) into eq. (6), the unknown relation d d from eq. (6) can be substituted by known k)/k~ values. One only has to use the measured shifts of the time distributions of alphas and fragments caused by changes of the detector bias Uo. This
MEASUREMENT
I
5cm
I
252Cf- Source Detector (Active Area) Photo - Multiplier (RCA 8850) Scintillator Optical Grease
Coupling
Collimat6r Pie xiglas FlangeAluminium Foil-
Fig. 1. Arrangement of the detectors (surface barrier detector inside the vacuum chamber, photomultiplier outside).
efficiency for low energy quanta (100 keV and below). Some of its properties are listed in table 1. More information is given in refs. 15-17. The scintillator is situated in a borehole of a plexiglass flange. To reduce the losses of scintillation light, PM and scintillator are matched to the plexiglass flange with "optical coupling grease". The aluminum foil prevents electric coupling from the PM to the SBD-branch. The collimator prevents alphas and fragments from entering the detector too obliquely. The distances between the 252Cf-source and the surface of the SBD and the scintillator are about 6.5 mm and about 1.5 mm, respectively. Fig. 2 shows a block diagram of the electronic system used for the correlated analyses of plasma delay differences and particle energies. The two constant-fraction discriminators can be switched to leading-edge discrimination. Because of the limited time resolution (< 1.5 ns fwhm; exact values are not determined until now) one cannot expect separated peaks for ~-particles and fission fragments. This is the reason why we analyse the particle energies too. By means of the
252Cf-S°urce "~-C
gives for the unknown plasma delays of alphas and fragments the two equations d tpf--
d
tp= =
=
531
OF PLASMA DELAYS
"
' ' ['SBD
measured value,
(8)
\d-xJ
d In this second method for the determination of tpj the approximations from eqs. (1)-(3) only affect the correction tdp~.
3. Experimental set-up In fig. 1 the coupling of the (low noise) photomultiplier (PM) to the vacuum chamber is shown. We use a plastic scintillator with especially high
I
Timi eCalibI rator I ~DDelay ~tc
~cStOp ]
TABLE 1 Some properties of the scintillator. Type Light output Decay time Total absorption efficiency / of 1 mm thick discs of NE 102 A relative to NE 140= 1
Dimensions
NE 140 (tin-loaded plastic scintillator, 5% Sn by weight) 48% relative to anthracene 2.5 ns
I nt,r'oo. I
0.18 (14.4 keV) 0.14 (23.9 keV) Diameter 3 mm
20 ram,
thickness
ComputeI r ( TR 86 )
Fig. 2. Block diagram of the electronic system.
532
H. H E N S C H E L
time--energy matrix the time distributions of alphas and fragments can be separated completely and the time distributions of heavy and light fragments in part. The scintillation detector records (in the chosen geometry) about 27°.% of the fissions and about 1% of the ~z-disintegrations detected by the SBD. An important point is the correct adjustment of the constant-fraction discriminators, because in the PM-branch as well as in the SBD-branch the pulse height spectra are very wide: from 35 mV (threshold) to about 3.5 V and 0.1 V (o~-pulses) to about 1.5 V (fragment pulses). To find the optimal walk adjustment, an attenuator (50 [2 impedance, 0-41 dB in 1 dB-steps) was introduced before the two discriminators. It allows attenuations up to a factor 112 without changing the transit times. We changed the walk adjustment until the position and shape of the time distribution, after pulse attenuation, showed no distinct changes. In the PM-branch it was achieved that the peak shifted less than 120 ps after attenuation by a factor 112. In the SBD-branch this was less than 80 ps after attenuation by a factor 14. By means of changing the threshold we guaranteed that the discriminators " s a w " always nearly the same (upper) part of the pulse height distribution (e.g.: unattenuated, threshold 300 mV; attenuation by a factor 10, threshold 30 mV). 4. Measurements, discussion of the results Because our computer interface was not yet completed, we began with some one-dimensional measurements (using a multichannel analyzer; see fig. 2) to test the whole system and to see if the plasma delays of fission fragments have in fact the magnitude determined by the indirect determination of ref. 11 (about 3.5 ns). The accuracy of determining the peak position is still low (<60 ps). We investigated three detectors. a) A self-fabricated detector. The detector is fabricated of n-type silicon with a resistivity of about 400 12cm. With a detector bias Up = 57 V we measured at first with low thresholds in the PM- and in the SBD-branch (solid curve in fig. 3). Then in the SBD-branch we raised the threshold above the ~zpulse height (dashed curve in fig. 3). It can be seen that alphas and fission fragments yield two distinct peaks, although they have only small time-of-flight differences. The distance between the peak centers is about 27.5 channels or, with
A N D R. S C H M I D T
Detector a (/~00 f2cm) , U o = 5 7 V A
12
e-
:=:' 10 ,.Q
-A l p h a s * Fragments ( /*430 Events ) . . . . . F r a g m e n t s ( Z 7 4 0 Events~ FWHM =1.36ns ) 71.4 p s / C h a n n e l
~6 f
r
= 6
GI
,,',2
io .5
13o
Time ( C h a n n e l s )
Fig. 3. Time distribution for s-decays and fissions, measured with the self-fabricated SBD. The mean (electric field strength (see text) is about 10.7 kV/cm.
the sensitivity of 71.4 ps/channel, 1.96 ns. If one also takes into account the mean flight time difference of 252Cf fission fragments and alphas (about 190 ps) one gets tpj--tp=d d = 1.77 ns. The fwhm of the fragment peak is about 19 channels (= 1.36 ns). This is an upper limit of the time resolution for fissions. With a detector bias UD = 77 V alpha peak and fragment peak are centered around channels 99
Detector b ( 3 0 0 ~ c m ) , Uo = 50V 16
- - A l p h a s + F r a g m e n t s ( 10940 E v e n t s ) . . . . F r a g m e n t s ( 6 3 6 0 E v e n t s : W H M =l.&&ns)
14
67 ps / Channel
12
<
I0 8 6 /,
LU
2
N .,.,,%.,,,, "-'
0 131 161.5 Time (Channels)
Fig. 4. Time distribution for ~z-decays and fissions, measured with an Ortec SBD. The mean electric field strength (see text) is about 11.9 keV/cm.
MEASUREMENT OF PLASMA DELAYS TABLE
Detector
16 14
c (1000 Q c m )
--U
=30V
( 9 ? 6 0 Events
....
U =80V
( 9 6 0 0 Events
75.1 ps / Chrtnne[
UD (V)
i
;'iA
30 50 80
i:
8
tO In
tdof - tdc~(ns)
6
IM
!
80
lOO
12o 14o Time ( Channels )
i
v
160
180
Fig. 5. Time distributions for ~z-decaysand fissions, measured with an Ortec SBD. The mean electric field strengths (see text) are about 5.5 kV/cm (full line) and 9.6 kV/cm (dashed line). and 126 respectively. Their distance has decreased only by about 0.5 channels ( = 3 6 p s ; within the still limited accuracy of peak position determination). b) Ortec-detector BF-70-400-60 (300 12cm). With this detector the same m e a s u r e m e n t s as with the self-fabricated detector were carried out. T h e resultant time distributions for Uo = 50 V are shown in fig. 4. T h e distance between the peak centers is about 30.5 channels or, with the sensitivity of 67 ps/channel, 2.04ns. Taking into account the mean time-of-flight difference from fragments and alphas gives a plasma delay difference tp/--tp~dd = 1.85 ns. The fwhm o f the fragment peak is 1.44 ns. W h e n raising the detector bias to 70 V, the peak centers shifted to channels 130 (alphas) and 159 (fragments). This corresponds to a peak distance of TABLE 2
Peak positions for different bias voltages (case c). UD (V)
Peak position (channel) .f
a~
(ps) f
901 113
939 376
Peak shifts
12 ~0
-g ,e-
3
Plasma delay differencesand variations for different bias voltages (case c).
I.I1
'-
533
2.63 2.59 2.33
1.94 ns and, correcting for the mean flight time difference, to a plasma delay difference of 1.75 ns. c) Ortec detector CA-025-200-100 (1000 Qcm). Time distributions for three different bias voltages (30, 50 and 80 V) were measured. Fig. 5 shows the distributions obtained with 6'0 = 30 V and UD = 80 V. T h e results of the peak position determinations are tabulated in table 2. The plasma delay differences and the plasma delay variations with the detector bias, both derived from table 2, are compiled in table 3 (sensitivity 75.1 ps/channel). One can now try to calculate even from these few and crude results the plasma delays t~; and t ~ , as described in sect. 2. At first we employ method II, which makes use of the plasma delay differences and the delay changes with the bias voltage. For the step from 30 V to 50 V and 50 V to 80 V we take the mean plasma delay difference (2.61 ns and 2.46 ns; see table 3) and the electric field strength and the depletion depth at the mean voltages (40 V and 65 V). For the depletion depths at 40 V and 65 V one gets with eq. (3) 107/~m and 137/am. For the field strength we have chosen the value at the centroids of the ionization tracks (with the distances x~ and x; from the front side). In ref. 18 it was pointed out that d E / d x vs x can roughly be approximated by straight lines for alphas as well as for fragments, but with opposite sign of the slope. Therefore we can write x s ~ l R j and x ~ R ~ , R = r a n g e of the ions in silicon, x~-~ ~R~ was already used in TABLE 4 Plasma delays and field strengths (method II).
UD (V) F~ (kV/cm)Ff(kV/cm) t ~ (ns) tapf(ns) (kt~/ka~) 30 50 80
121 109 107.5
158.5 146 141
40 65
5.78 7.87
7.11 9.20
9,04 0.83
11.58 3.22
1.57 4.53
534
H. HENSCHEL AND R. SCHMIDT
TABLE 5 Plasma delays and proportionality constants /61 (method 1). UD (V)
k~ (ns. V/m)
k)~ ins. V/m)
tda (ns)
tdf (ns)
tdf- tda (ns)
40 65
1.61 × 106 3.22× 105
2.54× 106 1.46x 106
2.79 0.41
3.57 1.59
0.78 1.18
refs. 10, 19 and 20. In silicon we have R I ~ 15/am and R ~ 3 5 / a m and thus x j ~ 5 / a m and x~ ~ 24/am. The resultant plasma delays are compiled in table 4, together with the field strengths at Xr and x~, and the ratio kid/k~ calculated with eq. (7). The most striking result is the small value of kP/k~ at the lower field strength. Therefore it does not hold that tp~t~1, and the calculated plasma delay for fission fragments at UD = 40 V becomes uncertain. On the other hand one only can use the (At~/AUD)-Values of table 3 to calculate k~ and k~ [eq. (5)] and therewith, using eq. (1), the plasma d (method I). With the field delays t ~ and tpj strengths of table 4 one gets the results shown in table 5. The congruity between the results of the both evaluation methods is not very good, as is to be expected. At present one should only regard the measured plasma delay differences and perhaps the calculated plasma delays at Up = 65 V from table 4. To have a better survey all the measured plas-
• Detector a (t.00 ~ c m l = Detector b (300 Qcm} o Detector c (1000 ~ c m )
3.0
2.5 t-
,-r
2.0 1¢
I
,¢ 1.5
1.0
I
I
5
I
10 Electric
Field
15 (F)
(kV/cm)
Fig. 6. Plasma delay differences between 252Cf fission fragments and alpha particles as a function of the mean electric field strength (see text).
ma delay differences are plotted via the field strengths at a distance ½(x~+xj)-~ 15/am from the front side of the SBD (fig. 6). Of course a representation against such a mean field strength is a little arbitrary, but xI and x, are small compared to d. Therefore the field strengths at xI and x~ do not differ very much (see table 4*). These preliminary results show a relatively sharp decrease of the plasma delay difference between 7.5 and 11 kV/cm, a fact that should perhaps be regarded when SBDs are used for time-offlight measurements. The most important result of our present measurements is that the existence of a plasma delay for fission fragments in SBDs also came out from a direct measurement. It agrees well with the plasma delay derived from an indirect determination (ref. 11; under similar conditions here about 3.2 ns, there about 3.5 ns). After this the question why with SBDs one gets time resolutions of less than 100ps (e.g. f w h m < 6 5 ps with fission fragments ref. 2) becomes still more interesting, because with ref. 10 one should expect an appreciable "plasma time jitter" (for alpha particles about 10-15% of their plasma delay). In ref. 11 it was pointed out that for fission fragments this "plasma time jitter" (better: "plasma delay jitter") should be at most 3% of their plasma delay ~. How can it be explained that the duration of this (long-lasting) plasma state shows such little fluctuation? We intend to carry out very soon more exact, systematic two-dimensional measurements (time and particle energy) in a greater range of electric field strengths. Additionally we want to examine the temperature dependence of the plasma delay. * Besides this, we used the resistivities given by the manufacturer (Wacker Chemitronic and Ortec) for the calculation of the field strengths. These can differ from the values measured in the middle of the detectors (e.g. refs. 11 and 21). t Perhaps this indicates: Alphas and fragments have different "plasma delays", but (nearly) the same "plasma delay jitter"?
MEASUREMENT
OF PLASMA
Appendix Test of the adjustment of the constant-fraction discriminators With detector c (at LID = 30 V) we made comparative measurements by means of successive switching of the discriminators to leading-edge triggering (LE-triggering). After changing from constant-fraction mode (CF-mode) to LE-mode at discriminator 1 (PM-branch) the shape of the time distribution remained nearly the same (as in fig. 5). Only the distance between the peaks increased by about 680 ps (compared with CF-triggering in both branches). That is what we expect if this CFdiscriminator was adjusted correctly: The PMpulses following u-decays are in general smaller than those following fissions. Therefore (LEmode!) the start of the TPHC is more delayed for ,z-decays than for fissions and, as a consequence, the shift of the ~z-peak to shorter times is greater than that of the fragment peak. Their distance increases. With LE-discrimination in the SBD-branch (and CF-discrimination in the PM-branch) there remains only one (broad) peak. Again this is to be expected, if the CF-discriminator was adjusted correctly: Owing to the smaller height of the acDetector c ( 1000 Qcm ), Uo =30V 16
--
Alphas + F r a g m e n t s
....
Fragments
"1 12 .,
t-
~
8
.c kit tI
U 2 0
, 60
535
pulses (fragment energy/a~-energy= 15:1) the stop of the TPHC is delayed much more for a~-pulses than for fragment pulses. Thus the plasma delay difference (if existing) becomes partly compensated (or over-compensated). If one raises the threshold in the PM-branch above the heights of the pulses following the ~z-decays, only the right part of the peak is left (fig. 7). Though as a result of the time walk in the stop branch of the TPHC alphas should be registrated later, the alpha events are still situated in the left part of the peak. In both cases the peak shifts have the magnitude of the expected time walks [pulse rise times from 10% to 90% : PM-branch about 2.3 ns; SBDbranch: u-pulses about 5 ns, fragment pulses about 8 ns. Thresholds during the LE-mode: PMbranch about 10 keV, SBD-branch about 1 MeV. Inherent walk of the discriminators: < 300 ps from × 2 to ×20 threshold]. All this indicates a correct adjustment of the CF-discriminators. The measured time differences should therefore be a result of the different plasma delays of fission fragments and alpha particles in SBDs. We thank Prof. Dr. R. Brandt and Dr. H. Jungclas (Kernchemie, University Marburg) for the preparation of the 252Cf-source.
References
75.1 ps/Channet
~o
DELAYS
ibo
80 Time (Channels)
Fig. 7. Time distributions for u-decays and fissions, determined with a CF-discriminator in the PM-branch and LE-discriminator in the SBD-branch. Full line: low threshold in the PM-branch; dashed line: high threshold in the PM-branch; only fissions are detected (to allow a better comparison, the data collection time in the second case was longer).
l) G. L. Miller, L. Brown and P. F. Donovan, IRE Trans. Nucl. Sci. NS-7 (1960) 185. 2) H. Meyer, IEEE Trans. Nucl. Sci. NS-13 (1966) 180. 3) p. A. Tove, W. Seibt and W. Leitz, Nucl. Instr. and Meth. 51 (1967) 304. 4) A. H. Krulisch and R. C. Axtmann, IEEE Trans. Nucl. Sci. NS-14 (1967) 58. 5) A. Alberigi Quaranta, A. Taroni and G. Zanarini, IEEE Trans. Nucl. Sci. NS-15 (1968) 373. 6) p. A. Tove, W. Seibt and K. E. Sundstr(Sm, Proc. IAEA Syrup. on Nuclear electronics, lspra (1969). 7) W. Seibt, K. E. Sundstr6m and P. A. Tove, Nucl. Instr. and Meth. 113 (1973) 317. 8) R. N. Williams and E. M. Lawson, Nucl. Instr. and Meth. 120 (1974) 261. 9) A. Alberigi Quaranta, A. Taroni and G. Zanarini, Nucl. Instr. and Meth. 72 (1969) 72. 10) M. Moszyfiski and B. Bengtson, Nucl. Instr. and Meth. 91 (1971) 73. 11) H. Henschel, H. Hipp, A. Kohnle and F. G6nnenwein, Nucl. Instr. and Meth. 125 (1975) 365. 12) Ortec application Note 41. 13) Ortec application Note 40. 14) G. Dearnaley and D. C. Northrop, Semiconductor counters Jbr nuclear radiation (Spon, London, 1966). 15) Z. H. Cho, C. M. Tsai and L. A. Eriksson, IEEE Trans.
536
H. HENSCHEL
Nucl. Sci. NS-22 (1975) 72.
16) L. A. Eriksson, C. M. Tsai, Z. H. Cho and C. R. Hurlbut, Nucl. Instr. and Meth. 122 (1974) 373. 17) j. Becker, L. Eriksson, L. C. Moberg and Z. H. Cho, Nucl. Instr. and Meth. 123 (1975) 199. 18) E. C. Finch, Nucl. Instr. and Meth. 121 (1974)431.
AND R. SCHMIDT
19) C. A. Ammerlaan, R. F. Rumphorst and L. A. Ch. Koerts, Nucl. Instr. and Meth. 22 (1963) 189. 20) M. A. EI-Wahab and M. Sakka, IEEE Trans. Nucl. Sci. NS-24 (1977) 117. 21) O. Meyer and H. J. Langmann, Nucl. Instr. and Meth. 3,1 (1965) 77.
INSTRUMENTS
AND METHODS
151 ( 1 9 7 8 )
529-536
; Q
NORTH-HOLLAND
P U B L I S H I N G CO,
A DIRECT MEASUREMENT OF PLASMA DELAYS IN SURFACE BARRIER DETECTORS* H. HENSCHEL and R. SCHMIDT
Abt. Angew. Kernphysik, Strahlenzentrum der Universitiit Giessen, Giessen, W. Germany Received 26 July 1977 A method for the measurement of the difference between the plasma delays caused by fission fragments and alpha particles in surface barrier detectors (SBDs) is presented. With SBDs fabricated from n-type silicon of resistivities ranging from 300 I2 cm to 1000 g2 cm and applied voltages from 30 V to 80 V the resultant plasma delay differences are about 2 ns. This is approximately the plasma delay from fission fragments, if one assumes that the plasma delay caused by alpha particles is several times smaller.
1. Introduction Surface barrier detectors (SBDs) show even for very heavy ions (e.g. fission fragments) a good relative energy resolution (about 1%). Using them as stop detectors in time-of-flight spectrometers however becomes uncertain if velocities of ions with very different masses and energies have to be determined with the same relative accuracy as their energies. Heavy ions create along their ionization tracks a plasma of high charge carrier density (average densities are about 1016 charges/cm 3 for 5 MeV alphas and about 1018 charges/cm 3 for fission fragments). The applied electric field cannot penetrate this plasma at once. Pulse formation first begins when the plasma becomes diluted by charge carrier diffusion and, in the main, by the iLnfluence of the electric field. Further the pulse rise times are longer compared with light particles, because the charge carriers only gradually come under the influence of the electric field. The duration of these plasma effects is of the order of 1-100 ns, depending on the electric field strength and mass and energy of the detected ions. This is the reason why timing with SBDs in many cases gives an insufficient time resolution. The first hint of the existence of "plasma times" was given by Miller et al. t) in 1960. They ,obtained (with 252Cf fission fragments) pulse rise times of 10 ns instead of the expected 2 ns. In the following years numerous experimental determinations of "plasma times" were carried out2-8). Mostly the plasma time was calculated by quadratic subtraction of the whole instrumental rise time from the measured rise time of the voltage pulses (t~ = /2meas_ tinstr). 2 In refs. 6-8 "plasma times" were * The work is supported by the "Deutsche Forschungsgemeinschaft".
derived from the width of current pulses by linear 6'7) or quadratic s) subtraction of the residual parts. In all these investigations of the current or voltage pulses an extension of the charge collection time by the plasma is determined, that means the time from the beginning of the pulse formation to the moment, when all charge carriers are under the influence of the electric field. Alberigi Quaranta et al. 9) showed for the first time (as far as we know) that there exists another plasma effect, a "plasma delay". They found a retardation between particle impact and pulse formation in the detector which can be measured directly, contrary to the above defined "plasma time". For its determination one needs a start pulse (for the time-to-pulse-height converter) and, besides the particles whose "plasma delay" is to be measured (in ref. 9 deuterons and alphas), others which cause a several times smaller delay, to fix the time-zero mark. In ref. 9 the start pulse is derived from the pulsed accelerator. The time-zero mark is delivered by protons in cooled SBDs. In ref. 10, where the "plasma delays" caused by natural alphas are determined, the start pulse is produced in a scintillation detector by the coincident conversion electrons. In ref. 11 plasma delays of 252Cf fission fragments were determined. The start pulse was made by the second fragment in another SBD. In the last two references no comparison with particles causing relatively small plasma delays was possible and hence no fixation of a time-zero mark. Therefore direct determinations of plasma delays were not practicable. For this reason it is interesting to carry out direct measurements of the "plasma delay" for heavier particles than deuterons and alphas too. Today determinations of the plasma delay are of practical interest, because after the wide-spread
530
H. HENSCHEL AND R. SCHMIDT
use of constant fraction discriminators (in the amplitude-and-rise-time-compensated triggering mode, see for example ref. 12), it should be mainly the "plasma delay" which injures timing with SBDs, whereas changes of the rise time of voltage pulses (as a result of different plasma times) should hardly affect the time resolution.
get a more exact value for tpj.d As done by the authors of refs. 10 and 11 we start from the equation (1) tpa = kd/F
2. Principle of measurements, treatment of data We are measuring the difference of the plasma delays from fission fragments (t~f) and alphas (tdp~) of a 252Cf-source (about 36 times more alpha decays than fissions," in future we write tp'J for "plasma delay", tp for "plasma time"). The moments of alpha decays and fissions are indicated by a scintillation detector which detects the prompt low energetic ~,'s and X-rays following the alpha emission ( 2 5 2 C f ~ 2 4 8 C m ) and the prompt neutrons and y's emitted from the fission fragments respectively. The photomultiplier pulses start the time-topulse-height converter (TPHC), whereas the pulses of the SBD to be examined stop it. As done by Alberigi Quaranta et al. 9) we make use of the fact that the plasma delay of a light particle (here alphas, there protons) is small against that of heavy ones (here fission fragments, there alphas). For "plasma times" it was found tpj-~ 4 tp~ (ref. 2). Similar relations are expected for the "plasma delays", because the ratio of the constant k~ found in ref. 11 to the constant k~ found in ref. 10 is about 4.8:1 [k is the particle typedependent constant of the equation tp = k/F" ( F = electric field strength, n ~ 1) which seems to be valid for plasma times 2,4-s) as well as for plasma delaysX°)]. In ref. 13, too, it is expected that the plasma delays caused by alphas should be several times smaller than those caused by fission fragments. Therefore the time difference measured in our experiment should be approximately equal to the "plasma delay" of fission fragments*. Nevertheless one should try to determine t ~ to
F(x) =
*
As an alternative method the comparison of the plasma delays of fission fragments and alpha particles being emitted in a ternary fission event (LR alphas) was taken into account: Coincident detection of alphas and fragments in different SBDs to start and stop (resp. stop and start) a TPHC. The LR alphas, however, have kinetic energies of about 25 MeV and, as a consequence, ranges of about 300/lm in silicon (compared with about 15 ~m for fission fragments). Because of these very different conditions electric field strengths in partially depleted SBDs) we gave up this (simpler) method.
The constant k d can be determined (following refs. l0 and 11)from the variation of tap with the electric field strength (resp. the applied bias voltage UD). If one uses the relations
d-x d-x , eeo/2p a
(2)
(e.g. ref. 14) with d the depth of the depletion layer, x the distance from the entrance window, t and g0 the relative and absolute permittivity, p the resistivity and/~ the charge carrier mobility of the detector material, a for abbreviation, and d = x/(2 eeo#P Up) = X/(2 a UD),
(3)
one gets a tp
ak a x/(2 a Uo) - x'
_
t3td -dUD
½akd x/(2a)
(4) (5)
4 U D [ X / ( 2 a U D ) -- X] 2"
With different values for the centroids of the ionization tracks of alphas and fission fragments (x~,xj) it follows from eq. (4)
tpaf = kdf ( d- x~,']2 tpa,
k~a \ d - x J
'
(6)
and from eq. (5)
A tapf _ kdf ( d - x,'~ 2 '
\d-x~ resp.
k~ = Atap~k d - x J
(7)
With eq. (5) it is now possible to calculate the constants k~ and kid, using the measured shifts of t~ with Uo (zlt~/AUo). These constants allow thereafter [with eq. (1)] the calculation of t ~ and t~f as a function of the electric field strength (method I). Better results for t~f one should get, however using the direct measured difference tpl-tp~.d d Substituting eq. (7) into eq. (6), the unknown relation d d from eq. (6) can be substituted by known k)/k~ values. One only has to use the measured shifts of the time distributions of alphas and fragments caused by changes of the detector bias Uo. This
MEASUREMENT
I
5cm
I
252Cf- Source Detector (Active Area) Photo - Multiplier (RCA 8850) Scintillator Optical Grease
Coupling
Collimat6r Pie xiglas FlangeAluminium Foil-
Fig. 1. Arrangement of the detectors (surface barrier detector inside the vacuum chamber, photomultiplier outside).
efficiency for low energy quanta (100 keV and below). Some of its properties are listed in table 1. More information is given in refs. 15-17. The scintillator is situated in a borehole of a plexiglass flange. To reduce the losses of scintillation light, PM and scintillator are matched to the plexiglass flange with "optical coupling grease". The aluminum foil prevents electric coupling from the PM to the SBD-branch. The collimator prevents alphas and fragments from entering the detector too obliquely. The distances between the 252Cf-source and the surface of the SBD and the scintillator are about 6.5 mm and about 1.5 mm, respectively. Fig. 2 shows a block diagram of the electronic system used for the correlated analyses of plasma delay differences and particle energies. The two constant-fraction discriminators can be switched to leading-edge discrimination. Because of the limited time resolution (< 1.5 ns fwhm; exact values are not determined until now) one cannot expect separated peaks for ~-particles and fission fragments. This is the reason why we analyse the particle energies too. By means of the
252Cf-S°urce "~-C
gives for the unknown plasma delays of alphas and fragments the two equations d tpf--
d
tp= =
=
531
OF PLASMA DELAYS
"
' ' ['SBD
measured value,
(8)
\d-xJ
d In this second method for the determination of tpj the approximations from eqs. (1)-(3) only affect the correction tdp~.
3. Experimental set-up In fig. 1 the coupling of the (low noise) photomultiplier (PM) to the vacuum chamber is shown. We use a plastic scintillator with especially high
I
Timi eCalibI rator I ~DDelay ~tc
~cStOp ]
TABLE 1 Some properties of the scintillator. Type Light output Decay time Total absorption efficiency / of 1 mm thick discs of NE 102 A relative to NE 140= 1
Dimensions
NE 140 (tin-loaded plastic scintillator, 5% Sn by weight) 48% relative to anthracene 2.5 ns
I nt,r'oo. I
0.18 (14.4 keV) 0.14 (23.9 keV) Diameter 3 mm
20 ram,
thickness
ComputeI r ( TR 86 )
Fig. 2. Block diagram of the electronic system.
532
H. H E N S C H E L
time--energy matrix the time distributions of alphas and fragments can be separated completely and the time distributions of heavy and light fragments in part. The scintillation detector records (in the chosen geometry) about 27°.% of the fissions and about 1% of the ~z-disintegrations detected by the SBD. An important point is the correct adjustment of the constant-fraction discriminators, because in the PM-branch as well as in the SBD-branch the pulse height spectra are very wide: from 35 mV (threshold) to about 3.5 V and 0.1 V (o~-pulses) to about 1.5 V (fragment pulses). To find the optimal walk adjustment, an attenuator (50 [2 impedance, 0-41 dB in 1 dB-steps) was introduced before the two discriminators. It allows attenuations up to a factor 112 without changing the transit times. We changed the walk adjustment until the position and shape of the time distribution, after pulse attenuation, showed no distinct changes. In the PM-branch it was achieved that the peak shifted less than 120 ps after attenuation by a factor 112. In the SBD-branch this was less than 80 ps after attenuation by a factor 14. By means of changing the threshold we guaranteed that the discriminators " s a w " always nearly the same (upper) part of the pulse height distribution (e.g.: unattenuated, threshold 300 mV; attenuation by a factor 10, threshold 30 mV). 4. Measurements, discussion of the results Because our computer interface was not yet completed, we began with some one-dimensional measurements (using a multichannel analyzer; see fig. 2) to test the whole system and to see if the plasma delays of fission fragments have in fact the magnitude determined by the indirect determination of ref. 11 (about 3.5 ns). The accuracy of determining the peak position is still low (<60 ps). We investigated three detectors. a) A self-fabricated detector. The detector is fabricated of n-type silicon with a resistivity of about 400 12cm. With a detector bias Up = 57 V we measured at first with low thresholds in the PM- and in the SBD-branch (solid curve in fig. 3). Then in the SBD-branch we raised the threshold above the ~zpulse height (dashed curve in fig. 3). It can be seen that alphas and fission fragments yield two distinct peaks, although they have only small time-of-flight differences. The distance between the peak centers is about 27.5 channels or, with
A N D R. S C H M I D T
Detector a (/~00 f2cm) , U o = 5 7 V A
12
e-
:=:' 10 ,.Q
-A l p h a s * Fragments ( /*430 Events ) . . . . . F r a g m e n t s ( Z 7 4 0 Events~ FWHM =1.36ns ) 71.4 p s / C h a n n e l
~6 f
r
= 6
GI
,,',2
io .5
13o
Time ( C h a n n e l s )
Fig. 3. Time distribution for s-decays and fissions, measured with the self-fabricated SBD. The mean (electric field strength (see text) is about 10.7 kV/cm.
the sensitivity of 71.4 ps/channel, 1.96 ns. If one also takes into account the mean flight time difference of 252Cf fission fragments and alphas (about 190 ps) one gets tpj--tp=d d = 1.77 ns. The fwhm of the fragment peak is about 19 channels (= 1.36 ns). This is an upper limit of the time resolution for fissions. With a detector bias UD = 77 V alpha peak and fragment peak are centered around channels 99
Detector b ( 3 0 0 ~ c m ) , Uo = 50V 16
- - A l p h a s + F r a g m e n t s ( 10940 E v e n t s ) . . . . F r a g m e n t s ( 6 3 6 0 E v e n t s : W H M =l.&&ns)
14
67 ps / Channel
12
<
I0 8 6 /,
LU
2
N .,.,,%.,,,, "-'
0 131 161.5 Time (Channels)
Fig. 4. Time distribution for ~z-decays and fissions, measured with an Ortec SBD. The mean electric field strength (see text) is about 11.9 keV/cm.
MEASUREMENT OF PLASMA DELAYS TABLE
Detector
16 14
c (1000 Q c m )
--U
=30V
( 9 ? 6 0 Events
....
U =80V
( 9 6 0 0 Events
75.1 ps / Chrtnne[
UD (V)
i
;'iA
30 50 80
i:
8
tO In
tdof - tdc~(ns)
6
IM
!
80
lOO
12o 14o Time ( Channels )
i
v
160
180
Fig. 5. Time distributions for ~z-decaysand fissions, measured with an Ortec SBD. The mean electric field strengths (see text) are about 5.5 kV/cm (full line) and 9.6 kV/cm (dashed line). and 126 respectively. Their distance has decreased only by about 0.5 channels ( = 3 6 p s ; within the still limited accuracy of peak position determination). b) Ortec-detector BF-70-400-60 (300 12cm). With this detector the same m e a s u r e m e n t s as with the self-fabricated detector were carried out. T h e resultant time distributions for Uo = 50 V are shown in fig. 4. T h e distance between the peak centers is about 30.5 channels or, with the sensitivity of 67 ps/channel, 2.04ns. Taking into account the mean time-of-flight difference from fragments and alphas gives a plasma delay difference tp/--tp~dd = 1.85 ns. The fwhm o f the fragment peak is 1.44 ns. W h e n raising the detector bias to 70 V, the peak centers shifted to channels 130 (alphas) and 159 (fragments). This corresponds to a peak distance of TABLE 2
Peak positions for different bias voltages (case c). UD (V)
Peak position (channel) .f
a~
(ps) f
901 113
939 376
Peak shifts
12 ~0
-g ,e-
3
Plasma delay differencesand variations for different bias voltages (case c).
I.I1
'-
533
2.63 2.59 2.33
1.94 ns and, correcting for the mean flight time difference, to a plasma delay difference of 1.75 ns. c) Ortec detector CA-025-200-100 (1000 Qcm). Time distributions for three different bias voltages (30, 50 and 80 V) were measured. Fig. 5 shows the distributions obtained with 6'0 = 30 V and UD = 80 V. T h e results of the peak position determinations are tabulated in table 2. The plasma delay differences and the plasma delay variations with the detector bias, both derived from table 2, are compiled in table 3 (sensitivity 75.1 ps/channel). One can now try to calculate even from these few and crude results the plasma delays t~; and t ~ , as described in sect. 2. At first we employ method II, which makes use of the plasma delay differences and the delay changes with the bias voltage. For the step from 30 V to 50 V and 50 V to 80 V we take the mean plasma delay difference (2.61 ns and 2.46 ns; see table 3) and the electric field strength and the depletion depth at the mean voltages (40 V and 65 V). For the depletion depths at 40 V and 65 V one gets with eq. (3) 107/~m and 137/am. For the field strength we have chosen the value at the centroids of the ionization tracks (with the distances x~ and x; from the front side). In ref. 18 it was pointed out that d E / d x vs x can roughly be approximated by straight lines for alphas as well as for fragments, but with opposite sign of the slope. Therefore we can write x s ~ l R j and x ~ R ~ , R = r a n g e of the ions in silicon, x~-~ ~R~ was already used in TABLE 4 Plasma delays and field strengths (method II).
UD (V) F~ (kV/cm)Ff(kV/cm) t ~ (ns) tapf(ns) (kt~/ka~) 30 50 80
121 109 107.5
158.5 146 141
40 65
5.78 7.87
7.11 9.20
9,04 0.83
11.58 3.22
1.57 4.53
534
H. HENSCHEL AND R. SCHMIDT
TABLE 5 Plasma delays and proportionality constants /61 (method 1). UD (V)
k~ (ns. V/m)
k)~ ins. V/m)
tda (ns)
tdf (ns)
tdf- tda (ns)
40 65
1.61 × 106 3.22× 105
2.54× 106 1.46x 106
2.79 0.41
3.57 1.59
0.78 1.18
refs. 10, 19 and 20. In silicon we have R I ~ 15/am and R ~ 3 5 / a m and thus x j ~ 5 / a m and x~ ~ 24/am. The resultant plasma delays are compiled in table 4, together with the field strengths at Xr and x~, and the ratio kid/k~ calculated with eq. (7). The most striking result is the small value of kP/k~ at the lower field strength. Therefore it does not hold that tp~t~1, and the calculated plasma delay for fission fragments at UD = 40 V becomes uncertain. On the other hand one only can use the (At~/AUD)-Values of table 3 to calculate k~ and k~ [eq. (5)] and therewith, using eq. (1), the plasma d (method I). With the field delays t ~ and tpj strengths of table 4 one gets the results shown in table 5. The congruity between the results of the both evaluation methods is not very good, as is to be expected. At present one should only regard the measured plasma delay differences and perhaps the calculated plasma delays at Up = 65 V from table 4. To have a better survey all the measured plas-
• Detector a (t.00 ~ c m l = Detector b (300 Qcm} o Detector c (1000 ~ c m )
3.0
2.5 t-
,-r
2.0 1¢
I
,¢ 1.5
1.0
I
I
5
I
10 Electric
Field
15 (F)
(kV/cm)
Fig. 6. Plasma delay differences between 252Cf fission fragments and alpha particles as a function of the mean electric field strength (see text).
ma delay differences are plotted via the field strengths at a distance ½(x~+xj)-~ 15/am from the front side of the SBD (fig. 6). Of course a representation against such a mean field strength is a little arbitrary, but xI and x, are small compared to d. Therefore the field strengths at xI and x~ do not differ very much (see table 4*). These preliminary results show a relatively sharp decrease of the plasma delay difference between 7.5 and 11 kV/cm, a fact that should perhaps be regarded when SBDs are used for time-offlight measurements. The most important result of our present measurements is that the existence of a plasma delay for fission fragments in SBDs also came out from a direct measurement. It agrees well with the plasma delay derived from an indirect determination (ref. 11; under similar conditions here about 3.2 ns, there about 3.5 ns). After this the question why with SBDs one gets time resolutions of less than 100ps (e.g. f w h m < 6 5 ps with fission fragments ref. 2) becomes still more interesting, because with ref. 10 one should expect an appreciable "plasma time jitter" (for alpha particles about 10-15% of their plasma delay). In ref. 11 it was pointed out that for fission fragments this "plasma time jitter" (better: "plasma delay jitter") should be at most 3% of their plasma delay ~. How can it be explained that the duration of this (long-lasting) plasma state shows such little fluctuation? We intend to carry out very soon more exact, systematic two-dimensional measurements (time and particle energy) in a greater range of electric field strengths. Additionally we want to examine the temperature dependence of the plasma delay. * Besides this, we used the resistivities given by the manufacturer (Wacker Chemitronic and Ortec) for the calculation of the field strengths. These can differ from the values measured in the middle of the detectors (e.g. refs. 11 and 21). t Perhaps this indicates: Alphas and fragments have different "plasma delays", but (nearly) the same "plasma delay jitter"?
MEASUREMENT
OF PLASMA
Appendix Test of the adjustment of the constant-fraction discriminators With detector c (at LID = 30 V) we made comparative measurements by means of successive switching of the discriminators to leading-edge triggering (LE-triggering). After changing from constant-fraction mode (CF-mode) to LE-mode at discriminator 1 (PM-branch) the shape of the time distribution remained nearly the same (as in fig. 5). Only the distance between the peaks increased by about 680 ps (compared with CF-triggering in both branches). That is what we expect if this CFdiscriminator was adjusted correctly: The PMpulses following u-decays are in general smaller than those following fissions. Therefore (LEmode!) the start of the TPHC is more delayed for ,z-decays than for fissions and, as a consequence, the shift of the ~z-peak to shorter times is greater than that of the fragment peak. Their distance increases. With LE-discrimination in the SBD-branch (and CF-discrimination in the PM-branch) there remains only one (broad) peak. Again this is to be expected, if the CF-discriminator was adjusted correctly: Owing to the smaller height of the acDetector c ( 1000 Qcm ), Uo =30V 16
--
Alphas + F r a g m e n t s
....
Fragments
"1 12 .,
t-
~
8
.c kit tI
U 2 0
, 60
535
pulses (fragment energy/a~-energy= 15:1) the stop of the TPHC is delayed much more for a~-pulses than for fragment pulses. Thus the plasma delay difference (if existing) becomes partly compensated (or over-compensated). If one raises the threshold in the PM-branch above the heights of the pulses following the ~z-decays, only the right part of the peak is left (fig. 7). Though as a result of the time walk in the stop branch of the TPHC alphas should be registrated later, the alpha events are still situated in the left part of the peak. In both cases the peak shifts have the magnitude of the expected time walks [pulse rise times from 10% to 90% : PM-branch about 2.3 ns; SBDbranch: u-pulses about 5 ns, fragment pulses about 8 ns. Thresholds during the LE-mode: PMbranch about 10 keV, SBD-branch about 1 MeV. Inherent walk of the discriminators: < 300 ps from × 2 to ×20 threshold]. All this indicates a correct adjustment of the CF-discriminators. The measured time differences should therefore be a result of the different plasma delays of fission fragments and alpha particles in SBDs. We thank Prof. Dr. R. Brandt and Dr. H. Jungclas (Kernchemie, University Marburg) for the preparation of the 252Cf-source.
References
75.1 ps/Channet
~o
DELAYS
ibo
80 Time (Channels)
Fig. 7. Time distributions for u-decays and fissions, determined with a CF-discriminator in the PM-branch and LE-discriminator in the SBD-branch. Full line: low threshold in the PM-branch; dashed line: high threshold in the PM-branch; only fissions are detected (to allow a better comparison, the data collection time in the second case was longer).
l) G. L. Miller, L. Brown and P. F. Donovan, IRE Trans. Nucl. Sci. NS-7 (1960) 185. 2) H. Meyer, IEEE Trans. Nucl. Sci. NS-13 (1966) 180. 3) p. A. Tove, W. Seibt and W. Leitz, Nucl. Instr. and Meth. 51 (1967) 304. 4) A. H. Krulisch and R. C. Axtmann, IEEE Trans. Nucl. Sci. NS-14 (1967) 58. 5) A. Alberigi Quaranta, A. Taroni and G. Zanarini, IEEE Trans. Nucl. Sci. NS-15 (1968) 373. 6) p. A. Tove, W. Seibt and K. E. Sundstr(Sm, Proc. IAEA Syrup. on Nuclear electronics, lspra (1969). 7) W. Seibt, K. E. Sundstr6m and P. A. Tove, Nucl. Instr. and Meth. 113 (1973) 317. 8) R. N. Williams and E. M. Lawson, Nucl. Instr. and Meth. 120 (1974) 261. 9) A. Alberigi Quaranta, A. Taroni and G. Zanarini, Nucl. Instr. and Meth. 72 (1969) 72. 10) M. Moszyfiski and B. Bengtson, Nucl. Instr. and Meth. 91 (1971) 73. 11) H. Henschel, H. Hipp, A. Kohnle and F. G6nnenwein, Nucl. Instr. and Meth. 125 (1975) 365. 12) Ortec application Note 41. 13) Ortec application Note 40. 14) G. Dearnaley and D. C. Northrop, Semiconductor counters Jbr nuclear radiation (Spon, London, 1966). 15) Z. H. Cho, C. M. Tsai and L. A. Eriksson, IEEE Trans.
536
H. HENSCHEL
Nucl. Sci. NS-22 (1975) 72.
16) L. A. Eriksson, C. M. Tsai, Z. H. Cho and C. R. Hurlbut, Nucl. Instr. and Meth. 122 (1974) 373. 17) j. Becker, L. Eriksson, L. C. Moberg and Z. H. Cho, Nucl. Instr. and Meth. 123 (1975) 199. 18) E. C. Finch, Nucl. Instr. and Meth. 121 (1974)431.
AND R. SCHMIDT
19) C. A. Ammerlaan, R. F. Rumphorst and L. A. Ch. Koerts, Nucl. Instr. and Meth. 22 (1963) 189. 20) M. A. EI-Wahab and M. Sakka, IEEE Trans. Nucl. Sci. NS-24 (1977) 117. 21) O. Meyer and H. J. Langmann, Nucl. Instr. and Meth. 3,1 (1965) 77.