i i o . X' . . . I OFL. ORNLP 247| . . j FERTISEE MICROCOPY RESOLUTION TEST CHARY NATIONAL BUREAU OF STANDARDS -1963 1?) ORNL-P.2471 Conf-660301-5 COSTI PRIC.. SM/76-13 s Lice; ww_SE MASTER 529 H.C. $ 1.00; MN LEGAL NOTICE SEP 26 1966 This roport mo propared us wo account of Govembrat sponsored work. Nolther the United Butas, por the Commission, nor way pochod rcting on baball of the Comm on: A. Makes any warranty or represenlaton, expressed or implied, with rupect to the uccu- racy, completeness, or wefulness of the information contained in blo report, or that the wo of way lasoraatior, apparatus, wolbod, or process Uklosed in this report may not talringo printoly owned righus; or B. Assumos nav labslues with respect to the wo of, or for dengue routing from the use of any information, apparatus, molbod, or process discloud b this report. As wed in the above, "persoa ucting on behall of Wo Commissioo" laclides way um. ploys or cornictor of the Commission, or employee of much contructor, lobo extent tha: soch employo. Úr contractor of the Commissioa, or ediploy of rich coatractor preparos, disseminatos, or provides access to, any informatica gurowat to wo employmeat or coatract with the coamission, or kis omployment with such contractor. RELEASED FOR ANNOUNCEMENT IN WUCLEAR SCIENCE ABSTRACTS PROTON-RECOIL NEUTRON SPECTROMETRY WITH ORGANIC SCINTILLATORS* V. V. Verbinski, W. R. Burrus, R. M. Freestone, and R. Textort REVISED ABSTRACT The possibility of using an organic scintillator as a proton-recoil neutron spectrometer is very attractive because of its high efficiency, large size, high speed, and gamma-ray discrimination capability. Its disadvantages, including nonlinear light yields to heavy particles and smeared step-like responses to monoenergetic neutrons, have been largely overcome in the two spectrometry techniques reported here. Response functions were obtained for NE-213 scintillators from Monte Carlo calculations for pulse-height distributions for monoenergetic neu- trons of 0.1 to 20 MeV. Measured values of detector resolution and light yields for protons, alpha particles, and carbon ions were used in the calculation. Proton leakage from the scintillator was considered. Abso- lute experimental checks of the Monte Carlo calculations were made at 2.66 and 3.4.43 MeV with the associated particle method. These showed better than 5% overall agreement. In one spectrometry technique the family of response functions is used as input data for a spectrum unfolding code which unfolds experi- mental pulse-height distributions to obtain complex neutron spectra. The code is based on "quadratic" programming and utijizes the known non- negativity of the neutron spectrum. An energy range of 0.7 to 18 MeV has been covered with this method when pulse heigint distributions correspond- ing to 0.4 to 18 MeV proton-recoil energies were used as input to the code. In the other technique, utilizing time-of-flight methods, each response function was integrated above the discriminator level to obtain the scintillator efficiencies necessary for calculating energy spectra. *Research sponsored by the U. S. Atomic Energy Commission under contract with the Union Carbide Corporation. TOak Ridge Central Data Processing Facility: With both experimental techniques, a modified Forté gamma-ray discriminator was used. It accepted neutrons from 0.25 to 20 MeV and operated effectively with 5 mr/hr of 8° Co gammas irradiating a 5-cm by 5-cm diam scintillator (50 mr/hr for a 2-cm diam by 1 cm scintillator). A few techniques are presented for eliminating two sources of spurious counts from the gamma-ray pulse-height distribution, thus making it possible to measure neutron and gamma-ray dose simultaneously with a single detector. Several proposed methods of neutron and gamma ray dos imetry are outlined. 1. INTRODUCTION The development of fast-neutron spectrometry techniques with moder- ately large organic scintillators has been undertaken at the Oak Ridge National Laboratory because the large area and high efficiency of these scintillators make possible a large variety of experiments not feasible with most other neutron detectors. Among the many recent applications of these proton recoil scintil- lators are nuclear reaction studies, cross section measurements, and neil- tron transpc:t and scattering measurements. Both time-of-flight and steady state methods are feasible with these l'ast scintillators, but the steady state me'chod is of much greater use in neutron dos imetry, and will therefore be the principle subject of this paper. In the steady state technique, neutrons irradiate an organic scin- tillator which is viewed by a high gain photomultiplier tube, the output of which provides a signal proportional to the scilitillator's light out- put, and also a signal that identifies a neutron-induced event. Thus. it is possible to obtain a neutron pulse-height distribution in the presence of gamma radiation. This pulse height spectrum is then unscram- bled with a computer code to provide the neutron flux vs energy of the radiation field. A scintillator response function is defined as the pulse height distribution for monoenergetic neutrons of energy En, normalized to one incident neutron. The values of En must cover the energy range of interest. A family of such pulse height distritutions for a number of values of Er is referred to here as a response matrix. This matrix is renormalized to one incident neutron/cm and used as an input to the inscrambling code which then operates on the pulse height distribution. The output is an energy spectrum with realistic statistical errors based on counting statistics and response function uncertainties. The normalized pulse height distributions are used directly in time- of-flight work where the scintillator efficiency must be accurately known. At each value of En, the efficiency is simply the area of the pulse height distribution above the experimenter's bias setting. The conversion of the neutron energy spectrum to neutron dose can be accomplished by weighting with energy dependent dose conversion factors, but a large computer is required to obtain the energy spectrum, at present. Proposed methods of converting the pulse-height distribution to dose are discussed, in which the data can be weighted channel by channel with pre- computed conversion coefficients. Since neutrons are nearly always accompanied by gamma radiation, it is advantageous to measure both the neutron and gamma dose simultaneously. Some proposed methods of mixed neutron-gamma dosimetry are outlined for the present spectrometer, along with some suggestions for further improv- ing the technique. WIM . 2. THE SCINTILIATORS Although a number of scintillator types and sizes have been calibra- ted in the past, it was decided to concentrate on a single one for a calibration accurate enough for use in unscrambling pulse height distri. butions. This was the NE-213 liquid scintillator, * almost 5-cm by 5-cm *Obtained from Nuclear Enterprises Ltd., Winnipeg, Canada. diam (80 cc £ 1 cc volume), glass encapsvlated, deoxygenated, and covered with a loose fitting aluminum foil reflector. It is a mixture of xylene, slowly decaying component of light which is fractionally greater for recoil-protons and other ions than for electrons, thus making gamma-ray pulse-shape-discrimination possible. NE-213 has a slightly greater hydro- gen content but lower total density than stilbene. It also has a slightly lower light yield and pulse-shape-discrimination capability. The good transparency of NE-213 has made operation with a 12.5 cm right cylinder of the scintillator (mounted on an XP1040 photomultiplier tube) quite feasible for fast time-of-flight work. Stilhene and the plastic NE-150 scintillator* were also tried. The NE-150 gave poor results with the present pulse-shape-discrimination circuit, while the stilbene crystals were expensive, cracked easily, and showed a 15% anisotropy for 3 MeV neutrons. An anisotropy of 26% hos been reported for 6.95 MeV alpha particles [1]. Perhaps the greatest difficulty in obtaining useful results with organic scintillators and in comunicating these results stems from the nonlinearity of the light output vs particle energy. It is extremely nonlinear for protons and heavier ions, while it is approximately linear for electrons. The work of Flynn et al (2) gives the linearity of some toluene-base organic scintillators for electrons. The point of normali. zation to the present work is at one "light unit," which is the upper edge of a *°co pulse height distribution, linearly extrapolated to the base line. In terms of this unit, the light output fur recoil protons, Vp, for alpha particles, Va, and for carbon-recoils, Vc, is given in Table I. These data may be sensitive to photomultiplier tube base design, as dis- cussed below, and also to the amplifier pulse wicth. A narrow amplifier pulse (< 1.Ousec) with fast rise time will give a smaller pulse for neutro. than gamma-events, as compared to a wide pulse, because the neutron event produces a larger fraction of slow-light emission. 3. ELECTRONICS In Fig. 1 is shown, a photomultiplier tube base diagram with a block diagram of the remaining electronics. An event in the scintillator pro- duces a small voltage pulse on dynode 1l which is proportional to the light yield (linear), and large voltage pulses at dynode 14 and anode, to which the pulse-shape-discrimination (PSD) circuit is connected. Simultaneous outputs from the discriminator circuits of the PSD and linear amplifiers cause the coincidence circuit to gate on the analyzer, labeling the event as a neutron event which is stored in one half of the analyzer's memory. All unlabeled pulses, which we may refer to as garma- ray events, can be stored in the other half of the analyzer (see Sec. 10). All electronic components were standard except that (i) the 6810A photomultipliers were hand selected, basically for high gain, (ii) the pulse-shape-discriminator amplifier had all RC coupling time constants substantially increased and the overload characteristics (clipping diodes) altered, and (iii) che tube base was of original design. The PSD discriminator circuit, based on one of the designs of Forté et al [3], was modified in terms of resistor values, type and length of *Obtained from Nuclear Enterprises Ltd., Winnipeg, Canada. * F the 1 . AO . ---- . S S Table I. NE-213 Scintillator Light Output vs Energy for Protons, Vo, Carbon Recoils, V, and Alpha-Particles, for One light unit is the linearly extrapolated edge of the ºco pulse height distribution. ENERGY (MeV) va 0.10 0.14 0.20 0.30 0.40 0.60 0.78 1.00 1.40 2.00 3.00 6.00 7.80 10.00 12.00 14.00 17.00 20.00 0.0061 0.0087 0.0134 0.0237 0.0362 0.0675 0.102 0.152 0.259 0.461 0.866 1.31 0.00104 0.00135 0.00179 0.00251 0.00319 0.00436 0.00535 0.00657 0.00864 0.0116 0.0166 0.0219 0.0323 0.0420 0.0540 0.0654 0.0777 0.0988 0.121 0.00164 0.00224 0.00320 0.00490 0.00675 0.0108 0.0150 0.0210 0.0337 0.0552 0.110 0.182 0.407 0.727 1.32 1.98 11.00 2.30 3.30 4.58 5,79 7.03 2.74 9.00 11.10 4.03 5.44 La - - · T : .. delay line stub, and location and type of diodes. The principle of one !Q- tion is as follows: The slow component of each photomultiplier pulse 18 attenuated relative to the Past component by a delay line differentiator at the anode, so that by adding an excess of the anode's fast signal to the dyłode 14 signal, the PSD circuit produces large, positive pulses for neutrons. The 6810A tubes were operated at 2200-2300 volts with the highi- current voltage divider string of Fig. I to giuvide large currents for the nonlinear circuit elements (the diodes) of the PSD circuit. This resulted in a threshold for recoil-proton acceptance of about 0.25 MeV, a 50% acceptance at about 0.35 MeV, and nearly 100% at 0.45 Mey. For the larger pulses, considerable saturation of the fast component at dynode 14 and anode is necessary for the PSD circuit to be eftective up to about 20 MeV. This pulse height is roughly equal to that on a 12 MeV electron (which cannot be totally absorbed in straight flight through the 5 cm scintillator). The small linear signals from dynode ll, which had to cover about a 500; 1 pulse height range for 0.25 - 20 MeV recoil protɔns, were plagued witn pickup from dyniodes 12, 13, 14, and anode. This las reduced to a tolerable level by (1) shielding the linear dynode 11 signal with a coaxial cable, (ii) using a copper ground bus which was connected to the chassis only at a point near the output of dynode ll, (111) branching the ground bus separately from here to lower and higher dynodes, and (iiii) locating the PSD circuit on the side of the tube base opposite to dynode il, and near the ground bus and chassis. 4. MEASURING TECHNIQUES For measui nents of neutron spectra, by means of unfo.låing pulse- height distribu.ons, pulse height spectra were measured at two gain set- tings of the linear anplifier, using an accurate 10:1 gain ratio. For the run at each gain setting, the analyzer live time and the total run- ning time were recorded and used for corrections to source-strength monitors (i.e., a current integrator attached to a Faraday cup, when an accelerator beam is used to pro luce neutrons ) due to analyzer dead time. For initial calibration prior to any measurement, a precision pulser is used and the pulse decay times so adjusted that its pulse shape at the linear-amplifier output matched that for gamma-ray events. After setting the zero channel and checking the linearity of the amplifier- analyzer combination with this pulse shape, tests for photomultiplier pulse height nonlinearity are made with both a °°Co source and 15 MeV neutron source at a number of photomultiplier tube voltages to determine where pulse-height saturation begins. After the proper operating voltage is chosen, the *° Co spectrum is used for gain checks, and runs are maäe with, say, a Po-Be neutron source at two gains. A gain ratio of 10:1 is used for these runs. The two runs are plotted on logarithmic paper and corrected for the difference in absolute channel widths and any analyzer live-time differences. A slight adjustment of zero setting may be necessary to make the high and low gain plots agree in the overlap region, depending on the amplifier pulse shape and the analyzer mode of operation. This shift is due to the slight but noticeable difference in pulse shape between the precision pulser (or gamma-ray) output and neutron-pulse output. - - : 7 TEN 27 134 After optimizing the PSD potentiometer setting at dynode 14 and the PS.)-amplifier discriminator setting for best gamma-neutron discrimination, the spectrometer is used to obtain a neutron induced "proton-recoil spec- trum" at the two standard gain settings (10; 1 gain ratio), covering a useful range of 0.03 to 1 light unit in steps of 0.005, and 0.4 to 10 light units in steps of 0.05. 5. DATA HANDLING The output of the analyzer is used as input to a binning program. This program combines the two runs, made at a 10;1 gain ratio, and rebins the data into a standard 113 bin format, with about 2.5 bins per scintil- lator resolution width. Tiie gain corrections, background subtraction, and normalizations are made in the binning program. A properly calculated normalization factor will result in an output from the unscrambling code that has the desired final form (i.e., neutz'olis MeV-cm-sec, or millibarns/ MeV-steradian, etc.). The unscrambling code has, as fixed input data, the detector-response matrix with which it operates on each properly normalized 113 bin pulse height distribution to give the correct absolute neutron flux and the statistical error as a function of neutron energy. 6. OLTAINING A RESPONSE MATRIX A family of pulse height distributions, each normalized to one neu- tron incident on the scintillator, was obtained by a Monte Carlo code, 05S, oť R. Textor (to be published). The results of this code were based upon, and ver.fied by, numerous experimental measurements. To begin with, a family of 15 pulse height distributions for mono- energetic neutrons was obtained experimentally from 0.2 to 17 MeV for (p,t), (a,d), and (a,t) neutrons with the ORNL 5 MeV Van de Graaff gener- ator. Since the lower portion of each curve was generally contaminated by room scattering and unwanted reactions, only the upper edge of each pulse height distribution was used. The half-height of this dge or "step" was plotted vs En, the neutron energy, to obtain an initial recoil- proton light curve, Vo(E). This curre was used as input to the 05S code and the pulse height Listributions were calculated at each of the energies used in the experiments. The Vo(E) curve and smearing were then separat- ely adjusted to obtain à match to the upper edges of the experimental curves, and the pulse height spectra recomputed with the correct, Vo(E). The Vo(E) curve was obtained with natural alpha sources placed in the scintillator, and it was extrapolated with the aid of 14.43 MeV associated-particle experimen is using the 5 cm by 5-cm diam NE-213 scin- tillator in which a number of b2c/n,a) events occur. The V.(E) curve for carbon-recoils was obtained from published data [4], from a double-scintillator carbon scattering experiment with the associated-particle technique, and from the 14.43 MeV associated particle measurement. In a precision experiment of Love (5), 3(a, n)He reactions were produced at deuteron energies below 100 keV, and some of the 'He ions were detected with a totally depleted silicon surface-barrier diode placed at about 1200 to the incident deuterons. This labeled some of the neu- trons produced in a solid angle element centered at about 55°. A disk- shaped NE-213 scintillator intercepted these neutrons, which were counted in coincidence with the associated "He ions. The ratio of coincidence be counts to "He counts gave the absolute neutron efficiency abuve some very low neutron-pulse bias setting, when great care was taken to reject unwanted counts from the silicon dinde. Also, the resulting pulse height distribution was free of all but 14.43 MeV neutron events. The results, along with a similar experiment with