The energy distribution of antimonyberyllium photoneutrons

Journal ofNuclear Energy, Vol.24,

pp. 123 to 132.

Pergamon

Press 1970. Printed inNorthern Ireland

THE ENERGY DISTRIBUTION OF ANTIMONYBERYLLIUM PHOTONEUTRONS” M. LALOVIC and H. WERLE Institut fiir Neutronenphysik

und Reaktortechnik, Kernforschungszentrum 75 Karlsruhe, Germany

Karlsruhe,

(Received 11 September 1969)

Abstract-The proton-recoil proportional counting method with gamma-neutron discrimination has been used to measure the neutron spectrum of an antimony-beryllium source located in a lead gamma-shielding. Since the variations in the neutron spectrum caused by the lead and the source material are small, the photoneutron spectrum could be determined with good accuracy by a Monte Carlo neutron transport code. It turned out that the distribution of photoneutrons from the primary group has a width of 3.5 f l-5 keV and an average energy of 26.0 f 1.3 keV. In addition to this primary neutron group a second neutron group at 363 & 15 keV and with an intensity of 5.1 & 1 per cent as compared with the primary group could be established. The neutron spectra for a series of representative source geometriesare calculatedand compared with previouslypublishedresults.

1. INTRODUCTION NEUTRONS from antimony-beryllium sources are extensively used in cross section measurements (BELANOVA et al., 1966; MILLER and POENITZ, 1969; SCHMITT and COOK, 1960) and for importance measurements in fast reactors (ROSE and ABSALOM, 1957; TUTTLE, 1963). Therefore, knowledge of the neutron energy distribution is of great interest. A number of studies have been made, but, in general, only an average neutron energy has been determined. So far, as we know, only two attempts were made to measure the neutron energy distribution directly. In the cloud chamber measurements of HUGHES and EGGLER (1947) the statistical accuracy is very poor. HANSON(1949) measured the proton-recoil distribution in a proportional counter and compared it with that from monoenergetic neutrons of a Van de Graaf-accelerator. But it was impossible to receive detailed information on the differential neutron spectrum, since the intense gamma radiation of the antimony completely masked the proton-recoil distribution. The intense gamma radiation of this source is the main problem in the measurement of the differential neutron spectrum. With the y-n-discrimination technique for proton-recoil proportional counters (BENNETT,1967) the serious difficulties from the gamma background can be reduced to a great extent, because a good separation in mixed fields-up to a gamma-to-neutron ratio of about hundred-is possible. Therefore, the gamma-shielding material can be reduced to an amount, which does not markedly change the neutron spectrum and makes a reliable and accurate determination of the source spectrum possible. 2. EXPERIMENTS The pulse shape discriminationtechnique, developedby BENNEIT(1967) was used to reject the gamma-induced fast electrons in the proton-recoil proportional counter. The technique is based upon the large difference in specific ionization (ion pairs per cm of track) between a recoil proton and a fast electron of the same total ionization. The electron crosses the counter along a chord and the radial extension of the distribution of ionization is, on the average, extended over a substantial fraction of counter radius. A recoil proton, by contrast, has negligible range so that its ionization is highly localized. Therefore, the primary electrons of a recoil proton arrive simultanously at the anode and * Work performed within the association in the field of fast reactors between the European Atomic Energy Community and Gesellschaft fiir Kemforschung m.b.H., Karlsruhe. 123

124

M. LALNIC and H. WERLE

the resultant pulse rises much more rapidly than is the case for gamma-induced fast electrons, where the primary electrons, because of the finite drift velocity for electrons in gases, arrive over a long time interval. The electronic system (STRAUSSand BRENNER,1965; LARSEN, 1966) used for pulse shape discrimination is shown in Fig. 1. Each pulse from the counter is analyzed by separate amplifiers. The fast, 200 nsec-delay line clipped amplifier produces a pulse with peak height proportional to the rate-of-rise of the incident pulse; the second is a conventional RC-amplifier and produces a pulse with peak height proportional to the initial ionization. An analog pulse height computer takes the ratio of the peak heights of the two pulses, which is proportional to the radial specific ionization of the event Detector Lin-Ampl. HVL-LIP0

Pre-Ampl.

without

pulse -shape

PH A ND - 2200

discrimination

Lin-Ampl. HVL-L120

B _ Int. Discr. HP- 5583 A

with pulse-shape FIG. 1 .-Block

PH A

B

ND-2200

Coinc.

.

Strobe

t

di$crimina!ian

diagram of the electronic system.

and is large for a recoil-proton pulse and considerably smaller, on the average, for a gamma-induced electron pulse. The ‘ionization’ pulse and the ‘specific ionization pulse are simultaneously recorded in a two-parameter pulse height analyzer (128 ionization groups x 32 specific ionization channels). For any given ionization level events resolve into two distinct regions, as can be seen in Fig. 2. The broad peak at low specific ionization values corresponds to gamma-ray events, while the narrower one at higher values of specific ionization is caused by proton-recoil events. Measurements in the mixed gamma-neutron field of the Sb-Be-source were compared with those from a pure gamma source and in this way the overlap of events of gamma origin with that part of the specific ionization spectrum associated with proton-recoil events was corrected separately for each ionization group. Thus, an integrated recoil proton number, corrected for gamma background, was determined for each value of ionization. This corrected proton pulse height distribution was used to evaluate the neutron spectrum. Reliable gamma-corrected recoil proton numbers can only be received, if the ratio of events of gamma origin to events of neutron origin is not too high. In fact, with our discrimination system, the limit for this ratio lies around 100. Therefore, lead shielding had to be used to suppress the intense gamma flux of the source. As is shown in Fig. 2 with 10, 15 and 20 cm of lead shielding between source and detector reliable proton numbers can be received down to about 15, 12 and 8 keV, respectively. The source consisted of 22984 g of antimony powder homogeneously mixed with 657 g of Be powder and contained in an aluminium capsule (height 3.6 cm, diameter 2.3 cm). At the time the measurements took place the neutron source strength was about 5 x 10’ n/set. A cylindrical proportional counter (active length 25 cm, diameter 4.2 cm) filled with 2 atm of hydrogen and @ 1 atm of methane for quenching was used for the spectrum measurements. A trace of *He was added to provide an energy calibration and to check the energy resolution (FWHM = 8

Antimony-beryllium Proton

photoneutrons

Energy 23.3 keV

15.8 keV

E = 0.4 keV

3 110

Lead

Shielding

6-

Specific

o

Sb-Be

??

Pure

FIG. t.-Spectra per cent for the 764 keV-protons keV-*He-protons was compared this comparison an error of 0.8 between 10 and 30 keV. Below 40 keV no distortions relation

ionization

(channels)

Source Gamma

Source

of specific ionization for the three shielding systems. from the *He(n,p)T-reaction). The energy calibration from the 764 with the resonance dips of an iron transmission spectrum, and from keV was estimated for the energy calibration in the energy region of the proton spectrum by wall-effects occur. Therefore, the simple g@)=--

1 U(E) NT

dE

E u(E)

was used to determine the neutron flux (per energy), q(E), from the gamma-corrected recoil proton distribution (per energy), P(E). a(E) is the n,p-scattering cross section, N is the number of hydrogen atoms in the counter and Tis the counting time. A cut-off model of W (energy loss of fast protons in the filling gas per ion pair), i.e. E = Z + @4 @bDlLE et al.,1969) has been assumed to relate the

126

M. LALOVICand H. WERLE

proton energy E(keV) to the measured ionization Z(keV), but the flux values above 10 keV would change by a few per cent only if a strictly proportional relation had been used instead.

To have a good check on the consistency of measurements and Monte Carlo calculations three different lead shieldings have been used. The corresponding neutron spectra and one gamma-corrected proton spectrum are shown in Fig. 3. The softening of the spectra with increasing amount of lead is clearly visible. The position of the maximum in the neutron spectrum of the primary neutron group caused by the intense 1.692 MeV gamma-line shifts from about 24 to 225 and to 21 keV with 10, 15 and 20 cm of lead between source and detector respectively. In addition to detailed neutron spectrum measurements of the intense primary group the goal of the work was to prove directly the existence of the high energy neutron groups and, if possible, to determine their intensities. The simple amplifier system without y-n-discrimination (Fig. 1) could be used for these measurements since no gamma-induced events are detected in the counter above about 150 keV. The proton spectrum above 150 keV is shown in Fig. 3. A neutron peak at 363 & 15 keV is clearly visible, and there is also an indication of a neutron peak at about 550 keV, though the statistics in this energy range is very poor. It took about 10 hrs to measure the spectrum of the low energy neutron group for each shielding system and 50 hr were necessary to register the proton spectrum above 150 keV. The measured proton-recoil distribution corresponding to the 363 keV neutron group has been compared with a calculated one, also shown in Fig. 3. The theoretical proton distribution has been obtained in the following way: First, we took a neutron group at 363 keV as input for the Monte Carlo neutron transport code and calculated the corresponding leakage spectrum out of the lead-shielding. This calculated neutron leakage spectrum was used as input to a Monte Carlo proton-recoil code (BRANDL, 1968), which calculated the proton-recoil distribution. Further, we assumed that the background of higher energy neutrons varies as the proton distribution caused by neutrons with energy of 550 keV. From the relatively good agreement between measured and calculated proton-recoil distribution it can be concluded that, if there are any neutrons between 150 and 350 keV, their integral intensity is at least a factor of five smaller than that of the neutron group at 363 keV. From the integrated proton-recoil distributions of the 26 keV-neutron group and the 363 keV-group the intensity ratio has been determined to be &a kev/1z6liev = 5.1 rf 1 per cent. 3. MONTE

CARLO

CALCULATIONS

Lead shielding had to be used to reduce the intense gamma flux to tolerable levels. The variations in the neutron spectrum caused by the source and shielding material were calculated with the Monte Carlo method. The general purpose Monte Carlo neutron transport code MONTE (MOELLER, 1965), extended to treat problems with external sources, was used. In this code the energy is a continuous variable allowing detailed studies of neutron spectra. The history of a neutron is followed until it escapes from the system or until the neutron weight is reduced to a minimum value. Then Russian Roulett is used to decide if the neutron history stops or not. The boundaries of the source and of the lead shielding are described as second order surfaces. Microscopic cross sections are used for the calculations. They are described by polynomials in energy in the non-resonance region and in the resonance region by individual resonance parameters or by statistical distributions of resonance parameters.

Antimony-beryllium

.&.

0

Neutron

??

Proton

photoneutrons

lx

spectra spectra

L-

0 08 0

127

Lead Shielding 20.20.20 cm3

.2. 5 CD% 0 0 _

c 30

20

7 *25 cm=

20.20

OWQ

2.

0t

‘Ll 00

AC

10 Proton saectrum (Counts/gro$)

-..

-_._._

Energy

%ax., I

363215

(keV)

30

keV 20.20.20

cm3

??

-_-i

. Energy 0

Fig. 3.-Neutron

2

200

400

.

(keV 1 600

600

spectra of the primary neutron group and proton spectrum-above 150 keV.

128

M. LALOVIC and H. WERLE

Isotropic emission and a uniform spatial distribution of neutrons is assumed in the source. The atomic densities of the source with the aluminium container homogenized over the whole source volume are: Be 4.6695 x 1O22l/cm3, Sb l-215 x 1O22l/cm3, and Al 8.9293 x 1Orgl/cm3. In principle, the exact geometry of source, shielding, and detector could be taken into account with the Monte Carlo neutron transport code. But because only about 2 per cent of source neutrons impinge on the detector, much computer time would have been required for the series of calculations made to determine the photoneutron spectrum. Therefore, it was assumed that the spectrum of those neutrons which impinge on the detector may be replaced by the spectrum of all neutrons leaking out of that side of the lead pile which looks towards the detector. The difference between both spectra should be most pronounced for the smallest lead pile. But calculations showed that even in this case, for photoneutron energies of about 25 keV, the average energy of neutrons impinging on the detector is only about 0.1 keV smaller than that of all neutrons leaking out of the pile. The photoneutron spectrum was determined by claiming agreement between measured and calculated leakage spectra for the smallest shielding system (20 x 20 x 20 cm3) in the following way: For monoenergetic neutrons with energies Ej between 23 and 29 keV starting in the source the leakage spectra vij (per energy interval) were calculated. The average statistical error in vij is about 5 per cent. Then the least square method was used to determine the coefficients aj for the system of linear equations j$laj~II = $i,

i = 1, 12.

(2)

The values of pli, are given in Table 1. The coefficients {a,] represent the photoneutron spectrum, and the values in column nine give the corresponding calculated leakage TABLE 1 .-RESULTS OF THE LEASTSQUARE FIT ‘pi, Energy (keV) group i

,/’

1

2

3

4

5

6

7 Least square

/’ 1

: 3 5 6 7 8 9 10 11 12

/+ /i

10-14 14-17.5 17.5-21.5 21.5-22 22-23 23-24 24-25 25-26 26-27 27-28 28-29

Least square fit ~eficients

23

0.0126 0.0341 0.0984 0.239 0180 0.0

0.0512

24

0.0083 00314 0.0743 0,171 0187 0.194 0.0

a1690

25

26

0.0060 0*0300 O@XMI 0106 0.110 0.189 0.194 0.0

0.015 0020 0.040 0.076 0.107 0.130 0.204 0,184 0.0

0.2330

0.2250

27

0.0060 00174 00330 00580 0.0780 0.0847 0136 0177 0.177 0.0

0.2160

28

0*0083 00091 0.0330 0.0500 0.0660 0.0700 0,103 0.136 0164 0.174 0.0

0.1070

29

OGO40 0.0168 0.0260 oG#O 0.0370 0.0620 0.0740 0.115 0.114 0.166 0.175

0.0937

fit

Experiment

0.0093 0.0240 0.0540 0.105 0.118 0.137 0138 0.105 oG664 0.0341 0.0164

0.020 0.0 0.064 O-103 0.117 0.138 0.138 0.106 0.066 0.034 0.0165

Antimony-beryllium

photoneutrons

129

spectrum. The experimental leakage spectrum $i is given in the last column. The average energy of the photoneutron spectrum is 26.0 keV. The influence of several important parameters in the photoneutron spectrum on the leakage spectrum was studied. The results may be summarized as follows: The leakage spectrum is very insensitive to details in the shape of the photoneutron spectrum, provided the average energy and the width are fixed. Thus, the photoneutron distribution may be assumed to be symmetrical, what is done in the following. The leakage spectrum is also not very sensitive to changes in the width of the photoneutron spectrum. One has to change the width of 3.5 keV by about & l-5 keV to get remarkable alterations in the leakage spectrum. In contrast to that, any alteration in the average energy of the photoneutron spectrum causes an alteration of the average energy in the leakage spectrum of nearly the same amount. The total error of the average energy in the measured leakage spectra is estimated to be 1 keV, taking into account a 0.8 keV uncertainty in the energy scale, a 30 per cent error in the neutron flux below 18 keV caused by the increasing gamma background, a 10 per cent systematic error in the neutron flux between 10 and 30 keV caused by inaccuracies in the numerical evaluation and statistical uncertainties. The error in the average energy of the calculated leakage spectra is due to the statistical uncertainties only. Measured and calculated leakage spectra for the three shielding configurations are shown in Fig. 4. (The energy groups are not identical with those of Table 1.) The average statistical error in the calculated spectra amounts to about 2 per cent for the smallest system and to about 6 per cent and 10 per cent for the larger systems. All spectra are normalized to equal area and a symmetrical photoneutron spectrum with E, = 26 keV and a width of 3.5 keV has been used for the calculated spectra. The agreement between measured and calculated spectra is good, with exception of the largest system, where the statistics in the calculated spectrum is very poor. This can also be seen from the average neutron energies of measured and calculated spectra (Fig. 4), which differ by not more than O-1keV even for the largest system. To determine the average energies of the leakage spectra the region below 10 keV for the larger systems and below 14 keV for the smallest system had not been taken into account, because the flux is very low and the measurements are inaccurate there. The good agreement of measured and calculated leakage spectra for the two larger shielding systems with a photoneutron spectrum derived from the measurements of the smallest system is thought to be a good proof for the reliability and consistency of measurements and calculations. For experimental purposes it is primarily the spectrum of the emitted source neutrons which is important. The spectra for a series of source geometries were calculated with the Monte Carlo neutron transport code. The average energies of the source spectra, assuming average photoneutron energies of 26 keV and 24.8 keV (SCHMITT, 1960) with and without taking collisions in antimony into account, are summarized in Table 2. The source geometries are indicated in Fig. 5. The relative degradation of the average energy varies between 11 per cent and 3 per cent, depending mainly on the amount of beryllium. SCHMITT(1960) neglected

130

M. LALOVICand H. WERLE

-_ .12 I

Lead

__

Shielding

20*20*20

cm3

-

Measurement

----

Calculation

Y ‘;

.06

-

/ _ _____ ___

I 0

E,,.

--‘:

7 10

E av. meos = 22.5 t 1 keV

15

20

25

talc

= 22.4 f 0.3 keV

30

.12 20.20.25

cm3 av, meaS = 21.0 f 1 keV

.oe..

av,ca,c

0

10

15

20

25

.= 21.1 * 0.5 keV

30

20.20.30cm3 .08*

E av, meas : 13.5 2 1 keV

;-----------

I

____~ -

E

av.calc

= 19.6 *

0.5

keV

.04Energy

--___-J o-

10

FIG. 4.-Measured TABLE

15

20

25

( keV)

30

and calculated leakage spectra for the three shielding systems. Z.-AVERAGE

ENERGIES

(keV)

OF SOURCE

NEUTRONS

Spherical sources SCHMITT(

1960)

BELANOVA~?~.

~ (1966)

Homogeneous, cylindrical source used in this work

r = 12-7 mm d = 0.76 mm

r=lOmm d=5mm

r=12mm d=3mm

Energy of photoneutrons (keV)

24.8

26.0

24.8

26.0

24.8

26.0

24.8

26.0

Average energy of source neutrons (keV)

24.1 24.2*

25.1 25.3+

22.1

22.8

22.8

23.9

235

24.4

~~

* Collisions in antimony neglected.

Antimony-beryllium

Photoneutron

photoneutrons

131

Spectrum

z5 0 .z


Source

;s

-

10

spectra Spheiicol source

Homogenous,cyl.source

15

20

25

30

h;

O’D 0o

s-l-

PY u .-.2.

Measured

c = aI=

Leakage

lb

15

00 Spectrum

8

20

FIG. 5.-Comparison

0

25

30

of photoneutron spectrum, source spectra, and leakage spectrum.

in antimony and estimated that 88 per cent of the neutrons escape from his source without any collision, which reduces the photoneutron energy of 24.8 keV to an average source energy of 24-O keV. We received a value of 87.9 per cent in this case but a somewhat higher average source energy of 24.2 keV. In Fig. 5 some representative source spectra (antimony collisions taken into account) are compared with the corresponding photoneutron spectrum and the measured leakage spectrum of the smallest shielding system.

collisions

4. RESULTS The results of the present investigation may be summarized as follows: The distribution of photoneutrons from the first neutron group may be assumed to be symmetrical with a most probable, i.e. average energy of 26-O f l-3 keV and a width of 3.5 & l-5 keV. The average neutron energy of the source used in this experiment lies about 1.6 keV lower, because of the tail of slow neutrons caused by elastic scattering in the beryllium. There exists a second neutron group at 363 f 15 keV with

132

M. L~VIC

and

H. WERL~

an intensity of 5.1 5 1 per cent as compared with the tist group and a very weak third neutron group at about 550 keV is indicated in the measured proton spectra. The 26 keV-average energy of the photoneutrons from the first group agrees with the direct measured value of 24.8 k 2.4 keV (SCHMITT, 1960), but is higher than the value of 22.8 f 1-OkeV calculated from the measured y-ray energy of 1690.7 f O-4 keV (RYVESand BEALE,1967) and from the QBe(y, n)*Be-threshold energy of 1665.1 f 0.6 keV (MATTAUCHet al., 1965). Previously published values of source neutron energies are : WATTENBERG (1947) 24 & 15 keV, HUGHESand EGGLER(1947) 35 & 10 keV, HANSON (1949) 24 f 3 keV, ROBERTSet al. (1950) 20 f 4 keV, CULP and HAMERMESH (1954) 24.5 i 4 keV. The estimated width of 3.5 keV is in reasonably good agreement with an inherent maximum spread of f 1.4 keV due to the center-ofmass motion of the (9Be + y)-system. Energy and intensity of the second neutron group may be compared with the theoretical values of 375 keV and 5.4 per cent calculated by TUTTLE (1963). Concerning the neutron group at about 550 keV, one expects a very weak neutron group at about 530 keV due to the 2.26 MeV y-ray (LANDOLT-BOERNSTEIN, 1961). REFERENCES BE;~N;;~A

T. S., VAN’KOV

A. A. MIKHAILUS F. F. and STAVISSKIIYu. Ya. (1966) J. nucl. Energy

BEN&-~ h. F. (1967) Nucl. Sci. Engng. 27, 16. BRANDL V. (1968), EUR-Rep. No. 4154. d. CULP R. and HAMMERMESH B. (1954) Phys. Rev. 93,1025. HANSON A. 0. (1949) Phys. Rev. 75, 1794. HUGHESD. and EGGLER C. (1947) Phys. Rev. 72,902. LANDOLT-B~ERNSTEIN (1961) New Series I/l. LARSENR. N. (1966) Rev. Sci. Instr. 37, 514. MATTAUCH J. H. E., THIELEW. and WAPS~RA A. H. (1965) Nucl. Phys. 67, 32. MILLER L. B. and POENITZW. P. (1969), Nucl. Sci. Engng. 35,295. MOELLERU. (1965) KFK-Rep. No. 297. ROBERTSL. D., HILL. J. E. and MCCAMMON G. (1950) Phys. Reu. 80,6. ROSEH. and ABSALOMR. N. (1957) AERE-Rep. No. R/R. 2150. RYVES T. B. and BEALED. W. (1967), Znt. J. Appl. Radiation and Isotopes 18,204. SCHMITTH. W. (1960) Nucl. Phys. 20,220. SCHMITTH. W. and Coon C. W. (1960) Nucl. Phys. 20,202. STRAUSSM. G. and BRENNERR. (1965) Rev. Sci. Instr. 36,1857. TUTTLE J. G. (1963) AZ-Rep. No. 8549. WATTENBERGA. (1947), Phys. Rev. 71,497. WERLEH., FIEG G., SEUPERTH. and STEGEMANND. (1969) Nucl. Znsfr. Meth., 12, 111.