s ae a‘ylflm%‘" o 5 O -wn - ¢ 0 m i v sk e ' PTRIY DECTASSTFIED THIS DOCUMTHT WAS PRO nVITY ORDER 1979 KL D~15;?9, R.T. DUFF, 00C N AL T4 I Ll ot DA omE T NTINANY PER TOE L ~ AND IS L R, 1 w Saaa T 31¢/$s DATE / omwco_%;_ P. S. BAXIZR, DT IALS AEC RESEARCH AND DEVELOPMENT REPORT C 84 ~ Reactors=Special Features of O'fi’i N L | :mw Come ORNL=-3029 Alrcraft Reactors 63 - AIRCRAFT NUCLEAR PROPULS ION PROJECT ' SEMIANNUAL PROG RESS REPORT " FOR PERIOD ENDING OCTOBER 31, 1960 DOE 1979 REVIEW OF DECLASbIFIED REPORTS This Document 1s- Properily Declassified Revin;ed by P. §. Bal"erfléi ORNL Class 7 /?1 ification Officer OAK RIDGE NATIONAL LABORATORY operated by UNION CARBIDE CORPORATION ‘ for the o | U.S. ATOMIC ENERGY COMMISSION é‘ ' LEGAL NOTICE: ' r This report was prepared: as an uccount of Governmenf sponsored work.. Neither the Unl.ted‘ Siofes,‘ nor the. Commission,, nor ony. person octing, on behalf of the, Commission: A. Maokes any werrdnty or repre’senfation,. expressed or :mplrled with respect to, the: occuracy;, completeness., or usefulneSs of: the. ‘information: conralnedh in thls‘ report, - or that the use of any: mFormuf.lon,, upparcnius,, mefhod" or process disclosed -in this: l'eport may inot infringe privately: owned rights:; or - S : _ B. Assumes any, liabilities with respecf to the use of, or for dumages resulhng from +he wuse of ‘any information, epparatus, merhodr or-process disclosed in:-this. feport. As used in the: above, _pers.on,_acr,mgj on‘_rbe‘hulf) of the Commission!” ‘includes: any. employee or contractar of f_F\‘d Commission,..or e'fip-lb,y.e'ee-of such contractar, fo the extent that such employee or ‘contractor of the: Commission, or employee .of such contractor- prepares, disseminates, or provides. access. to, any information pursuant to his. amploymenr or conrrucf with the Commission; or his. employment with such contractor., = i @y > ORNL-3029 C-84 — Reactors—Special Features of Aircraft Reactors _ M-3679 (24th Ed.) This document consists of 232 pages Copy 2 of 234 copies. BSeries A. Contract No. W-7405-eng-26 ATRCRAFT NUCLEAR PROPULSION. PROJECT SEMIANNUAL PROGRESS REPORT for Period Ending October 31, 1960 Staff Oak Ridge National Laboratory | CLASSIF] CLASSHFICATION CANCELLED, DATE %/J S BT For The Atomic Energy Commission ;fig{ ‘;? C}i o B Date Issued ' Chief, Declaselfncatzon Branch fl DEC 161960 X% Oak Ridge National Laboratory Qak Ridge, Tennessee operated by - UNION CARBIDE CORPCORATION for the U. S. ATOMIC ENERGY COMMISSION FOREWORD o The ORNL~ANP progrem primarily provides research and development - support in reactor materials, shielding, and reactor engineering to &y organizatlons engaged in the development of air-cooled and liquid-metal- cooled reactors for aircraft propulsion. Most of the work described here 1s basic to or in direct support of investigations under way at Pratt & Whitney Alrcraft Division, United Aircraft Corporation, and General Electric Company , Airéraft Nuclear Propulsion Department. "W 4 S.[JW'ARY 8 5 4 6 3 9 0 4 0 B8 G B S B S G B B8 A B S EE S S S SIS S S S T O ENEPE SR PART 1, MATERIALS RESEARCH AND ENGINEERING la M—ATEB-IALSCOMPATIBILITY s % 25 e 08 S et s A A S SN et Reactions of Columbium and Columbium Alloys with Low-Pressure Gases 4 8 & 4 0 » 8 s ® 0 2 8 & 2 4 & T 5 PSP N B EE PSS e RS E S E S S sy sE S o . Reactions With OXYEEL « ... eeenvesrneencnsonsessnessonsnnens Reactions With Air .... 2 % 8 8 B 4 8 3 B9 S P S O S E D OO R LY SN e Reactions .Wj-th CH4 .......... o 0 9 & & 0 % 0 % 5 O 09 & 0 O B S PO S S0 B e P Tentative Conclusions Regarding the Reaction Rates of Columbium and Its Alloys with High-Temperature Gases ..... Compatibility of Columbium and Columbium-Zirconium Alloys W'ithUOz andUO2"BeO # e 6 5 9 8 8 8 0 L 8 8 B B E B 6 S L P 6 SO0 0 B S0 S EE A0 Effects of Oxygen Contamination ‘and Subsequent Exposure to Lithium on the Tensile Properties of Columbium and Cb—1% ZrAlloy .l'l.I‘....|....lI..ll.l.'...ll...llI..'.I..I..I..lll Effect of Time on the Corrosion of Oxygen-Contaminated Columbium by Lithium ....i0eeeveesnnsaccns ceresrcnssssasan e Effect of Oxygen Concentration of Tantalum on Its Corrosion Resistance to Lithiwn T " 8 9 8 8 8 P S & B S P AR B E P B S ST O RS ee SN See DiSSimilaI'—Metal MaSS-TI'aIleEI' StU.d.ieS T8 s BT NSRRI EE YRR NN e Tests of Structural Materials in Boiling Potassium ....cs0eese ReflwingcapsuleTeSts » 8 & ¢ 8 & 4 % 8B B 8 P T O OB 0SB R2 00 S T B 8 40 Boiling Loop Test sveeeersercenseeseosscssestssssnssnssnsnsns 2. AGING’ ST[JDIES OF COL‘[JI\[BI.UM“BASE ALLOYS ¢ e s s sty . , Wrought Material ....... T I Y Fusicon-Welded Material ..ccesesocsccessnssssssssassssnssssncssses 3. MECHANICAL PROPERTIES INVESTIGATIONS .ve.eesseosancaccossssvecs Tube«Burst Tests on Columbium-Zirconium AllOYS .viievesoscnese Effect of Gaseous Contaminants on the Mechanical Properties of Columbium .iesesesececncessnsosssncesnsnnsncnns 4. ALI’OYPREPARATION O'.'I...l....I..'.l..........ll........'I..II Electron~Beam Melting of Columbium-Zirconium AllOYS ...eveevss ) - O 1 wWw 10 13 14 14 18 23 23 28 37 37 39 41 41 Electron~-Beam Melting of Vanadium, Molybdenum, Tantalum, and Tungsten .e.. et ienecersscrssesoasossnssscasossessaes cem e Electron-Beam Ingot Melting .cveeeeesessscsesceosnans ceess e WELDING AND BRAZING ..... Ceevisaseraasasessasnas T Electron-Beam Welding .veeeevecrtonsoceans cecaseacsas crescan e In-Pile Tube-Burst Capsule Fabricalion .....ccceeveeenesccnns BERYLLIUM OXIDE RESEARCH .....vv00ue Sectbessescbanernanss oo Purification and Calcination Processes ........ cereeeennse sees Oxaiate Purification Process ..v.ieeeececess cessasranses .o Calcination of BeC,0,4-3H,0 to BeO ...... teesessenssna cesae Fused-Salt Precipitation Processes ............ ceesaesoens Solvent Extraction Process ........... Cesecrtassetietass e Phase Relationships in BeO-Metal Oxide Systems P BeO=Las03 .overirirteensnnsenssorcasonnans cssees et eann BeO-Cal ....vvennss e ieeanaaee Cesrasserassnnnae cersenna MgO0-BeO and Ce0,-Be0 ....cve00uv.n cersasunecee ctreesas . ENGINEERING AND HEAT TRANSFER STUDIES ...... tessennas savenns Boiling-Potassium Heat Transfer Experiment ....... cesrnesncn Thermal Conductivity of Lithium and Lithium Alloys' .......... RADIATION EFFECTS ....... crecsuna cranunn cetsctictsrsennreat s Irradiation of BeQ in the EIR ....... cesresetsterersasaacnans Determination of Helium in Irradiated BeO ....ceievieerennanes Creep and Stress-Rupture Tests Under Irradiation ....... e PART 2. SHIELDING RESEARCH INTRODUCTION lllll ® 8 3 & % 0 8+ 3 S 0 S F & DTS A eSS SN SRS E S S S E DEVELOPMENT OF REACTORS FOR SHIELDING RESEARCH ..... cevsrsnen Tower Shielding Reactor II (TSR-IL) ........ e enee e Critical Experiments .....c 000 .. Ceesses s tesesrs s s s ue Thermal~Neutron Flux Distributions ...... tesretssesseessnn - Power Distribution and Heat Flux in TSR-II Core .......... FlOWDistribution Studies * S & % 0 00 9 0 0 s S 8 s P S 5 8 5 8 0" A B b s a0 43 47 47 47 49 49 49 49 51 56 57 58 59 60 62 62 A 70 70 72 73 79 82 82 82 87 87 g2 10. S 11. Bulk Shielding Reactor IT (BSR-II) ........ Static Measurements ..... Dynamic Measurements .... Safety Evaluation of BSR-II Present Status DEVELOPMENT OF RADIATION DETECTION EQUIPMENT ..... Gamma~Ray Spectroscopy ...... Monte Carlo Calculations of Response Functions of Gamma~Ray Scintillation Detectors Experimental Study of the Gamma-Ray Response of Large NaI(T1l) Crystals Intrinsic Line Width in NaI(T1) Gamma-Ray Spectrometers Light Output of CsI(T1) Under Excitation by Gamma Rays .. The Model IV Gamma-Ray Spectrometer .. Unscrambling of the Scintallation Spectra ..... Neutron Spectroscopy ...e.vs. Use of Silicon Surface-~Barrier Counters in Fast- Neutron Detection and Spectroscopy .. Distribution in Energy of Alpha Particle-Triton Pairs Resulting from Neutron Bombardment of Lithium Fluoride .. A Neutron Chopper Spectrometer for the Bulk Shielding Facility ..... A Spherical Detector Shield for the TSF & & 5 & & & & 5 e s B e @ Experimental Determination of Flux Depression and Other Corrections for Gold Foils Exposed in Water ........ BASTIC SHIELDING STUDIES . The Spectrum of Prompt Gamma Rays from U227 Fission ...o..... Source Standardizstion ... T s 80 s Response Matrix Formulation for the Pair Spectrometer ... Data for Low-Energy Region ......c.... Remaining Analyses ..... . Monte Carlo Code for Deep Gamma~Ray Penetration Calculations 20 90 93 % 95 96 96 96 108 111 117 118 121 128 128 137 142 150 151 153 153 154 155 165 166 166 vii 12. 13. viii A The Single-Scattering Approximation to the Solution of the Gamma-Ray Air-Scattering Problem ... mating the Scattering of Gamma Rays from a Crinder — An IBM=-704 Monte Carlo Program for Esti- s 8 8 6 & 680 + 0 Cylindrical Medium ........ cesctiacarrsenns e Angular Distribution of Neutrons Emerging from Planar Surfaces of Diffusing Media .e.vivevenrences Description of Experiment ................ Results ....... Cesesietsastearsonennaseee Conclusions .eeeienseesencens Cerateriannase An IBM-704 Program to Determine Angular Distributions of Neutron Histories Generated by the 05R Code .. Development of an IBM-704 Analytical Code for Analysis of Axially Symmetric Reactor Shields ...vceseseseascennsnse Calculations of Inelastic Cross Sections for High~ Energy Particles Incident on Complex Nuclei Calculation of the Penetration of High-Energy Particles Through Shields .veevertiecsasasanss APPLIED SHIELDING DEVELOPMENTS ..vieivecensss .+ 56 808 8 000 0 s "¢ * o0 Preanalysis of Pratt & Whitney Divided Shield Experi- ment at TSF ...0ceeeen e ceesesesreaatenensans ILTSF Study of Secondary Gamma-Ray Production in y238.rig Configurations ..eeevensecscnranes . s 8 ¢ 0 ITSF Shielding Measurements for Aerojet Mobile Power ReaCtorm-l ooooooooo o s " e w0 S 0 8 % 8 &t e 8 e S}EE]‘—DDESIGN 2 5 8 5 BB S 0 3 2D S S 0B S eGSR e S e Optimization of a Reactor Shield .......... .o 176 177 179 180 183 184 187 188 190 193 194 194 196 199 201 201 w-’ ' ANP PROJECT SEMIANNUAL PROGRESS REPORT SUMMARY Part 1. Materials Research and Engineering 1. Materials Compatibility A comparative survey of the oxidation characteristics of numerous experimental columbium alloys was conducted at 1000°C and an oxygen pressure-of 5% 10”% mm Hg. The 300-min weight gains for all the alloys were of the same order of magnitude as those for pure columbium. The weight gain increased as the concentration of Zr, Be, Ce, and Al in the alloy was increased. ZE¥ach of these elements forms an oxide that is more stable than columbium oxide. The portions of the weight gain which accrued linearly and parabolically were significantly influenced by the alloying element. Linear reaction rates were promoted by Zr, Mo, Re, and Ti, while Y, Ce, Al, La, and Cr promoted parabolic rates. Oxidation studies of columbium and a few of the alloys were conducted in air at a pressure of 5 X 10”4 mm Hg and a temperature of 1200°C for exposure times up to 3000 min. The reaction rates were either linear or exhibited an incubation period up to 1000 min during which the rates were not measurable. | Several tests were made to determine the compatibility of columbium- zirconium alloys with UO,. Compacts of the metal in contact with U0, powder were aged for 100 hr at 2000 and 2400°F., No reaction was noted between the U0, and columbium alloys containing 0.5 to 5.0 wt % zirconium. There was also no reaction with pure columbium, but an extensive reaction occurred between a pure zirconium specimen and the UO, powder. Additions of up to 1200 ppm of oxygen to a Cb—1% Zr alloy were shown to have no effect on the room-temperature tensile strength or ductility of the alloy. When the oxygen-contaminated alloy was heat treated for 2 hr at 2912°F, it displayed unimpaired corrosion resistance to lithium at 1500°F. When the exposure to lithium of the oxygen- contaminated alloy was not preceded by heat treatment, the Cb—1% Zr alloy was attacked, and an increase in tensile strength and a decrease in elongation were noted. The attack of oxygen-contaminated columbium by lithium was shown to occur at a very fast rate. In tests at 1500°F, large removals of oxygen and large penetrations of the columbium occurred within the first 2 hr. Tests of tantalum metal indicated that attack at elevated temperatures by lithium was greatly promoted by oxygen concentrations in the tanfalum of as low as 300 ppm. When specimens of Cb—1% Zr alloy 0.040 in. thick were exposed to . NeK at 1700°F for 1000 hr, surface layers of CbC and Cb,N were formed | - to a total thickness of 0.001 in. When these layers were removed by - machining off 0.002 in. from the alloy, chemical analyses of the remaining - bulk alloy showed it to be still heavily contaminated with mass-transferred - carbon and nitrogen. Because of interest in boiling potassium as a working fluid for electrical generators aboard space vehicles, tests for determining the resistance of various structural metals to corrosion by boiling potassium are being made. Specimens of type 316 stainless steel, Haynes qlloy No. 15, and Inconel have been tested in capsules containing boiling potassium at temperatures around 1500°F. It has been found that the iron~based material (type 316 stainless steel) is more resistant to corrosion by potaséium than either the cobalt=-based Haynes alloy or the - nickel-based Inconel. Since a type 316 stainless steel thermal-convection loop had shown good resistance to corrosion by boiling potassium for 200 hr at a boiler temperature of 1500°F, a second loop was operated at . 1600°F for 3000 hr. Examination of the test specimens and representative | portions of the entire loop disclosed that only minor attack and mass transfer occurred during the 3000-hr test. 2. Aging Studies of Columbium-Base Alloys Studies were continued on the effect of small amounts of surface contamination on the aging behavior of Cb—1% Zr alloys. Specimens of the alloy were vacuum annealed for 2 hr at 2912°F and then held at 1700°F in an argon-filled columbium capsule or an evacuated quartz capsule or were wrapped in tantalum foil and held at 1700°F in an evacuated quartz capsule. The effects of aging in the various environments exhibited only minor differences. Samples from three heats of the alloy which‘were aged in evacuated quartz capsules exhibited no changes in stiength or elongation when their surfaces were removed by machining. Nine commercial heats of Cb—1% Zr alloys have been tested in the current aging studies. Of these nine heats, five showed an aging response after amnealing at 2912°F and four did not. In those heats which did not exhibit aging, the oxygen content was high relative to other impurities such as nitrogen and carbon. It is postulated that nitrogen and carbon react with the zirconium in the alloy to form compounds which will precipitate under proper conditions of time and temperature to cause aging. It is believed that oxygen may possibly have the effect of in- creasing the solubility of the zirconium nitrides and/or carbides in the alloy and thus inhibit aging. Another possibility is that oxygen will react preferentially with part of the zirconium and will precipitate in a form which will tend to nullify the effects of the nitride and/or carbide phases., Tensile tests were made of aged all-weld-metal specimens which were welded in a recently improved inert-atmosphere chamber. They confirmed the previous -evaluations of the aging phenomenon made on the basis of bend tests on welded specimens. In a comparative weld testing program with Pratt & Whitney, in which both laboratories studied the same heat of material, it was shown that weld aging was more than just a surface effect, since the machining off of surfaces from aged welds did not - reduce their brittle behavior. The data also showed that weld aging could be avoided by the use of a sultable preaging annealing procedure, With the common test heat (PGTF), weld specimens annealed at 2200°F for 1 hr no longer became embrittled by subjecting them to an aging tempera- ture of 1500°F for 100 hr. High-purity electron-beam-melted columbium- zirconium alloys were shown to be considerably less susceptable to aging than commercial alloyé. Alloys containing large amounts of oxygen were found to be very susceptible to aging after welding. xi - W5 % - 3. Mechanical Properties Investigations It has been shown that moderate additions of oxygen, nitrogen, and hydrogen strengthen columbium at room temperature. Oxygen present in solution in concentrations in excess of 3000 ppm causes serious embrittle- ment. Nitrogen in excess of 1000 ppm reduces the room-temperature bend and tensile ductilities of columbium to essentially =zero. 4. Alloy Preparation > Exceptionally pure columbium-zirconium alloys were prepared in the electron-beam furnace, They contained very low quantities (less than 50 ppm each) of oxygen, nitrogen, and carbon. It was then shown that - the carbon content of the alloys could be controlled by means of adding - carbon to the melts. Control of the zirconium content was not very accurate because of its high volatility. The reduction of oxygen and nitrogen contamination in vanadium, molybdenum, tantalum, and tungsten by melting in the electron-beam furnace was demonstrated. Some revisions of the furnace equipment were initiated to make it mofe suitable for the melting of large ingots. 5. Welding and Brazing It was demonstrated that an electron-beam welding technique adequately - maintained the purity of Cb—1% Zr with respect to contaminants such as oxXygen, nitrogen, carbon, and hydrogen. o 6. Beryllium Oxide Research ) Methods for precipitating beryllium from oxalate solutions were studied further. A precipitate of beryllium oxalateébe:yllium hydroxide glass was prepared, and when it was calcined a very high surface area BeO was formed. Studies of the reactions occurring during the calcination of BeC204-3H,0 have continued because the method of conducting the calcina- tion has been shown to have a most important bearing on the sinterability of the BeO which is obtained. In order to determine the phase changes, xii samples of BeC504°3H>0 were placed in porcelain crucibles and held at selected temperatures in stagnant air until "equilibrium" was attained. Equilibrium conditions were said to have been attained when the weight change of the sample over a 24~hr period was less than O;l%. Thé phases present were then determined by x-ray analysis at room temperature and by means of a polarizing microscope. The results so obtained were also corroborated by high-temperature x-ray analyses. The trihydrate was shown to decompose at temperatures as low as 50°C. The phase present between 50‘and 225°C was shown to be the monohydrate. The monohydrate began to decompose to BeO at 250°C. Four experiments were in progress to prepare approximately 1 kg of BeQ crystals. In each experiment, helium saturated at room temperature with water vapor was passed as a covering atmosphere over a mikture of LiF-BeF, (60-40 mole %) maintained at 800°C in a nickel reaction vessel. Melt agitation was achieved by bubbling dry helium through the molten mixture. When the melt composition became 65 mole % LiF because of the precipitation of BeO, a purified mixture of LiF-BeF, (30-70 mole %) was added to restore the original melt composition. Petrographic examination of a small portion of BeO prepared in these experiments indicated the formation of a desirably pure product. A liquid-liquid extraction process for purifying beryllium was found to give reasonably high separation factors from other metals. In the experiments, beryllium was extracted in the form of the acetylaceto- nate, and methods of converting this compound to pure BeO are being investigated. 7. Engineering and Heat Transfer Studies Fabrication of the major components of an apparatus for studying heat transfef to flowing, boiling potassium in a heated tube was com- pleted except for the boiler section. Preliminary thermal conductivity data were obtained on an essentially pure lithium sample (99.8 wt % Li) and in a lithium~silver alloy (3 wt % I1i). These data give values for the thermal conductivity of lithium xiii ranging from 20 Btu/hr:-ft:°F at 740°F to a value of 41.5 Btu/hr-ft-°F at 1780°F. Determinations made at 1520°F on the silver alloy indicated its thermal conductivity at that temperature to be 36.5 Btu/hr-ft-°F. 8. Radiation Effects Beryllium oxide irradiations were continued in the EIR while previ- ously irradiated specimens were being examined. A method was developed for detecting helium formed by neutron reactions with beryllium and . lithium in the Be0 bodies. An experimental assembly containing ten type 304 stainless steel . tube~burst specimens was tested at 1300°F in air in the poolside facility of the ORR. The data obtained in this experiment when compared with out-of-pile data indicate that neutron bombardment does not affect the time to rupture of this material under these conditions. Fabrication of equipment for columbium=-alloy tube-burst tests was essentially come pleted, and assembly of the first in-pile apparatus was begun. Part 2. Shielding Research - 9. Development of Reactors for Shielding Research A series of critical experiments was performed with the spherically shaped Tower Shielding Reactor II following final assembly and shakedown tests at the TSF. Prior to final assembly, the internal control region was modified to eliminate the aluminum plugs extending from the support structure énd to make the control plates solid instead of perforated. Experiments utilizing the pulsed-neutron technique showed that the com- bined worth of the solid control rods from the critical position to full insertion is -3.52 £ 0.12 dollars, that the worth of a single rod from the critical position to full insertion is -0.62 * 0.05 dollars, that the combined worth of the control rods from the fully withdrawn position to the fully inserted position is -4.77 + 0.14 dollars, and that the excess k of the reactor is 1.56% Ak/k. An excess k value determined by adding boric acid to the moderator water until the shim-safety rods - - L w5 5 . . . .\-vl_ could be completely withdrawn yielded a value of 1.48% Ak/k at 20°C. A measurement of the temperature coefficient resulted in a value of -1.24 X 10™% (Ak/k)/°F for a mean core temperature of 140°F, and void coefficients were found to be =6.3 X 10=7 (Ak/k)/cm® of void outside the control-mechanism housing and +1.06 X 107® (Ak/k)/cm?® of void. inside the housing. The ratio of prompt-neutron generation time to the effective delayed-neutron fraction was found to be 6.61 £ 0.16 msec, and an investi- gation of the importance of the U235 added on the spherical cover plates of the control-mechanism housing showed that the ratio of the percentage change in mass to the change in reactivity is 2.5; 1t was also verified that the reactor can be operated without either the external water reflector or the lead-boral shield and that no dangerous reactivity effects due to shield changes can occur with presently available shields. A limited number of measured thermal-neutron flux distributions agreed with corresponding calculated distributions. On the basis of this agreement, power distributions were calculated from the calculated flux- distribution data. The results indicated that the power generated in the single cylindrical element at the lower end of the core is 19.2 kw and does not change much with changes in the flux distributions, while the power generated in the spherical fuel plates on the control-mechanism housing varies from 113 kw for the cold, clean, critical case to a maxi- mum of 125 kw for 5-Mw operation. The average heat flux in the core, excluding the spherical fuel cover plates, was calculated to be 25 120 Btu/hr-ft2. Analyses of the results of tests with the stainless-steel-clad Bulk Shielding Reactor II at the NRTS SPERT-I Facility have been completed. Authorization by the Commission for operation of the reactor at the Bulk Shielding Facility has been requested. The SPERT tests were divided into static and dynamic measurements. The static tests included measurements of the worth of the four pairs of control plates, tempera- ture and void coefficients, and fiux shapes. The temperature coefficient was found to have an average value of 0.0136 dollars/°C for the range from 15 to 85°C, and extrapolation to lower temperatures indicates that the coefficient changes sign at about 12°C, The air-void coefficient XV for the core is -14 x 107% dollars/cm3 of moderator void. The dynamic measurements were made to investigate self~limited transients and transients that were terminated either by the period Safety system or by the level safety system alone. The peak power observed during the self=- limited transients was 226 Mw. Because a small permanent warping occurred in some of the fuel plates at that power, the tests with self- limited transients were then terminated, and the data that had already been collected were extrapolated on the basis of the similar APPR P18/19 | * core. In the remaining tests with transients it became apparent that a sharp rise in peak power occurs as the reciprocal period passes about - 100 sec"l, which corresponds to a period of less than 10 msec. The - excursions terminated by the level scram system were asymptotic for small inverse periods up to 100 kw, which is the scram point set on the level scram system. The period safety system is much more effective in the range tested than the level safety system, since a factor of at least 103 separates the peak powers for a given inverse period in the two cases. At the shortest period measured, the peak powers of self-limited and period-scram-terminated excursions differed by a factor of about 25, and those of the self-limited and level-scram-terminated excursions differed by about a factor of 12. In a safety evaluation of the reactor, 1t appears that the BSR-II is safer than the aluminum-clad BSR-I in the reactivity range up to 2.1 dollars of excess reactivity. 10. Development of Radiation Detection Equipment “» The Monte Carlo codes for calculating the'gamma-ray response func- tions of the scintillating materials Nal, xXylene, and Csl have been used to (1) compare calculated values for Nal crystals with experimental values, (2) compare calculations for Nal crystals with the Monte Carlo calculations of other investigators, (3) compare Nal and CsI calculations, (4) compare CsI calculated results with experimental results, and (5) in- vestigate the effects of scintillator dimensions on the pulse-height spectra of large Nal crystals. 1In general, the calculated and experi- mental results agree remarkably well, In the comparison of the calculated values for Nal crystals with those of Miller and Snow, it was found that the values differed by an amount which was equal to the contribution by bremsstrahlung effects, the ORNL data including bremsstrahlung and the Miller-Snow data apparently omitting it. The calculations for investigating the effects of scintillator dimensions clearly showed that a definite amount of control over the photofraction and over the shape of the tail spectrum can be obtained by care in the choice of crystal dimensions. Investigations of the gamma~ray response functions of a composite 8-in.-diam by 8-in.-long NaI(T1l) crystal have continued. Experimental spectra from Cst37 and Na?% sources have been measured and compared with calculated spectra developed with existing Monte Carlo codes. The re- sults are encouraging, but additional study of interface effects, reflec- tors, and background effects is desirable. In the use of NaI(Tl) scintillation crystals for gamma=-ray spectroscopy it is important to achieve a minimum line width in the pulse-height spéctrum resulting from monoenergetic gamma rays. The contribution of the scintillator to the line-width broadening has been studied, and it has been demonstrated that it is the result of the nonlinear response of NaI(T1l) to electrons. A calculation of the intrinsic broadening due to this effect was made and combined with a semiempirical function describing the broadening due to photomultiplier effects to give a calculated value of the over-all line width. At 661 kev the width was calculated to be 5.5% for a 2.5-in.=-diam by 2-in.-high crystal, which is to be compared with experimental values of 6 to 8%. It is.pointéd out that no account was taken of broadening due to imperfect crystal or optical effects and that the broadening must depend upon source configuration and crystal dimensions, the effects of which have not been investigated. _ In a series of experiments investigating the light output of CsI(T1) under excitation by monoenergetic gamma rays (10 kev to 2.6 Mev) it has been demonstrated that the light output of CéI(Tl) is not a linear function of the energy of the incident gamma ray. The magnitude of the departure from linearity has been measured for the particular crystal employed. The data will permit a calculation of light output per unit energy. “ xvii Assembly of the Model IV gamma-ray spectrometer has been completed, with only final calibration of the positioner and readout system yet to be accomplished. A series of preliminary investigations of the system performance, utilizing a 9-in.-diam by 12-in.-long composite NaI(T1) scintillator as the gamma-ray detector, is in progress. These experi- ments will assist in optimizing final performance of the spectrometer system. Comparisons with calculations are being used as guides to further improvement. _ ' . A method of automatically processing pulse-height data obtained as -the output of a scintillation spectrometer in a spectrum of gamma rays . is being developed. The method makes no prior assumptions about the spectrum being examined. Semiempirical functions are used to fit the experimentally determined response of the spectrometer to monoenergetic gamma rays. Simple power function fits, as a function of energy, to the parameters of the response function are used to develop the response matrix. Solutions to the result-matrix equations are being investigated. Special techniques are necessary to get useful results. Numerical tests using a nonnegativity constraint have been made, and tests are in progress using a smoothing function. The possible use of neutron-sensitive semiconductor detectors constructed by depositing a thin layer of Li®F between two closely ad Jacent silicon-gold surface-barrier counters has been investigated . further. Experimental pulse-height spectra have been recorded for monoenergetic neutrons ranging from "slow" to 14.7 Mev. Preliminary expériments examining the discrimination of these counters against gamma- ray backgrounds have also been made. It is concluded that fhese counters may offer advantages of simplicity of construction and operation, small size, and reasonably good resolution of neutron spectra above 1 or 2 Mev. A disadvantage is the low detection efficiency; present counters give: 3.4 x 1072 for thermal neutrons and 0.94 X 10~® for 2-Mev neutrons, both at normal incidence. Methods for improving the detection efficiency are being explored. The interest in paired silicon=-gold surface=barrier counters to record the sum of the energies of the tritons and alpha particles e xviii — ~ 3 (3 \ produced in a LiG(n,a)T reaction as a measure of the incident neutron energy has prompted the calculation of the shape and width of the distri- bution in total energy of the pairs reaching the silicon. Some preliminary results have been obtained. The preliminary design of a neutron chopper spectrometer for the BSF to be located underground adjacent to the pool of the BSR has been completed. The spectrometer will have a flight path of about 10 meters, which, with the chopper rotor currently proposed, will give energy. resolutions ranging from ~3% for energies below 10 ev, through 10% at 100 ev, to ~30% at 1000 ev. A stack of high-pressure BF3 counters will serve as the detector, with pulses going to a 256-channel pulse-height analyzer. Readout will be in the form of punched tape coded for the Oracle, which will transform the time-of-flight data to neutron energy spectra. A spherical detector shield fabricated for use at the TSF is being modified to include temperature control apparatus. The detector will primarily be used for radiation spectral measurements, both for gamma rays and neutrons, but it will also be used for dose measurements. Work has continued on the experimental determination of a correction factor to be applied to thermal-neutron flux measurements with thin gold foils. Results are presented as a plot of the correction factor as a function of foil thickness. A comparison of the experimental.results with variocus calculations indicates that the experimefit’is in good agree- ment with the work of Dalton and Osborn. 11. Basic Shielding Studies - In the continuing analysis of the experiment for determining the spectrum of gamma rays emitted promptly in the thermal-neutron-induced fission of U235, all experimental work and the combination of data, together with the corrections for all count-rate sensitive phenomena, have been completed for all energy regions. The source-strength deter=- minations required for spectrometer efficiency measurements have been obtained either by absolute coincidence counting techniques or by use xXix of a 47 high-pressure ionization chamber, and the pair spectrometer efficiency and response to monoenergetic gamma rays have been expressed in a response matrix. The response matrices for the Compton and single- crystal spectrometers, two steps which must precede the final solution for the gamma-ray spectrum, have not yet been formulated. The "conditional" Monte Carlo sampling method has been used in a calculation of the energy spectra and angular distributions from an isotropic point source of gamma rays in an infinite medium in which - energy degradation was permitted. The purpose was to test the applic- ability of this technique to problems involving several space variables, , - since the technique had previously been successfully used in a one- dimensional phase space computation. Differential energy spectra and angular distributions were computed for two cases: a l=Mev source in water and an 8-Mev source in lead, both for penetrations up to 20 mean free paths. Comparison with existing moments-method results showed agreement to distances of about 10 mfp. The result of a calculation using the single~scattering approximation, with no exponential attenuation or buildup, has been compared with the result of a Monte Carlo calculation of the dose rate and number flux in alr from a monoenergetic gamma~ray source. The comparison shows that the simple approximation is generally adequate for unshielded detectors. The energy specfra and dose rates resulting from scattering of a gamma-ray beam incident on a homogeneous right-circular cylinder can be obtained from an IBM-704 Monte Carlo program called Grinder. A maximum of 20 detector positions is permitted. Results are given as energy- spectra histograms from which dose can be computed. An important feature of the program is that computing time, rather than number of histories, is an input parameter. The code can be used to investigate the dose rate in a cylindrical crew shield resulting from scattering in various parts of the shield. Measurements of the angular distribution of low-energy neutrons emerging from the surface of hydrogenous slabs indicate that the angular distribution not only follows the ¢(u) = 1 +./3 p dependence given by Bethe for pure paraffin but also is indepéndent of the poison content for the range investigated, the slab thickness (after spectral equilibrium is reached), and the angle of incidence of the input neutrons. Measure- ments for a non~slowing-down medium (lead) indicate that the angular distributioh is more strongly peaked in the forward direction when the medium is poisoned. A general purpose Monte Carlo reactor code, identified as the Oak Ridge Random Research Reactor Routine (05R), was developed for the IBM~-704 computer. Because the output of the code includes the velocity components of the neutrons after each collision, it is adaptable to the investigation of the angular distribution of neutrons across plane boundaries. An additional IBM=-704 code is therefore being developed which will process the neutron histories generated by the 0O5R code. Upon completion this code will be used to predict the angular distributions of neutrons emerging from hydrogenous slabs in a geometry analogous to the experiments réported in the preceding paragraph. A series of codes, each of which can be used for a line-of-sight calculation of a sepafate component of the radiation from a reactor shield, are being written for the IBM-704 computer. These codes will be applicable to the preanalyses of Tower Shielding Facility experiments and also to general shield design. Thus far, three codes have been com- pleted: a code for the primary neutron dose component, another for the pPrimary gammauray dose component, and a third one for gamma-ray buildup. Calculations are being made of the differential energy and angular cross sections for the emission of particles from complex nuclei when | the latter are bombarded by pions or nucleons in the energy range from 50 Mev to 30 Bev. These cross seétions are needed for the design of shields for rocket vehicles. The lower-energy phase of the program (incident energy range from 50 to 300 Mev) is nearly complete. A code is being developed for calculations of the transport of high=energy particles in matter within a broad energy range. The code, which will be applicable for space vehicle shielding calculations, will be developed to be as general as possible in order to obtain all the N s desired data. Preliminary plansrindicate that for simplicity and useful- hess slab geometry should be the first shield geométry used. 12. Applied Shielding Developments An extensive calculational program has been undertaken for estimating the neutron and gamma-ray dose rates which will ™e measured in the planned divided shield experiment for Pratt & Whitney Aircraft Company at the Tower Shielding Facility. The TSF compartmentalized detector tank will be utilized for the experiment as a crew compartment and the radiation " source will be the TSR-II encased in the Pratt & Whitney uwranium~-lithium hydride shield. This preanalysis is an attempt to predict the dose rates - by using basic principles and the best calculational tools presently - available. A summary of the approach is given., A study of the shielding properties of a depleted uranium and normal 1ithium hydride slab combination has been completed for Pratt & Whitney Aircraft at the Lid Tank. The experiment was designed primarily to examine the secondary gamma-ray production in the shield for the Pratt & Whitney 1ll-c reactor and to obtain pertinent information regarding the attenuation properties of the shield as a function of thickness and position of the uranium within the shield. 1In addition to the mockup measurements, the neutron and gamma-ray attenuation characteristics of IiH and LiH-U combinations were examined. - A series of experiments was performed at the LITSF to evaluate the - shield configuration proposed for a mobile power reactor (identified as the ML-1) being designed by the Aerojet-General Corporation. Thermal- | - neutron flux, fast-neutron dose, and gamma-ray dose measurements were made behind slab mockups of the proposed shield with several possible variations, and the post-shutdown decay characteristics of a mockup were also determined. The results are summarized. 13. Shield Design Preliminary consideration has been given to the design of a reactor shadow shield of optimum shape. The source is considered to be a point xxii N s L gsource of monoenergetic gamma rays, and the shield and source are assumed to be embedded in water. Only air-scattered gamma rays and those suffering a single collision in water and scattering toward the detector are con- sidered. Several simplifying assumptions are made to facilitate the calculation. iilllllI'..T xxiii 1“' PART 1. MATERIALS RESEARCH AND ENGINEERING l" 1. MATERTALS COMPATTBILITY Reactions of Columbium and Columbium Alloys with Low=Pressure Gases Both the fabrication and the use of columbium and columbium alloys are hampered by the presence of small amounts of gaseous impurities, especially oxXygen. Studies are therefore under way to obtain information on tolerable impurity levels in protective atmospheres, reaction rates, and reaction mechanisms for columbium and its alloys. Reactions with Oxygen The reaction rates of unalloyed columbium with oxXygen have been determined, as reported previously,® at 850, 1000, and 1200°C in the O, pressure range 3 X 1077 and 5 X 10™? mm Hg. The oxidation of unalloyed columbium at an O, pressure of 5 X 10™% mm Hg and 1000°C was found to proceed at a linear rate for a short period and then to change to a parabolic rate. The linear reaction rate was associated with internal precipitation of Cb0O, and the parabolic rates were associated with the formation of a Cb0O,; scale that was protective. A comparative survey of the oxdidation characteristics of numerous experimental columbium alloys has now been éonducted at 1000°C and an oxygen pressure of 5 X 10™* mm Hg. The materials tested included simple binary compositions in which some of the alloying additions weré thermo- dynamically more stable than LiOp, alloys in which some of the alloying - additions were solid solution strengtheners, and some advanced alloys which have been developed by other installations. The results of the. tests are shown‘in Table 1.1 in terms of the weight 1ncrease of the specimens due to oxygen pickup. The portions of the 300-min weight gain that acecrued linearly and parabolically were determined from the rate curves in order to obtain 1"ANP Semiann. Prog. Rep. April 30, 1960," ORNL-2942, pp. 3—7. estimates of the amount of internal oxidation and scale formation for each composition. The Weight increases for most of the alloys were greater than those for commercial unalloyed columbium under the same Table 1.1. Oxidation Rates of Columbium Alloys Exposed at 1000°C to Oxygen at a Pressure of 5 X 10™4 mm Hg Nominal Composition Weight Gain Weight Gained Weight Gained at (st %) in 300 min at Linear Parabolical ’ (mg/cm? ) Rate (%) Rate (%) High-purity Cb 2.20 50 50 Commercial Cb 1.30 30 70 Cb—1 Y 1.55 0 ' 100 Cr0.5 Zr 1.40 50 50 Cb—1 Zr 1.60 60 240 Cb—3 Zr 2.80 86 14 Ch5 Zr 3.55 80 20 Cb-0.25 Be 1..30 45 55 Cr0.50 Be 1.82 35 65 Ch—1.0 Be 2.20 70 30 Ch-0.25 Ce 1.05 0 100 Ch0.50 Ce 1.50 0 100 Cb—1 W 3.95 60 - 40 Ch—5 W _ 3.20 50 50 Cr-0.5 Hf 2.20 4.5 55 Cb—1 Hf 2.50 35 65 Cbh-3 Hf 2.30 60 40 Cb—1 Mo 3.43 50 50 Ch—5 Mo 1.76 35 65 Cb~7 Mo 2.54 40 60 Cb—3 Th 1..80 50 50 Cb0.25 Re 3.00 50 50 CrL.50 Re 2.80 50 50 Cb-1.0 Re 1.72 - 100 Cb—1 Re—2 Ce 2.7 50 50 Cb—2 Re—2 Ce- 3.44 60 40 Cbh—5 W—2 Ce 1.96 50 50 Ch-2 La~1 Re 3.70 0 100 Cr0.5 Al 1.00 0 : 100 Cb—-1.5 Al 1.20 0 100 Cbh2.5 Al 2.20 65 35 Cb—33 Ta—0.7 Zr (F-82)% 1.96 30 70 Cb—15 W—5 Mo—l Zr (F-48)* 1.26 100 Cb—10 Ti~10 Mo (D~31)* 3.10 100 Chr»0.9 Cr 2.4 0 100 ¥Commercially available alloys. test conditions. -However, an analysis of the reaction rates shows that alloying significantly influenced the type of oxidation which occurred. It may, therefore, be feasible to reduce the deleterious effects produced by slightly contaminated environmenté by a sultable combination of alloying agents. It was noted that elements such as Y, Ce, Al, La, and Cr promoted parabolic reaction rates, while elements such as Zr, Mo, Re, and Ti promoted iinear reaction rates. The alloys which exhibit parabolic reaction rates would tend to be more resistant to embrittlement than the alloys which oxidize at a linear rate, sinée the oxide is confined principally to the surface in the former case. Reactions with Air Oxidation studies of columbium and its alloys in low=pressure air have been conducted at a pressure of 5 X lO_"4 mm Hg and a temperature of 1200°C for exposure times of up to 3000 min. The materials tested to date have included high~purity Cb, Cb—7% Mo, Cb—5% Zr, and Cb—10% Mo—10% Ti! The reaction rates which were observed were linear or exhibited an incubation period up to 1000 min during which the reaction rates were not measurable. The average reaction rates for 300 min, assuming linear rates, are given in Table 1.2. The oxidation rate of' unalloyed columbium at 1000°C was 0.12 mg/cm?® in 300 min. Table 1.2. Oxidation Rates of Columbium Alloys at 1200°C in Air at a Pressure of 5 X 10™% mm Hg Composition Weight Gain in 300 min (mg/cm? ) High~purity Cb 0.19 Cb—7% Mo 0.28 Cb—10% Mo—10% Ti 0.65 Co—5% Zr 0.18 Reactions with CHy Tests have been conducted to determine the carburization rate of unalloyed columbium in methane, primarily as a method of adding carbon to specimens for aging studies. Sheet specimens were exposed to the gas at pressures between 1 X 10=% and 8 x 10~2 mm Hg and temperatures of 1200 and 1300°C. The reactions, which were assumed to be a result of carburization, proceeded at a rate of about 3 X 1072 mg/cm?® in 300 min. A slower average rate obsexrved in a test period of 180C min indi- cated that the carburized surface retarded the reaction. - Tentative Concluslons Regarding the Reaction Rates of Columbium and Its - Alloys with High-Temperature Gases For several of the binary alloys systems, it was observed that the reaction rates increased as the concentration of the alloylng element increased. Alloying elements that showed this trend were Zr, Be, Ce, and Al. Inasmuch as these elements form oxides that are more stable than columbium oxide, it is concluded that these elements increase the "gettering" ability of the alloy or produce morphological changes in the oxide scale. The fact that increasing the alloying element concenw- tration increased the extent of internal oxidation in zirconium, beryl=- 1ium, and aluminum-containing alloys suggests that "gettering" was enhanced. ‘ . - Alloys which exhibited only parabolic oxldation rates have been assumed to have formed a surface layer of Cb0O, as the principal reaction product. In accordance with oxidation theory, such reactions are, in - general, relatively independent of the oxygen pressure, and therefore alloys of this type would be preferred for exposure to oxygen-contaminated atmospheres. i For an alloy to be considered oxidation resistant, the reactor rate must be no greater than 5 x 10”2 mg/em® in 300 min. It is therefore readily apparent that the alloys discfissed here oxidize at a rate more than 10% times that for conventional oxidatlion~-resistant alloys. From F the standpoint of gas-metal reaction rates, greater promise is seen in the use of these alloys in carburizing or nitriding2 atmospheres. The reaction rate of columbium was 10 times slower with CH; than with oxygen, even though higher temperatures were used. Similarly the reaction rates of columbium and its alloys were about 10 times slower in air than in oxygen at the same pressure. -Compatibility of Columbium and Columbium~Zirconium Alloys with UQ, and UQ,~BeQ Several tests have been made to examine the compatibility of columbium=zirconium alloys with UO,. A compact, contalning pure columbium, several columbium~zirconium alloys, and pure zirconium allv in contact with UO, powder, was hot rolled and sectioned. The compati- bility specimens thus obtained were aged for 100 hr at both 2000 and 2400°F. Metallographic examination showed extensive reactlon of the pure zirconium with the U0, at both temperatures. However, no reaction was noted between UO,; and pure columbium or the columbium alloys which had 0.5, 0.75, 1.0, 2.0, or 5.0 wt % Zr. Columbium alloys of 10, 20, 30, 50, 70, 80, 90, and 95 wt % Zr are being made for further tests with both U0, and a U0,—65% BeO mixture. Effects of Oxygen Contamination and Subsequent Exposure ‘ to Lithium on the Tensile Properties of Columbium and Cb—1% Zr Alloy A series of columbium and Cb~1% Zr alloy sheet tensile specimens 0.04O in. thick were contaminated with oxXygen énd subsequently exposéd to lithifim for 100 hr at 1500°F. The oxygen was added at 1832°F at an oxygen pressure of 0.1 p. The results of the columbium tests were reported previously,3 and the data obtained recently for the Ch—-1% Zr alloy ere presented in Table 1.3. 2W. M. Albrecht and W. D. Goode, Jr., "Reactions of Nitrogen with Niobium," BMI~1360 (1959). 3"ANP Semiann. Prog. Rep. April 30, 1960," ORNL-2942, p. 7. Table 1.3. Effect of Oxygen Additlons and Subsequent Exposure to Lithium on the Room~Temperature Tensile Properties of Cb—1% Zr Alloy Yield Maximum Specimen Congzi%iztion Treatment Prior gi?ziéih Strength, Eigng?;;?n Depth of No. (ppm) to Test (psi) 0.2% Offset o () Attack _ PP P (psi) a8 (mils) 1 90 Heat treated 2 hr 38 690 35 625 14.0 2 290 at 2912°F in 37 280 33 265 9.0 3 510 vacuum 38 835 33 230 11.0 4 780 35 730 30 065 11.0 . 5 1150 35 780 28 645 14.0 6 85 Heat treated 2 hr 35 990 32 055 11.0 - 7 440 at 2912°F in 33 535 29 820 9.5 . 8 470 vacuum and aged 32 430 28 415 10.5 _ 9 1020 in argon 100 hr 35 775 29 310 12.0 10 1200 at 1500°F 32 730 27 210 12.0 ) 11 90 Heat treated 2 hr 32 505 28 680 13.5 0 12 425 at 2912°F in 32 715 28 795 8.5 0 13 450 vacuum and aged 32 320 28 405 11.5 0 14 1060 in Ydthium 100 hr 33 105 26 740 10.5 0 15 1610 at 1500°F 3). 930 24 135 13.5 0 16 280 Aged in lithium 63 715 58 800 6.0 0 17 870 100 hr at 1500°F 62 715 60 040 2.5 11 18 1050 46 815 46 815 1.0 14 19 1290 43 785 43 785 0.5 19 It may be seen in Table 1.3 that addition of up to 1200 ppm oxygen did not affect the room~temperature tensile strength or ductility of the Cb—1% Zr alloy before exposure to lithium. No corrosion or change in the tensile properties was observed when exposure to lithium was preceded by a thermal treatment at 2912°F for 2 hr. However, when no such treat- ment was given prior to exposure to lithium, the Cb—1% Zf alloy was gttacked, and an increase in tensile strength and a decrease in elonga= tion were noted. The results presented previously3 for unalloyed columbium indicated that similar treatment had little effeét of its corrosion resistance or tensile properties. 1In the case of specimen No. 16, no attack was observed, but the tensile strength was doubled and the elongation halved. The high strengths of specimens 16 fhrough 19 can be attributed to the effect of the oxygen contamination. Subsequent heat treatment at 2912°F apparently causes the strength of the alloy to decrease. : - The results obtained for the heat-treated Cb—1% Zr alloy are com- pared with the earlier results for similariy treé,ted columbium specimens in Fig. 1.1. These data illustrate the sensitivity of bxygen-contaminated columbium rtc‘> corrosion by lithium and the relative ir}sensitivity of | oxygen-contaminated Cb—1% Zr alloy to attack in the heat-treated con&i- tion. R ORNL—LR—DWG 50444 60 : Cb HEAT TREATED I IN ARGON FOR 100 hr AT 1500°F 50 >Y 2 . Cb=1% Zr = //{ ///’;EAT TREATED 's 40 7 ~T—IN ARGON FOR — < \ i . / 100 hr AT 1500°F I —‘-—-—A--- ------ g e s e e e e o ‘ g -/ -1y QO 30 —#"xs — . w Iz - N~Cb1 % Zr o i \ EXPOSED TO LITHIUM v '\ FOR 100 hr AT 1500°F 520 7w S TN ) \\ & \ " 10 | Cb EXPOSED TO LITHIUM /™ _ FOR 100 hr AT 1500°F e 0 0 500 1000 1500 2000 2500 OXYGEN CONCENTRATION (ppm) Fig. 1.1. Effect of Oxygen Contamination and Exposure to Lithium on the Room-Temperature Tensile Strength of Columbium and Cb-1% Zir Alloy. - Effect of Time on the Corrosion of Oxygen-Contaminated Columbium by Lithium A1l previous studies of the corrosion of columbium by lithium have Been conducted at 1500°F for 100 hr. Since the rate of attack should give some insight as to the mechanism involved, the variation in attack with time was measured. A sheet of 0.040=in.~thick columbium was annealed 2 hr at 2732°F in order to obtain stress-free material with equiaxed grains. Eight specimens 3/4-X 3 in. were prepared and approxi- mately 1000 ppm O, added to each at 1832°F and an O, préssure of 0.1 p. The specimens were then held for 6 hr at 2372°F under vacuum prior to exposure to lithium at 1500°F for 1, 2, 3, 16, 24, 100, 250, and 500 hr. The results of these tests are described in Table 1.4. Table 1.4. Effect of Time on Corrosion of Oxygen-Contaminated Columbium by Lithium at 1500°F . Oxygen Comeentration (ppm) Weight Attack (mils) Time (hr) - Change : Before Test After Test Change (mg/ln.z) Maximum = Average 1 820 820 ) 0 =-0.2 4.5 3 2 1500 1100 =400 0.2 5.5 3 3 1300 730 =570 0.5 5.0 2 16 1100 480 620 ~1.6 3.0 2 24 1000 320 -680 -1.8 5.0 3 100 1200 490 ~710 ~2.0 6.0 5 250 1300 240 ~1060 -2.0 3.5 2 500 1000 240 =760 -2.1 2.0 1 These results indicate that attack by the lithium proceeds very rapidly, as indicated by the specimens exposed for 1, 2, and 3 hr, and that lithium continues to getter oxygen from the columbium for very long times, as indicated by the oxygen~concentration data. The variation observéd in depth of attack as a function of time is not believed to be significant and is attributed teo variations in oxygen concentrations and other variables in the test procedure. The rapidity with which attack " by the lithium takes place suggests that the mechanism probably is not one which is diffusion controlled. Further studies will consider the effect of temperature and stress on the corrosion process. Effect of.Oxygen Concentration of Tantalum on Its Corrosion Resistance to Lithium It has been shown that the attack of columbium by lithium increases as the oxygen concentration of columbium increases.* In order to determine 4E, E. Hoffman, "The Effects of Oxygen and Nitrogen on the Corrosion Resistance of Columbium to thhlum at Elevated Temperatures,' ORNL~2675 (Jan. 16, 1959). 10 various additions were made to 0.040-in.~thick specimens, and the spécimens were‘feéfed'in-lithium for lOO.hr at 1500°F. Oxygen Waé added at 1832°F, and the specimens were held 2 hr at 2372°F to obtain a more‘finiform.distribution of the oxygen prior to testing in lithium. The results of the tests in lithium are summarized in Table 1.5, and the corroded specimens are shown in Fig. 1.:2. Table 1.5. Effect of Oxygen Additions to Tantalum on Its Corrosion Resistance to ILithium Oxygen Concentration Depth of Attack (ppm) (mils) 80 0 300 Complete 700 Complete 900 Complete It may be seen from the data of Table 1.5 that complete attack occurred when the tantalum contained 300 ppm oxygen. This is very much lower than the oxygen concentration in columbium that produced complete attack. The solubility of oxygen in tantalum is considerably less than the solubility of oxygen in columbium at the test temperature, and it is believed that this factor contributes to the more severe corrosion observed. The attack of the specimen containing 300 ppm oxygen appeared to occur along crystallographic planes rather than along the graln boundaries as is usually the case in heavily attacked columbium. However, the specimens containing 700 and 900 ppm oxygen showed both transgranular and grain boundary attack. The characteristic "corrosion front" was not evident, since, when corrosion was observed at all, it had already proceeded through the entire specimen. It should also be noted that, in the case of columbium, the attack could be observed metallographically on the sample in the as=polished condition, whereas tantalum did not show 11 PHOTO 52071 (¢) 0.07 % O, (d) 0.09 % O, E B woesg §E g E k| 2 e e | E F T 50X 1 s B miesp B R [ EE ] EEE Iu 01 |— = - = ‘:- © ~ b Fig. 1.2. Effect of Oxygen Concentration of Tantalum on its Cor- rosion Resistance Upon Exposure to Lithium at 1500°F for 100 hr. HNO3-HF-glyceria. 12 Etchant : any ev1dence of attack 1n .the as- pollshed condltlon, but etchlng readlly revealed the extent of the attack. o | R - " In order to determine the effect of heat treatment at 1500°F for 100 hr in the'absence of lithium, samples were cut from the four speci- mens and freated in'argon.' The metallographic appearance of these samples was similar to that of the as-received tantalum, and therefore the micro- " structures shown in Fig.'l.2 are assoclated with the exposure to lithium. Dissimilar-Metal Mass-Transfer Studies Earlier results® indicated that columbium and Cb—1% 7r alloy speci- mens tested in NaK contained in type 316 stainless steel capsules picked up carbon and nitrogen by mass transfer from the stainless steel. Metal- lographic examination of columbium and Cb—1% Zr alloy test specimens after exposure to NaK in type 316 stainless steel capsules at 1700°F for lOOO hr revealed two surface layers approx1mately 0.001 in. in total thlckness. X-ray diffraction analyses have identified the layers as CbC and CbjN. Room-temperature tensile tests on these specimens indicate that the in- crease 1n carbon and nitrogen concentration-is accomplished by an in- crease in the tensile strength and a corresponding decrease in the ductility of both the columbium and the Cb—1% Zr alloy. In order to determine the extent to which carbon and nitrogen diffused into the 0.040-1in. -thick test specimens, approximately 0.002 in. was machined off both sides, and the specimens were. chemically analyzed. The re- sults of these tests are shown in Table 1.6. It would appear that, in addition to the formation of wvery thin carbide and nitride layers, there was also bulk contamination which would tend to explaln the marked ' changes in tensile properties which were observed previously.5 S"ANP Semiann. Prog. Rep. April 30, 1960," ORNL-2942, pp. 9—12. 13 Table 1.6. Carbon and Nitrogen Analyses of Columbium and Cb—-1% Zr Alloy Specimens Exposed to NaK in Type 316 Stainless Steel Capsules at 1700°F for 1000 hr Specimen Number Carbon Nitrogen p Specimen Condition Content Content and Type (ppm) (ppm) A-1696, Cb—1% Zr alloy Before test 190 160 After test 420 850 After layers machined off 310 520 A-1697, columbium Before test 92 80 After test 510 2000 After layers machined off 150 590 A-1697, Cb—1% Zr alloy Before test 128 160 After test 770 1500 After layers machined off 290 1300 A-1713, Cb-1% Zr alloy Before test 128 160 After test - 910 1200 After layers machined off 230 Tests of Structural Materials in Boiling Potassium Refluxing Capsule Tests Refluxing capsule tests of iron-base, nickel-base, and cobalt-base materials in boiling potassium have been conducted because of interest in boiling potassium as a working fluid for electrical generation aboard space vehicles,© These screening tests have all been run in the tempera- ture range 1500 to 1600°F. In the initial tests, the test coupons were suspended in various regions of the ca.psule.7 It was suspected, however, that the largest amount of metal solution (corrosion) occurred where the condensing rate was highest, and therefore in subsequent tests the corrosion specimens were suspended in these regions. This was achieved by lining the inside of the refluxing capsules with tight-fitting, sleeve- type insert specimens, as shown in Fig., 1.3. The results of 500C-hr ®"ANP Semiann. Prog. Rep. April 30, 1960," ORNL-2942, p. 13. ""ANP Semiann. Prog. Rep. April 30, 1960," ORNL-2942, p. 1l4. 14 UNCLASSIFIED ORNL-LR-DWG 502144 o e o - LT AL N ATY Y eSS N N 7 Iy 4 Y i Y15 75 Vi 7 i i , ? , H & 15 BN I N < BN " Al ?g :\‘ i . PR n - tag TN 14! M-t N f’ “\ LN i W Ay . LN gir e .k I % R M B N & i3 B . NATFA -~ 4] 7~ i N S3 ///\ V) ~ Fig. 1.3. Refluxing System for Testing Structural Materials in Boiling Potassium. tests of specimens of type 316 stainless steel and Haynes alloy No. 15 (50% Co—20% Cr—15% W—10% Ni—2% Fe, nominal composition) are given in Table 1.7. A greater loss of'material, as determined by weight-change measure- ments, was found to have occurred on the vapor-zone samples of the Haynes alloy than on the stainless steel. No attack was found metallog;aphically on the Haynes alloy vapor-zone specimens, and thus uniform removal of material is ihdicated. A slight roughness was found on the stainless steel that consisted of depressions less than 0.5 mil deep at the grain boundaries. 15 Table 1.7. Weight Changes of Structural Metal Specimens After 500 hr of Exposure to Boiling Potassium in Refluxing Capsule Tests Approximate . . . s . Temperature* Weight Specimen Material Position of Specimen ) Change of Specimens /. 2 ooy (mg/in.?) Type 316 stainless Top of vapor zone (cap) 1475 -1.1 steel Liquid~vapor interface 1490 +1.4 Iiquid region (boiler) 1500 +2.4 Haynes alloy No. 25 Top of vapor zone (cap) 1475 ~4 .6 Liquid~vapor interface 1490 +1.7 Liquid region (boiler) 1500 +2.4 ~ *Temperatures were approximated by comparing the outer wall temperature readings with those recorded from a thermocouple well inside a similar refluxing capsule. These two tests were conducted in capsules that had thermocouples attached on the outer surface, and, since there was a substantial tem- perature drop across the container wall, estimates of potassium tempera- tures were subject to considerable error. Consequently the third capsule wag designed with a thermocouple well and a movable thermocouple, as ‘shown in Fig. 1.3, in order to more accurately determine the temperatures in various regions of the capsule. The specimen for the third test was constructed of Inconel, and the test was run at a boiler temperature of 1600°F for 1000 hr. The con- figuration of the test unit and data on weight change as a function of specimen position are presented in Fig. 1.4. The weight-change déta show results similar to those observed in previous refluxing capsule tests of other materials. A greater weight loss (-10 mg/in.?) was found for the Inconel located in the vapor region than for similar specimens of cobalt- base and iron-base materials. The grain-boundary attack to a depth of approximetely 10 mils found on the Inconel cap specimen is shown in Fig. 1.5. In summary, type 316 stainless steel 1s the most corrosion resistant of the materials tested in boiling potassium. 16 UNCLASSIFIED FLAT CAP SPECIMEN 30 17 M~ o < M Te ] o 3 - ® o - 1 ™ o N a1 ® £ _ \ ! / £ _ ® o & \ E O w . o 2 - g : I ) - I : o \. o .\ m I 4 A A A A 4 r % A 3 r A A 1 F A A 0 A £ A 1 } \\\\\\\\M\ L A \\ A A \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ \\\ e L o o o A e A e o o o “ SO /7 NN 4_7 /,# N ,wr/ 4_/ DN ////W//////// //_7/////////47 T \ [ ] ] | | I | I A1 7 OO N o o SN NSNS SIS S NN R NN SN NN NN L LI T o T s A P S A g T T T S A A \\ A e T S S e A 3 1] @ 0 Q T ) o > 3 7 -— T A U o H o ‘U by ‘HOdVYA T ‘w9 ‘amnonn s 5 I - Test System Configuration and Weight-Change Results for Inconel Specimen Tested in Refluxing Capsule for 1000 hr at a Boiler Temperature of 1600°F. MOVABLE THERMOCOUPLE/‘ Fig. l.4. UNCLASSIFIED Y-36914 - con > < o ' = = @) .1 * ©» 008 - < ) . - Fig. 1.5. Vapor Side of Inconel Cap Specimen Tested in Refluxing Capsule. Etchant: aqua regia. Boiling Loop Tests The examination of the first type 316 stainless steel loop, which operated for 200 hr, showed that there was very little corrosion; there- fore a second loop test of 3000 hr duration was run. The test con- figuration and conditions of the second test are shown in Fig. 1.6. The most significant modifications made in the test system® for the second test were (1) enlargement of the boiler section pipe diameter to permit more heat input and as a result to obtain & higher wvapor flow rate, (2) installation of & pressure sensor to continuously monitor the potassium vapor pressure, and (3) suspension of the type 316 stainless 8" ANP Semiann. Prog. Rep. April 30, 1960," ORNL-2942, p. 15. 18 UNCLASSIFIED ORNL-LR-DWG 47744 12-in. HEATER: ' 1545°F __INSULATION HOT AIR—=— _-Ya-in- SCHED 40 PIPE PITOT TUBE = 1600°F INSULATION —1" & . THERMOCOUPLE ' LOCATIONS o F 1600°F - COOLING FINS ° \ T . 6 ft 12-in. HEATER - COOL AR d S = 15'80°F PINNED (N 1310°F fi .:. T~ LIQUID LEVEL LOOP LiQuiD LEVEL—__E ' 8-in. HEATER ‘ 1410°F { f?' —1615°F - % : - 21 4 . ‘ ¢ 2-in. SCHED 40 5in. 8-in. HEATER PIPE PRESSURE SENSOR 12-in. HEATERS || 8-in. HEATER NaK FILLED TUBE TO TENSILE SPECIMEN PRESSURE TRANSMITTER COLD TRAP - Fig. 1.6. Thermal-Convection Loop for Testing Type 316 Stainless . Steel in Bolling Potassium. steel sheet tensile specimens in the loop as shown in Fig. 1.6. The tensile specimens were used to determine the room-temperature mechanical properties“Of the alloy after exposure to potassium during the loop test. The calculated potassium mass flow rate was 200 g/min‘and the vapor flow rate 50 ft/sec. The calculation was based on the air-cooler heat balance, and the assumption was made that no liquid potassium carry- over‘from the boiler to thevcooler occurred. | The results of weight-change measurements on the'tehsile-test specimens are given in Table 1.8. With the exception noted, the weight - | 19 Table 1.8. Welght Change and Mechanical Property Data on Type 316 Stainless Steel Sheet Tensile Specimens Followlng Exposure 1n a Boiling~Potassium Loop for 3000 hr Mechanical Property Data Tempera~ Weight S?figfgen ey ture Change Yield Tensile Elongation (°F) (mg/in.?) Strength Strength in 2-in. | (psi) (psi) Gage (%) , x 103 x 103 _ 1-A Boiler (liquid) 1600 -11.0 23.9 78.6 56 1~B Boiler (liquid) 1600 -10.6 23.8 78.0 58 2~A Boiler (liquid-vapor 1580 -8.9 23.7 79.3 57 interface) _ 2-B Boiler (liquid-vapor 1580 8.7 . 22.0 80.7 56 ‘ interface) 3 Vapor (hot leg) 1600° +8.2° 21.8 79.5 61 4 Vapor (cooler) lA‘BO‘b 4.1 22.4 80.5 56 5 Cold leg (1iquid) 1310° +29.6 32.3 86.2 43 6 Cold leg (liquid) 1410° +18.1 29.1 86.5 24 Control-l Vacuum 1600 26.7 80.1 55 -2 Veacuum 1480 30.7 90.9 48 =3 Vacuum 1410 30.6 90.8 48 =4 Vacuum 1310 30.5 88.4 52 aSpecimen dimensions: 5 X 0.25 x 0.040 1in. bExterior wall temperatures; other temperatures were determined by means of thermocouples proJjecting into the center of the pipe 1n wells. 11 cSpecim.en mechanically damaged during-tést by "bumping; weight. damage affected its losses occurred on the specimens from the hot-liquid and vapor regions of the loop, and the weight gains were observed on the specimens in contact with liquid potassium in the cold-leg section of the loop. The surfaces of the specimens that were suspended in the boiler and cold leg of the loop are shown in Fig. 1.7. The specimen from the boiler region showed evidence of slight surface dissolution, whereas the cold-leg specimen was covered with a mass-transfer deposit to a depth of approxi- mately 3 mils. The maximum attack observed in the loop occurred on the pipe wall of the boiler in the vicinity of the liquid-vapor interface and was in the form of small subéurface voids to a depth of 2 mils. 20 a® UNCLASSIFIED Y-36737 NI PLATE INCHES }ififif' i Fig. 1.7. OBurfaces of Type 316 Stainless Steel Specimens from (a) Boiler (Liquid at 1600°F) and (b) Cold-Leg (Liquid at 1310°F) Regions Pollowing 3000 hr of Exposure in Boiling Potassium—Type 316 Stainless Steel ILoop Test No. 2. Specimens were nickel plated following test to facilitate metallographic preparation. Etchant: glyceria regia. The results of the room tempersture tensile tests are slso given in Table 1.8. The control specimens were heat treated in vacuum for 3000 hr in order to provide a base for evaluating the combined effects of heat treatment and exposure to potasgsium. No large differences in the mechanical properties of these specimens were found; however, the yield strengths of the specimens from the hotter regions are slightly lower than those of the specimens from the cocler regions and the control specimens. X-Ray analyses of the surfaces of the tensile specimens from the boiler section of the loop revealed the presence of some alpha-iron, which is attributed to preferential leaching cf austenite stabilizing elements, such as carbon and nitrogen. The cold-leg specimens were found to be covered with chromium carbide (Cr;3Cg). Chemical analyses 21 of the crystalline deposits from the cold-leg specimens indicated high chromium and nickel content. Chemical analyses were also made of turnings machined from the inside pipe wall in the cooler region and the cold-leg region below the cooler. Turnings machined in 3-mil increments to a depth of 9 mils were taken from each region in order to determine whether concentration gradients existed in the wall as a result of the exposure to potassium. The turnings were analyzed for Fe, Ni, Cr, Mo, and C. The only signifi- cant modifications in composition were a slight depletion in chromium content of the surface of the pipe in the cooler section and a considerable change in the carbon content of the pipe in the cooler and cold-leg sections. The variation in the carbon contents is given in Table 1.9. . Considerable carbon depletion occurred in the cooler, and carbon enrich- ment occurred in the cold leg. The carbon depletion noted in the chemical results was further substantiated by the metallographic appearance of the depleted regions. The slight changes in mechanical properties noted in Table 1.8 may be explained in terms of the alterations in composition effected by mass transfer. Table 1.9. Carbon Content of Turnings Machined from the Wall of the Type 316 Stainless Steel Loop Following 3000 hr of Exposure to Potassium - Location of Turnings Carb?EtC%?tent i Top of cooler (1545°F) 0%=3 mils 0.03 . 36 mils 0.02 69 mils 0.02 Cold leg (1310°F) 0*-3 mils 0.55 3~6 mils 0.37 6—9 mils 0.25 - As-received pipe 0.08 *Inner surface of pipe wall. 22 - 2. AGING STUDIES OF COLUMBIUM-BASE ALLOYS Wrought Material Specimens of a special heat of Cb—1.25% Zr alloy (S233C-1) which contained less than 350 ppm total interstitials were énnealed for 2 hr at 1600°C énd then aged under thrée different conditions. Two specimens were aged in an argon-filled columbium capsule, two were aged in an efiacuated guartz capsule, and two were wrapped in tantalum foil before aging in an evacuated quartz capsule. All specimens were aged 100 hr at 927°C and then tensile tested at 927°C. The tensile-test results are presented in Table 2.1, While it is evident that there are slight differences in the strengths and elongations of the specimens aged under the three different conditions, it may be noted that the strengths are 33 to 65% higher and the ductilities are 50 to 65% lower than for the as-annealed material, indicating aging. Table 2.1. Effect of Various Containers on the Tensile Properties of Aged Cb~1.25% Zr Alloy (S23SC-1) Ultimate ] Elongation (%) Tensile Specimen Tensile Yield Condition Strength Streg%th In ~ In ~ (psi) ~ (psi 2-in. Gage 2 1/2-in. Gage Annealed 2 hr at 1600°C 24 945 11 464 12 9.5 25 390 14 270 8.5 7 Aged 100 hr at 927°C in 34 870 33 095 4.5 4.5 argon-filled columbium 32 165 28 980 e 2.5 capsule Aged 100 hr at 927°C in 39 980 38 940 4 3 quartz capsule 33 715 30 505 5 4.5 Tantalum wrapped and 38 960 36 645 5 4 aged 100 hr at 927°C 33 920 31 415 3.5 3 in quartz 23 In the current aging studies, nine commercial heats of Cb—1% Zr alloys have been tested. Of these nine heats, five showed an aging response and four did not, as indicated in Table 2.2, which also lists the heats and gives their chemical analyses. In the four alloys that did not exhibit aging, the oxygen content was high relative to the other impurities. When the impurity concentrations are compared as ratics of oxygen to carbon and oxygen to nitrogen, a definite correlation can be seen, the only exception being the oxygen-to-carbon ratio of heat S8FW. Another way of presenting the same data is shown in Fig. 2.1 where the oxygen-to-zirconium ratio is plotted against the nitrogen-to-zirconium . ratio. As may be seen, the alloys which age can be separated from the - nonaging alloys by a transition line or zone. A similar transition line - can be drawn on a plot of the oxygen-to-zirconium ratio versus the carbon-to-zirconium ratio, except for the data for heat S8FW, An analysis of these data indicates that a heat which has a high oxygen content relative to its carbon or nitrogen content will not exhibit aging. Therefore it should be possible to add oxygen to a heat which is known to age and, by this addition, to inhibit the aging reaction. Table 2.2. As-Received Chemical Composition of Columbium-Zirconium Alloys and Remarks on Effects of Annealing at 1600°C for 1 hr and Aging at 927°C - Chemical Composition ' le Ratio of Ratio of . fleat No. o, C Zr 0y to C 0, to N, Remarks (ppm) (ppm) (ppm) (wt %) ) PFYU 145 160 210 1.07 0.7 0.9 Definite aging PGVE 106 115 120 0.98 0.9 0.9 Definite aging S4KW 200 200 300 1.0 0.7 1.0 Definite aging S238C-1 120 85 140 1.25 0.9 1.4 Definite aging S8FW 260 160 70 0.75 3.7 1.6 Definite aging S15EC 900 140 170 0.76 5.3 6.4 No aging PGTF 320 46 120 1.14 2.7 7.0 No aging S16EC-1 1300 180 190 0.60 6.8 7.2 No aging S24EW 900 100 500 0.77 1.8 9.0 No aging 24 UNCLASSIFIED ‘ 042 ORNL-LR-DWG 48878R In order to test this hypothesis, ® ® HEAT S24Ew | SISEC various amounts of oxygen were & 040 added to heats S23SC-1 and PGVE. '_ . . E NO AGING OCCURS 7 These heats were then annealed 0.08 - S 7 2 hr at 1600°C, aged for 100 hr z HYPOTHETICAL 4 ‘ 2 006 TRANSITION LINE .~ at 927°C, and tensile tested at N 7 o /,’ 927°C. The tensile-test results fJQO4 ‘,’/ S8FW—] are presented in Table 2.3 and o PGTF 7 . i o o AGING OCCURS Fig. 2.2. If the "aging effect" 5 002 7 SAKW— o o « 7 PGVE ¢ PFYU can be defined as the difference 7 * s . ol” $235C-1 between a property of the material 0 0.01 0.02 RATIO OF N, TO Zr CONCENTRATION (ultimate tensile strength, yield strength, or elongation) in the Fig. 2.1. Plot of Ratios of Impurity Concentrations to Zirconium Content for Eight Columbium—Zirconium property in the aged condition, Alloys Showing Transition Line Between . Heats that Age and Those that Do Not i1t can be stated that the Age. addition of oxygen to the Cb—1% annealed condition and that Z2r alloy reduces the aging effect in proportion to the amount of oxygen added. It is postulated that nitrogen and carbon react with the zirconium in the alloy to form compounds which will precipitate under proper conditions of time and temperature to cause aging. It is believed that oxygen may possibly have the effect of increasing the solubility of the zirconium nitrides and/or carbides in the alloy and thus inhibit aging. Another possibility is that oxygen will react.preferentially with part of the zirconium and will precipitate in a form which will tend to nullify the effects of the nitride and/or carbide phases. Exploratory tests were conducted to determine the carburization rates of unalloyed columbium in methane as an aid in the preparation of specimens for aging studies. The data from these tests are reported in Chapter 1 of this report. Studies have been continued on the effect of surface contamination on the aging behavior of Cb-1% Zr alloys. Three heats of Cb-1% Zr alloy 25 9¢ Table 2.3. Effect of 0, Contamination on the Aging Reaction in Heats PGVEa and SZBSC—la bf Cb—1.25% Zr Alloy As Annealed After Aging 100 hr at 927°C Aging Effect . . Change in . Heat Total O, O1vi™a% vi14 Rlongation Total 0, Orvi™% yie1q Elongation Ultimate ClorgS in Tensile Tensile . Yield Contentb Strength Contentb Strength Tensile (ppm) Strength (psi) (ppm) Strength (psi) Strensth Strength (psi) P PP (psi) g (psi) (psi) S235C-1 120 25 167 12 867 10.5 120 39 980 32 220 4.5 +14 823 +19 350 315 23 365 12 875 g.5 347 34 270 30 944 6.0 +10 905 +18 069 329 23 055 15 230 6.0 339 32 430 27 795 5.5 +9 375 +12 565 403 26 475 14 990 6.5 417 34 165 29 075 4.5 +7 690 +14 085 586 26 730 15 120 3.5 595 34 270 22 845 7.0 +7 540 +7 725 777 27 865 14 755 g.5 g00 29 690 17 695 g.5 +1 825 +2 940 PGVE 110 24 610 10 445 8 110 33 920 28 440 6 +9 310 +17 995 620 32 315 15 820 12 650 29 846 14.5 -2 475 ®See Table 2.2 for composition. These 0, values were determined by weight change and represent the 0, content before aging. UNCLASS!IFIED ORNL-LR-DWG 49571R2 T T (x103) T ; Cb-1.25% Zr ALLOY \ HEAT NO. S23SC -{ , 36 * ANNEALED 2hr AT 1800°C , AGED 100hr AT 927°C TESTED AT 927°C 32 5 - ® ® 28 . AS AGED no B < YIELD STRENGTH, 0.2 % OFFSET (psi} 20 \ N S ~ 16 AS ANNEALED fi o o | L 4 Q 12 0 004 002 0.03 004 005 006 007 RATIO OF O, TO Zr CONCENTRATION Fig. 2.2. Effect of Oxygen Con- tamination on the As-Annealed and As- Aged Yield Strength of Heat S23S5C-1 of Cb-1.25% Zr Alloy. (PFYU, PGTF, and PGVE) were used to test the effect of machining off the outer layers of the tensile specimens after aging. Two 0.040-in.-thick and two 0.045-ifi.-thick speéimens were stamped from sheets of each heat,. These were vacuum annealed for 2 hr at 1600°C (2912°F) and then aged in quartz for 120 hr at 927°C (1700°F). The 0.045-in. specimens were then machined on both sides to 0.040 in. All specimens were tensile tested at 927°C. The results of the tensile tests are given in Table 2.4. The machined specimens of heats PFYU and PGVE showed slight decreases in Table 2.4. Effect of Surface Contamination Removal on Tensile Properties of Aged Columbium-Zirconium Alloys Alloy . . Ultimate . Elongation Heat Tengllg.igecimen Tensile Strength tield S?gength in 2-1in. No. onaition (psi) (psi Gage (%) PFYU As annealed 36 400 11 Aged, unmachined 52 000 5 Aged, unmachined 54 000 : 5 Aged, machined | 46 520 46 680 - 9.5 Aged, machined 47 680 46 470 10 PGTF As annealed ‘ 35 500 25 640 10 Aged, unmachined 30 220 14 890 - 14 Aged, unmachined 30 610 17 165 14 Aged, machined - 29 740 19 355 - 12 Aged, machined 30 745 21 975 11 PGVE As annealed 24 610 10 445 8 Aged, unmachined 36 200 30 785 5 Aged, unmachined 36 200 31 315 6.5 Aged, machined 33 065 25 605 7 Aged, machined 33 092 25 855 6 ¥A11 specimens annealed 2 hr at 1600°C. Aged specimens were heat-treated 120 hr at 927°C in evacuated quartz capsules. All specimens tensile tested at 927°C. 27 tensiie strength, whereas the tensile strength of heat PGTF did not change as a result of machining. However, for heats PFYU and PGVE, the ultimate tensile strength was significantly higher in the as-aged and machined condition than in the as-annealed condition. Only heat PFYU showed an increase in elongation as a result of machining. Fusion-Welded Material Aging studies on welds of Cb-1% Zr alloys have been continued, with primary emphasis on the bend characteristics of aged welds. In addition to bend tests, some all-weld-metal tensile tests have been performed, and welds with various degrees of contamination have been evaluated. A1l the test welds have been made in an inert-atmosphere chamber that is an improved model of the one described previously.l The major changes consist of a larger vacuum system and an improved welding carriage. The ovef-all procedure for test specimen preparation is described below: 1. Shear 0.060-in. (nominal) sheet specimens to proper size. 2. Clean specimens thoroughly in acetone. | | 3. Butt specimens together in welding carriage. 4. DPlace carriage in chamber and carry out the following purging procedure: ' a. evacuate chamber to 10~° mm Hg, b. back-fill chamber with high-purity helium, c. re-evacuate chamber to 10~% mm Hg, d. refill with high-purity helium. 5., Make square-butt fusion weld .in specimen. 6. Shear welded specimen to appropriate size. 7. Wrap specimen in tantalum foil and encapsulate in quartz and evacuate. 8. Age specimen. 1"ANP Semiann. Prog. Rep. April 30, 1960," ORNL-2942, Fig. 2.4, p. 24. 28 Drawings of typical tensile and bend specimens and the welding conditions for producing them are shown in Fig. 2.3. For the bend specimens, the full material thickness is used. For the tensile speci- mens; the longitudinal axis of the weld coincides with that of the specimen, the gage length is machined to a width such that it is all- weld metal, and both surfaces of the specimen are machined off to a final thickness of 0.040 in. All data obtained in this reporting period have confirmed the previously reported2 bend behavior of welds aged at temperatures ranging from 1500 through 1800°F. Preliminary room-temperature tensile tests have been run using subsize all-weld-metal specimens of the type described in Fig. 2.3. The tests were all made on specimens from heat S23SC-2. This material was found by chemical analysis to contain 120 ppm 02, 85 ppm N, i4O ppm C, and 1.20 wt % Zr. The results of these tests are presented in Fig. 2.4 as yield and tensile strengths vs aging time at 1700°F. The shapes of these curves indicate the presence of an aging reaction, but the limited number of tests does not permit the determination of exact values for these properties., These data show that fhe most severe aging occurs between O and 100 hr at 1700°F. Previous bend-test data® are in agreement with these results. Because it is quite difficult to obtain an accurate measurement of elongation on the subsize tensile specimens used here, ductility values are not used as indicators of agihg. This does'not appear to be a detriment, however, since both the tensile and yield strengths are suitable indicators, as shown by these tests. | In conjunction with the weld-aging studies, a cooperative weld- testing program was agreed upon by Pratt & Whitney Aircraft bivision, CANEL, and the Oak Ridge National Laboratory. In this program welds were made in the same heat of material (PGTF) at both Laboratories. Samplés were then exchanged and suitable tests were performed on them. The results of the CORNL tests are summarized in Table 2.5. The data indicate that, if weld specimens are aged at 2200°F for 1 hr, they will 2Ibid., Fig. 2.11, p. 35. 29 UNCL ASSIFIED ORNL-LR DWG 53610 TYPICAL WELDING CONDITIONS TYPE OF SPECIMEN CURRENT SPEED WIDTH OF WELD BEND 100 amps 5 in./min, ~ e in. TENSILE 170 amps 12 in./min. ~ %5 in. ALL DIMENSIONS ARE IN INCHES 4D 0.250 DY) / & N T .00 | | 3 —] en—3/ 1,50 - A6 AN ) - B~ . —tf- 3 TENSILE BEND Fig. 2.3. Drawings of Typical Tensile and Bend Specimens and the Welding Conditions for Producing Them, remain ductile after subsequent aging at 1500°F for 100 hr. In order to check further as to whether the weld aging observed at ORNL was merely a surface contamination effect, two specimens (8 and 9) were aged for 100 hr at 1500°F and then 0.002 in. was machined off each surface before bending. These specimens wefe brittle, but they did bend somewhat further than unmachined specimens. The data of Table 2.5 indicate that the aging reaction is more than just a surface effect, as shown by the fact that specimens whose surfaces were machined were still brittle. Also, the data show that the selection of a suitable preaging anneal, in this case 1 hr at 2200°F, may avert the detrimental effects of subsequent aging. This may prove to be a most important practical consideration if it develops that it is impossible to avoid aging by adjustment of the chemical composition of the alloy. The weld aging beha#ior of three high-purity special heats of electron-beam-melted columbium-zirconium alloy which were prepared at 30 UNCLASSIFIED {(x103) ORNL- LR-OWG 53498 66 - 62 / S n 58 54 ULTIMATE TENSILE STRENGTH ({psi) 50 46 LA \ YtELD STRENGTH {psi} )\ -—\\ -\ -‘\fl‘b 0 - 10 20 30 40 50 60 70 80O . 30 100 AGING TIME AT 1700°F (hr} Fig. 2.4. Influence of Aging Time at 1700°F Upon the Yield and Tensile Strength of All-Weld-Metal Specimens of Cb—1% Zr Alloy from Heat S235C-2. 34 31 Table 2.5. Results of Weld Aging Tests on Specimens from Heat POTF™ . Room- Specimen $£:§;2§n Preaging Aging Postaging Temperature No. at Treatment Treatment Treatment Bend~Test Behavior 1 P&wWw Held 100 hr at . Brittle 1500°F : 2 P&W Held 100 hr at Brittle 1500°F 3 P&wWw Held 1 hr at Held 100 hr at ‘ Duectile 2200°F 1500°F 4 PA&W Held 1 hr at Held 100 hr at Ductile ' 2200°F 1500°F , 5 ORNL Ductile 6 ORNL Held 100 hr at Brittle 1500°F 7 ORNL Held 100 hr at Brittle 1500°F 8 ORNL - Held 100 hr at 0.002 in. machined Brittle : 1500°F off surfaces 9 ORNL Held 100 hr at '0.002 in. machined Brittle _ 1500°F off surfaces 10 ORNL Held 1 hr at Held 100 hr at Ductile 2200°F 1500°F 11 ORNL Held 25 hr at Ductile 1700°F 12 ORNL Held 1 hr at Held 25 hr at Ductile 2200°F 1700°F ®See Table 2.2 for composition. ORNL has been studied. Welds were made in specimens from each heat and then aged at 1600°F for various lengths of time. The ductility of the aged welds was evaluated by bend testing. The results of this study are presented in Table 2.6. Previous data® on aging at 1600°F showed that specimens became fully brittle in 25 to 100 hr of aging. Thus, it may be seen that this electron- beam-melted material is considerably less susceptible to aging than the alloys tested previously. It seems reasonable to ascribe the relative insensitivity of these specimens to aging to their high purity. The results of chemical analyses of the three heats are presented in Table 2.7. It is interesting to note that heat No. 37, the heat'purest from the standpoint of N, and O, contamination, was also the heat which showed, 32 Table 2.6. Effects of Aging at 1600°F on Bend Behavior of Special ORNL Electron-Beam-Melted Columbium- - Zirconium Alloy Heat Zirconium Content Aging Time Room-Temperature No. (%) (hr) Bend Behavior 24 3.95 Ductile Ductile ’ 10 Ductile - 25 Ductile 50 Slight cracking in ] heat-affected zone i 100 Ductile | 250 Ductile 34 1.06 2 Ductile 5 Ductile 10 Ductile 25 Ductile ’ 50 Slight cracking in ' . heat-affected zone 100 Ductile 250 Ductile - 500 Ductile - 37 0.79 2 Ductile = 5 Ductile e 10 Ductile with = cracking 25 Brittle 50 Ductile 100 Ductile 250 Ductile 500 Ductile 33 Table 2.7. Chemical Analyses of ORNL Electron-Beam- Melted Columbium-Zirconium Alloys Chemical Composition - Heat No. Zr (wt %) N, (ppm) O, (ppm) C (ppm) 24 3.95 34 45 120 34 - 1.06 37 70 90 ) 37 0.79 10 26 110 comparatively, the most embrittlement upon aging at 1600°F. The reason for this behavior is not immediately obvious, but it does appear that even minute amounts of impurities may contribute to the aging of weld metal. ‘ A metallographic study was made of the welds used in these tests, and the microstructures of the welds that showed aging reactions were compared with those of the ductile welds. The samples for metallographic examination were prepared by a staining technique that beings out pre- cipitated oxides, nitrides, and carbides in colors gquite distinct from that of the matrix. A standard etching procédure was also used. ' Previous metallographic examinations had indicated that welds which showed an aging reaction by their brittle behavior also showed a definite difference in microstructure when compared with unaged welds. This ' ;\ difference is evident as small, but noticeable, areas of precipitate : 51‘~ phase in the matrix and at grain boundaries of aged and brittle welds. When no aging_reactidn has occurred, either in under- or overaging, the precipitate is absent. In the present case, absolutely no differences could be seen in the specimens which were studied. This indicates that some aging may take place even though it cannot be detected by the light microscope. In an attempt to obtain additional information on the aging be- havior of wrbught material contaminated with oxygen, two welds were 34 ' - contaminated with oxygen and aged. The welds were made in the same heat of material (XM-339) containing 120 ppm O;, 170 ppm Ny, 190 ppm C, 1 ppm Hy, and 0.87% + 0.05% Zr. One weld was contaminated before welding and one was contaminéted after welding. The final 0, content of 621 ppm, as calculated by weight change, was identical for both welds. The results of this aging study are shown in Table 2.8. The behavior of the material contaminated before welding was poorer than that of the weld contaminated after welding, which, however, also exhibited generally erratic brittle behavior. The values of the Op~to-N, and O,-to-C ratios are important in predicting whether aging will or will not occur, and in these welds these ratios were 4.36 and 3.90, respectively. Both of these numbers fall in a region where some inconsistencies in the results have been observed. Thus it seemed desireable to invéstigate materials containing higher ratios. Preliminary experiments were performed on welds in a single'héat (CB64) of columbium-zirconium alloy which had a high as- received oxygen content. The vendor's analysis of the_base metaifis 900 ppm Op, 100 ppm Na, 500 ppm C, 16 ppm Hy, and 0.77% Zr. The welds were aged at 1500 and 1600°F, at times from 1 to 250 hr, and all were brittle at room temperature. Again thé effect of high oxygen content in the base metal before welding was evident. The Op-to-Np; and 0z~to-C Table 2.8, Effect of Aging at 1500°F on Oxygen-Contaminated ,l' Welds . | . . Room-Temperature Specimen Treatment Aging Time Bend Behavior Contaminated before welding As welded Brittle 5 hr Brittle 25 hr Brittle 100 hr Brittle Contaminated after welding As welded Ductile with cracking 5 hr ‘Brittle 25 hr Ductile with cracking 100 hr Brittle 35 ratios were 9.0 and 1.80, respectively. It may be concluded on the basis of these data that it is probably impossible to make a nonaging weld in wrought material containing a high oxygen content. Further, it appears doubtful that it is possible to make a weld which is even ductile in the as-welded condition if a high-oxygen~content base metal is used. 36 3. MECHANICAL PROPERTIES INVESTIGATIONS Tube-Burst Tests on Columbium-Zirconium Alloys Tube-burst tests are being performed on columbium-zirconium alloys to obtain baseline data for comparilison with the results of in;pile tests (see chap. 8 of thils report). Specimen preparation is under way, and the problem of machining the'specimens to the rigld tolerances required for tube-burst tests appears to have been solved. The quality of the specimens is steadlly improving. | The equipment used previocusly for tests at 1500°F 1s being modified for tests at 1800 and 2000°F. A schematic drawing of this equlpment is shown in Fig. 3.1. The chamber consists of a type 309 stainless steel tube with a water Jacket at one end and an evacuation tube at the other. Thermocouples, sheathed in stainless steel, and a pressure line pass through a metal plate at the top of the chamber. The vacuum is also measured at the top of the chamber. The columbium alloy specimen is wrapped in tantalum foil and in- serted in a zirconium cup. Even though the zirconium cup acts as an excellent getter at these temperatures, further protection is provided by the tantalum foil. The furnace, which 1s rated at 2200°F, 1s used to heat the chamber and specimen tc temperatures of up tc 2000°F. The vacuum that can be maintained at 1800°F is about 0.1 p. With sultable gettering, this vacuum appears to be adequate. Tests are now In progress to determine whether a pure helium atmosphere can be malntalned in this chamber for extended times. | Preliminary tests have shown that some redevelopment of the pres- surizing system will be necessary. A specimen which failed after 7.1 hr at 1800°F 1s shown in Fig. 3.2. A gas pressure of 1725 psi was requlred to rupture the specimen. At these high pressures the volume of gas in the pressure system must be minimized to prevent excessive "olow out” when failure occurs. In another approach to the equipment problem, an effort is being made to develop an Internal heating system. A small high-temperature 37 furnace will be mounted inside a water-cooled chamber to minimize the outgassing problem. This system can be used for testing above if the need should arise. UNCLASSIFIED ORNL-LR-DWG 53499 PRESSURE GAGE = = ARGON PRESSURE SOURCE %L/PRESSURE LINE VACUUM ‘ GAGE — =\ | e—SHEATHED THERMOCOUPLE EESSRSE RS SSSSSIRISY 7 ] _d ’ j A_—WATER JACKET /] —&zz’ ] 4—MARSHALL FURNACE 1~ COLUMBIUM SPECIMEN Y SN N e NS SR 1~ TANTALUM FOIL S S, ZIRCONIUM CUP --—STAINLESS STEEL CHAMBER ,./ 1 1 V] [ i VACUUM PUMP Fig. 3.1. Schematic Drawing of Columbium Tube-Burst-Test tus. ' ' 38 2000°F Apparaf UNCLASSIFIED ¥=-36078 Fig. 3.2. Columbium0.6% Zirconium Tube-Burst Specimen Which Failed After 7.1 hr at 1800°F in Vacuum with an Internal Pressure of 1725 psi and a Tangential Stress of 22 500 psi. Effect of Gaseous Contaminants on the Mechanical Properties of Columbium The presence of relatively small amounts of certain elements in columbium and its alloys can markedly alter the strength and formability of the materisl. Creep tests of initially pure columbium in controlled atmospheres of argon, oxygen, and nitrogen and room-tempersture bend tests of specimens contaminated with oxygen, hydrogen, nitrogen, and carbon have been reported previously.lsz The present studles extend the previous work and include the results of tensile tests in determining the specific effects of oxygen, nitrogen, and hydrogen on the mechanical properties of columbium at room and elevated temperatures. The details of this work have been presented elsewhere? and will be only summarized here. Moderate additions of oxygen, nitrogen, and hydrogen strenghten columbium at room temperature. Oxygen present in solution in concen- trations in excess of 3000 ppm causes serious embrittlement. Surface oxides formed by annealing in oxygen &t 400°C cause cracking when the 1"ANP Semiann. Prog. Rep. Oct. 31, 1959," ORNL-2840, pp. 44-47. 2"ANP Semiann. Prog. Rep. April 30, 1960," ORNL-2942, pp. 37-40. 3H. E. McCoy and D. A. Douglas, "Effect of Various Gaseous Con- taminants on the Strength and Formability of Columbium" presented at Columbium Metallurgy Symposium, Lake George, New York, June 9-10, 1960 (to be published). 39 material is tested in bending. Oxygen concentrationé of the order of 2000 ppm increase the mechanical strength of columbium at elevated tem- peratures but do not appear to reduce its ductility seriously. Nitrogen present in excess of 1000 ppm reduces the room-temperature bend and tensile ductilities of columbium to essentially zero. The tensile strength and creep strength of this material at 1000°C are in- creased by nitrogen, and the observed reductions in ductility are not as serious as at room temperature. However, heavy surface cracking has been noted that may cause problems in fabrication. Hydrogen concentrations in excess of 500 ppm seriously embrittle columbium at room temperature. The absorption rate of hydrogen in columbium seems to be greatest in the temperature range 500 to 600°C, where a stable hydride, CbHg.gg, is formed. The formation of this hydride was observed to be quite rapid, and it is formed din all speci- mens cooled through the temperature range 500 to 600°C in the presence of hydrogen or water vapor. Hydrogen and water vapor increase the creep rate of columbium at 982°C, although no reduction in ductility is Oobserved. 40 o 4. ALTQOY PREPARATION Electron~-Beam Melting of Columbium-Zirconium Alloys - The role of the impurity elements oxygen, nitrogen, and carbon in the aging phenomena observed in columbium~zirconium alloys is being studied for both wrought materials and weldments, as mentioned in the preceding chapters of this report. In order to make a systematic étudy of the effect of the Impurities, high~purity Cb—l% Zr samples and samples containing controlled amounts of oxygen, nitrogen, ahd carbqn are reqfiired. In the high~vacuum, eleétron-beam—melting process, purification of columbium with respect to oxygen and nitrogen proceeds by volatilization of the suboxide and nitride at rates which depend on such variables as initial composition, pressufe, temperature, and molten surféce_area. It would be very diffiqult, thérefore, to obtaln a specified final oxygen or nitrogen composition in the alloys. Carbon, on the.ofher hand, is removed only in the presence of oxygen, as CO, and shbuld remain in the melt if the alloy has been deoxlidized previously. Accordingly 1t was decided to try to prepare columbium alloyé éonfainifig i fit %QZr ana | varying amounts of carbon with the lowest possible oxygen and nitrogen contents. The alloys desired with higher oxygen or nitrogen content ,could‘then be prepared by controlled contamination of the high~purity material. Sufficlent quantities of each composition to prepare at least 20 tensile specimens (~0.040 in. X 3/4 in. X 5 in.) were desired. Previous experience had shown that a 300-g sample could be rolled to the proper dimensions to yield the desired number of specimens. Therefore melts of thls slze were selected. The columbium melting stock was in the form of as-reduéed pellets containing 1730 ppm oxygen, 520 ppm nitrogen, and 210 ppm carbo@. The unalloyed pellets were first electron~beam melted to yield columbium which contained relatively low amounts of oxygen and nitrogen. Then 1.5 wt % Zr and O, 100, 200, and 400 ppm by weight of carbon was added 41 to the four approximately 300~g melts. The zirconium was added as strips of rolled-iodide zirconium and the carbon as crushed spectrographics electrode carbon. After electron-beam melting, the alloy buttons were rolled into strip, sheared, and remelted to achieve homogeneity. The remelted buttons were then rolled to 0.040-in.-thick sheet and sampled for chemical analysis. The results of the analyses are presented in Table 4.1. The chemical analyses indicate the high degree of purification with respect to oxygen, nitrogen, and carbon achievable by the electron~-beam melting of columbium. In the cases where carbon was added to the alloy, the results indicate promise for achieving preselected compositions, although further experimentation will be required. If 20 ppm is.taken to be the residual carbon, the retention of the added carbon in melts 107, 108, and 109 can be calculated to be 90, 80, and 87.5%, respectively. By first determining residual carbon and using 80 to 90% as the retention factor, it should be possible to calculate the amount of carbon to be added to yield a desired final carbon content. The variations in the zirconium analyses were larger than desired, or anticipated, and they point out the\difficulty of controlling alloy compositions when there is a wide difference in vapor pressures of the constituents. Attempts were made to hold the melting conditions for the four samples constant so that wniform loss of zirconium would occur. The aging of tensile specimens prepared from this material is in progress. Table 4.1. Chemical Analysis of Columbium-Zirconium~Carbon Alloys Zirconium added: 1.5 wt % Zirconium Carbon (ppm) Oxygen . Nitrogen Melt No. Analysis Analysis Analysis - (wt %) Added Analysis (ppm) (ppm) 110 0.882 0 20 46 12 107 0.663 100 111 30 9 108 0.575 200 180 22 6 109 0.742 - 400 370 24 11 42 Electron=Beam Melting of Vanadium, Molybdenum, Tantalum, and Tungsten ~ Buttons of the refractory metals, vanadium, molybdenum, fiantalum, and tungsten were electron-beam melted and analyzed to determine the degree of purification achieved. . The results of chemical analyses are summarized in Table 4.2. Table 4.2. Results of Chemlcal Analyses of Electron-Beam=Melted Refractory Metals . for Oxygen and Nitrogen Results of Vacuum Fusion Analyses Metal , Oxygen (ppm) Nitrogen (ppm) Vanadium | Before melting 700 130 After melting 4'7 12 Molybdenum ' Before melting = = - 29 .10 After melting 10 - <5 . Tantalum Before melting 24 16 After melting 12 <5 Tungsten Before melting 130 <5 After melting 50 <5 These analyses i1llustrate the effectiveness of electron~beam melting for purifylng the refractory metals and confirm the work of Smith at the Stauffer-Temeécal Company.l It should be noted that, although several buttons of each metal have been melted, the data of Teble 4.2 are based on single analyses. 1H. R. Smith, Jr., pp. 221-35, in ”Vacuum.Metallurgy {ed. Dy R. F. Bunshan), Relnhold New York, 1958. 43 Electron~Beam Ingot Melting Although the ability to produce high-purity refractory metal and alloy buttons in sizes up to 350 g is of considerable value, the develop~- ment of fabrication and Jjoining technology and comprehensive evaluatiofi require larger quantities of material. The techniques required for melting ingots in the electron bombardment furnace are therefore being developed. The ingot~-melting procedure may best be described by referring to the cutaway drawing of Fig. 4.1. The electron beam is accelerated from the gun and focused through an aperture onto the top of the ingot. The melting stock, in the form of powder or pellets, i1s fed from the charging tank to the melt by means of a Syntron Vibra-Flow Feeder. As metal 1s added to the melt, the ingot 1s withdrawn to maintain the molten surface at a constant level. | Two operational difficulties have been encountered. Because of nonuniform magnetic fields around the furnace, it has sometimes been difficult to focus the electron beam through the aperture and onto the ingot. When the focus coil was adjusted to minimize the pickup of stray electrons on the aperture, the impact points at the ingot would be off to one side. The aperture is required to enable the gun~chamber pumping system to maintain the pressure around the electron gun below the arc= discharge threshold in the event of severe outgassing from the melt. At the suggestion of the furnace manufacturer, the NRC Equipment Corporation, a second focus=coil assembly was installed below the 4-in. vacuum valve shown in the drawing. The drift tube was extended 9 in. to provide the necessary space. Now the upper focug coil has the sole function of getting the beam through the aperture with minimum loss. The lower coil, operating wilthout the restrictions on movement imposed by an aperture, can be used to direct and focus the beam on the desired impact point. The result is a much more effective and efficient use of the available electron-beam power. A significant increase in beam power density at the melt is also available. bty ; UNCLASSIFIED ORNL-LR-DWG 4579 COOLING AIR I ELECTRON GUN i fl — . CATHODE RS ELECTRON BEAM ‘ GUN CHAMBER FOCUS COIL ION GAGE ’ 2 o i T G “ ¢ S 7 . N A [ =7 - CHARGING - . e - ' TANK _ _ .~ 4-in.VALVE W APERTYLE . 4-in DIFFUSION PUMPAY 1 (ISt Al 4-in. [N R ” NS “ 4 (P vaLve \ RER - - e LA “1 AR |’ . 6-cfm ROTARY PUMP-{|| (R 2 NN & . : :l:‘. \ ~M~ | . 5/%' :, : '.! “' LW = O el B | | | Ao (It »’Afifl— £ A /- \ e . 10- in. VALVE CLiH y e V] : =31 ! ,.F 'S : < SYNTRON VIBRA-FLOW ION GAGE alezd 220 N FEEDER " OPTICAL BAFFLE ot Gk O\ = ;p‘f_ SIS MELTING CHAMBER COLD TRAP 2L A2 & S =" Al g // > / - , : & ‘S . ‘ HS SiGHT PORTS 10-in.DIFFUSION i;:‘|‘. ‘ ' INGOT 1 , PUMP Al > LR waTer cooLeD MoLD Al = o i ' ' . 2 e | U/ UP ASSEMBLY i : 7 U MOLD C S LR —= T 5 7? O » o / i . ' - /“U‘I ' | - “;}, - ||| w ' . l gl A | | 4 o ' 10-in.VALVE 7 ‘ - o ‘ OPTICAL BAFFLE oy 1 M COLD TRAP iyl ' Il ’ | * \ ' » 10-in. DIFFUSION PUMP | > "y | l ~~ INGOT RETRACTING 1 . : /3~hp GEAR MOTOR -' ‘ A0 p S \ MECHANISM ' . P ! 2 = Wi | e e = ol ‘ A % ¥ I 4 D Fig. 4.1. Electron Bombardment Furnace. "Qn 45 The second difficulty has been the deposition of vaporized metal on the sight ports. The depesit usually becomes thick enough to obscure visibllity of the melt within a few minutes. A mirror assembly has been tested with some success. A plain window-~glass mirror remained bright and reflecting until it became too hot, softened, and warped. Pyrex, Quartz,‘and'Vycor plates were then tried as mirrors, but the columbium deposit tended to spall and produce a rough, nonreflecting surface after a short time. A similar result was encountered with stainless steel - mirrors. In spite of these difficulties, the mirror system appears | attractive enough to warrant further study to determine what mirror - materials and conditions will yield bright deposits. Other possibilities being considered to alleviate the sight~glass coating problem are the projection of images through pinholes and the use of replaceable glass shields for the sight ports. In spite of the difficulties encountered, the electron bombardment furnace has demonstrated the capability of melting 3-in.-diam columblum ingots. The future work will be directed first toward solving the sight giass problem.and then to reducing the mechanics of ingot melting to a standard routine. The preparation of‘specific metals and alloys will follow! It is anticipated that considerable development effort may be required to achieve good control over the alloy compositilon. 46 5. WELDING AND BRAZING Electron-Beam Welding Preliminary studies have been performed to investigate the effects of electron-beam welding on the chemical composition of the columbium- zirconium alloy. Two electron-beam fusion welds were made in the same heat (S235C-4) of Cb—1% Zr alloy, with one weld made in a single pass- and the other made by remelting for each of five passeé.- Thefweld metal of each specimen was then analyzed chemically, and the analytical data were compared with the data for the unwelded base metal. The analytical data are presented in Table 5.1. Within the limits of analytical error, no change in chemical composition was produced by the electron-beam welding. Table 5.1. Chemical Analyses of Cb—1% Zr Base Metal and Electron-Beam Weld Metal (Heat S23SC-4) Material Oxygen Nitrogen Carbon Hydrogen Zirconium Analyzed (ppm) (ppm) (ppm) (ppm) (vt %) Base metal 130 100 50 (50-50 mole %) melt mole %) melt mole %) melt . Al <10 <10 25 B - <5 Trace <10 Ba - <5 - None detected L <25 . Bi ' <5 None detected None detected ) Ca 50 , <20 . Co <5 50 <5 cr - .20 - ‘ , 150 o 600 - Cu 10 . <10 <5 Fe <20 150 200 K <20 <50 : <10 Li 100 500 1000 Mg ' 15 . <10 15 Mn ~10 20 20 Na <10 100 <10 Ni - 20 4000 ' 400 .81 45 200 100 In a previous experiment from which U0, crystals coated with BeO were obtained, it was observed that only the small crystals of UO, (up to 20 u across) were clad. Subsequent experiments also showed that near stoichiometric UO, could be prepared by the reaction of water vapor with UF, dissolved in LiCl-NaCl (75-25 mole %). Preparations in this solvent were more satisfactory because of the relatively high solubility of the solvent in water. It might be presumed that by adding selected particles of U0z (up to 20 u) to a melt éontaining BeF,, the water-vapor precipita- tion might provide a better yield of coated crystals. - Accordingly, - approximately 200 g of UO, was'prepared by reaction of water vapor with UF, dissolved in LiCl-NaCl (75-25 mole %) at 700°C. In an attempt to reduce the particle size of UO, crystals, helium saturated at room tem- perature with water Vapdr was bubbled through the molten salt mixture. The recovered U0, product was observed by petrographic examination to . | 55 consist of single crystals of U0, with a refractive index of 2.36. These crystals were free of inclusions and ranged in particle size to 200 u. Approximately one-third of the UOp crystals were between 10 and 20 u. X-ray diffraction examina?ions further showed that_the unit-cell-lattice parameter, a,s was 5.470 A. The results of spectrographic analyses are presented below: Tmpurity Content ' Impurity Content . (ppm) (ppm) Ag <1l K <50 ' N Al ; 15 Li 30 . - B 0.1 Mg <5 ) Ba <10 Mn <5 ' - Be <0,01 Na 20 Ca <50 Ni 425 ’ Ccd <0.1 P <100 Co <1 Si <10 Cr <6 Sn <10 Cu <2 v 1 Fe A <10 In a subsequent experiment, a portion of this prepared U0, will be added to LiF-BeF, (60-40 mole %) for the precipitation of BeO. Solvent Extraction Process There are a number of processes reported in the literature for pre- paring reactor and refractory-grade beryllium oxide containing 50 to 5 a* 500 ppm total impurities. For special reactor applications and sinter- ing studies it may be desirable, however, to prepare beryllium oxide of much higher purity than is now available. Toward this end, a liquid- liquid extraction process is now under investigation. An extraction process could be carried out with plastic equipment at room temperature in a closed system, and, with favorable separation factors, there is no theoretical limit to the degree of purity which can be achieved. Th¢ system under study consists of an aqueous phase equilibrated with a. carbon tetrachloride phase. The CCl, phase contains acetylacetone >R. E. Moore, "Purification of Beryllium Compounds: A Literature Survey," ORNL-2938, June 1, 1960. ' . 56 ' ' - (2,4-pentanedione) to complex beryllium as beryllium acetylacetonsate, Be{ CH3COCHCOCH3) 5, and the agueous phase contains ethylenediaminetetra- acetic acid to complex and retain metal Impurities. This method for purifying beryllifim-has been reported in the literature,6-7 but separa- tion factors are not available, In the present study 50-ml quantities of aqueous solutions prepared to contain 0,1 molar metal nitrate and 0.12 molar disodium ethylene- diaminetetraacetate were equilibrated with equal volumes of organic solutions containing 92 vol % CCl, and 8 vol % acetylacetone. In most cases the aqueous solutions used actually contained lessAthan 0.1 molar metal ion because solids precipitated after pfeparation of the solutions. Separation factors at various pH values were obtained in many cases. Preliminary data are reported in Table 6. 2. Additional extraction studies with sclutions at pH 6.2 originally containing 0.2, 0.3, 0.4, and 0.5 molar Be(NC3), gave separation factors (aqueous to organiec) of 0.00385, 0.011, 0.028, and 0.022. It is obvious that aluminum and magnesium fiill be the hardest to remove in this scheme. A stepwise process calculated to reduce these elements to the lowest practicable levels should remove all other im- purities with higher separation factors. - The successful use of the process for preparation of ultra-pure BeO depends on the development of a procedure for converting beryllium acetylacetonate into a compound such as Be(OH), or BeC,0,+3H,0 without intreducing contaminants. This aspect of the purification procedure- is under study. Phase Relationships in BeO-Metal Oxide Systems Investigations of phase relationships in BeO-metal oxide systems have continued.® These investigations were stimulated by the observation 6J. A. Adam, E. Booth, and J., D. H. Strickland, Anal. Chem. Acta., 6, 462 (1952). 7I. P. Alimarin and I. M. Gibalo, J. Anal. Chem. U.S.S.R. (Engl.), 11, 405 (1956). | 8"ANP Semiann. Prog. Rep. April 30, 1960," ORNL-2942, pp. 52-54. 57 Table 6.2. Liquid-Liquid Extraction of Metal Ions PH Separation Factor, pH Separation Factor, Metal Ion Values Aqueous to Organic Metal Ion Values Aqueous to Organic Bet? 6.5 0.025 crt3 1.9 >1620 4.6 <0.01 5.3 10.2 AL+3 5.6 0.364 6.0 2.3 4.5 0.423 Cat? 1.8 11.7 4.0 400 s+ L1 ;'g 3'3 5.6 199 ] 41 10.6 6.3 99 5.0 8.5 Nat o 2 43,4 i 6.0 8.3 6.0 191 - 6.9 7.2 cat= - 1.6 4.5 - Mgt? 1.4 1.73 3.9 28.4 - 3.0 1.69 Kt 5.8 770 . g‘é f{'g'g Bat+2 3.9 918 5.9 1.46 cut? 3.7 3000 6.7 1.63 6.3 150 +3 Nit? 2.1 >2000 Fet™ & 62 4.6 >2400 6'3 6‘45 5.5 >935 srt? 3.7 >1780 +2 Zn 2> o0 Si (silicate) 4.2 37.2 6.0 >300 : P (phosphate) 6.8 395 that minor "impurities” in BeO powders have considerable effects on sinterability.9 The information developed may also be useful in the evaluation of the mechanical and physical properties of BeO. Data con- cerning the binary systems, BeO-La»03, BeO-Ca0, Mg0O-BeO, and CeO,-Bel - were obtained. Be0-La,05 Thermal analysis data for several mixtures of BeO-La;03 heated in air and cooled at approximately 200°C/min from 1500°C showed a thermal arrest at 1250°C for compositions rich in BeO and at 1330 and 1445°C for La,03-rich compositions. "ANP Semiann. Prog. Rep. Oct. 31, 1959," ORNL-2840, p. 75. 58 Be0-Cal Thermal analysis data were obtained for compositions between 10 and . 90 mole % increments. The samples were heated in air and cooled at- approximately 200°C/min. These data showed thermal arrests at approxi- mateiy 1365°C. Cooling rates were observed to shift the arrest tempera- ture in the composition range of approximetely 45 to 50 mole % BeO. Upon cooling the 45 mole % Be0—55 mole % Ca0Q specimens at approximately l500°C/min; an arrest was noted at approximately 1365°C. When this specimen was cooled at about 70°C/min from the same temperature, the - arrest was observed in the vicinity of 1450°C. f Samples of the same composition as those used for thermal analyses were quenched in mercury after heating to approximately 1500°C. The ] results of the examination of these samples by means of the polarizing microscope and x-ray diffractometer are shown in Table 6.3. Optical examinations of the microstructures of the samples showed a matrix of iéotropic material, with well-developed growth and BeO crystals on the Table 6.3, Identification of Phases in BeO-Cal Mixtures Heated to Supersolidus Temperatures and Quenched in Mercury BeO Content Phases Presen.ta o (mole %) - | 90 BeO + CayBe;05° ' ' g0 ' BeO + CapBes0s5 . : 70 ‘ ' BeO + CayBes0j5 60 C&2B8305 . 50 Ca0 + CajyBe30s5 40 Ca0 + CapBe30s5 30 Ca0 + CayBe30s5 20 Ca0 + Ca,Be30s5 10 . Ca0 + CayBes0s ®As determined by optical and X-ray analysés, bC&2B8305 is the approximate formula for what appears 1o be a metastable compound formed from liquid of 60 mole % BeO—40 mole % Ca0 composition. - 59 BeO-rich side of the 60/40 composition and Ca0O crystals on the CaO-rich side. Solid-state reactions which occur as a result of heating mixtures of BeO and Ca0 in air for nine weeks at 1000°C were reported previously.8 These samples were crusted over with a vivid green material. X-ray dif- fraction analysis revealed the presence of three phases — Ca0O, BeO, and a cubic BeO-Ca0 compound — in all samples. The strongest indication of the compound was found in those samples which contailned large amounts of the green crust. The effect of atmospheric conditions on thekformatidn of the green compound is being investigated. Several samples of beryllia-rich mixtures were treated hydro-thermally for one week in an autoclave at 6450 psi at a temperature of approximately 500°C. These samples are currently being examined. MgO-BeO and CeO;-BeO In the consideration of materials for a high-temperature reactor, the ability of a fuel element to retain nuclear fuel under reactor- operating conditions is very important. The many desirable properties of U0, as a fuel and BeO as a moderator recommend the use of these materials for a fuel element; however, UO; will readily volatilize from the simple BeO-UO; mixture at elevated temperatures. The addition of a sultable third oxide to chemically combine with or form extensive solid solution with the UO; will retard the loss of UO,. The choice of the third oxide phase is difficult, since the phase relationships which exist between BeO, UO,, and many of the other refractory oxides are not known. Furthermore, the stability of.these combinations of these materials with various gaseous environments has not been investi- gated. A program to investigate these matters has been initiated, and preliminary phase studies have been made in the systems MgO-BeO and Ce0;-BeO. A series of experiments in which MgO-BeO mixtures were fired to 1800 to 1900°C in air and to 1450°C when sealed in platinum indicated 60 no solid-state reaction between these oxides. At the higher tempera- tures, the rate of MgO volatilization, at least in air, did not seriously impede normal quenching techniques. The work at higher temperatures in- dicates that MgO and BeO form a.simple eutectic system with a eutectic temperature of 1860 £ 20°C. Similar firings'in the Ce0,-Be0 system also indicate a simple eutectic reaction at 1775 = 20°C. Experimentatiofi is in progress. to determine the eutectic composition in both these binary systems. Thermogravimetric equipment is bresently being fabricated to test these and other potential fuel element materials in various gases at temperatures of 1800 to 1900°C. Data obtained using this equipment will provide information about the stability of these materials and also pro- vide a means of investigating the mechanism of UO, volatilization. 61 - 7. ENGINEERING AND HEAT TRANSFER STUDIES Boiling-Potassium Heat Transfer bkxperiment The fabrication of components for the boiling~potassium heat fransfer system shown in Fig. 7.1 héé been continued. -All major loop components, with the exception of the boiler section, have been fabricated. The lower portion of the loop (between the levels A-A and B-B in Fig. 7.1) has been installed. Subassemblies of the remaining units are being made in the welding shops in order to minimize the number of field welds required. Electrical instrumentation (indicators, controllers, recorders, etc.) is approximately three~fourths complete. The major remaining fabrication problem, as mentioned previously,1 is the development of a satisfactory technique for brazing the 2 X 5-in. copper blocks of the test section to the inner,3/8-in.-o.d. stainless steel boiling tube and the 5-in.-i.d. stainless steel ocuter jacket. Since the boiler has been designed to achieve a maximum heat flux of 5 x 107 Btu/hr-ft2 at the inner tube surface, a good thermal bond between | the copper blocks and the stainless steel boiling tube is mandatory. For example, the‘existence of a 0.001-in. helium-filled gap at maximum flux conditions will result in an additional temperature drop in excess of 200°F. This is sufficient to raise the temperature of the copper at its outer surface to about 1850°F. The melting temperature of copper is 1980°F, and therefore a gap as large as 0.002 in. would cause de- struction of the ocuter portions of the copper blocks. ©Since the copper blocks will also be used in determining the boiling-tube surface tempera- ture (by extrapolétion of the femperatures measured along a radial line in the blocks), the magnitude of this additional thermal resistance at the inner copper-to-stainless steel interface must be accurately known. Such a determination would necessitate a difficult and lengthy auxiliary 1YANP Semiann. Prog. Rep. April 30, 1960," ORNL-2942, pp. 64—71. 62 UNCLASSIFIED ORNL-LR~DWG 46124A VACUUM == VENT == CONDENSER SEPARATOR PRESSURE _ : ELEMENT SURGE é ~ VAPOR TANK THERMGCOUPLE THERMOCOUPLE | : ' -— * : CONDENSER . , BOILER THERMOCOUPLE - THERMOCOUPLE - " A —_— —a . MIXING MIXING CHAMBER CHAMBER THERMOCOUPLE = ' -TUBE THERMOCOUPLE LEVEL J HOLD INDICATOR TANKS : L~ THERMOCOUPLE O/ O/ THERMOCOUPLE FLOWMETER LIQUID- RESERVOIR -t 8 SUMP FREEZE - TANK VALVE . CoLD TRAP Fig. 7.1. Schematic Flow Diagram of Liquid-Metal Boiling Experi- ment. ‘ ‘ - experimental study, which would be complicated further by the high probability that the contact resistance would not be circumferentially uniform. | ‘Accordingly, much effort has been expended on the problem of achieving a good thermal bond between the copper blocks and the stainless steel boiling tube. The diff'iculties encountered appear to be due to oxidation of the stainless steel by oxygen arising either from outgassing - 63 of the copper, surface contamination of the metals, or residual air in vacuum and inert gas systems. The best results have been obtained by nickel plating both the copper and steel surfaces prior to brazing. Developmental work in an effort to obtain an adequate bond will continue. Preliminary testing of 2-in.-long, 5-in.-i.d. clamshell heaters using exposed Kanthal A-1 wire coils have indicated that the required power output of 2.3 kw per heater (corresponding to a flux of 5 X 10° Btu/hr-ft2 at the inside surface of the boiling tube) can be achieved with a maximum element temperature of 2350°F. The heater life appears to be satisfactory under these conditions. Thermal Conductivity of Lithium and Lithium Alloys Preliminary data have been obtained on the thermal conductivity of an essentially pure lithium sampie (99.98 wt % 1i) and of a lithium- silver (3-97 wt %) eutectic mixture. The results are summarized in Fig. 7.2. | The design of the experimental apparatus used for these studies, shown partially disassembled in Fig. 7.3, was based on a one-dimensional steady- state heat-flow model, with the heat being transferred axially down a long cylinder from an upper heat source through a liquid sample and a thermal-flux meter to a lower heat sink. The sample container and the heat meter were formed integrally from a 1l.5-in.-o0.d. type 347 stainless steel bar bored out at the upper end to a wall thickness of 1/16 in. to accommodate a 3-in.-deep.liquid-metal specimen. Three thermocouples, located at the cylinder center line in thin-walled thermowells, which were equally spaced and rotated 120 deg between successive levels, were used to measure the temperature gradienf in the sample. (The thermo- couple spacings in the sample and the heat meter are indicated by the data points of Fig. 7.4.) The temperature distribution in the heat meter was determined by four pairs of thermocouples axially spaced at l/2-in. intervals in small holes drilled into the solid rod. At each level, one thermocouple extended to the center line, while the other 64 UNCLASSIFIED ORNL-LR-DWG 53500 45 | | —— LITHIUM (REF &) —-— CANEL LITHIUM (REF 5) A T 40 ——~" LITHIUM (REF 8) . ° " e LITHIUM (REF T) A & A LITHIUM (ORNL) = ® SILVER-LITHIUM ALLOY (ORNL) - ™~ 3 35 a E 4 —-——-—-_ 2 ‘_/’ K 30 = 2 },/ I & =T © ""'-.\ / A ;‘25 T Py ~ Z A ~~ Ll ‘s\ E N < 20 4 < \ \l N 15 ' 200 400 600 80O 1000 1200 1400 1600 1800 TEMPERATURE (°F) ‘Fig. 7.2. Summary of Experimental Measurements of the Thermal. Conductivities of Lithium and a Lithium-Silver (3-97 wt %) Alloy. (diametrically opposite) was located 1/8 in. from the outer surface of the cylinder. The thermocouple pairs were displaced by 90 deg with respect to the thermocouples at adjacent levels. Radial guard heating was effected by a set of nine cylindrical heaters that were fabricated from tantalum wire wound on a ceramic core (see Fig. 7.3). These heaters were of various lengths and were manually controlled to generate an axial temperature distribution at the inner heater surface which matched that aloneg the axis of the test cylinder in order to minimize the radial flow of heat. The test cylinder and guard heaters were concentrically positioned within a second heater whose function was to provide the desired temperature level; the entire system was contained within a bell jar and blanketed with argon. Initial measurements were made with a lithium-silver alloy -con- taining 3 wt % (32.5 mole %) lithium; this mixture melts at 1150°F. A typicél axial profile along the sample and heat meter is given in the 65 - UNCLASSIFIED ~ PHOTO 34734 — EXPANSION POT \ HEAT SOURCE Nor SAMPLE CONTAINER e NS 2 GUARD [ . HEATERS —— Fig. 7.3. Axial Heat-Flow Apparatus for the Determination of the Thermal Conductivities of Liguid Metals. 66 *l' i UNCLASSIFIED upper curve of Fig. 7.4. In ORNL-LR - DWG 53501 contrast with the profiles?:3 600 .—-/ observed in the determination 1500 . ,/f’/ of the thermal conductivity of //fE;:M—QUERAuDY s01id materials using a similar 1400 s device, these data show no con- ¥ //”// tact thermal resistance at the g 1300 “ rw”"”’r' solid-liquid interface. This E oo I“/Tfifiam Suggegts complete wetting of E ‘,/“, the type 347 stainless steel 1i00 ,A(/ surface by the liquid metal. The average thermal con- 1000 ductivity for this alloy ———HEAT METER&+—|5AMPLE (based on five runs) was calcu- 9000 E l ; a 5 6 lated to be 36.5 Btu/hr-ft.°F * AXIAL DISTANCE (in) at a mean temperature of 1515°F, Fig. 7.4. Typical Axial Tempera- ture Profiles Obtained in the Determina- tion of the Thermal Conductivities of the data around the average of Lithium and a Lithium-Silver (3-97 wt %) 7. Alloy. with a maximum deviation of Since data on the thermal conductivity of silver at temperatures above 932°F (mp, 1742°F) are not available, a direct comparison of these results with the values for silver is not possible. While it is difficult to estimate the conductivity of solid silver* near the melting point, a value in the range 190 to 240 Btu/hr-ft-°F does not seem unreasonable. Thus, the addition of a small amount of lithium appears to drastically reduce the conductivity. This same phenomenon has been observed with-solid binary alloys in the range for which they exhibit 2"ANP Semiann. Prog. Rep. Oct. 31, 1959," ORNL-2840, p. 61. >"MSR Quar. Prog. Rep. Jan. 31 and April 30, 1960, ORNL-2973, p. 25. “C. R. Tipton, ed., "Reactor Handbook," 2nd Edition, Vol I, "Materials," p. 827, Interscience Publishers, New York, 1960. 67 solid solfitions. Austin® points out that as little as 2 at. % nickel dissolved in copper results in a 60% reduction in the conductivity of the copper. It is significant to note also that the higher the con- ductivity of the solvent metal, the greater is the decrease in con- ductivity caused by the presence of a few foreign atoms. Data for the silver-gold system indicate that this is true even though the con=- ductivities of both pure materials are high (k for gold at 32°F is 172 Btu/hr-ft-aF); the addition of 35.2 mole % gold to silver reduces the conductivity to about 48 Btu/hr-ft-°F. Some experimental difficulties were encountered in similar measure- ments with lithium. Inspection of the apparatus following a shutdown forced by a power failure disclosed that a number of thermocoupleslhad been displaced from their nominal positions. A study of the data obtained during the period of high-temperature operation (about 1000°F) indicated that the thermocouple displacements may have occurred early in the life of the apparatus. This may account in part for the unex- pectedly wide scatter in the lithium conductivity results. A typical axial profile is shown as the lower curve in Fig. 7.4. It may be seen that the "best' line drawn through the heat-meter temperature data results in an interface temperature about 10°F greater than that observed from the extrapolation of the temperature curve in the sample region. In view of the uncertainties in temperatures, this may reflect excessive guard heating in the region of the heat meter and hence a gréater heat flow in the meter than in the sample. On the basis of the lithium-~silver resfilts, it has been assumed that the interface temperature indicated by the sample thermocouples is correct; the heat- meter profile has been drawn to accord with this assumption. The lithium conductivity calculated for this run vas 40.0 Btu/hr-ft-°F at a mean temperéture of 1280°F, The results of these‘preliminary measurements are given in Fig. 7.2. While doubt exists as to the magnitude of the conductivity, there does °J. B. Austin, "The Flow of Heat in Metals," pp. 30-35, American Society for Metals, Cleveland, Chio, 1942. 68 seem to be agreement with the data obtained at CANEL® (calculated from electrical conductivity measurements) and with Russian data”? as to the upward trend of the conductivity with temperature. The data of Miller and Ewing8 are also shown in Fig. 7.2; the agreement with the CANEL results is not unexpected because the Miller”afid'EWihé aéta were used to estimate the Lorenz number for converting the electrical measurements to equivalent thermal values. For completeness,_thé_data of Webber et al.’ have been included in Fig. 7.2. The.presenfqreguitS‘are_approximately 25% higher than the values estimated by CANEL,.bfititHis discrepancy may relate to the pre#iously mentioned thermocouple displacements and the subsequent excesSive heating in the heat-meter region of the apparatus. | A number of modifications have been made in the conductivity apparatus, including a more positive system for assuring the locations of the thermocouples and an altered system for guard heating. Measure- ments will be resumed shortly. 6"Muclear Propulsion Program, Engineering Progress Report, Jan. 1, 1960-March 31, 1960," Pratt & Whitney Aircraft, CANEL Operation, Report PWAC~601, p. 92. (Secret). 7S. S. Kutateladze et al., "Liquid-Metal Heat Transfer Media," p. 3, Atomic Press, Moscow, 1958, translated by Consultants Bureau, New York, 1959. 8R. R. Miller and C. T. Ewing, Naval Research Laboratories, Washington, D.C., private communication. °H. A. Webber et al., "Determination of the Thermal Conductivity of Molten Lithium," Trans. Am. Soc. Mech. Engrs. 77, 97 (1955). 69 8. RADIATION EFFECTS Irradiation of Be0O in the ETR An assembly of 21 BeO speciméns in seven capsules (assembly ORNL- 41-5) is being irradiated in the ETR and has accumulated in excess of 3364 Mwd of exposure. The infiegrated fast flux is estimated to exceed 5 X 104209 neutrons/cm2 for at least two of the capsules. FEach BeO speci- men is a right cylinder, 1 in. long and 0.800 in. in diameter. The tem- peratures within the assembly at the-center of each BeO specimen range from 110 to 950°C depending on the coolant or on the vertical position in the reactor flux. A marked decrease in thermocouple failure has been achieved with this assembly, the fifth in the series, and it has been possible to extend the scheduled in-pile period to several times that' of earlier tests. Extensive use of thermocouple Jjunctions integrally fused with the thermocouple wire sheath is believed to be responsible for the improved performahce. The assembly ifradiated in the immediately preceding éxperiment, test ORNL—41—4,7received the most extended irradiation to date. This assembly was unique in that it consisted entirely of direct water-cooled specimens (i.e., no buffer gas). The 12 specimens were enclosed in a single capsule. The capsule was not equipped with thermocouples, since it was expected that it would be at the reactor water temperature.' In the ORNL-41-3 experiment, a similar capsule in contact with reactor coolant water déveloped an internal temperature of 110 to 120°C. Low-temperature capsule ORNL-41-4 is equipped with internal fast- and thermal-neutron flux monitors. It was in the H-10 position for 6257 Mwd and received an estimated fast-neutron exposure of 9.58 x 1020 neutrons/cmz. The unit is being held at the ETR for possible reinsertion and continued exposure if the current irradiation‘test is prematurely terminated. The third assembly (ORNL-41-3) was removed after an exposure of 2230 Mwd., It contained l-in.-long cylindrical BeO specimens of three diameters: 0.428, 0.636, and 0.800 in. The operating temperatures were 70 120 to 1100°C for the various capsule units, which included 12 BeO samples. The fest-neutron exposure has heen estimated to be 3.41 x 1020 neutrons/cfiz, and it will be checked by analysis of the fiux monitors. Dieassembly of this unit is under way. Postirradiation examination of the second assembly (ORNL-41-2) has been delayed by hot cell difficulties. Partial observations of two speci-~ mens from one capsule indicated that the specimens were sound and with- out cracks and that they resisted mechanical damage by an extrusion device used in the hot cell. The irradiation exposure of these specimens was . reported to be 1783 Mwd. The integrated thermal flux for the hottest capsule was 4.7 X 1020 neutrons/cm?®. The fast flux was calculated from the thermal flux and was estimated to be 2.9 X 102° neutrons/cm?. A contract has been made with Battelle Memorial Institute to provide hot cell services for examination of the BeO specimens from assemblies 41-2 and 41-3. This work will include opening the capsules, measuring the gross changes in size of specimens, and examining them metallographi- cally. This additional help should reduce the lead time on disassembly and inspection of the remaining irradiated units. A summary of the irradiation conditions of this series of tests is given in Table 8.1. The typical temperature values given are representa- tive of normal operation. Higher temperatures for short periods of time have also been recorded during the progress of certain temperature- control tests. It seems probable that the BeQ in assembly 41-2 maintained its integrity, as was the case with assembly 41-1, which had received a con- siderably lower irradiation dose but operated at similar temperatures. Similarly, a review of the operating temperature history of assembly 41-3 yields no information which might indicate any major change in the BeO. This is also true, to date, for experimental assembly 41-5, and it may be inferred for the uninstrumented assembly ORNL-41-4 on the basis of the performance of similar capsules in assemblies 41-3 and 41-5. 71 Table 8.1. Summary of Irradiation Condlitions for BeO Specimens Inserted in ETR . Integrated Flux . Radiel Experimental Accumulated ?E}i ;ZWE_SR (neutrons/cm?) g:fiiii Temperature Assembly Exposure of 175 My Temperature Difference No. (Mwd ) (hr) Thermal Fast (>1 Mev) E()fc) Across Capsule Neutrons Neutrons (°c) % 1020 x 1029 ORNL~41-1 ' 411 56 2.2 1.4 722 47 ORNL~41-2 1783 262 4,72 2.98 Capsule 1 : 219 b 3 388 b 5 ' 500 b ORNL~41-3 2230 314 3.41° Capsule 1 458 a 3 858 b 5 1025 167 7 120 20, ORNIL-41-4° 6257 850 9.58° ~120 ~20 ORNL=41~5 3364 ~i0) 5,15 Capsule 1 by d 2 737 b 3 _ 827 35 4 900 130 5 ' 944 60 6 950 107 7 110 20 aFlu.x value for hottest capsule. b'Z[‘he:c'mocm‘lple failed. cEstimated; to be checked by analyses of flux monitors. dActual value uncertain. eAssembly consisted of 12 specimens in one capsule instead of usual three specimens per capsule, Determination of Helium in Irradiated BeO A method has been developed for the determination of helium in irradiated beryllium oxide pellets. Helium is formed from neutron reactions in beryllium and also from the reaction of neutrons with the lithium present as an impurity. Since the helium is entrapped in the pellet, physical treatment of the sample must be such that no loss of gas can occur prior to actual.analysis. In order to accomplish this, the pellet is placed in an evacuated reaction flask containing a mixture of potassium fluoride-hydrogen fluoride and is fused. The reaction flask is constructed of copper and the inner surfaces are silver plated. 72 The flask is held at a temperature of 120 to 130°C for approximately 24 hr to insure complete disintegration of the beryllium oxide; about 225 g of the fluoride mixture is used for each 15-g pellet. The gases evolved-aré collected in a gas—transfer loop, which is connected to a Perkin-Elmer vapor fractometer 154-C. The fractionation column, which is 8 ft high and 3/8 in. in outside diameter, contains 35-50 mesh Linde 5A molecular sieves. The gas retention volume is of the order of 85 ml. The column operating temperature is 100°C, and nitrogen at a flow rate of 50 cm3/min is used as the carrier gas. The gases evolved during fusion are analyzed with the gas chromato- graph, The helium contents of two ~15-g samples that had undergone similar irradiation were 2.7 and 3.1 ppm. A lower limit of detection of 0.1 ppm of helium seems to be attainable. Creep and Stress-~-Rupture Tests Under Irradiation Studies of the effect of neutron bombardment on the stress-rupture characteristics of structural materials are being continued in the CRR. The design of the test system and the results of the experiments to date have been described in previous reports in this series. In-pile test- ing of type 304 stainless steel has continued, and preparations are being made for a columbium-alloy tube-burst experiment. An experimental assembly containing ten type 304 stainless steel tube-burst specimens was tested at 1300°F in air in the poolside facility of the ORR. The data obtained in this experiment when compared with out- of-pile data indicate that neutron bombardment does not affect the time to rupture of this material under these conditions (see Table 8.2). Results of earlier experiments indicated that at 1500°F, the time to rupture of this material was reduced by a factor of 2 by neutron bombard- ment. Testing at 1500 énd 1600°F is in progress. No further in-pile'tests of Inconel were conducted, but out-of-pile test data wefe obtained for Inconel heat No. 2, which is type CX-QOO Inconel. A comparison of the ifi—pile and out-of-pile data is presenfed in Table &8.3. It may be seen that neutron bombardment reduced the time 73 Table 8.2. Comparison of Results of In-Pile and Out-of-Pile Tube- "~ Burst Tests of Type 304 Stainless Steel in Air Irradiation Dose Tim= to Tem?fr?ture S?re§§ Specimen at Rupture Rupture F PsSL No. (MW—hI') (hI') 1300 13 000 15-2 2 000 130a 15-3 12 470 197 15-6 2 380 144 15-8 1 975 122 126 oP 144, 127 - ob 144 11 000 15-4 8 250 512 15-7 7 750 480 15-10 7 459 462 124 0 549 128 0 465 2 000 15-1 22 100 1680 15-5 ' 25 200 >185og 15-9 25 200 >1850 123 0 1077 . 125 0 1700 1500 6 600 o7 0 184 5 200 0 1150 5 000 0-17 5 400 337 9-8 8 300 518 4 200 0 1650 4 000 7-3 12 300 895 7-6 8 500 673 -aIrradiated at 1300°F approximately 8300 Mw-hr prior to the applicetion of stress. bZe:r'o irradiation dose indicates out-of-pile test. CTest concluded at this time. to rupture by a factor of 2. In addition, the out—of;pile data showed that the time to rupture is shorter by a factor of 3 to 5 for heat No. 2 than for previously tested Inconel heat No. 1, which is not type CX-900. Preparations for future in-pile tests of Inconel have proceeded. Additional test specimens of heats No. 1 and 2 are being machined and welded. Some of these spécimens may receive annealing treatments prior to testing. Also, specimené from the six special heats of Inconel described in Table 8.4 will be tested. These materials were obtained T4 Table 8.3. Comparison of Results of In-Plle and Out- of-Pile Tube~-Burst Tests of Inconel Heat No. 2 (CX-900 Specification) at 1500°F in Air Stress Specimeh Irradiation Dose Time to ( ) N at Rupture Rupture psl ‘ O. | (Msw~hr) _ (hr) 6000 89 0& 80.3 5000 . 90 0 114.4 14-1 1726 66 14-5 1803 70 4000 91 0 319 14-6 3536 ‘ 181 14-9 2476 108 14-10 : 1106 35 3000 92 0 478 14=4 9075 ' T 443 14-7 8835 431 14-8 6209 - 299 ®ero irradiation dose indicates out-of-pile deta. Table 8.4. Special Heats of Inconel Obtained from International Nickel : Company for In~Plle Testing ‘ I e et o : a Experiment Heat Manuf;gizrer 5 . Boron Concentrations (ppm) Designation Designation Soluble Insoluble Total MIR heat No. 1 _ X 8962 30 | 1 31 CX-900 heat No. 2 NX 5757 41 1 42 Special heats 1 OB <10 2 2B 20 3 - 10B 100 4 4B11 (B! enriched 98.5%) 40 5 6Bl1l (B1 enriched 98.5%) 60 6 ~ 6B10 (B0 enriched 95%) “The soluble and insoluble analyses were made at ORNL; where only total concentration is listed, the figure was supplied by the manufacturer and is nominal. ‘ ‘ 75 for studying the effect of boron impurity concentration (especially BO) on the in-pile time to rupture, ductility, and fracture characteristics of Inconel. Fabrication of eqguipment for columblum-alloy tube-burst tests is essentiélly complete and assembly of the first in-pile apparatus has begun. The specimens are presently being welded. The furnace to be used in the first in-pile experiment consists of a tantalum form and heater winding insuvlated with Al,03. A second type of furnace utilizing a - sheathed heater element will be used in later experiments. The sheathed- element furnace has the advantage of less mass and greater radiant heat : - transmission and will therefore allow lower temperature operation of the specimen. The first experimental assembly will contain eight specimens of Cb—1% Zr. The test temperatures will range from 1600 to 2000°F. The stresses will be based on the test temperature and the results of the out-of-pile tests now under way. 76 . PART 2. SHIELDING RESEARCH INTRODUCTION The Laboratory shielding reséarch program for the ANP project con~ tinues along three major lines: (1) the development of equipment for experimental investigations, (2) the collection of basic shielding data, and (3) applied research. The first category includes the development and construction of reactors and radiation detectors designed primarily for obtaining experi- mental data that are amenable to analysis. The reactor that most nearly fits these requirements is the Tower Shielding Reactor II, which, after a long period of development involving many uniqué design problems, is now ready for applied research at a power level of 100 kw and can be used for 5-Mw operation once approval has been received from the AEC. Because of the spherical shape of the TSR-II, 1t will emit a spherically symmetric flux which will avoid numerous analysis problems that have been inherent in past shielding experiments. Furthermore, the 5-Mw power level, because of the greater neutron and gamma-ray intensities being emitted (the TSR-I was limited to 500 kw), will permit more detalled (e.g., spectral) data to be collected. Another reactor for applied research 1s being installed at the Bulk Shielding Facility and is identified as the BSR-II. This reactor, which is really a stainless~steel~clad core that can be used interchangeably with the aluminum-clad core of the BSR~I, has two major advantages over the BSR~I. It 1s much more compact, and thus will be more useful with shield mockups, and it has a harder gamma-ray spectrum more nearly approaching that of anticipated ANP reactors, The development of radlation detection equipment is dictated by the need for accurate and detalled experimental shielding data, which, of course, is limited by the capability of the instruments being used. For shield design applications it is particularly important to know the energies and the directions of the radiation belng detected; conseguently, considerable emphasis has been placed on the development of gamma-ray and neutron spectrometers and appropriate collimators. The application 79 of scintillating crystals, such as NaI(Tl), to gamma-ray spectroscopy has been under experimental and theoretical study for some time, and recent studies of surface-barrier counters for fast-neutron spectroscopy have indicated quite promising results. In addition, large spectrometer housings with radiation collimators have been constructed both at the Bulk Shielding Facility (BSF) and the Tower Shielding Facility (TSF), One at the BSF is the gamma~ray Model IV spectrometer. It 1s made of lithiated lead, the lithium being used to reduce capture gamma-ray production, and it could be constructed only after a technique for distributing the_lithium in the lead had been developed. A neutron- : chopper spectrometer is currently being designed for use with the BSR to measure low-energy neutron spectra in shields; such data are important for predictions of the production of secondary gamma rays. At the TSF a, large spherical detector shield is being fabricated for use in making dose measurements, neutron spectral measurements, and gamma-ray spectral measurements. In addition to the spectral studies, foll detector studies are being made to determine the effects of the foils themselves on the flux of radiation being measured. Although ANP shielding research is "applied," by definition, it must have a foundation of basic research on which to base the methods of applied research. For example, a knowledge of the spectrum of gamma rays emitted in the fission of U?>?° is required for an accurate calcula- tion of the gamma-ray sources in a reactor; this has been a subject of gstudy at the BSF for some time. Although the experimental data have not yet been completely analyzed, a preliminary spectrum is already beilng used for shielding design applications. Methods for calculating the attenuation of gamma rays by shield materiasls are also required. Un- fortunately, most currently available methods (e.g., the moments method) are limited to extremely simple geometries, and efforts are ndw being made to develop procedures for more complicated geometries, with emphasgis being placed on the Monte Carlo method, which is much more flexible than the moments method. Recently a Monte Carlo code for deep gamma-ray penetration calculations was successfully developed after several attempts were discouragingly unsuccessful, 80 A basic understanding of neutron processes is also being sought. In a recent experiment at the BSF the angular distribution of low-energy neutrons emerging from hydrogenous slab shields was observed to'agree with theoretical predictions for a slab of pure paraffin of .a specified thickness. This problem is also being investigated computafionally by an analysis of the results obtained with a Monte Carlo code (the 05R code) that generates neutron histories for various sources and media. Other basic studies include the development of codes for analysis of special types of reactor shields, as well as investigations of the interactions of high-energy particles for possible application to the degign of space-~vehicle shielding. Finally, applied shilelding research, the third category listed above, covers various experiments with shield mockups at the Lid Tank Shielding Facility (LISF), the BSF,.and the TSF and analyses or pre- analyses of the results of the experiments. It further covers the application of available shielding data to specific shield designs. Preparations are being made currently for a series of experiments at the TSF with a lithium-hydride—depleted-uranium reactor shield which was designed by Pratt & Whitney. The TSR~II will be encased in the shield, and radiation intensitles will be measured both around the re- actor shield and in the TSF cylindrical detector shield. Preparatory to the experiment, a preanalysis 1s being performed. The LTSF has similarly been engaged in work for Pratt & Whitney. The most recent series of experiments, which is described here and is being analyzed by Pratt & Whitney, has been an investligation of the production of secondary gamma rays in U~LiH slab configurations. . Measurements have also been made at the TSF on a shield configuration for Aerojet-General. 81 9. DEVELOPMENT OF REACTORS FOR SHIEIDING RESEARCH Tower Shielding Reactor II (TSR~II) It was previously reported1 that final assembly and shakedown tests of the TSER-11 were under way and that a series of critical experiments with the complete assembly was being initiated. vBefore the experiments were begun, however, the control-mechanism housing in the internal reflector region was modified in an attempt to lncrease the control-rod worth and to.simplify the design. The original design consisted of a ll-in.-diam central aluminum ball and 2~in.-diam aluminum plugs extending from the ball through holes in the control plates, as shown in Fig. 9.la. In the new design the diameter of the central ball has been increased to 11 1/2 in. (see Fig. 9.1b), and the control plates are unperforated. The thickness of the control plates has been reduced from 1 to 1/2 in., but the new plates contain approximately the same amount of BsC as the perforated set. The effect of these changes is discussed below in con=- Junction with the description of the latest series of critical experi=- ments. Critical Experiments In an earlier series of experiments at the Critical Experiments Facility,? 1t was found that with 230 g of U?2° placed on the outer skin of the sphefiéal control-mechahism housing the regulating rod was worth O.B%AAk/k. In the final-assembly, which was used for the most recent series of experimenfis, new cover plates containing 233 g of U233 were used for the housing, and, as discussed above, the contents within the housing were modified. The combined effect of these changes was an increase of the total rod worth from 0.3 to 0.43% Ak/k. N "ANP Semiann. Prog. Rep. April 30, 1960," ORNL-2942, p. 11l. 2"ANP Semiann. Prog. Rep. Oct. 31, 1959," ORNL-2840, p. 146. 82 £8 SHIM-SAFETY ROD UNCLASSIFIED PHOTO 51383 REGULATING ROD CENTRAL SPHERE RCRVER PLATE COVER PLATE r ALUMINUM PLUG CENTRAL SPHERE L] " 3 ol l 7 Bt v W (o) PSR . . v 4 Yuat | Fig. 9.1. TSR-II Control Mechanism Housing. (a) ll-in.-diam model with aluminum ,plugs. (b) 11 1/2-in.-diam model with plugs eliminated. The shim-safety rod calibration with the complete assembly was divided into a measurement of the shutdown k and the excess k. The shut=- dcwn k measurement was made by the "pulsed~neutron' method previously used for similar measurements on the BSR-II.? Basically it consists of pulsing neutrons into a reactor, elther at delayed critical or with a known amount of negative reactivity, and measuring the decay of the neutron density at the surface of the core following the pulse. The interpretation of the experiment is then based on the inhour equation. The neutrons pulsed into the TSR-II were supplied by a 300~kv accelerator and an accelerafior target positioned close to the TSR-II core (via a channel through the shield). The decaying neutron pulse was observed with a fission chamber located at the surface of the core about 105 deg from the target, and pulses from the fission chamber, after amplification and shaping, were fed to an 18=channel time analyzer. Measurements were made for three positions of the control rods. In the first measurement the combined worth of the control rods from the critical position to the position of full ingertion was =3.52 = 0.12 dollars. In the second measurement the worth of a single rod from the critical position (with all rods withdrawn equal amounts at critical) to full insertion was ~0.62 £ 0.05 dollars. In the third measurement the combined worth of the control rods from the fully withdrawn position (boron poison was added to the core until it was just critical) to the fully inserted position was =4.77 + 0.14 dollars. The results of the experiment are plotted in Fig. 9.2. The excess k measurements were made by adding boric acid to the water until the shim-safety rods could be completely withdrawn. The resulting change in reactivity, computed with the BNU code on the IBM- 704 computer, was 1.344% Ak/k at 26.5°C. The total excess at 20°C was determined by adding to this amount 0.085% Ak/k, which was the amount still available in the regulating rod, and 0.048% Ak/k, which is the temperature correction from 26.5 to 20°C. This gives a total excess of 1.48% Ak/k at 20°C. 3Tbid., p. 113. 84 The excess k was also com= UNCLASSIFIED as | | | | I 2|~0|—osc|3—37n2 puted by taking the differerce 0 A MEASURED BY PULSED-NEUTRON METHOD, | in shutdown k, as measured by ’ & Bgqs = 0.00B (ASSUMED) a5 1\ the pulsed~-neutron-source tech= 10 \ nique, with and without a boron x 4 ‘ S N PREVIOUSLY CBTAINED TSR-1I SHIM solution in the reactor. The 8225 X ROD CALIBRATION CURVE ;20 N excess obtained, assuming a 2z, N 5 N 1 2 N value of 0.008 for Bere? which o N ™ is the value measured in the 1.0 4 BSR~I, was 1.56%. The total 05 ~ 7 . ‘ \\\\M shim-safety~rod worth obtained 0 N 0 02 04 06 08 1.0 1.2 1.4 16 18 20 by the plllSEd-neutrOIl measure= SEPARATION DISTANCE FROM FUEL (in.) . ments was 3.82%, as shown in Fig. 9.2. Reactivity Worth of . _ Y TSR-IT Shim-Safety Rods. Fig. 9.2. The value obtained before the final alterations to. the control~mechanism housing were made was 2.55%. A measurement of the temperature coefficient of the reactor resulted in a value of -1.24 X 10=% (Ak/k)/°F for a mean core temperature of 140°F. This value agrees quite well with a calculated coefficient of -1.34 x 10~* (Ak/k)/°F at 140°F but differs somewhat from a previously reported? experimental value of -8.5 x 10”7 (Ak/k)/°F. The difference is probably due to the lesser volume of water in the new control=- mechanism housing region, a region which has a positive temperature éoefficient. _ Void coefficients of reactivity were measured both for voids in the water annulus between the control-mechanism housing and the core region and for voids in the water pockets behind the control plates inside the control-mechanism housing. Styrofoam was used to simulate the voids. The coefficient is ~6.3 X 1077 (Ak/x)/cm® of voild outside the control- mechanism housing and +1.06 x 107% (Ak/k)/cm® of void inside the control- mechanism housing. The total volume of water inside the control-mechsnism housing is 20 000 cm’. 41.. B. Holland and C. E. Clifford, "Description of the Tower Shielding Reactor II and Proposed Experiments, ORNL-2747, p. 19 (1959). 85 The pulsed=-neutron technique described above was also used to measure the ratio of the prompt-neutron generation tlme to the effective delayed~neutron fraction, Z/Beff, of the TSR-II. The resulting value was 6.61 * 0.16 msec, which, assumping a Borr value of 0.008 (the value measured for the BSR-I), gives a value of 53 pusec for I. A calculated value previously reported5 was 50 péec. An experimental investigation of the importance of the U235 gdded on the spherical cover plates of the control-mechanism housing showed that the ratio of the percentage change in mass to the reactivity change, (aM/M)/(2k/k), is 2.5. This indicates that the fuel in the shell region is very important, as would be ex?ected, and thus should have a con~- siderable effect on the neutron flux distribution and control-rod worth. The reactivity effects of shield changes were determined both calculationally and experimentally, as shown in Table 9.1, The results indicate that the reactor can be operétedfleither unreflected or without the lead-boral shield and that no dangerbfis reactivity effects due to shield changes can occur with the presently available shields. 51bid., p. 21. Table 9.1. Reactivity Effects of ShieldAVariations on the TSR-IT Reactivity Change (% Ak/k) Shield Configuration Ixperimental Calculated Normal, except that there was no water -0.08 | ~0.26 reflector outslide the reactor pressure ' vessel Lead~boral reflector removed from -0.38 ~1.73 normal configuration 7 Water reflector replaced by beam shield +0.17 +0.65 (504 water and 50% lead) adjacent to reactor pressure vessel Water removed from beam shield +0.37 +1.57 86 Thermal-=Neutron Flux Distributions Following the substitution of the fuel-containing platés for the original, unfueled cover plates of the control-mechanism housing, new calculations of the thermal-neutron flux distribution throughout the core were made. IFor these-calculations, the control plates were repre- sented by a continuous boral shell with the boron concentration adjuéted so that the transmission of neutrons through the shell was equivalent to the ratio of the area not occupied by the control plates to the fotal area for the particular control radius under cohsideration._ This approxi~ mation resulted in a value of k that was 5% higher than the experimental value and a thermal-neutron flux distribution through the core that | agreed in shape with the measured distribution, except at the inner surface of the core. | Additional calculations were later made with the control region set ét the same separation from the fuel as was observed in the experi- ments and the boron concentration in the control region adjusted to | make the calculated flux shape agree with the experimental-results.. These gave an effective multiplication factor of 1.011 and the thermal- neutron fluxes shown in Fig. 9.3. Curve A represents the calculated | flux for the clean, cold critlcal condition, and curve C shows the corresponding measured flux. Curve B shows the calculated thermal~ - neutron distribution for higher power operation. For the latter calcu- lation, it was assumed that the control plates were withdrawn from 0.85 to 1.49 in. to compensate for reactivity losses due to temperature rise and xXenon poisoning, and the boron concentration in the control shell was adjusted to produce a change in k of +1.26%. (The measured change is 1.2%.) Power Distribution and Heat Flux in TSR=II Core It is assumed that the power distribution in the TSR~IT core is the same as the thermal-neutron flux distribution, and the agreement between the calculated and measured fluxes for the cold, clean eritical condition 87 UNCLASSIFIED 2-01-060-93 B4C CONTROL PLATE REGION CASE e CASE Al AND Hp0 BORAL Al AND H50 (x10"3) Al Hp0~l 1.8 FUEL \ A CALCULATION B CALCULATION; HIGHER-POWER 7 - OPERATION C EXPERIMENTAL ; GOLD FOIL DATA c.8 ! 0.6 ——— i — ] THERMAL NEUTRON FLUX (neutrons-cm_z- sec” ) 0.4 \ 0.2 0 , | | \ | u 0 10 20 30 40 50 60 70 RADIUS (cm) - Fig. 9.3. Thermal-Neutron Flux Distributions in the TSR-II. noted above gives sufficient confidence in the calculations to encouragé the use of the fluxes calculated for operating conditions to predict the power distributions for operating conditions. Aécordingly, power distri- butions for 5-Mw operation have been calculated. The power generated in the single cylindrical element at the lower end of the core was calcu=- lated to be 19.2 kw; it did not change much with changes in the flux distribution. The power generated in the spherical fuel plates on the 88 - - control-mechanism housing varies from 113 kw for the cold, clean critical condition to a maximum of 125 kw for 5-Mw operation. Corresponding heat flux values throughout the core have been calcu=~ lated. The average heat flux in the core, excluding the spherical fuel cover plates, is 25 120 Btu/hr-ft?. The heat flux values for the spherical elements can vary from 91 500 to 101 000 Btu/hr-ft? over the life of the core. Exclusive of the cover plates, the maximum heat fluxes in the core vary from 30 500 to 34 750 Btu/hr.ft?. These fluxes occur in the infiermost fuel plate of an annular element. The maximum flux in the outermost fuel plate of an annfilar element can vary from 20 600 to 21 050 Btu/hr.-ft2. The heat flux values for the spherical cover plates are much higher than those for the other plates because of the flux peaking and because all the heat flux is assumed to flow outward frem.the cover plates. Flow Distribution Studies The power generation in the annular fuel elements changes when the ~control rods are withdrawn, but the peak values do not approach those obtained before the addition of the U22°-~loaded cover plates. Since - the annular baffle plates were designed for earlier, more highly peaked, power generation distributions, and since the power generation varies so little in the outer plates, these baffle plates should be adequate for the expected range of power generation. Further analysis of the high heat flux in the spherical cover plates will be required, however; in order to determine whether the water flow 1s adequate over the whole cover plate surface, The baffle plate design for the lower and upper central elements 1s not affected by changes in thermal-neutron flux distribution, since a uniform flow distribution suffices in all cases. An acceptable flow distribution has been achleved for these two reglons, although studies are continuing in an effort to reduce fhe pressure drop through the upper fuel element baffle plate. 89 Bulk Shielding Reactor II (BSR=II) The progress of the design, fabrication, and testing of a U0z~ stainless steel core for the Bulk Shielding Facility reactor has been reported periodically, and at the time of the last report6 the core had just undergone a series of static and dynamic tests at the SPERT-I Facility of the National Reactor Testing Station. The resulting experi- mental datq have since been analyzed and are summarized here. Static Measurements The loading configuration which was employed for the static tests of the BSR~II is shown in Fig. 9.4, with the four pairs of control plates referred to as rods 1, 2, 3, and 4. The reactivity worth of rod 1 as a function of its vertical position is plotted in Fig. 9.5 along with the combined reactivity worth of rods 2, 3, and 4 operated as a gang. These measurements were made by polsoning the reactor with a boron solution in the SPERT tank, a method which obviated the need for compensating motion of other poison rods which would have perturbed the flux shape. In each measurement the rod (or rods) not being evaluated was kept at its upper -limit. The temperature coefficlent of the BSR~IT, measured for the condi- tion of uniform temperature change throughout the core and reflector, was found to have an average value of 0.0136 dollars/°c for the range from 15 to 85°C. Extrapolation to a lower temperature indicates that the coefficient changes sign at about 12°C, becoming positive for lower temperatures. Vold coefficient measurements made with magnesium strips simulating the voids indicated that, for uniform void fractions of up to 4% of the moderator volume, the coefficient showed a linear dependence, with a value of =8.0 x 10™% dollars/cm® over the range. In these measurements the control-rod elements and a transient-rod element (see below) could 6"ANP Semiann. Prog. Rep. April 30, 1960," ORNL-2942, p. 103. 20 a} Ya FUEL ELEMENT CONTROL ROD ELEMENT NO. 1 Fig., 9.4, TRANSIENT ROD UNCLASSIFIED - 2-01-058-0-558 Core Configuration Used in Static Tests of the BSR-IT at the SPERT-I Facility. 91 REACTIVITY (dollars) -0.5 —2.0 -2.5 Fig. 9.5. UNCLASSIFIED 2—01-058—-0—-559 = .,/ iy N /"/ / NO. 1 ROD \/, / A / 7 / v e ) / . RODS NO. 2,3 AND ,/// /#L"/’_4 GANGED 4 5 6 7 8 9 10 1 12 13 14 ROD POSITION ABOVE FULL INSERTION (in.} Reactivity of BSR-IL Control Rods as a Function of Distance Above Full Insertion. not be voided, and it was necessary to make separate measurements with magnesium strips adjacent to these elements to obtain data which could be graphically interpolated across the elements. The resulting average magnesium-vold coefficlent is ~11 X 107 dollars/cm3 or 0.46 dollar for each 1% of moderator void. An experiment for determining the'comparative effects of magnesium voids and air voids (the air being contained in Styrofoam) in one fuel element has indicated that the air-void coeffi- cient for the core is =14 x 10~% dollars/cm® or 0.6 dollars for each 1% of moderator void. 92 The flux shape in the BSR~II has been mapped by a total of 186 gold foils exposed in and adjacent to the core while the reactor was at a power level of 500 w and the four control rods were moved in gang to maintain a constant flux level. The results of the flux mapping were comparable with results obtained in a calculation made with the use of the PDQ code. Dynamic Measurements For the dynamic tests the reactor was made critical with the tran- sient rod (a "double-ended" device with a lower poison section and an upper fuel section) fully inserted and rods'2, 3, and 4 fully withdréwn; i.e., only rod 1 was used. The transient rod, whose reactivity worth was about 5.5 dollars (considerably more than: the 2.07 dollars required for a maximum transient), was then'withdréwn to almost its full length, shutting dbwn the reactor. The core was then left without a source until the neutron flux dropped to a level of about 1 count/min in the fission chamber, which represented a power level of about 10~7 w. While the flux was decaying, the regulating rod (rod 1) was withdrawn a measured amount determined from the reactivity curve (Fig. 9.5) and the inhour equation to produce an excursion of the desired period. When the flux had»décayed to the level indicated above, a sequence timer was actuated, which, in turn, started two 36-channel oscillograph recorders and released the transient rod. As the transient rod moved downward, the reactivity went from negative through critical to the predetermined positive amount, which was reached when the transient rod was seated. The transient then occurred when triggered by a random neutron. The transients were termi- nated either by the control system or by the manual release of the rex maining three controls after the peak of the power burst has been reached. As reported_previously,6 the first series of dynamic tests involved self-limiting transients, and at a peak power of 226 My a small permanent warping occurred in some of the fuel plates loCatéd near the éenter of the reactor. The self-limiting tests were then terminated, and the data that had alresdy been collected were extrapolated on the basis of the similar APPR P18/19 core, which had previously been tested at the SPERT-I 93 Facility. Before the transient testing was resumed, a complete second channel of safety instrumentation was I1nstalled to providé a safety margin against a scram-system failure. The remaining series of measurements consisted of two types: (1) tests in which the full safety system was operating, which invariably resulted in the scram being initiated by the period safety system, and (2) tests in which the period safety system was rendered inoperative and the level safety system alone terminated the excursion. The results of these tests are shown in Fig. 9.6 as plots of peak power versus. reciprocel period. It is apparent from the curves that a sharp rise in peak power occurs as the reciprocal period passes about 100 sec™t , which corresponds to a period of less than 10 msec. The excursions terminated by the level scram system are asymptotic, for small inverse periods, to 100 kw, which is the scram set point on the level scram syétem. The period safety system is much more effective in the range tested than the level safety system, since a factor of at leést 103 separates the peak powers for a given inverse period in the two cases. For comparison, a curve for the self~limiting tests is also shown in Fig. 9.6. It is apparent that, 1n the rénge tested, the control system was extremely effective. At the shortest period measured, the peak powers of self-limiting and period-scram~terminated excursions differed by a factor of about 25, and those of the self-limiting and level-scram~terminated excursions differed by about a factor .of 12. Safety Evaluation of BSR~IT In order to ascertain whether the BSR~II core will be as safe for operations at the BSF as the BSR-I core, the curves of peak power versus reactivity obtained for the BSR~II in the SPERT-I tests were compared with similar curves obtained for an aluminum~clad core which had physical characteristics similar to those of the BSR~I and which had previously been tested at the SPERT~I Facility. The self=limited peak powers in the BSR~II are higher by about a factor of 3 than those in the aluminum- clad core; however, in the region of competence of the control system 9% -y UNCLASSIFIED 2-01-058-0-520R! PERIOD {msec) 100 10 i 100 1000 Mw 100 Mw 10 Mw B A Mw = o a x < w a 100 kw 10 kw § i0 100 1000 RECIPROCAL PERIOD (sec ') Fig. 9.6. Peak Power of BSR-II as a Function of Inverse Period for Controlled Transients. the peak powers that will occur in the BSR~II with the control system operating are considerably lower than the self-limiting excursions In the BSR=I for the same reactivity step. ' Thus, if reliance on the BSR~II safety system is postulated, the BSR~IT is safer than the BSR-I in the reactivity range up to 2.1 dollars of excess reactivity. Beyond this range, where the. usefulness of the safety systems of both reactors becomes exceed= ingly limitéd, the difference in peak power of a factor of 2 or 3 may be significant but is probably not very decisive. Present Status Piecemeal return of the BSR~IT elements to ORNL has been completed, and the reactor awalts final authorization by fhe Comw mission before further use at the BSF. A supplement to the original safeguards report7 is being prepared requesting approval for power operations with realistic excess reactivity loadings. A small amount of corrosion has occurred around brazed joints in the elements that is poséibly due to overheating in the process of brazing.' It may be that leSer temperature control during element fabrication will eliminate the corrosion. 7E. G. Silver and J. Lewin, "Safeguard Report for a Stainless Steel Research Reactor for the BSF (BSR~IIL)," ORNL-2470 (1958). 95 10. DEVELOPMENT OF RADITATION DETECTION EQUIPMENT Gamma-Ray Spectroscopy Monte Carlo Calculations of Response Functions of Gamma-Ray Scintillation Detectors Monte Carlo codes for calculating the response functions of three scintillation materials, NaI? xylene, and CsI, were discussed previ- ously.l’2 The Nal and Csl codes have since been used to compare calcu- lated responses of the two crystals, as well as to compare calculated responses with experimental responses. The latter comparison not only indicates the utility of the codes but aiso ailds in the evaluation of background effects nonexistent in the codes. The results of the Nal code have also been compared with the results of another Monte Carlo‘ calculation. In addition, the Nal code has been used to investigate the effect on the response functions of varying the dimensions of the crystals. Comparison of Nal Calculations with Experiments. In order to further verify the accuracy of the codés, calculated pulse-height spectra (plotted as counts per Mev, with the data normalized to a total response of one count) for a 3- by 3-in. Nal scintillation crystal have been compared with experimental pulse-height spectra for various source energies ranging from 0.478 to 7.48 Mev. In the comparisons with Be” and Cs'37 gamma rays, the calculated data show a deeper dip Jjust below the photOPeak and deviate to a certain extent at the smaller pulse heights.» This deviation can be accounted for partially by background effects which were not taken into consideration in the experiment and partially by the‘poorer statistical accuracy of the calculation in this region. At sofirce energies above 1 Mev, the deviation between calculation and experiment persists at small pulse heights, with a definitely increasing trend with increasing source energy. This is apparent, for example, in Fig. 10.l, the experimental 1"ANP Semiann. Prog. Rep. April 30, 1960," ORNL-2942, p. 87. 2"ANP Semiann. Prog. Rep. Oct. 31, 1959," ORNL-2840, p. 97. 96 % . i a UNCLASSIFIED 2-01-059-498 PRESENT | LAZAR, CALC. ET AL, PHOTOFRACTION C.1476 0.107 INTRINSIC EFF, 0.3976 e LAZAR, £T AL., EXP'T, ORNL-2076 N2 (p,ay) €2, £, =1.2 Mev A CALCULATED COUNTS PER Mev ANALYTIC ZERO VALUE Nal LQ.B cm 3'|n.-'—| O 1.0 2.0 3.0 4.0 5.0 PULSE HEIGHT (Mev) 3in. Fig. 10.1l. Comparison of Cal- culated Pulse-Height Spectrum with Experimental Values of Lazar and Willard for 4.43-Mev Gamma Rays In- cident on a 3- by 3-in. NaI Secintil- lation Crystal. Calculated values are normalized to a total response of one count. values being those of Lazar and Willard.? At large pulse heights the calculated and experimental data for the higher source energies 'agrée remarkably well. This also is apparent in Fig. 10.1. Comparison of Nal Calculations with Miller-Snow Calculations. The results produced with the Nal code have also been compared with Monte ~Carlo calculations by Miller and 4 for several source energies. Snow At low source energies the agree- ment 1s good,_but at high source energles a definite disagreement is noted, particularly at the higher energy pulse heights. This is demon- strated in Fig. 10.2, which shows the comparison for a 6.13-Mev source and a 6- by 6-in. Nal crystal with perfect resolution. (The lower- energj pulse-heights are not shown, since they disagree only slightly.) As an aid in analysis of the dif- feréfices, a. calculation was per- formed in which the effects of all secondafj radiation were neglectéd, the curve of which is also shown in Fig. 10.2. It is evident that the Miller-Snow results follow the éurve, neglecting secondary'effects, up to the second escape peak. At this 3N. H. Lazar and H. B. Wlllard Phys. Semiann. Prog Rep. March 10, 1956, ORNL-2076, p. 55. 4W., F. Miller and W. J. Snow, Rev. -Sci. Instr. 31, 39 (1960). 97 UNCLASSIFIED 2-01-059-4996 MILLER, | PRESENT 6.3 Mev ET AL, CALC. PHOTOFRACTION | ©.363 0.247 INTRINSIC EFF. 0.853 0.857 e MILLER, £7 AL. A PRESENT CALC. COUNTS PER Mev ALL SECCNDARIES NO SECONDARIES 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 PULSE HEIGHT {Mev) Fig. 10.2. Comparison of Present Calculated Pulse-Height Spectrum with Calculated Spectrum of Miller and Snow for ©6.13-Mev Gamma Rays In- cident on a Zero-Resolution Nal Scintillation Crystal. Curves are normalized to total response of one count. point there is a discontinuity in the spectrum, as well as another at the first escape peak. The data from the present calculatiocn, in con- trast, present a smooth curve, generally above the Miller-Snow results. The effect of the discrepancy is a 50% or a 30% difference in the peak- to-total ratio (photofraction), depending upon which data are used as a basis for comparison. The difference in the two calculations appears to be almost entirely due to the difference in the contribution of the " bremsstrahlung. The Miller-Snow data show little or no bremsstrahlung effects below the second escape peak, and yet they show some effect b3 o8 -y immediately above that peak. There does not seem to be a reasonable explanation for this or for the discontinuity at the first escape peak. Comparison of Nal and CsI Calculations. The CsI code was developed as a result of the interest evinced recently in the possibility of achieving better defined pulse-height spectra by'the use of CsI rather than Nal as a scintillator. In order to investigate the relative merits of the tfio scintlllators, calculations were performed with both the CsI code and the Nal code to obtain the response functions of the two crystals to 0.662-, 1.368-, 2.754~-, and 8-Mev gamma rays. For these calculations the two crystals were assumed to have linear dimensions that were inversely proportional to the crystal densities. On this basils, the volume and weight of the CsI crystal were assumed to be 0.665 and 0.816, respectively, of the volume and weight of the NaI crystal. The calculated results for the 0.662-Mev source are shown in Fig. 10.3. The CsI crystal, shown in solid outlines, is taken to be 5 in. in diameter and 3 in. long. The corresponding Nal crystal is indicated by the dashed outline. The source was assumed to be a point isotropic source at a distance of 12 in. from the crystal, with the gamma rays collimated so as to impinge upon a l/2-in.-diam circle centered on the face of the crystal. Both curves are normalized to one source photon, rather than at the photopeak, as might be suspected at first glance. The intrinsic efficiencies and photofrac- tions are seen to be very nearly equal, with a slight edge favoring CsI. Both cases were run with 2000 source particle histories. The broadening coefficients for both cases were obtained from experimental data for the two materials. Comparison of CsI Calculations with Experiments. Calculations with the Csl code were compared with experimental measurements made with a 5-in.-diam by 3 1/2-in.-long CsI crystal. The results for a 0.662-Mev cs?37 source are shown in Fig. 10.4. The crystal and source geometries are indicated on the figure, and the curves were normalized at the photo- ~peak. The area between the two curves indicates the experimental back- ground effects attributed to backscattering and collimator effects. 99 UNCLASSIFIED 2-04-0359-492 Csl Nal 0.9214 0.8424 4.510 0.9207 0.8173 3,677 INTRINSIC EFFICIENCY PHOTOFRACTION DENSITY (g/cc) B8 o COUNTS PER Mev | 7.788cm 6.35¢cm l ’ 30.48 cm i -z — 0635cm ‘,L Cst I.‘..—Nal | I o 762cm 9.346 cm ) o 0.2 0.4 06 0.8 1.0 PULSE HEIGHT (Mev) Fig. 10.3. Comparison of Calculated Nal and CsI Pulse-Height Spectra for 0.662-Mev Gamma Rays. Data normalized to one source photon. A similar comparison for 1.368- and 2.754-Mev gamma rays from a Na% source 1s presented in Fig. 10.5. Pulse-height spectra for the two energies were calculated separatély, both normalized to one source photon, and then added together. The combined calculated spectrum was then normalized to the experimental spectrum gt the higher of the two photopeaks. As before, the experimental backgroundreffects can clearly hbe seen. The hint of a meximum on the calculated curve at 1.73 Mev corresponds to the second escape peak below 2,754 Mev, while the maximum at 0.86 Mev shows the first escape peak below 1.368 Mev. The peak at 0.5 Mev on the calculated curve is spurious and can be attributed to a random high point in the unbroadened 100 UNCLASSIFIED 2-04-059-494 COUNTS PER Mev EXPERIMENTAL CALCULATED 0 o 0.2 03 04 0.5 06 07 0.8 PULSE- HEIGHT (Mev) Fig. 10.4. Comparison of Calculated Pulse-Height Spectrum with Experi- mental Values for O.662-Mev Gamms Rays Incident on a CsI Scintillation Crystal. Data normalized at the photopeak. UNCLASSIFIED 2-01-059-495 3 m . EXPERIMENTAL COUNTS PER Mev 6‘- CALCULATED © o 0.5 1.0 1.5 2.0 2.5 30 PULSE HEIGHT (Mev} : Fig. 10.5. . Comparison of Calculated Pulse-Helght Spectrum with Experi- “mental Values for 1.368- and 2.754-Mev Na?% Gamma Rays. Data normalized at photopeaks. | - e - 101 spectrum. The corresponding peak in the experimental data is a back- ground effect. Calculations to'Investigate the Effects of Scintillator Dimensions on the Pulse-Height Spectra of Large Nal Crystals. A previously reported5 investigation of the effects of the scintillator dimensions was limited to source energies below 2 Mev, because the Nal code did not at that time include secondary effects. Since the code now considers annihilation radiation and bremsstrahlung, the calculations have been repeated for source energies of 2.75- and 7.5-Mev gamma rays incident on an Nal crystal whose standard dimensions were taken to be 9 in. for the diameter and 10 in. for the length. The effects of varying the length and diameter of the crystal were calculated, as well as the effect of including an axial well in the end of the crystal. Because of the idealized geometry adopbted in the code, the contribution to thevtail of the experimental spectrum due to gamma rays degraded in energy by scattering in the walls of the col- limator and source container is conspicuously absent from the computed spectra. Nevertheless, the calculations do show relative effects. The calculated results for the 2.75- and 7.5-Mev monoenergetic gamma rays incident on the standard crystal are shown in Fig. 10.6. The re- sults havé beenlsmoothed by the use of a Gaussian smoothing function derived from experimental observation. Each curve represents the sum of the points which correspond to the Gaussian full-energy peak plus the histogram lines whiéh represent the predicted tall spectrum. Because of the smoothing, the tail extends above the full gamma-ray energy in each case. The low intensity of the spectrum at low energies suggests that- such a tail would not be experimentally observable. The heavy arrow ad jacent to the diagrams indicates the position and direction of the in- cident collimated beam, chosen to be 2 ecm in diameter for all these cal- culations. The 2.75-Mev data from Fig. 10.6 are plotted on a linear scale in’ Fig. 10.7 to give an idea of area. The solid lines represent the same °"ANP Semiann. Prog. Rep. Oct. 31, 1959," p. 99. 102 . Y UNCLASSIFIED 2-04-058-0-542 : o CASE 05 {— m—) P/T=0.734t0006 CASE C 02 P/T =0.57+0007 04 005 o S FRACTION OF ALL INTERACTIONS PER Mev 00 - SMOOTHED TAIL SPECTRUM ® SMOOTHED PEAK SPECTRUM 0,005 - GRAPHICAL FIT TO SUM OF PEAK AND TAIL 0002 0.004 o 10 2.0 30 40 50 60 70 80 90. GAMMA-RAY ENERGY ABSORBED IN CRYSTAL (Mev) Fig. 10.6. .Estimated Smoothed Pulse-Height Spectra for 2.75- and 7.5-Mev Gamma Rays Incident on a 9-in.-diam, 10-in.-long, Nal Crystal. smoothed peak and tail spectra shown on the previous figuré, while the broken lines are the sum of the smoothed peak and tail. Interestingly, the intensity of the predicted tail under the photopeak is enough to raise the peak ~2%. This increment would not be experimentally identified because it occurs so close to the full energy. The contribution of the smoothed tail also slightly shifts the apparent position of the photopeak. The small rectangles shown on the figure represent the unsmoothed calcu- lated energy loss spectrum, the height of the réctangle corresponding to the estimated standard deviation calculated by the code. Such features as the single escape peak at ~2.2 Mev show very clearly in such a pre- sentation; therefore, all the results of the calculations are presented 103 2-01-055-0-513 as unsmoothed spectra. The large a5 block shown centered around the 40 5 photopeak has an area equal to that of the photopeak and is intended to O VALUES t ERRORS FOR 35 UNSMOOTHED SPECTRA —— SMOOTHED PEAK AND TAiL =—--SUM OF SMOCTHED PEAK AND TAIL peak fractions. The primary purpose assist in comparison of tall and o o ] f in presenting Figs. 10.6 and 10.7 I [ | AREA QOF BLOCK = AREA OF PEAK is to clarify the graphical nomen- N v clature, which will be consistent in the comparisons which follow. The effect of a well through | | 1 275 Mev n FRACTION OF ALL INTERACTION PER Mev which a beam of 2.75-Mev gamma rays enter the crystal i1s demonstrated in Fig. 10.8. The path length of e e e . L —F == — fl the gamma rays through the crystal 05 1 bl | is kept constant. The "no second- » I o ' . " . o b | ?5fl aries” cases are those cases in o 05 1.0 1.5 2.0 2.5 3.0 3.5 GAMMA-RAY ENERGY ABSORBED IN CRYSTAL (Mev) which no brems Strahlung 1s assumed Fig. 10.7. Smoothed and Un- and pair production processes are smoothed Estimated Spectra. considered as pure absorption. Although the curves for these cases agree at low energies, it is clear that the use of a well has a strongly marked effect on the section of the gspectrum just below the photopeak. The difference is satisfactorily interpreted, both in approximate magnitude and in energy, as arising from the escape from the crystal of radiation Compton-scattered through angles near 180 deg by material close to the surface of the crystal. The histograms which include the secondary effects also demonstrate the marked influence of the well, the photofraction increasing from 73 to 83% with the inclusion of the well. Again the differences are all above 2 Mev, the curves mingling at low energies. The reduction in the single escape peak is especially noteworthy. | From a similar set of cases for 7.5-Mev gamma rays, shown in Fig. 10.9, it is immediately clear that the "no secondaries" condition 104 .i »r UNCLASSIFIED 2-01-058-0-515 P/4 B A 0.8 T L I | E | Lo —_——— ) — ———-+—— H ! ! 275 ) 0.7 |— Mev te—10 in.— o B = 06 |- === —-—-—1— gin. x 2.75 B _ 4 w0 =z : o 5 03 | ! l =4 5 CASE SYMBOL SECONDARIES P/T 5 B YES 0.734 £ 0.006 :,l 04 — —_— NO C.84 N E --——- YES o.s3o¢o.oos£ o — NO 0.875 3 03 } - O = o w $_* ,£ 0.2 E J I L‘| i I 3 X . —F 8 ;E&F i L™ e g L:L 0 0 05 1.0 1.5 20 2.5 30 GAMMA-RAY ENERGY ABSORBED IN CRYSTAL ( Mev) Fig. 10.8. Effect of a Well in a Nal Crystal on the Estimated Energy Loss Spectrum for 2.75-Mev Photons. produces a bad estimate. The well again produces a sizeable improvement in the histograms calculated with all secondaries, the photopeak value increasing from 57 to 70%. The effect of changing the - crystal length is illustrated ifi Fig. 10.10. Crystal B, of coufse, is the standard 10-in.-long crystal discussed earlier. Cxrys- tal B is a c¢rystal with a well and a 10-in. penetration, while crystal G is a crystal with a well and a 6-in. penetration length., When the spectrum from crystal E is compared to that from crystal G, a slight differ- ence is noted in the region near the full energy of 2.75 Mev, but the major difference occurs-at low energies, where the é-in. crystal yields a much higher predicted intensity. This is precisely what is expected from the "analytical zero" concept. The analytical zero, or analytically calculated zero-energy intercept of the pulse-height spectrum, may be calculated by considering the escape out of the far end of the crystal of Compton gamma rays scattered through very small angles, since this is the only process in idealized geometry by which one can get such a small amount of energy deposited in the crystal. TFor the simple case of an axially collimated beam, the analytical zero is Just (L —¢) - I(o) = ——— 2n lrrg — € m 2 0 0 J 2 B2 105 where € = intrinsic crystal efficiency, =1 - e_uz, ! = path length of gamma-ray beam through crystal, n, = electron density of the crystal, UNCLASSIFIED 2-01-058-0-546 10 T f 09 0.8 C e —e—— @ @——— ——f— - 7.5 o Mey 0.7 10in. P/2 > s £ oe— T = e O e Q £ YES 0.83 0.005 I . g 03 - NONE NO 0.875 T T Q [ T YES 0.88 £0.005 1 @ NONE NO 0.9 b I_}[. 0.2 ; | | I B 04 : i UL . : L‘IL 'f .J ' | L S v e St T £ 0 0 05 1.0 1.5 2.0 2.5 GAMMA—-RAY ENERGY ABSORBED IN CRYSTAL (Mev) Fig. 10.11. Effect of Diameter of a Nal Crystal on the Estimated Energy Loss Spectra. 108 control over the photofraction and over the shape of the tail spectrum can be obtained by care in the choice of crystal dimen- sions., ‘Efficient design of spectrometers would seem to dic- tate taking advantage of calcu- lations similar to the above, perhaps altered to fit the region of parameters appropriate to a given problem. Experimental Study of the Gamma- Ray Response of Large NaI(T1) Crystals Experimental investigations of the response of Nal scintil- lation crystals to gamma rays have been carried out with the 8-in. -diam by 8-in.-long com- posite crystal described previ- ously6 and have been greatly 6Ibid., p. 1l16. s - assisted by the availability of the IBM-704 Monte Carlo Nal code discussed in the preceding section. The source gamma rays have been the 0.662-Mev gamma, ray from Cs'37 and the 1.368- and 2.754-Mev gamma rays from Na2#%, The résults for the 0.662-Mev source are shown in'Fig; 10.12. Curve A is the calculated distribution, in histogram form; Curve B is the ex- perimental distribution; and Curve C i1s the anticoincidence distribution when the gamma rays are collimated through a l/2—in.-diam hole along the afiis of an additional 4-in.-diam, 2-in.-thick crystal mounted on'the end of the large crystal. The curves are,individually‘normalized to unit area and then normalized in amplitude at 0.662 Mev. The shape of the experimental curves below 0.2 Mev is due to the electronic characteristics of the system. The use of the anticoincidence arrangement has nearly eliminated the small back-scatter escape peak at 0.45 Mev. This indicates that a well in the end of the large crystal would improve the peak-to-total ratio. The discrepancy evident between calculational and experimental UNCLASSIFIED o oRNCLASSIFIED results below 0.6 Mev may be . explained by postulating that (B CALCULATED DATA SINGLE CRYSTAL DATA the tail of the measured distri- © ANTICOINCIDENCE DATA 3 w bution is greatly increased by o the contribution of gamma rays o degraded in energy by scattering Q (3] -in the source container and in the walls of the shield before 'C). n reaching the crystal. This dis- o 0.05 tribution would not appear in FRACTION OF ALL INTERACTIONS { per Mev) the calculations. The explané— tion is strengthened by the fact: 002 001 OO O ray nerevien S %7 ®® that in other comparisons in- ~ volving uncollimated source gamma Fig.- 10.12. Comparison of the Ex- perimental and Calculated Responses of an 8-in,-diam by 8-in.-long NaI(Tl) gamma rays of the present case Crystal to Colllmated 0. 662-Mev Gamma ' Rays from Cs?t rays rather than the collimated the discrepancy has not occurred. 109 The response of the 8-in.-diam crystal to 1.368- and 2.754-Mev Na?4 gamma. rays 1is given in Fig. 10.13 and is again compared with calculated results. Thé effect of the gamma rays degraded in energy by scattering is again very prominent. The small peak in the experimental data (curve B) is probably due to ammihilation radiation from the lead of the collimator and shield wall and, of course, does not appear in the calculation. Measured resolutions of the experimental crystal for various energies, as well as the comparison of measured and calculated photofractions, are shown in Table 10.1. It is, of course, impossible to reach the calculated values under experimental conditions because of the idealizations inherent in the calculation. The results obtained with the 8-in.-diam by 8-in. - long composite crystal have been much more satisfactory, however, than those obtained with any of the previous approaches. Additional study is needed of possible_effects of the interface between the parts of the com- posite crystal 'and the effectiveness and placement of reflecting materials UNCLASSIFIED ORNL-LR-OWG 47562 1368 Mev 2.754 Mev @ CALCULATED DATA ® EXPERIMENTAL DATA FRACTION OF ALL INTERACTIONS {per Mev) Q5 10 15 20 25 30 GAMMA-RAY ENERGY (Mev) Fig. 10.13. Comparison of the Experimental and Calculated Responses of an 8-in.-diam by 8-in.-Long NaI(Tl) Crystal to Collimated 1.368-Mev and 2.754-Mev Gamma Rays from NaZ24, 110 *3 > Table_lO.l. Cofiparison of Experimental and Calculated Results for the Response of an 8-in.-diam by 8-in.-Long NaI(T1l) Crystal FPhoton Measured Measured a Calculated a Energy Resoclution Photofraction Photofraction (Mev) (%) (%) (%) 0.662 11.8 75 - 92.4 0.899 8.8 66 1.368 - 8.6 61 g2.7 1.840 7.3 60 2.754 6.3 52 68.3 gPhotofraction is defined as the ratio of the area under the total absorption peak to the total area under the pulse-height distribution curve. bValues of the measured photofractions are reproducible to <3%. - in the crystal package. The comparisons with calculated responses in- vite further study leading to optimization of collimators and shields. Intrinsic Line Width in NaI(Tl) Gamma-Ray Spectrometers In the use of NaI(Tl) scintillation crystals. for gama-ray spec- troscopy, it is of considerable importance to achieve a minimum line width in the pulse-height spectrum resulting from monoenergetic gamma rays. Much attention has been devoted to the various factors contributing to the line width of the full-energy peak, reviews of which are available 758 elsewhere. Briefly stated, the over-all line width is the sum of two "W. E. Mott and R. B. Sutton, "Handbuch der Physik," ed. by S. Flligge, - Vol. 45, p. 86, Springer, Berlin (1958). 8C. E. Crouthamel (ed.), "Applied Gamma-Ray Spectrometry," Chap. 2, Pergammon Press, New York {(1960). | 111 contributions. The first is the statistical fluctuation in the pro- duction of photons in the scintillation crjstal by the incident gamma "ray and in the escape of such photons to the cathode of the photomulti- plier tube. The second is the fluctuation in the production of photo- electrons. at the photomultiplier cathode and their subsequent multiplica- tion in the dynode string of the photomultiplier. It is possible to separate the contribution of the photomultiplier from that of the crystal by experiments in which the photomultiplier cathode is illuminated by an external pulsed-light source. Results of these experiments7'9 indicate that the contribution from the crystal is surprisingly large, which can- not be reasonably accounted for on the basis of inhomogenieties in the crystal, optical coupling, etec. For example, in one experiment® the over-all line width of 0.661-Mev gamma rays was found to be 7.7%, which was composed of a 4.0% contribution from the photomultiplier and a 6.6% contribution from the scintillator. (Line width is expressed here as AE/E, where AE is the full width at half maximum of a peak occurring at pulse height E.) The origin of the line broadening from the scintilla- tion crystal has been in question for several years, and the discussion which follows reports an investigation of the effect on the line broadén— ing of the nonlinear response of NaI(Tl) to electrons. Scintillation Response of NaI(Tl) to Electrons. The scintillation response of NaI(Tl) to incident gamma rays results from the interaction of the gamma rays with the atomic structure to eject electrons which are subsequently stopped in the crystal. In this investigation it was impor- tant to obtain the dependence of the response of the crystal on the energy of these electrons. Such information is not directly obtainable from experiments; however, it can be deduced from experimental studies of the response of the crystal to gamma rays. The studies of the pulse- height vs energy relationship for gamma rays on NaI(Tl) have shown °G. @¢. Kelley et al., IRE Trans. Nucl. Sei., NS-3, 57 (1956). 112 a distinct nonlinearity of behavior that is most pronounced at energies ‘below several hundred kev. This is apparent from Fig. 10;14, in which the pulse height per unit energy, L/Ey, for a 2.5-in.-diam by 2-in.-long crystal is plotted -as a function of E7. The ordinate is arbitrarily | normalized to unity at 661 kev, and the data points are unpublished re- sults of experiments at Argonne National Laboratory.lo (The data points: at the three lowest energies are tentative values. The calculated curve shown in the figure is discussed later.) 19Unpublished results of W. W. Managan, I. S. Sherman, S. I Baker, and A. J. MacKay, Argonne National Laboratory. UNCLASSIFIED ORNL—-LR-DWG 52576 .20 145 ' /] ”\'i EXPERIMENTAL / . \\“\ \//CALCULATION o \ p N EXPERIMENTAL POINTS b FROM MANAGAN , efal. ‘ ‘ \‘. 5 ~ o w e ® o o Y - L /E,, PULSE HEIGHT PER UNIT ENERGY e o o |® N 0.95 0.90 - 1 2 5 10 20 50 100 200 500 1000 2000 - 5000 - Ey, GAMMA RAY ENERGY (kev) Fig. 10.14. Comparison of Calculated and Experimental Values of the Pulse-Height Response of a 2.5-in.-diam by 2-in.-High NaI(Tl) Crystal to Gamma Rays. 113 Since the observed nonlinear response to gamma rays can be accounted for by a nonlinear response to electrons, the response of the crystal to low-energy electrons can be calculated from the experimentally observed low-energy gamma-ray data. This is relatively easy to do, since the main features of the gamma-ray response curve for the energy range below 100 kev can be attributed to the response of the crystal to photoelectrons ejécted by the first collisions of the incident photons and the subse- quent electrons that appear as a result of the photoelectric event. A simplified model was adopted to account for the transitions that take place as a result of a photoelectric event, and values of L/Ee Vs Ee were determined. A smooth fit to the resulting data i1s shown in Fig. 10.15 for 1 = Ee < 70 kev. For Ee > 70 kev, the curve was first drawfi in an arbitrary fashion and then adjusted by calculations described below. The ordinate of the final L/Ee Vs Ee curve was arbitrarily normalized to unity at 3 Mev. UNCLASSIFIED ORNL-LR-DWG 52577 i 1 \ N S / 110 \ N , LIGHT INTENSITY PER UNIT ENERGY 4 e 1.05 L/E ™ oy 1.00 1 2 5 10 20 50 100 200 500 1000 2000 5000 . €g, ELECTRON ENERGY (kev) Fig. 10.15. Response of NaI(Tl) to Electrons, as Deduced from Gamma-Ray Data. 114 Calculations of the Intrinsic Line Broadening and Pulse HeightAper Unit Energy for Gamma Rays. It is now possible to demonstrate that the nonlinear response of NaI(Tl) to electrons results in an intrinsic broadening of a monoenergetic gamma-ray line. It is first assumed that all electrons originating in the crystal are stopped, and only those monoenergetic gamma rays which enter the crystal with an energy somewhat greater than 100 kev and terminate their history in a_photoelectric event are considered. Since the histories of the gamma rays vary considerably, the accumulated light output will vary from one gamma ray to another, | because of the nonlinear response to the different Compton electrons and photoelectrons generated in each history. The variation in the‘accumulafied light output is reflected in the spread in the pulse-height spectrum of totally absorbed, monocenergetic gamma rays. In order to calculate the intrinsic broadening, the Monte Carlo code which has been discussed in the preceding papers was altered to include the nonlinear response to the electrons. The calculated pulse-height spectrum was obtained in two forms. The first was a spectrum accounting only for the nonlinear response of the electrons, and the second was a | version of the first which had been altered to include the broadening resulting from photomultiplier statistics. The full width at half maxi- mum of the photomultiplier broadening was assumed to be given by AL = A\ffl} where L is fhe pulse-height and A is a constant chosen according to typical experimental values. For a good photomultiplier, AL/L may be about 4% for the case of 66l-kev gamma rays on NaI(Tl); in the calculation described here, A was chosen such that AL/L was 3.83% at 661 kev. The final electron response curve, shown in Fig. 10.15, was the re- sult of an iterative procedure in which the trial function, L/Ee Vs Ee used in the Monte Carlo code was changed until the resulting L/E7 Vs E7 curve agreed with the experimental curve. The calculated curve shown in Fig. 10.14 is the result of the last iteration. The calculated pulse-height spectrum for the total absorption peak for 66l-kev gamma rays 1s shown in Fig. 10.16. The histogram indicates the extent of the intrinsic broadening and the smooth curvé shows the 115 o0 ORNIVCLASSIEED 78 same peak after the photomulti- plier broadening has been folded 200 INTRINSIC in. Although the histogram is BROADEN ragged to a certain extent be- 100 cause of the statistical nature of the Monte Carlo calculations, 5.5% RESOLUTION 50 the general features of the data INTRINSIC BROADENING AND PHOTOMULT'PLIER BROADENING can be easily explained. The leading peak at the lower pulse- N o height values is due to first- collision photoelectric events, COUNTS (arbitrary units) and the second peak is caused by 3 the buildup of second-collision photoelectric events. The valley between the two peaks is real and not a result of statisties. A measurement of the contri- ~bution of the intrinsic broadening to the total line width cannot 15 125 135 145 155 165 PULSE HEIGHT be obtained directly from the . histogram presented in Fig. 10.16, Fig. 10.16., Calculated Pulse- Height Spectru_m in the Region of the since it does not have a Gaussian Total Absorption Peak for 66l-kev Gamma Rays Incident on a 2.5-in.-diam by 2- in. -High NaI(T1) Crystal. Source-to- resolution can be calculated crystal distance = 10 cm. shape. However, its effective. from the expression - 2 _ p2y1l/2 Ry (R Rp) where R = 5.48% is the resolution of the line, including the photomulti- plier broadening shown in Fig. 10.16, and Rp = 3,83% is the resolution of the photomultiplier broadening alone. The result is an intrinsic resolution, R of 3.92%. By similar means, the intrinsic resolution I) 116 *q has been calculated for other gamma-ray energies, up to 2.5 Mev, and has been found to go through a maximum in the 400 to 500 kev region, reaéhing a value slightly above 5%. The intrinsic resolution is about 2.7% at 0.15 Mev and 1.5% at 2.5 Mev. The calculated minimum line width of 5.5% for 66l-kev gamma rays incident on a 2.5-in.-diam by 2-in.-high crystal (source 10 cm from crys- tal face) is to be compared with experimental values of the over-all line width in the range 6 to 8%. It should be pointéd out, however, that the intrinsic broadening must depend on the source configuration and crystal dimensions. This dependence has not been inVestigated, and it is dif- ficult to estimate how sensitive the above results will be to changes in the crystal size and source-to-crystal separation distance without further calculations. A study of these effects will be made in the future. Light Output of CsI(Tl) Under Excitation by Gamma Rays A series of experiments designed to investigate the light output of CsI(T1l) under excitation by monoenergetic gamma rays has been eésentially completed. These experiments represent the first portion of a program- for investigating the fundamental scintillation properties of CsI(T1), with particular emphasis on the investigation of the nature of the energy transport process in the crystal. A theoretical model relating the light output of an activated inorganic scintillator to its stopping power for particles of a given energy and to the density of activator sites has been proposed which predicts that the light output of the alkali halides under excitation by charged particles is not a linear function of the particle energy. As reported in the preceding section, such nonlinearity has been experimentally observed for gamma rays incident on NaI(Tl). In order to examine the prediction of the theory that such behavior is characteristic of other members of the alkali halide group, CsI(T1l) was chosen for these experiménts. The fundamental scheme of the experimental program consisted of the measurement of the light output of CsI(T1l) for gamma rayé of various energies relative to the output for a standard gamma ray, which was chosen 117 to be the 662-kev gamma from the decay of Cst37. The CsI(Tl) crystal used was grown by Harshaw Chemical Co., and it contained about 0.1 wt % thallium. ' The experimental results, encompassing a range of gamma-ray energiles from 10 kev to 2.6 Mev, are shown in Fig. 10.17, where the pulse height per unit energy, *L(Ey)/Ey, is plotted as a function of E . From these results the nonlinearity of the light output of CsI(Tl) as a function of energy for gamma-ray excitation has clearly been demonstrated. The magnitude of the departure from linearity hés been measured for the par- ticular crystal employed, although somewhat different results may be ob- tained for crystals of other sizes. The data will permit a calculation of the light output per unit energy, L(Ee)/Ee, for CsI(T1l) in the same fashion as that performed for NaI{Tl) and described in the preceding section. The Model IV Gamma-Ray Spectrometer Assembly of the BSF Model IV gamma-ray spectrometer has been com- pleted. PFinal calibration of the positioner and readout system 1s yet UNCL ASSIFIED ORNL -LR-DWG 53817 116 a2 >~ {08 ¥ E)- ~ {04 100 096 10 20 50 100 200 500 {000 2000 4000 Er( kev) Fig. 10.17. Relative Light Output of CsI(Tl) as a Function of Gamma- Ray Energy 118 sy to be accomplished. A series of preliminary investigations designed to optimize performance of the system is now in progress. For these in- vestigations a special loading of the Pool Critical Assembly is being.used that gives a long, thin slab of cold elements that are expected to provide a spectrum containing a minimum of scattered gamma rays. The goal of these preliminary experiments is the measurement of the spectrum. The program will include investigations of (1) backgrounds while the spectrometer is in a mixed field of radiation; {2) the relative effect of neutrons enter- ing the shield through the collimator; (3) the effects of collimator size and geometry, in order to make the final experimental result independent of these factors and to minimize the tail of the pulse-height distribution caused by scattering in the collimator; (4) the possible necessity for and effect of an exterior conical void és a part of tfie collimator sYstem, in order to remove the large lead shield from the region where gamma rays enter the detector; (5) the reproducibility of the data taken with the. multiphototube arrangement necessary with the large crystal; and {6) methods of exactly reproducing a particular power level in the reactor used_fqr calibration. | The 9 3/8-in.-diam NaI(Tl) scintillator that was to have been used in the spectrometer was broken while in transit from the manufacturer, but a new crystal has been obtained that consists of a composite formed by Joining a 9-in,-diam, 7-in.-long crystal with another crystal 9 in. in diameter and 5 in. long. The final configuration is a right cylinder, 9 in. in diameter and 12 in. long, with the connection between the two crystals being accomplished by an optically‘consistent medium. Gamma rays are collimated into a l-in.-diam, 2-in.-deep well drilled axially into one of the 9-in.-diam faces. A slightly smaller compbsite crystal of this type has been tested at the BSF, as mentioned in the preceding section entitled "Experimental Study of the Gamma-Ray.Response of Large NaI{Tl) Crystals.'" Such a crystal dispenses with the conical end which characterized the 9 3/8-in.-diam crystal and which proved to be detri- mental to its response in that it produced doubly peaked distributions for monoenergetic gamma rays of energy as low as 1 Mev. 119 shown in Fig. 10.18. A typical response curve obtained with the new composite crystal is The - measured pulse-height distribution is shown for the 2.73-, 6.13-, and 7.12-Mev gamma rays resulting from the decay of Nie, Contamination of the distribution is evident in the peak at 0.511 Mev due 20 counts per channel/sec Fig. 10.18. UNCLASSIFIED 2-04-058-557 N46 6.43 Mev fl 1 0.541 Mev 16 2.73 Mev fidrhmfiwmmmfi”#fi# & \:12 Mev N‘IG 20 40 60 80 CHANNEL NUMBER to Gamma Rays from the Decay of »N1©. 120 100 120 140 Response of a 9-in.-diam by 12-in.-Long NaI(T1) Crystal e to an annihilation gamma ray originating in the lead of the spectrometer housing and in a peak at 1.36 Mev due to the decay of g01? in the water of the pool. A detailed study of the response of the crystal as a function of energy has not been performed, but resolution is ~11.5% at 0.662 Mev, with a peak-to-total ratio (photofraction) of ~78% at the same energy. This resolution is better than that attained with the earlier conical- end crystal, while the peak-to-total ratio is somewhat lower than the value of 92% computed by a Monte Carlo calculation. This is to be ex- pected, since the experimental distribution is probably strongly in- fluenced by those gamma rays degraded in energy by scattering in the collimator and source before entering the crystal. Unscrambling of the Scintallation Spectra The experimental data from the observation of gamma-ray spectra by scintillation crystal spectrometers are obtained in the form of a distri- bution or spectrum of pulse heights. Unfortunately, the intensities of the scintillation pulses within the crystal are not directly related to the energies of the incident gamma-ray photons, but rather to the energies of the electrons resulting from the complicated photon interactions with the crystal. Also, statistical "smearing” is introduced into the results by the always nonideal components of the spectrometer system. For these reasons, the experimental data must always be "unscrambled” in order to - obtain a useful gamma-ray energy spectrum. A method that requires no prior éssumptions conéerning the spectrum being analyzed is under develop- ment to perform such unsérambling with the aid of the IBM-704 computer. In principal, the incident photon energy spectrum can be obtained from the measured pulseéheight distribfitibn by setting up and solving a system of Simultaneous linear eguations relating the incident spectrum Lo the measured pulse-height distribution. These equations have the form 121 a.x. +ta, x. . +... +a, x =Db 1171 122 In n 1 alle + a22x2 + ... F aZan = b2 . . . (l) amlxl + am2x2 + s, + amnxn = bm or, in matrix notation, Ax = b. In this system, the components of b represent the measured data, the pulse-height distribution from a malti- channel analyzer. The components of x represent the unknown number of photons within corresponding energy intervals. Thus, for monoenergétic radiation, the pulse-height distribution is given by a certain column of the matrix A. Physical reality demands that the elements of x and b be nonnegative. (One cannot have a minus number of photons or a minus number of counts.) The matrix A is not symmetric for scintillation spectrometers. There are two conspicuous difficulties involved in setting up and solving the formal matrix equation for the unknown elements of x. The first is that n® elements of the matrix must be determined (if n 1s the nunber of unknown elements of x) and n must be quite large (50 to 300 for typical gamma-ray spectrometers) in order to avoid obscuring fine spectral detall by the discreteness of the matrix. It is-virtfially impossible to measure all these n? elements. At best, a few calibration spectra are measured with nominal accuracy. These data must be extended and augmented by some sort of interpolation, based on their physical interpretation, - | to provide sufficient information to permit the solution to proceed. The second difficulty is that large-order systems of equations (assuming that they can be formulated and solved at reasonable expense) lead to results which are inordinately sehsitive to small, statisfiical errors in the experimental data. For example, in a typical 100 by 100 system, statistical errors are often a factor of 1010 larger than the - desired solution. | One solution to these difficulties lies in reducing the order of the system to a more modest value. However, this method leads to an unwanted 122 w coarseness if n is chosen too small relative to the inherent resolution of the spectrometer. Another approéch involves the calculation of a few parameters of the solution, e. g., the moments, or the first few terms in a series expansion, etc. The latter method may cause some shift in the position of a peak or may introduce "overshoots" into the result. The methods described below reduce these difficulties and alsc give an estimate of the spectrum directly. - Matrix Generation Method. A program has been initiated to obtain a complete set of response functions for a spectrometer consisting of a 3- . by 3-in. NaI(Tl) crystal supplying a 256-channel pulse-height analyzer . ~ with a full-scale energy calibration of 2 Mev. Such a spectrometer is typical of those used for analysis of radiation from reactors and from isotopes, and techniques developed for its use should be adaptablé with little change to other spectrometers. Standard spectra have at the present been obtained for eight isotopes, ranging from M08 o Y88, Each standard spectrum has been fitted to the following semiempirical function: 1/2 1 + b3 exp [(x — bs )/b4] y = baf . \ exp [(x - b3)/b4] + {l ~ by exp [(x - b3)/b4:| o - o () o fix- ] -3} . ) 2 ‘ + § : b, exp --[(x - bi)/bk] , (2) photo peaks ‘ : and escape peaks vhere y = total counts, x = channel number, and b, are fit parameters. k Each photopeak or escape peak is assumed to be Gaussian., The Compton distribution is fitted by the first function in Eq. 2, which is the gain function (vs frequency) of a shunt-compensated video amplifier. The . 123 second function in Eq. 2 is a "fill-in" for the valley between the photo- peak and thé Compton distribution. The well-known energy relationships between the positions of the photopeak and the escape peaks, etc., were not used ifi the fitting, since it was desired to avold any detailed assumptions concerning the relation between energy and channel number. Additionally, the first 20 analyzer channels were excluded from the analy- sis because of instrument nonlinearity. The parameters shown have been fitted to empirical functions of either the photopeak positidn or the Compton position, and the Compton position has also been fitted as a function of photopeak position. It is thus possible to reconstruct a standard spectrum by specifying the channel nunber of the photopeak center. Also, if the channel number is known as a function of energy, then specification of the energy of the incident photon will determine the response function shape. Figure 10.19 compares a response function, reconstructed by specifying the channel number of the photopeak center, with the experimental spectrum. A_complete set of elements for a response matrix can be generated from channel 20 to 200 or more. Such a matrix is being used as a "repre- sentative”" matrix for testing the methods described below. Equation Solution Methods. Seferal techniques are being studied which circumvent the traditional difficulties involved in straightforward matrix inversion. An error analysis shows that the large spurious com- ponents in the traditional results are due to rapidly fluctuating eigen-. values of A, which appear statistically with large amplitudes. It has been pointed outll that this difficulty can be alleviated by solving, not for the unknown x, but for x*¥ = Sx, where S is a smoothing matrix with a "smoothing width" comparable to the resolution of the experimental apparatus. The smoothed solution is obtained directly from the original set of equations by replacing the matrix A by AS™1, A method is being studied 11y, R. Burrus,. "Unscrambling Scintillation Spectrometer Data,"” IRE Trans. on Nuclear Sci., NS-7 (2-3), 102 (1960). 124 UNCLASSIFIED 4 - - 5x10 ORNL-LR-DWG 52817 . 5 w - =z 2 Qo - Q 2 2 . s o - - 103 5 2 102 20 30 40 50 60 70 8O 90 100 10 CHANNEL NUMBER ' Fig. 10.19, Comparison of Experimental and Reconstructed Spectra for Mn’%. which may give the matrix A5t as a modification of the matrix element - - | ‘ fitting procedure without actually computing the inverse of S. Another error-reducing technique recently tested applies the physical requirement of the non-negativity of the elements of x. vThis technique automatically avoids spurious components which would have resulted in " negative excursions and reduces the errors from ~1019 larger than the desired solution to something of the order of the desired solution. In essence, these methods use a programing techniquel2 to obtain approximate solutions to the eguation 125au1 I. Gass, "Linear Programing, Methods and Applications," McGraw-Hill Book Co., New York (1958). - . - 125 Ax —b=1r |, x=0 , . (3) such that some function of the residual vector, r, is a minimum. In other words, by allowing some small residual, consistent with the statis- tical accuracy of the data, an approximate solutlon is obtained which has no physically impossible negative values. It is possible to formulate the problem as one of linear programing, so that a function of the fol- -lowing form is minimized: F = W, :E: i i k k=1,2, or « -, (4) r- ]l [For limitingly large values of k, the largest weighted residual ik (or residuals) receive all the welght; thus k = « corresponds to minimizing the largest residuals so that all the weighted residuals lie within a ~ band of twix., which is as small as possible.] The weight factors, Wes are chosen according to the statistical errors of the ith element of b. Experimental Results. Experiments are in progress with synthetic test cases, using the SCROL codel? for linear programing with the IBM-704 computer. A 19 X 19 matrix of the form _;/2 1/4 P 1/4 1/2 1/4 1/4 1/2 1/4 _ /4 12 . 13yilliam Orchard-Hays, "SCROL, A Comprehensive Operating System for Linear Programing on the IBM-704," C-E-1-R, Inc., Washington, D. C. 126 which is known to be very poorly conditioned, has been used in a pre- liminary test. Table 10.2 compares the results from the usual matrix inversion technique with the results from the programing technique with k =« for a simple assumed spectrum. An error of 0.01% was artificially introduced in the tenth element of b. This example dramatically shows the extreme. sensitivity of the usual inversion technique to such a small (from an experimentalist's standpoint) error. It is also evident from thé example that the sensitivity to small errors is reduced to a reason- able value by the use of the programing technique, Because of the highly artificial nature of the test, however, it does not necessarily indicate Table 10.2. Comparison of Matrix Inversion Technique with the Linear Programing Method (k = «) for : a Simple Synthetic Test Case b = Ax ox El;zent X (exact b* x = A"lpx (programing ’ value) method ) . 1 0 0 0 -6 680.0 0 2 0 1 000.0 1 000.0 - 13 200.0 0 3 0 6 000.0 6 000.0 -19 403.2 0 4 0 16 000.0 16 000.0 25 139.2 0.66 5 64 000.0 26 000.0 26 000.0 33 732.8 63 998,13 6 - 0 30 000.0 30 000.0 34 659.2 - 2.34 7 64 000.0 26 000.0 26 000.0 25 796.8 63 998, 56 8 0 16 000.0 16 000.0 40 806.4 0.12 9 0 6 000.0 6 000.0 ~42 398.4 0.33 10 0 1 000.0 1 000.1% 42 934 .4 0 11 0 0 0 -42 398.4 0 12 0 0 0 40 806.4 0 13 0 0 0 -38 203.2 0 14 0 0 0 34 659.2 0 15 0 e 0 -30 26%7.2 0 16 0 0 0 25 139.2 0 17 0 0 0 -19 403.2 0.11 18 0 0 0 13 200.0 0 19 0 0 0 -6 680.0 0 ®An error of 0.01% is artificially introduced here. 127 that similar behavior may be expected in actual practice. In fact, with k = 1, the programing method was fortuitously completely insensitive to the error shown in Table 10.2. (The k = « case is at a disadvantage compared with k = 1 or 2, since the solution with k = « may be slightly nonunigue because of the freedom of "play" between the upper and lower bounds of the weighted residuals. The 17th element of x in Table 10.2 shows this effect.) | | At present the smoothing technique and the programing method are being combined. Tests using a "more representative” NaI(Tl) response matrix are being prepared. Neutron Spectroscopy Use of Silicon Surface-Barrier Counters in Fast-Neutron Detection and Spectroscopy The possible use of a neutron-sensitive semiconductor detector con- structed by depoéiting a thin layer of Li®F between two silicon-gold surface-barriér counters was mentioned previously.14 Neutrons are de- tected by observing the & + T pair resulting from the Li®%(n,q)T reaction; pfilses from the two counters are added and the summed pulse is amplified and recorded bn a_multichannel analyzer. Since the sandwich geometry permits similtaneous detection of both reaction products, the magnitude of the resulting summed pulse should Be proportional to the energy of the incident neutron plus the reaction @ value (4.78 Mev). Silicon-gold surface-barrier counters have been used as alpha- parfiicle spectrometers with extremely good energy resolution, e.g., 0.25% for collimated alpha particles with energy of about 5 Mev.1”? The pulse-height distribution to be expected from monoenergetic neutrons on 14" ANP Semiann. Prog. Rep. April 30, 1960," ORNL-2942, p. 117. 157. L. Blankenship and C. J. Borkowski, "Silicon Surface-Barrier Nuclear Particle Spectrometer,” paper presented at the Seventh Scintil- lation Counter Symposium, Washington, D. C., February 1960 (to be published in IRE Trans. .Nuclear Sci., 1960). 128 a sandwich detector of the type described here will exhibit a much broader peak, however, since the @ + T reaction products are subject to energy loss in the Li8F and gold layers before reachifig the sensitive volume of silicon. The energy loss in a particular event depends on the angle at which the reaction products are emitted; since the alpha particles and tritons may be emitted at any angle, the energy loss is a - variable from some minimum value up to the total energy available. In a practical counter, with a LiSF layer of finite thickness, this wvariable energy loss will govern the width of a monocenergetic neutron peak, with a much smaller contribution from the inherent line width of the silicon- counter. Counter Construction. is presented in Fig. 10.20. A schematic diagram of the sandwich counter The device consists essentially of two silicon counters seated in thin fluorothene sheets, which, in turn, are CLAMPING NUT —~a_ & GUIDE PIN SEPARATION OF COUNTER FACES ~0.002 in. POSITIONING SCREW 1 H oi?rla::;m.m“;:\::‘:flm{é UNCLASSIFIED ORNL-LR-DWG 48685 ELECTRICAL LEAD SILICON FLUOROTHENE ALUMINUM POSITIONING SCREW SILVER PAINT COUNTER GUIDE PIN VIEW OF FACE ¥y 0 1 INCH Fig. 10.20. Construction Details of a Sandwich-Type Silicon - Surface-Barrier Counter. SILICON~-GOLD - mounted in light aluminum rings. A layer a LiSF, of nominal thick- ness 150 pg/em?, is vacuum- evaporated onto one of the counters prior to assembly., The two counters are brought close together, adjusted to make the faces parallel, and rigidly clamped by means of three clamping nuts and three positioning screws, as shown. In practice, it is desirable to separate the counters by as small a distance as possible in order to avoid losing counts from an "edge effect" to be described below. An attempt is made to achieve a separation of about 0,001 in., at the same time taking extreme care that the counters do not touch and introduce small scratches on the delicate gold face. 129 Electrical contact to the gold surface of each counter was made by a narrow streak of silver paint leading from the periphery of the gold surface to the aluminum ring (see Fig. lQ.20). The aluminum rings were then electrically grounded. A reverse bias of approximately 30 v was applied through a 500 000-ohm resistor, and the signal was fed through a coupling cépacitor to a low-noise preamplifier. The signal was then fed to a DD-2 double delay-line amplifier with a 1.2-usec clipping time, and the amplified signal was analyzed in an RIDL 400-channel analyzer. The linearity and zero position of the analyzing equipment were checked between experimental runs with a mercury-relay tube-pulse generator whose signal was fed to the preamplifiér input. Detection of T(p,n)He3 Neutrons. In a first set of experiments-in which the sandwich counters were used, neutrons of known energy were produced by the T(p,n)He3 reaction. The ORNL 5-Mv Van de Graaff generator was used to accelerate the incident protons. The target consisted of a layer of ZrT, of nominal thickness 1 mg/cmz, which had been evaporated onto a platinum backing. The sandwich counter was placed in all cases ~at O deg with respect to the proton beam and was located about 1 in. in front of the target. The plane of the LiSF layer was perpendicular to the direction of the photon beam. Neutron energiles were calculated from published ta‘bles,16 taking into account the energy loss of protons passing through the ZrT layer. The full-energy spread of neutrons in- cident on a sandwich counter, arising from both proton energy loss and the angular acceptance of the counter, was small compared with the neutron-peak width in the pulse-height spectra. ©Slow neutrons (thermal and epithermal) were obtained by moderating with blocks of paraffin the fast neutrons obtained from the ‘I'(p,n)He3 reaction or from a polonium- beryllium. source. | A typical pulse-height spectrum from slow neutrons is shown in Fig. 10.21. The full width at half maximum was measured as 0.27 Mev, An approximate calculation of the slow-neutron pulse-height spectrum, 16J. L. Fowler and J. E. Brolley, Jr., Revs. Mod. Phys. 28, 103 (1956). 130 ol ¥ ORNL TR R IED based on the energy loss of the 2000 T charged particles in Li®F and SLOW NEUTRONS " COUNTER NO.1 ' gold, indicated that the slow- 4000 neutron peak should occur at a pulse height corresponding to 4.6 Mev (rather than the Q value 3000 0.27 Mev —_ of 4.78 Mev). The spread in / the slow-neutron peak is thus 0.27/4.6 or 5.9%. The shape of 2000 / the slow-neutron (or fast-neutron) COUNTS/CHANNEL SINGLE-PARTICLE ] peak is expected to be non- 1000 EVENTS \ }/ Gaussian and asymmetric by virtue \ . _flwjfk — of the energy-loss processes re- ™) 0 ) . _ © 10 20 30 40 50 60 70 80 90 100 {10 sponsible for the peak width. A PULSE HEIGHT low-energy tail that contributes Fig. 10.21. Pulse-Height Spectrum from Slow Neutrons Measured with Sili- con Surface Barrier Counter. peak down to zero pulse height counts in the spectrum from the is expected in all cases. Pulse-height spectra were measured for 1.09-, 1.99-, 3.00-, and 3.50-Mev neutrons. The results for the 1.99- and 3.50-Mev cases are shown in Figs. 10.22 and 10.23. In every case a well-defined fast- neutron peak was oObserved. A subsidiary slow-neutron peak was also ob- served in the spectrum that arose from the presence of a small number of degraded, low-energy neturons. These low-energy neutrons apparently were slowed in the immediate vicinity of the counter, since the relétive intensities of the fast- and slow-neutron peaks were essentially un- affected by enclosing the counter in a cadmium shield. A possible neutron moderator was a blob of Apiezon wax, which secured the pig-tail lead on the back of each counter. An additional effect noted in the fast-neutron spectra was the presence of a very large and steep background below the slow-neutron peak. The energy corresponding to the cutoff of this background was véry closely correlated with the incident neutron energy, indicéting that it was due to recoil protons following (n,p) scattering. 131 UNCLASSIFIED ’ . s . ORNL-LA-DWC 48767 Additional experiments have 500 confirmed that this is the case 1.99-Mev NEUTRONS COUNTER NO.1 and that the source of recoil 400 protons is the thin peripheral RECCIL PROTON BACKGROUND band Of polystyrene pI'eViO_usly described. 0.28 I:\dev \ In Fig. 10.21, a clearly o Q o defined hump occurs below the COUNTS/CHANNEL n Q 8 - slow-neutron peak and is labeled "single-particle events." Simi- SLOW-NEUTRON PEAK 100 lar maxima were observed in the fast-neutron spectra. They arise from events predominantly 30 40 50 60 T BO 90 00 O {20 {130 140 150 . . PULSE HEIGHT occurring near the edge of the Fig. 10.22. Pulse-Height Spectrum Li°F layer. As a result of the from-1.99-Mev Neutrons Measured with finite separation distance be- .Silicon Surface Barrier Counter. tween the two counters, there is a certain probability that an event will occur in which'the.alpha particles and tritons are emitted at a small angle to the plane of the Li¢F layer, and therefore one type of particle will be counted while the other escapes entirely. Thus the pulse-height spectrum from monoenergetic neutrons should, in principle, display three peaks: one due to the similtaneous capture of both alpha particles and tritons to give the full-energy peak, one due to capture of the alpha particles and escape of the tritons, and one due to capture of the tritons and escape of the alpha particles. In practice, however, the peak due to alpha-particle capture and escape of the tritons is expected to océur at a low pulse height and to be very broad and smeared; this follows from the large energy loss of the alpha particles in penetrating the Li®F and gold at small angles. In a practical case then, only two distinct peaks would be expected, as in Fig. 10.21. It may be seen that the slow-neutron peak, equivalent to 4.6 Mev deposition, occurs at . pulse-height 97; 132 UNCLASSIFIED . . ORNL-LR-DWG 48800 the triton-capture peak, equiva- lent to slightly less than 2.73 o T 3.50-Mev NEUTRONS COUNTER NO. | Mev, should occur just below 200 ' fl pulse-height 58, in agreement _——RECOIL PROTON BACKGROUND with the experimental result. . 50 ! . The alpha-capture peak is not present. The "edge effect" dis- o OAZWW\\T cussed here can be minimized in COUNTS/CHANNEL 100 / ’ two ways: (1) by minimizing SLOW-NEUTRON * the separation distance of the PEAK 50 °.:> }j/- ~ two counters and (2) by evapo- . , . rating the Li®F layer over an % ; .“{'\Wb—:’s‘ * ‘ .S . area less than that of the sili- 60 70 80 90 {00 {0 {20 130 114C ({50 160 70 ;80 . PULSE HEIGHT con counter, leaving a blank Fig. 10.23. Pulse-Height Spectrum strip around the border. from 3.50-Mev Neutrons Measured with . 3 Silicon Surface Barrier Counters. Detection of D(dzn)He' and T(d,n)Hfie4 Neutrons. In a second set of experiments neutrons were produced by either the D(d,n)He3 or the T(d,h)He4 reaction, with the deuterons being accelerated in a 200-kv Cockcroft-Walton accelerator onto a thick target of either ZrD or ZrT. In these experiments the silicon counters were constructed with fluorocarbon compounds to eliminate or, afi ieast, reduce the effects of the polystyrene and Apiezon wax used‘in earlier counters: .An additional feature was the use of a fast-coincidence circuit (resolving time of 0.1 psec) requiring simultaneous events in the two counters comprising the sandwieh. With this circuit, Lie(n,a)T events in which the alpha particles and tritons are stopped in separate counters are-fecorded, but those events (largely due to background effects) tfiat are stopped in only one counter are eliminated. A coincidence count was recorded only for those events in which the pulse amplitude from each counter corresponded to an energy deposition of 1.6 Mev or more. 133 500 450 400 350 300 250 200 COUNTS PER CHANNEL - 150 - 100 50 500 450 400 350 300 250 200 COUNTS PER CHANNEL 150 400 50 Fig. 10.24. UNCLASSIFIED ORNL-LR~ DWG 51266R{ 2.95~-Mev NEUTRONS (5) WITH COINCIDENCE SLOW-NEUTRON PEAK & 10 20 30 40 50 60 0 80 90 PULSE HEIGHT UNCLASSIFIED ORNL- LR-DWG 5{265R! 2.95-Mev NEUTRONS " (@) WITHOUT COINCIDENCE SLOW-NEUTRON PEAK /N 1V \ N T —— . 10 20 20 40 50 &0 70 80 €K PULSE HEIGHT Pulse-Height Spectrum from 2.95-Mev Neutrons with and With- out a Coincidence Circuit. 134 Pulse-height spectra resulting from D(d,n)He? neutrons of energy 2.95 Mev are presented in Fig. 10. 24. measured both with and without The spectra shown were the coincidence requirement, as indicated on the figure. The continuous background observed in Fig. 10.24a, without coinci- dence, is attributed to events resulting from 1l4-Mev neutrons which arise from the T(d,n)He% reaction due to a small amount of tritium present as a contami- nant in the target and beam tube. In particular, the background ob- served above the D(d,n)He? peak must be due to higher-energy neutrons, since the (n,charged particle) reactions in the counter which might contribute to a back- ground are generally endothermic - and could not give counts at The pulse-height spectrum with co- - such a high pulse height. incidence (Fig. 10.24b) may be seen to be much cleaner, in- dicating that the background events of Fig. 10.24a occur pre- dominantly in one counter only. Pulse-height spectra result- ing from 1l4.7-Mev neutrbns from the T(d,n)He* reaction are presented in Fig. 10.25. In this case the spectrum may be seen to be significantly different from those obtained with lower energy neutrons in that the Lié(n,a)T fast-neutron peak represents only a-small fraction of the total.counts recorded. Two pulse-height spectra are shown in Fig. 10.25, one from a sandwich counter with the Li®F layer and one from a sandwich counter without Li®F. The two counters were otherwise eésenn tially identical in construction and were exposed to the same nfimber of neutrons during the recording of the pulse-height spectra. Clearly, the very large background below the Lié(n,a)T peak arises from (n,charged particle) events in the silicon and gold and is not associated with the Li®F. Furthermore, the spectra of Fig. 10.25 were recorded with the coincidence reguirement, so the observed background counts must cor- respond to simultaneous events in both counters. There are a number of (n, charged particle) reactions UNCLASSIFIED 05 ORNL-LR-DWG 5263t 11 both silicon and. gold that are energetically possible at WITH LITF LAYER 14 Mev and which can contribute to the background, e.g., (n,p) 1o WITHOUT LI%F LAYER and (n,Q) reactions. The magni- -tude of the observed background indicates that the major contri- 14.7-Mev NEUTRONS bution must arise from silicon, since it constitutes the bulk of COUNTS PER CHANNEL 5 . o the counter. The results of Fig. 10.25 indicate that this Li%(n, a) T PEAK n 2 0 type of detector is of limited 0.45 Mev use for neutrons in this energy range. Response to Gamma-Ray Back- 0 20 40 60 80 100 120 140 160 PULSE HEIGHT | grounds. Preliminary experiments Fig. 10.25. Coincidence-Pulse- were carried out to examine the Height Spectra from 14.7-Mev Neutrons effect of gamma-ray backgrounds from Silicon Surface Barrier Counters with and Without & LiSF Layer on the neutron-counting properties 135 of the detectors. A counter was placed adjacent to a paraffin block in . which a polonium-beryllium source had been placed, and the pulse-height spectrum from slow neutrons was recorded. The slow-neutron peak was centered at channel 100 and had a fraction width of about 6%. A 0.6- curie Co®9 gamma-ray source was then placed 5 in. from the detector and the spectrum was rerun. No significant change was observed in the pulse height, width, or count rate in the slow-neutron peak. A steep back- ground occurred, however, at low pulse heights, which cut off at about channel 20. On.the basis of this experiment, it is tentatively concluded that these counters offer rather strong discrimination against gamma-ray backgrounds. The large positive Q values of the Li%(n,a)T reaction, of course, provides a,built-in bias against any background event up to 4.6 Mev. . - Conclusions. From the results of Figs. 10.23, lO.24, and 10.25; as well as other results not presented here, it is concluded that the full -width at half meximum ranges from 0.28 to 0.45 Mev. These figures should not be taken exactly, however, because they'are derived from only one set of experiments with relatively few fotal counts in each spectrum. The important points to note are that the neutron-energy spread is relatively constant, of the order of 300 kev, and that the neutron-energy resolution improves rapidly with increasing energy. The fractional width AEn/En decreases from spproximately 30% at 1 Mev to approximately 10%. at 3 Mev and 3% at 14 Mev. It is of interest to calculate the shape and width of the fast-neutron peak to determine whether the experimentally observed péak width approaches the'thebretical value. Such a calculation has been initiated, as reported below. The pulse height-vs-energy relationship, according to the present results, is apparently linear or very nearly so, as expected. It isv not possible to precisely determine this relationship on the basis of the present work, since the total energy deposited in the counters by a neutron of energy En is not simply En + Q, but is less than this by an amount that depends on the thicknesses of the Li®F and gold layers -and on the angular distribution of the reaction products. As indicated 136 =t previously, thermal neutrons result in the deposition of about 4.6 Mev, In the absence of detailed calculations for fast-neutron cases, the energy deposited in a fast-neutron event cannot be determined accurately. As a first approximation, however, a plot of pulse height vs.(En + 4.6 Mev) can be reasonably fitted with a straight line through the origin; including the 1l4-Mev point. Future Application. The experiments summarized here indicate pos- sible future application‘of sandwich counters of this type to certain . problems involving the detection and spectroscopy of fast neutrons. The principal advantages of the sandwich counter appear to be 1ts simplicity .of construction and operation, its small size, and its reasonably good resclution for neutrons above 1 to 2 Mev. A further advantage lies in the fact that the counter accepts neutrons from any direction, eliminating the need for neutron collimation. The detection efficiency is‘limited by the thickness-of the LiSF layer, all50-ug/cm2-thick layer offering a detection efficiency of 3.4 x 1072 for thermal neutrons and 0.94 X 1076 for 2-Mev neutrons when normal incidence of the incoming neutron is assumed. These efficiencies are, of course, rather low and may limit the use of this technique to experiments with sufficiently high neutron intensities. The efficiency can be increased by the use of thicker LiSF layers at the expense Of broadening the peak width. Another pos- sibility for increasing efficiency is that of stacking, say, 5 to 10 counters together. Distribution in Energy of Alpha PartlcledTrlton Palrs Resultlng from Neutron Bombardment of thhlum Fluoride The employment of paired silicon-gold surface-barrier counters separated by a layer of Li8F to record the sum of the energiesfof the tritons and alpha particles produced in an Li6(n)a)T reaction is reported in the preceding section. In this section the problem of calculating ‘the shape and width of the distribution in total energy, N(W) of the pairs reachlng the 31llcon is discussed. | A uniform beam of neutrons of energy E 1s assumed to be 1n01dent in the 7z direction normal to the plane of the Si-Au-LiF-Au-Si sandwich. 137 The (n,d).reaction takes place at some point z (0 _determined by Egs. (l),_< z =T z=0 (2), and (3) if v () z =T; AB) =B (V) 2z =0; 2B =E (V) 0 0 . v (W) z =T LB =E_ (V) VI (aW), =0 > (W) min @ G Wmax(w) AEa = an(\‘,f) Zz = O VIiI (DW)b z =0; AE, = E, (¥) z =T; AB, =E (¥) 0 0 V. (W) AE = E_ (V) AE =E_ (V) VIIT (E_+Q) >a0 > (W), {4 " %o ¢ 9 v (W) AR =E (V) AE = E (V) max‘ y S o ao tO *For q;c(EO) < ¥ < /2, both particles emerge in the forward hemisphere; i.e., 0 < X < 7/2. Table 10.4. Calculated Values of the Minimum - and Most Probable Energy Losses for Various Neutron Energies Minimum Energy Most Probable Neutron Energy Loss (kev) Energy Loss (kev) (Mev ) () i (?) ) () Gold thickness: 50 pg/cm?; Li®F thickness: 150 pg/cm? 0 53.6 53.6 210 210 0.60 49.1 - 54.1 236 167 2.00 43.4 51.2 239 135 8.00 31.5 39.3 210 80.0 14.00 25.8 32.5 179 59.7 Goid thickness: 75 ug/cmz; Li®F thickness: 150 p,g/cm2 0 63.9 63.9 220 220 0.60 60.7 63.1 247 176 2.00 54.0 59.1 250 142 - 8.00 41.0 45.0 219 85.6 14.00 3.2 37.3 187 64.5 of an isotropic reaction cross section, dL, is shown in Fig. lO.27; The maximum occurs at an energy loss of 210 kev for a gold thickness of 50 p.g/cm2 and an Li®F thickness of 150 ug/cm2. The width at half maximum is approximately 220 kev. Calculations of N(W) for higher energy neutrons are in progress. A Neutron Chopper Spectrometer for the Bulk Shielding Facility The preliminary design of a neutron chbpper spectrometer facility for the Bulk Shielding Faéility, to be located underground adjacent to the pool of the BSR, has been completed; construction of the experimental facility is expecfied to be completed in late 1961. It wiil be used for' 'neutron speétréi measurements which should either substantiate thevcalcu- lational methods or direct further study. Following initial alignment eXperiments, it is proposed to make spectral measurements in pure and 142 UNCLASSIFIED ORNL-LR-DWG 52621 0.9 . ' ;///"’v \\ THERMAL NEUTRONS 0.8 _ 0 =50 ug/cm? /{- \\ 7 =150 pg/cm? 0.7 : WL AN 0.3 1 - AN 0.2 N (AW) (arbitrary units) \\h—__; 0.4 — 0 50 400 150 200 250 300 350 400 450 500 550 600 650 700 750 AW (kev) Fig. 10.27. Calculated Energy Loss Distribution for Thermal Neutrons Incident on a Silicon Diode Detector. mildly borated water, for purposes of comparison with existing data from other installations, and measurements of the energy distributiofi Qf neutrons diffusing from the shield-air interface of a highly poisoned hydrogenous shield. Spectral measurements within a shield will also be made, and spectra from several regions of the BSR will be observed. Spectrometer Design. A schematic diagram of the chopper spectrometer is shown in Fig. 10.28. 1In the proposed arrangement, thé BSR is near the wall of its pool, adjacent to a metal window, through which the fieutrons_pass into an experimental cell housing a shield sample or ard - reactor matrix. Neutrons penetrating the sample are collimated to a rotating chopper, thence through an exit collimator to an evacuated drift tube, ana finally are intercepted by a stack of BFz-filled pro- portional counters. A boral-lined beam-catcher tube, terminated fiith paraffin to reduce back scattering, catches the neutrons which penetrate . The stack of counters. The flight path is about 10 meters,>which, with 0.065-in. slits in an 18-in.-diam rotor sPinnifig.at 10 000 rpm, will give an energy resolution of about 10% at 100 ev, 3% below 10 ev, and 30% at 1000 ev. 143 771 TO ANALYZER UNCLASSIFIED 2-01-058-0-545 BEAM-CATCHER TUBE INPUT MONITOR “Losn PARAFFIN DETECTOR STACK Schematic Diagram of Fig. 10.28. Facility. EVACUATED DRIFT TUBE COLLIMATOR NOTE: CHCPPER IS SHOWN AS IF ROTATING IN VERTICAL PLANE, FOR ILLUSTRATION ONLY, POOL = ' AIR = '—O =S=———¢ “—__Jl—"? gox| BSR EXIT ENTRANCE 26 CHOPPER COLLIMATCR THIN METAL WALL—»] REACTOR MATRIX OR SHIELD SAMPLE Proposed Neutron Chopper Spectrometer for the Bulk Shielding The mechanical chopper, which draws heavily upon the design ex- perience of the ORR fast-neutron chopper, was designed to yield maximum transmission of neutrons consistent with the resolutions noted above. The rotor and the collimators are made of Monel. A large-diameter Monel rotor will produce a sharp neutron burst over a'wide range of neutron energies. A small and reasonably constant value of edge leakage as a function of neutron energy is important for accurate calculation of the chopper transmission function. The rotor is suspended from a thin quill for operation at high rotational speeds, while fixed bearings will be clamped on for low-speed operation to prevent the serious resonance oscillations noted in a pilot model. _ . Pulses from the stack of high-pressure BF3-filled proportional counters terminating the flight path will go to a 256-channel time-of- flight analyzer that is gated "on" with a "start" pulse from the chopper. Each of the 256 channels can have a preselected window width of 2.5, 5, 10, 20, 40, or 80 psec to cover the time-of-flight range corresponding to a neutron energy range of from 0.005 to 1000 ev. The several rotor speeds needed to cover this range can be maintained to better than one part in 500 by a specially designed speed control. The analyzer will have a punched-tape readout coded for the Oracle, which will transform the time-of-flight data to neutron spectra in conformance with the factors discussed below. The total count rate of all counts entering the analyzer will be monitored. The neutron flux éntering the chopper will also be moni- tored by a small, thin-walled, low-pressure BF3 counter, and the outputs of the two monitors will be compared to deteét the sources of possible drifts and errors, such as reactor power-level changes and rotor-collimator misalignment. The input monitor will also serve to normalize the several pieceé of the spectral curve obtained at the several rotor speeds required. Energy Spectrum from Time-of-Flight Spectrum. The neutron energy, E, in usec/meter, is related to the measured time of flight, T, by the 2 E (ev) = (’72-3) , | (10) expression T 145 and if all other factors are constant with energy, (const.) 2 , | (11) = R1E where dN/dT vs E is the true time-of-flight spectrum. 1. Energy Dependence of the Detector. The stack of BF3 counters will tend to saturate in the thermal-energy region, but elsewhere it will have a l/v response. The detector stack will be calibrated over the entire energy range with a single low-pressure BF3 counter. In the epithermal region, the correction for counter response will be — Al d.N/ E ) ) (12) 81E where dN’/dT would be the observed time-of-flight spectrum if the trans- mission of the rotor were energy-independent and if the background counts were properly subtracted. | 2. Chopper Transmission Function. After the background is properly subtracted, the time-of-flight spectrum dN”YdT observed with a l/v detector is related to the true time-of-flight spectrum by T dt art ? (13) 1C where C% is the chopper transmission function. For the forward-directed neutron flux, from Eq. (11), dN _ &~ (14) for a l/v detector. DNote that in the region of the spectrum where dN/dE = (const.)/E = (const.) 12, 146 (const.) 12 =1 3T | (15) T or an’’ a-:i_— = (COIlSt.) CT . _ (16) Thus, under the conditions of a l/fi detector response, a forward directed flux with a l/E energy dependence, and proper background subtraction, the exact shape of CT is displayed on fihe time-of -flight analyzer. Since the conditions can be very nearly obtained, the calculated shape of CT can be checked. The final check and adjustment of CT will be made by measuring a spectral region at a number of rotor speeds. In general, only the correct shape of CT will produce a single-valued curve,. 3. Background., In a stable system, the background is a periodic function of rotor position because of neutron leakage through the array of rotor slits. As the slits are rotated, the amount of Inconel in the path of the collimated neutron beam will vary. The time dependence of this variation in leakage will be symmetric about the "burst" time (the time when the slits are fully open) if the slit geometry is symmetric about the longitudinal center line of the slit. For this reason, the proposed rotor will have symmetric slits. The time-of-flight analyzer is gated "on" many channels before the burst time so that the background is displayed. The very fast leakage neutrons, those that traverse the flight path within a burst width, can be subtracted out, channel by channel, counting from the burst time. The bulk of the leakage arises from these neutrons because the rotor is relatively transparent to high-energy neutrons. Slower-neutron leak- age can be estimated in many ways, e.g., by moving the neutron counters close to the chopper, by inserting a cadmium foil in the beam, or by changing the chopper angular velocity, in each case noting the change" of shape of the displayed background curve. 147 4. Relationship Between the Forward-Directed Flux and the Total Flux. A re-éntrant,hole in a reactor matrix or a shield slab, as shown in Fig. 10. 28, is‘considered in determining the relationship between the forward- . directed flux and the total flux. The chopper spectrometer measures the forward-directed flux crossing the z = zg plane at the bottom of the hole. This forward-directed flux is made up of neutrons scattered from various depths, z, behind the zy plane. The directed neutron flux, ¢(u,E,zO), is not, in general, equal to the total flux ¢(E,zo) at the same point. Where the diffusion-theory approximations are wvalid, the . following relationship holds:17s18 ’ ) A, (E) O¢ - ¢’(U-:E: Zo) = (COIlSt.) d)(E) Zo) 1l + — (EJ Z) J (17) ¢(E,zo) dz : Z=20 where p = cos6 and 6 is the angle to the normel of the z, plane; Rt(E) is the transport mean free path at energy E. Most of the previous time- of-flight measurements were made under conditions where this approxima- tion was valid; z_ is located near the center of a reactor matrix, near o the center of a moderator volume with a pulsed-neutron source also near the center, or the moderator cross section is fairly independent of energy, A E) = const. Note that such configurations could be used to ( t obtain a l/E forward-directed flux from a l/E total-flux region. This could then be used to obtain C <’ as discussed earlier. o . The relationship of Eq. (17) is not expected to be valid, how- .ever, where a source is located exterlor to the shield and where the ' shield i1s hydrogenous and contains a strong l/v poison. Here B¢/az and Bkt(E)/BE are both large, and the diffusion-theory approximations 17R, E. Slovacek and R. S. Stone, "Low-Energy Spectra Measurements," 8 paper presented at the Conference on Neutron Thermalization, Gatlinburg, Tennessee, April 1958. 18, g. Campbell et al., Proceedings of Second United Nations In- ternational Conference, Geneva, 1958, 15/p/10. 148 ; completely break down. The problem was investigated using'transport theory with the Goertzel-Selengut approximation [J(E,27) ~ J(E z) ], By retaining terms to 52¢/az , the expression becomes ¢ 32¢ | ¢(u,E,29) = ¢(E,25) + ¢ — (E,z) +d — (E,272) , (18) , z/ / dz? ,_ : z’=z, z'=z, where 1 c = (a - U«) ) z (E) 1 d=2 [bZt(E)—ap+u], Zt_(E) 2 — z (E) g = =— }12 l — t — . 3 3 2 (E) =2 _(E) 02 b =-— . ' Zt(E) - ZS(E) M Et(E) = total macroscopic cross section, 2 (E) = macroscopic scattering cross section, by = average of the cosine of the scattering angle in the laboratory system, T n average of the square of the cosine of the angle between the z-axis and the directed flux. . Expression (18) reduces to that given by Campbell‘ef al.18 if the 32¢/az term is dropped if the scattering in the sample is symmetrlc in the laboratory system, and if p = 1. For a highly poisoned, hydrogenous 149 medium, however, the scattering is the least symmetric, and the second derivative of the flux is quite strong in the geometric arrangements required. | Since Eq. (18) contains known functions of y, E, and the material constants, only the gradient of the flux and its second derivative need be measured to obtain the enérgy dependence of the total flux from the measured energy dependence of the directed flux. These flux plots are now being made for a mixture of methyl borate and methanol with a value of C = zs/'zt equal to that for LiH. 5. Hole Perturbation. The presence of a re-entrant hole might be the bottom of the hole. However, expected to perturb the spectrum at z 19 ’ the work done at General Atomic witg both poisoned and unpoisoned polyethylene demonstrates that as the hole diameter is increased to several inches no spectral change is observed. Of course, the situation should be somewhat worse with very high poison content because Kt becomes much smaller at low energies. The effect of hole perturbation can be determined by the "hohlraum" technique, in which the shield is sliced through the zg plane at the bottom of the re-entfant hole and a change of spectral shape-is looked for as the separation "w" between the two shield halves is increased. If the spectrum does change with w for small values of w but then reaches an equiiibrium value for larger values of w, one of these larger values will be used for all subsequent measure- ments. The hohlraum technique is applicable only if the flux in the zg "piane is constant over a distance of several w's. A gold-wire flux plot will be used to determine the applicability of the technique. A Spherical Detector Shield for the TSF The spherical detector shield described previously?® has been fabricated for use at the TSF. This shield is now being modified to 197. R. Beyster et al., "Measurement of Low Energy Neutron Spectra," GA-1088 (Nov. 13, 1959). 20"ANP Semiann., Prog. Rep. Oct. 31, 1959," ORNL-2840, p. 138. 150 include provisions for heaters and thermocouples, since rough tests in- dicated that it will be subject to large and unpredictable variations in temperature when in operating position. A constant temperature will be maintained by warming the shield to slightly higher than ambient temperature. The detector shield will house a 9-in.-diam, 12-in.-long, right- cylindrical, NaI(Tl) scintillation crystal for gamma-ray spectral measure- ments. The crystal has been purchased and should be delivered in the near future. Neutron spectral measurements in connection with the TSR-IT will originally be made with a variety of foils. Later in the experimental program a thorough comparison between the silicon-gold'diode detectors: (see above) and an Li®I scintillation crystal neutron detector is planned, with the final choice of a detector for neutron measurements resting upon this comparison. The detector shield will also be used during dose measurements. Experimental Determination of Flux Depression and Other - Corrections for Gold Foils Exposed in Water Additional experimental data have been obtained in the investigation aimed at determining a correction factor for thermal -neutron flux measure- ments made with gold foils in water. Foils ranging in thickness from 40 ug/cm2 to 493 mg/cm2 have been exposed in the LTSF at a position where the flux was isotropic, as required by the theoretical model,zl.and the resulting data were fitted with straight lines using the method of least squares. In the case of the thick foils, the data were fitted to fourth- order polynomials by a least-squares techhique employing the Oracle. The resulting correction factor, 0/¢; where ¢0 is proportional to the unperturbed flux and ¢ is proportional to the flux calculated from measurements with a foil of a given thickness, is plotted in Fig. 10. 29 ?1"ANP Semiann. Prog. Rep. March 31, 1959," ORNL-2711, p. 124. 151 Values of ¢, were determined by 5 , |mfifi$fifl?&mm an extrapolation to zero th%ck- - /$_WWWE - ness of measurements made with N / very thin (<1.7 mg/cm?) foils. / The error associated with the L7 . / points in the figure is ~3%. / mmm&\;7 Since this experiment was *° o /, ; //I {. begun a new calculational ap- 15 / ’// I | proach to the problem was made $/% // ///’ { I by Dalton and Osborn.?? 1In h4 ./’ ) l/fiUmE_ - this method the Boltzmann equa- ; ‘/ /// { x’i;: : tion describing the perturbed /( //' f{;jfi’““‘fiwns - neutron population is converted 12 /,//l f{é;fi;( : - into an integral equation. The /Q/ ’;fl" DALTON AND OSBORN angular integrations areper- ! }/75? 1 ] .formed analytically, and the o / | spatial integrations are carried 0 ] 2 4 6 8 10 12 out numerically with the aild of THICKNESS {mils) : the IBM-704. From the curve Fig. 10.29. Flux Depression and . . . - Self-Shielding Correction Factors for of Fig. 10.29, which also dis- .Gold Foils as a Function of Foil Thick- plays the results of the older ness: Comparison of Experimental with recipes of &I e, 23 Ritchie,24 Theoretical Results. ‘ ‘ - Bothe, 2% and Tittle, 2% it appears that the experimental results agree best with the results of ‘Dalton and Osborn. . An extension of Dalton's work to regions of greater - thicknesses in now in progress. 223, R. Dalton and R. K. Osborn, Trans. Am. Nuclear Soc. 3, 284 (1960) . 237, H. R. Skyrme, "Reduction of Neutron Den51ty Caused by an Absorb- 1ng Disc, " MS-91. 4R, H. Ritchie, Mealth Phys. Div. Ann. Prog. Rep., July 1, 1958 " . ORNL-2806, p. 133. 25y. Bothe, Z. Physik 120, 437 (1943). 260, W, Tittle, Nucleonics 8 (6), 5(1951); Nucleonics 9 (1), €0 (1951). 152 11. BASIC SHIELDING STUDIES The Spectrum of Prompt Gamma Rays from U225 Fission "An experiment is in progress, as mentioned previdusly,l for deter- mining the spectrum of gamma rays emitted within about 5 X 10=8 gsec after the thermal-neutron fission of U235, mThis spectrum.is not only of potential interest for comparison with the predictions of‘fission theories but also of immediate practical interest in the design of efficient radiation shields for fission reactors. In fact, a preliminaryl 2 of part of the spectral data above 400 kev has alfeady been analysis applied to shield design.3 | | As reported'previously,4 three spectrometers hasve been used in the investigation. For the energy range above 400.kev, pair and Comptofi spectrometers were employed, and, for the energy range below 400 kev, a single~crystal spectrometer was used together with a time-of-flight technique to exclude unambiguously the neutron-induced backgroufidé, One step in the final analysis of the data is the determination of the absolute efficiencies and response functions for the spectrometers, and it is this step which is currently in progress. It,cOnSists of two phases: +the standardization of the monoenergetic sources used to calif brate the spectrometers and the determination of the respbnsé functibns of the spectrometers for these sourcés. The first phase hfis been com= pleted for all the spectrometers, but the second phase has been completed only for the pair spectrometer. 1"ANP Semiann. Prog. Rep. April 30, 1960," ORNL-2942, p. 104. 2F. C. Maienschein et al., Proceedings of Second United Nations International Conference, Geneva, 15, 366~72 (1959). 3See, for example, H. Goldstein, "Fundamental Aspects of Reactor Shielding,” p. 58, Addison Wesley, 1959; A. W. Casper, "Comparison of Bulk Shielding Reactor Centerline Measurements in Water with Predictions," APEX~504 (Nov. 1958); A. F. Avery et al., "Methods of Calculation for Use in the Design of Shields for Power Reactors," AERE-R-3216, p. 35 (1960). ' 4"ANP Semiann. Prog. Rep. Oct. 31, 1959," ORNL-2840, p. 119. 153 Source Standardization Because of the difficulties in calculating the absolute efficiencies of the Compton and pair spectrometers from basic cross-section data, it 1s essentlal that they be determined experimentally at a number of points throughout the 0.4- to 8~Mev energy range. "This is done by observing the pulse-height spectra resulting when sources of known disintegration rates are placed in the region normally occupied by the U23% rission chamber during the'experiments.‘ However, finding sources of known strength was one of the major problems of the investigation. If was desirable to standardize the sources by- absolute counting techniques, using available coincidence equipment, but for this technique to be applicable the sources could not be strong. On the other hand, because of the low efficiencies of the péir and Compton spectrometers, convenient spectrometer efficiency measurements could not he made unless the source strength was of the order of 107 gamma. rays per second, and sources of this strength, even though they had favorable decay schemes, could not be standardized by absolute counting techniques.5 Furthermore, it was essentially impossible to find radiocactive sources with con=~ veniently long decay periods in the important energy region above 2.8 Mev. In an attempt to overcome these difficulties, several approaches were used. In some cases an initially strong source with a conveniently variable strength was used to study the spectrometer response while the source was strong, and it was later standafdized after it became suffi- ciently weak. High~pressure 47 ionlzation chambers having a wide dynamic range were then usually used as monitors to relate the intensities of the weak standardized sources to the strong-source intensities of the same material. | | ~ The ion chambers were also used to calibrate sources composed of materials other than those used in any of the absolute standardizations, ’The strongest attainable B**(p,y)C'? source, for example, had an intensity small enough to allow standardization, but a spectrometer response .- function with only poor statistical accuracy was obtained by counting for two~thirds of a day. 154 a procedure which was possible provided the gamma~ray energy dependence of the response of the ion chambers could be determined. The various sources which have been standardized by absolute counting are listed in Table 11.1. Response Matrix Formulation for the Pair Spectrometer - Of the two spectrometers used for the energy region from 0.4 to 8 Mev, the response matrix has been formulated only for the pair spectrometer, The following procedure, which is also being used for the other spec- trometers, was used for the formulation. Pulse-=height spectra obtained from the spectrometer ufion exposure to fission gamma rays were corrected for background effects, counts lost because of electronic limitations,: and the number of fissions observed during the spectrum accumulation. The resulting distribution of counts as a function of pulse height was then converted to a distribution of counts as a function of electron energy, Ee, where Ee = Ey'_ 2mgc?. The spectrometer gain and zero were measured with monoenergetic sources, and corrections for nonlinearity of the pulse=height analyzer were made in order to determine the electron energy corresponding to a given pulse height. For each region of the gamma=-ray energy spectrum, several pulse~height spectra were obtained. These were then combined to cbtain the number of counts, N,, in the 2 energy interval AEi, or bin i, located at energy EZ; " The remaining step in obtaining a gamma-~ray energy spectrum is the correction for the efficiency of the spectrometer or the probability that a gamma ray of energy ngill give rise to a count with a pulse height corresponding to the energy interval of a given bin 1. An ideal spec~ trometer would give rise to pulses corresponding to the energy of a single bin for a monoenergetic gamma ray, but scintillation spéctrometers are far from ideal in this respect. They give rise to a large tail of low=energy pulses, as well as a distribution of rulses into bins in the region of the correct one. The response of the pair spectrometer was determined experimentally by studying the pulse=height spectra produced upon the sequential exposure 155 9¢T Tgble 11.1. Summary of Sources Standardized by Absolute Counting Gamma-Ray Prima ' éfifizgxigazg Estimated Link to. Source or X=Ray .ry . & . Relative Used to Determine Sources Standardization Standardized Energy Standard Spectrometer . Method Source e s - (Mev) -1 Deviation - Efficiency (sec™) _ Sc46 . Cob0, Y88, Na?% 0.9-1.8 y-7 coincidence 10° 0.006 41 ionization chamber Nal4 1.37, 2.75 y=-y coincidence 10° 0.01 Direct Bil(p,7)Ccl? 4.4, 11.8 y=y coincidence 6 x 10% 0.015 Direct 7n®% 1.11 Scintillation counter 1.5 X 10° 0.011 417 ionization chamber of known efficiency Bg20? 0.28, 0.07 B-y coincidence, e-X 5 x 103 0.007 4 ionization chamber and ~ coincidence ' monitor counter ratio Ni6 6.1, 7.1 B=y coincidence 5 x 102 0.045% Monitor counter ratio Snli? 0.393, 0.0246 e=X coincidence 104 0.05% 41 ionization chamber and monitor counter ratio %Includes errors from published decay-séheme information. Ga of the spectrometer to sources giving rise to a total of 11 gamma rays of different energies. The resulting spectra were analyzed in terms of a "peak" area and a "tail" spectrum (see Fig. 11.1). The tail response per source photon, T(i,j), was taken as the number of counts observed in bin i upqn exposure to a photon of energy Eg in the photon energy interval AEg. The peak area per source photon, P(Jj), was taken as the sum of the observed responses between electron energy limits, c'j +d., . J which depended on the gamma~ray energy'Eg and were determined by fitting UNCLASSIFIED 05 2—~01—-058—-0-537 | COUNTS PER CHANNEL o w 102 "TAIL" FOR 2.754-Mev PHOTONS ENERGY ZERO EXTENT OF PEAK 0o 10 20 30 40 50 60 70 80 90 100 o 120 130 194G 130 ' CHANNEL NUMBER Fig. 11.1. Response of the Pair Spectrometer to the Radiation from Na?%: E_ = 1368 + 1 kev and 2754 = 1 kev. The circles show the experi - mental dgta, and the lines represent the predictions from the final tail fits, Three other Na2# spectra (2754 kev only) were analyzed for the fitting process. The statistical errors are shown at a few points. The electron energy was equal to approximately 19.25 times the channel number minus 5.9 for this run. Parameters used in the text in defining the peak are shown for both gamma~ray peaks. 157 the observed spectra. The peak area is located in the bin 1 = j, which includes the energy corresponding to the observed center of the peak, mj. Thus the over-all effect of the spectrometer, R(i,j), on a gamma- ray spectrum represented by ', photons per gamma-ray energy interval, J AEJ located at gamma=-ray energy EJ, may be written as : i . . j JZR(i,J) ryow) = 2 (R00) oy v (L0 T 08 <, 5 (1) where N i3 is the measured number of counts in the energy interval of bin i produced by a gamma ray of energy Eg . In principle, once the response matrix.R(i,j) has been determined, ‘ - Eq. 1 can be solved for the desired gamma~ray spectrum, FJ. Solution of the set of linear equations has been tried in the past6 with resulting serious difficulties because of divergent oscillations in the solution for an appropriately large number of energy divisions, i and j. It is expected that this type of difficulty will be avoided by solving for a spectrum smoothed by the energy resolution of the spectrometer. It was observed that the tail response of the pair spectrometer for monoenergetic sources could be fit reasonsbly well with a pair of ex~ ponential functions of electron energy. Thus the tall response becomes N _ P4 J Jial T(1,3) = P(3) [k] exp aj(E —m,) + 3 Il i | < 1 « + k5 exp a2Ee] AEe‘ for 0 < i< c‘j (2) - =0 - fori >e, , . and the total response is . os\ . J Jrel J Jol i R(1i,3) = P(J) {bij + [k exp ay(E - m,) + k5 exp asE )} 0BT . (3) 5Divergent oscillations in matrix inversion procedures have been reported in private communications from W. Zobel of ORNL and N. Scofield of the U. 5. Navy Research and Development Laboratory. 158 - | - At this point, the problem was divided into two parts: (1) determining P(j) and (2) determining the parameters kY and aY. Peak Efficiency Measurement. Values for P(j) were determined by exposing the spectrometer to each of the sources of gamma rays listed in Table 11.2. (Derived values for the N'® gamma rays were used because of the uncertainties associated with the experimental values.) For. the source measurements, the correction factors for counts lost due to the pair spectrometer electronics varied from 1.066 to 1.000 with errors which were unimportant (<0.006). In order to define the peak area, P(J), values of the constant c, and, dj had to be obtained as a function of Ei. The position of Cj (see Fig. 11.1) was chosen by observation of the monoenergetic response curves and was taken as | — - _ _— -5 J _ Cy =y =W, [1.31 — (8.5 x 10 )(Ey 1022)) , (4) where all quantities are in kev, and Wj is the measured resolution at Table 11.2. ©Sources Used in Calibration of Pair Spectrometer Source ' Ggigiégay Peak Efficiency Total Efficienc:ya (kev) (counts/photon) (counts/photon) x 1077 X 1077 7Znb3 1 114 - 0.00378 + 0.00053 0.0076 + 0.0011 Co®0 1173 0.010 * 0.0015 0.0164 + 0.0030 Na=? 1 276 0.0413 #* 0.0014% 0.0475 + 0.0018 Cob0 1 333 0.0682 * 0.0035 0.0868 £ 0.0050 Na2*% 1 368 0.0938 +* 0.001l4 0.121 * 0.0026 y88 1 840 0.511 * 0.010 0.566 + 0.011 Nal4 2 754 1.396 % 0.009 1.840 #* 0.013 glz 4 433 2.07° * 0.09 3.59 % 0.19 N16 6 135 2.32°> £ 0.09 4,922+ 0.20 Ni6 7 121 2.39P + 0.18 4.820 + 0.51 gl2 11 810 0.80 + 0.40 6.79 * 0.26 aProportional to Nal pair cross section. bAfter normalization. 159 full width and half maximum. The quantity wj is given approximately by W= 2.75 (Ei — 1022)1/2 — 4 for energies up to ~2.5 Mev, but for higher energies it increases more rapidly than the above expression because of broadening of the peak by bremsstrahlung losses. The quantity dj was taken as m, + 2w,. Large variations in the wvalue of dj cause only small variations in P(j), as may be seen by studying Fig. 11.1. The values obtained for P(j) are shown in Fig. 11.2 as counts/photon divided by a Bora approximation7 to the pair cross section for Nal in cmz/molecule. Division by this calculated (and incorrect) pair cross section was carried out simply in order to remove the very rapid dependence upon energy and permit reasonable interpolation. The value of P(j) decreases with increasing energy relative to the» pair cross section because losses from the peak become more important at higher energies. Positrons escape from the center crystal of the pair spectrometer and thus lead to lost counts. Bremsstrahlung losses by the electrons and electron escape give rise to pulses smaller than the peak pulse height. Because of.these and, perhaps, other factors, the tail appears to "swallow up" the peak at the higher energy until the two are virtually indistinguishable at 12 Mev. Tail Shape Fitting. The response curves obtained with eight gamma- ray sources were analyzed to find the tail shapes. These sources were not necessarily of known strength. For most of the gamma-ray energies, multiple runs were available for determination of the values of (kJ kg) and (ai,ag), which are the magnitudes and slopes, respectively, of the exponential fitting functions relative to the peak area. After a weighted combination of the tail constants was obtained, an interpolation vs energy was performed for determining three of the four constants. The fourth was determined by the constraint that the resulting value of 7Obtained by integration of the data given by P. V. C. Hough, Phys. Rev. 73, 266 (1948). 160 UNCLASSIFIED 2-01-058-0-538 & . 7“"“ / 1017 ) / counts . photon cm2 - molecule™!/ . ( / ~| & ~ o \ 2 N 10'6 5x 10 1 2 4 6 8 10 20 £, GAMMA-RAY ENERGY (Mev) FPig. 11.2. Variation of Pair Spectrometer Pesk Efficiency with Gamma-Ray Energy. The probability of producing a count in the defined peak, P(j), by a photon of energy, EJ, arising at the point of fission, has been divided by the Hough formulation of the Born approximation for the pair cross section of Nal. The larger of the two errors shown for the points at 6.13 and 7.12 Mev are supposed to reflect the effects of the normalization shown in Fig. 11.3. é; T(i,j)/P(J) equal the measured value, and then all four parameters were modified by a second iteration. A physical interpretation of the details of the fitting is difficult; because the slopes of both exponen- tials change sign as a function of energy. However, the fit achieved empirically is reasonably good, as indicated in Fig. 11.1. 16l The observed tail for the low-energy (1.368-Mev) gamma ray in Na2# was not used in obtaining the tail parameters. The dashed line shown in Fig. 11.1 was predicted by the fit. Experimental Determination of Pair Cross Section for Nal. Although the experiment was not designed to yield the pair cross section for Nal, it was possible to obtain relative values from the measured efficiencies. This process was carried out in order to promote confidence in the wvalidity of the measured efficiency data. The following types of corrections were applied to the observed pulse-=height spectra: 1. The tail response was extrapolated to zero energy before summing to determine the "total" response of the spectrometer. The extrapolated area was only a few percent of the total. 2. Attenuation of the gamma radiation in the source container was calculated by taking the geometry into account with a combination of analytical and numerical techniques. The geometry was "bad" and thus Compton scattéring from the source shell did not remove gamma rays. The attenuation varied from 1 to 14% as a function of gamma-ray energy. This effect cancels approximately in the peak efficiency determination, since the sources were in containers similar to the fission chamber. 3. Attenuation by a lithium neutron absorber (and its stainless steel cover) was in "good" geometry (attenuation of 13 to 26%). This effect exactly cancels in the peak efficiency determination, because the absorber is present in both fission and calibration data. 4. The energy-dependent attenuation of the gamma-ray beam in the Nal crystal was taken intoc account (8 to 15%). 5. Annihilation of positrons before slowing down constituted an appreciable effect. Fast annihilation produces either one quantum (unlikely) or two quanta, at least one of which does not have energy equal to 1 mocz, with a partly forward angular distribution with respect to the positron. Thus, fast annihilation will not, in general, lead to a palir spectrometer count. The positron spectrum used £o calculate the fast annihilation loss was taken to be intermediate between that predicted by the Born approximation and an assumption that all kinetic energy 162 resides in the positrons, since no exact spectral calculations are available. The loss due to fast annihilation varied from O to 19%. 6. Escape of positrons from the center crystal is the most diffi- cult correction to make. If the positron "escapes” from the Nal crystal but is annihilated in the aluminum can, a count may still be recorded in the tail response of the spectrometer. Therefore, attenuation in the can was considered in calculating the positron escape. As an indication of the order of magnitude of positron escape, the positron '"range" was taken as a fraction (0.5) of the "Nelms" range® and the Born approxima- tion positron spectrum assumed. The positron loss was later treated as a parameter in order to match the existing pair cross-section data. The change reguired from the above assumptions in order to achieve a fit was approximately 40%, well within the uncertainties. | 7. Finally, the total pair efficiency after correction for items 16 was divided by an integral of the Hough formulation of the Born approximation7 in order to give the relative pair cross section in Nal. The resulting data are shown in Fig. 11.3, together with a solid-line fit to other available data.g;lo The errors for the earlier data are similar to those shown. No points are shown in Fig. 11.3 that correspond to the two lowest energy points of Fig. 11.2, because the backgrounds for these low-energy sources were too large to permit a meaningful definition of the total efficiency. These backgrounds had a much smaller effect on the peak efficiency determination. In any event, fission data with the pair spectrometer are not available for energies below E7 = 1.4 Mev. The data appear to be reasonable, with the exception of the points at 6.135 and 7.121 Mev, which correspond to N6 gamma rays. Clearly an 8A. T. Nelms, "Energy Loss and Range of Electrons and Positrons," NBS-577 (July 30, 1958). See Table I; the ranges given ignore multiple scattering and bremsstrahlung losses. °G. W. Grodstein, "X-ray Attenuation Coefficients from 10 kev to 100 Mev," National Bureau of Standards Circular 583 (April 30, 1957). 10y, West, Phys. Rev. 101, 915 (1960). 163 A ¢ Pi.‘ UNCLASSIFIED 2-01-058-0-539 o 95\\ . $ N "FAIR/”BORN (ARBITRARY UNITS) o —— T —— THE CURVE IS DRAWN FROM AVAILABLE PAIR CROSS SECTIONS CF IODINE. | | | I l I i 2 5 10 15 £y, GAMMA-RAY ENERGY (Mev) Fig. 11.3. Variation of the Relative Pair Cross Section for Nal as Determined from the Measured Total Spectrometer Response. The total pair efficiency has been divided by the Hough formulation of the Born approximation for the pair cross section as a function of gamma-ray energy, but the absolute values of the ratio have no 51gn1flcance._ The solid line represents a fit to the available measured values®' ? for the pair cross section of I which carry errors comparable to those shown for the present data. The errors shown include all known contributions other than those due to positron escape. The points at 6.13 and 7.12 Mev were ad justed, as shown, to yleld a reasonable variation with energy. error was'made in the determination of the spectrometer efficiencies for this source, Two largely independent determinations led to the - indicated efficiencies within errors consistent with that shown in Fig. . 11.3. Careful re-examination of the data from these experiments has ) yielded no explanation for the discrepancy. Because the N6 experiment - and analysis required several months to complete, it was not repeated, but, instead, the apparent pair cross sections (relative to the Born cross section) were revised as shown. Then the peak-to-total ratios for this source were assumed to be correct, and the peak efficiencies were determined to correspond to the revised pair cross sections. These peak efficiencies are included in Fig. 11.2, together with errors esti- mated to include the uncertainties in normalization of the N'® data, as shown in Fig. 11.3. 164 Data for Low-Energy Region " The minimum gamma-ray energy covered by the multiple-crystal spece trometer data was approximately 0.4 Mev. In order to study lower energies, a single=-crystal spectrometer was used to examine the spectrum of gamma rays from fissions in an apparatus of low absorption. A large source- toFdetector separation distance was used so that a time~of-flight technique which would exclude neutron-induced backgrounds could be employed. A preliminary analysis of part of these data® demonstrated the presence of peaks in the pulse-height spectra, the most prominent of which could be identified as resulting from x-rays emitted by the light and heavy fission fragments. In a further analysis, data from several runs for each of the three overlapping energy regions were combined and correcfions made for random counts and other backgrounds, lost counts, analyzer nonlinearities, and the observed fission rates. After comparison within the three regions and in the areas of overiap, the data appeared to be consistent. - For the energy calibration required for the fission-data combination and for formulation of the single-crystal spectrometer response function, it was necessary to expose the spectrometer to sources of low=-energy gamma rays. Twenty gamma rays from 14 different sources were used, as well as the x-rays from somé of these sources. The response function remains to be formulated from these data. The absolute source strengths required to determine the peak efficiency, P(j), were obtained by the methods described above. The efficiency may also be calculated with .reasonable confidence. The energies for the x-ray sources were obtained from weighted combinations of energies for the K-shell transitions. The rather large attenuations in the apparatus used with the multiple-crystal spectrometer led to a careful design of minimum wall thickneés for the low-energy spectral measurements. Calculations were made for the attenuation of some 15 layers of material between source and detector. These calculations were equally applicable to either the monoenergetic sources or the fission-chamber radiation, since the sourceé were exposed at various positions within a replica of the fission chamber. 165 The gamma-ray cross sections used were those of Grodstein,9 except for = uranium, for which cross sections are available from recent experiments for the low-energy region.ll Most of the material was disposed in either "good" or "bad" geometry.- For the bad geometries, account was taken of the energy-angle correlation of Compton scattering, and a photon was considered to be removed from the peak response if the energy change exceeded the measured resolution (full width at half maximum). In order to check the calculated attenuations, eight sources of x-rays were placed at various positions within the fission chamber mockup, and the spectrometer response was observed. Comparison of the variation - of the measured response with the calculations demonstrated consistency by the usual X2 test for all but one of the sources. Since the fission chamber attenuation was dominant, especially for low energies, this - check is considered to validate the attenuation calculations. Remaining Analyses Two portions of the analyses remain to be completed. First the response matrix R(i, j) must be formulated for the Compton and single- crystal spectrometers from the response data and the known source strengths. Then Eg. 1 must be solved for the smooth gamma-ray spectrum and the associated uncertainty. Although they may be stated briefly, these steps are not trivial, especially since no satisfactory solution to the second problem has been demonstrated. Monte Carlo Code for Deep Gamma-Ray Penetration Calculations It was reported previously'? that the "conditional" Monte Carlo method does appear to be a suitable approach to the study of deep gamma- ray penetration, in spite of the fact that several earlier studies had indicated that further mathematical development would be necessary. The method has now been used to compute differential energy spectra and 11R. B. Roof, Jr., Phys. Rev. 113, 824 (1959). 12"ANP Semiann. Prog. Rep. April 30, 1960," ORNL-2942, p. 88. 166 | i .« ' differential angular distributions for two cases, the first being a l-Mev source in water, which is a weakly absorbing medium, and the other being an 8-Mev source in lead, which is a strongly absorbing medium. The separation distances in both media range from 1 to 20 mean free paths. . These particular cases were chosen because they were previously studied with the moments methods,13 and the results of the two methods could thus be compared. Expressions for the differential energy spectrum and the differential angular distribution are 1 yB) =2 dw yw,B 4(x,E) wf_l 8(z,0,B) (5) and rEg ,w) =2 ar ,w,E) ¢(r fl‘JE ¢(r (6) respectively, where . ¢(r:w:E)-= n(r,w,E) REE_) ’ n(r,w,E) = number of collisions per unit volume, steradian, unit energy, and unit time, E = energy of particle, w = cosine of the angle between the particle direction and the radius vector whose magnitude is r minus the distance from the source, | o ‘ Eqg = source energy. All the Monte Carlo results are given as histograms, where the quantities T plotted are those given by Egs. 5 and 6 scaled by 4flr2euo , where bo = k(Eo). | The differential energy spectra for the 1-Mev source in water are shown in Fig. 11.4, and the angular distributions for the same case are 13H, Goldstein and J. E. Wilkins, Jr., "Calculations of the Penetra- tions of Gamma Rays, Final Report,” NDA 15C=-41 or NYO=3075 (1954). 167 W s B _\o i Q o N, k= g 0.01 Fig. 11.4. Source in Water. 168 UNCLASSIFIED ORNL-LR-DWG 52588 MONTE CARLO MOMENTS METHOD Hol = 20 0.208 0.406 0604 0.802 4.0 £ (Mev) Differential Energy Spectra for a 1-Mev Isotropic Point s 1 " 9! 1 given in Fig, 11.5. The angular distributions for separations of 10 and 20 mean free paths are not included because they were noticeably irregular. The differential energy spectra and angular distributions for the 8-Mev source in lead are shown in Figs. 11.6 and 1l.7, respectively. In Figs. 11.4 and 11.5 the smooth curves are the comparable moments- method results.l? Figure 11.8 shows the differential energy spectrum of the 8-Mev source in lead at 20 mfp, as computed in three fashiéns. The solid-line histogram is the result shown in Fig. 11.6; the broken-line histogram represents Monte Carloc results obtained by using the moments-method absorption coefficients reported in ref. 13; and the smooth curve shows moments-method results. Figure 11.9 displays the angular distribution for the 8=Mev source in lead at 20 mfp, the solid-line histogram being taken from Fig. 11.7 and the broken~line histogram again representing the result obtained by using moments-methbd absorption coefficients. Table 11.3 gives a comparison of the two sets of absorptién coefficients. Table 11.3. Comparison of the Most Recently Obtained Linear Absorption Coefficients for Lead with Those Used in Moments-Method Calculations Lk, Linear Absorption Coefficient Energy — (Mev) (1) Most Recent® (2) Used in Moments= (1)/(2) _ ' Method CalculationsP 0.4 2.35 2.41 0.975 0.5 1.63 . 1.66 0.982 0.6 1.29 1.28 1.01 0.8 0.9%4 - 0.929 1.02 1 0.771 0.755 1.02 1.5 0.575 0.575 1.00 2 0.517 : 0.513 1.01 3 0.475 0.466 1.02 4 0.474 0.471 1,01 5 0.481 0.489 0.984 6 0.493 0.506 0.974 8 0.518 0.531 0.976 | aSee ref. ©. bSee ref. 13, 169 UNCLASSIFIED ORNL-LR—DWG 52589 10 amrieto’d (r,w) -2 10 1.0 08 0.6 0.4 0.2 0 -02 -04 -06 -08 -0 w Fig. 11.5. Differential Angular Distributions for a 1l-Mev Isotropic Point Source in Water. 170 UNCLASSIFIED % 2 ORNL- LR~ DWG 52550 MOMENTS METHOD MONTE CARLO 4m'zef‘LO_’¢> (r,E) 0.01 1.608 3.206 4.804 6.402 8.0 E (Mev) . Fig. 11.6. Differential Energy Spectra for an &-Mev Isotropic Point oource in Lead. 171 Figure 11.10 shows the buildup factors for both the 1l-Mev source in water and the 8-~Mev source in lead, as computed from Monte Carlo and moments methods. The buildup factor is the result of dividing the area under the differential energy-spectrum curve by the source energy and adding unity. UNCLASSIFIED ORNL-LR - DWG 52594 arrtelo’d (r,w) 0.2 0 -02 -04 -06 -08 -10 ‘W 0 08 06 04 Fig. 11.7. Differential Angular Distributions for an 8-Mev Isotropic Point Source in Lead. : 172 UNCLASSIFIED ORNL-LR-DWG 52592 100 50 20 10 amrtelo’(r,£) 0.04 1.608 ‘ 3.206 4. 8049 6.402 8.0 £ (Mev) : Fig. 11.8. Comparison of the Differential Energy Spectra for an 8=Mev Isotropic Point Source in Lead Calculated by (a) Monte Carlo Method with the Most Recent Absorption Coefficients, (b) Monte Carlo Method with Coefficients Used in Moments Method Calculations, and (e¢) Moments Method. All the data are averages of repeated computations, each with a relatively small number of histories compared with the total number of histories in the average. The differential data showed some fluctuation between repetitions, with the lead.déta much more stable than the water data, but the integrals of the data fluctuated much less. A1l the water data were computed with approximately 4000 histories per histogram, except for the 20-mfp results, which were obtained from 13 000 histories. The lead data utilized approximately 12 000 histories per case except for the 15 and 20 mfp computations, in which 40 000 histories were used. The larger number of histories in the lead 173 UNCLASSIFIED ORNL - LR-DWG 52593 amriel o’ (r,w) 1.0 0.8 0.6 Cc4 0.2 0O -¢2 -04 -06 -08 =10 w _ Fig. 11.9. Comparison of the Differential Angular Distributions for an 8-Mev Isotropic Point Source in Lead, Calculated by (a) Monte Carlo Method with the Most Recent Absorption Coefficients, and (b) Monte Carlo Method with Coefficients Used in Moments-Method Calculations., 174 -UNCLASSIFIED CRNL-LR—-DWG 52594 - 100 ”2°’W?‘ P MONTE CARLO USING . MOMENTS METHOD / COEFFICIENTS , . S : MOMENTS METy . ' ) ] 20 / / . y . AIONTE CARLO - x | 2 Pb & / a / 3 S 10 7/ 3 / MOMENTS METHOD / ,/ > s /L / : 4 /| / & / //MONTE CARLO 5 / / / /// / / ;/ - 2 / 1 0 2 4 6 8 10 12 14 16 18 20 Lo {mfp) Fig. 11.10., Comparison of Buildup Factors Calculated by Monte Carlo and Moments Methods for a l=Mev Ithropic Point Source in Water and an 8-Mev Isotropic Point Source in Lead. 175 calculations arise because the histories are terminated much more quickly in lead than in water. The results obtained in this test indicate that the conditional- sampling technique applied to the spacial part of the problem gave fairly accurate results for distances up to about 10 mfp. The Monte Carlo estimates for water became relatively inaccurate, compared with moments- method results, for distances greater than 10 mfp. It is emphasized that no importance sampling, etc., was performed to obtain samples in the momentum space. The Single=Scattering Approximation to the Solution of the Gamma~Ray Air-Scattering Problem It has been hypothesizéd that a good approximation of the scattered flux (or dose rate) in air is given by the single-scattered flux (or dose rate) with exponential attenuation and buildup neglected. If suffiéiently accurate, the simple single-sqattering approximation would be useful in computing skyshine from a reactor or other source. Neglect of attenuation, of course, tends to compensate for neglect of buildup. Such an approximation has been shown to be satisfactory in the case of neutrons scattered in air isotropically in the center-of-mass system in that the results are in good agreement with those of a Monte Carlo calculation.'% Since Monte Carlo calculations also exist!? for the air- - scattered flux and dose rate for gamma rays, a comparison of results obtained from the gamma-ray single~scattering approximation with Monte Carlo results has been made. A line beam of monoenergetic gamma rays emitted from a point source in a uniform, infinite medium of air was considered. The resulting dose 14, L. Keller, C. D. Zerby, and J. Hilgeman, "Monte Carlo Calcula- tions of Fluxes and Dose Rates Resulting from Neutrons Multiply Scattered in Air," ORNL-2375 (1958). 1R. E. Lynch et al., "A Monte Carlo Calculation of Air-Scattered Gamma Rays," ORNL-2292, Vol. I (1958). 176 "rate at a distance x from a source of energy E emitting photons at an angle ¥ from the source-detector axis is conveniently given by N C(E) D(E D(E, %) = & HLELDEY) where N is the electron density per cm’, and the functions C(E) and D(E,V¥) are defined by the expression - f %flc_’ (E,6) T(E,6) a6 = C(E) D(E,v) |, ¥ in which angle through which the photons are scattered, %g = Klein-Nishina scattering cross section, - T = conversion factor (given in ref. 2). The product C(E) D(E,¥) was computed numerically on the Oracle. The scattered flux was obtained by a similar method, which will not be detailed here. ‘ ‘The results of the calculations and the comparison with the Monte Carlo data have been published.'® They show that the simple dose-rate approximation gives very good fesults for all angles and photon energies, whilerthe flux approximation gives good results to about 60 deg for distances up to 10C ft. If the detector is shielded in any way, however, these results probably are not valid, since energy and angular distribution at the shield are important. Grinder — An IBM=704 Monte Carlo Program for Estimating the Scattering of Gamma Rays from a Cylindrical Medium A Monte Carlo program has been written for the IBM¥704 data- processing machine for the computation of energy spectra and dose rates 16p, K. Trubey, "The Single-Scattering Approximation to the Gamma- Rays," ORNL=-2998 (1960). 177 of gamma rays scattered from within a homogeneous right-circular cylinder to various positions outside the cylinder. The program can be used to investigate the dose rate in a cylindrical crew compartment resulting from scattering in various parts of the crew shield. As the program is presently written, the gamma-ray beam is incident upon the end of the cylinder and coaxial with the cylinder. A small change in the code would permit the beam to strike the end surface at any angle. The energy and radius of the incident gamma-ray beam, the dimensions and composition of the cylinder, and the detector positions (limited to 20) are specified in the input. An important feature of the program is that the computing time, rather than the number of histories, is an input parameter. The photon histories are traced strictly by an analogue procedurel”? (no biasing techniques are employed), and contributions to the detectors are computed by statistical estimation, i.e., the contribution to each detector is the probabllity that a photon will scatter toward the detector and arrive there uncollided. Thus each collision makes a contribution to each detector. The usual difficulty in statistical estifiation arising from scatterings occurring very close to a detector is not encountered because all detectors are outside the scattering medium. The energy spectrum at each detector position is given in histogram form in units of energy flux per unit incident photon flux per unit energy. There are 20 equal energy intervals of width (Ep — 0.01 Mev)/20. The energy spectra are multiplied by dose conversion factors and added . to yield an estimate of the dose rate (tissue) in rads per hour per unit incident photon flux. The output includes the number of casé histories, number of collisions, number of histories degraded below the energy and weight cutdffs, the energy albedo, the energy transmitted, the energy leaking from the lateral surface, time used for the cqmputation, initial and final random numbers, and the input parameters. The output is on magnetic tape for off-line printing. 17H. Kahn, "Applications of Monte Carlo," AECU-3259 (April 1954). 178 ir Angular Distribution of Neutrons Emerging from Planar Surfaces of Diffusing Media ' The design of a reactor shield would be greatly facilitated by a knowledge of the energy and angular distributions of the neutrons being transmitted through the shield. For example, the energy details of the flux within a shield are needed for the calculation of neutron-induced- gamma~ray fluxes in the shield materials, and the energy and angular distributions of the neutrons at the surface of a shield are required for calculations of gamma-ray doses resulting from neutrons captured outside a shield. Unfortunately, the problem of neutron diffusion near a boundary in a scattering and absorbing medium does not easily lend itself to an analytical solution, especially when the medium is hydrogenous and contains a strong poison. Slowing down is difficult to factor in, and strong flux gradients presumably make diffusion-theory approximations completely invalid. Two calculational methods now being developed for the problem are the NIOBE Code (Direct Numerical Integration of the Boltzmann Equation) at NDA and a Monte Carlo Code at ORNL, both of which will yield energy spectra and angular distribution data. In addition to the calculations, the problem is being attacked experimentally. A neutron chopper, which will be used to measure energy spectra, is now under construction at the ORNL Bulk Shielding Facility (see chap. 10), and an experiment to determine the angular distributions of low-energy neutrons emerging from slab shields is currently in progress. The results obtained thus far are reported here. The informa- tion is wvaluable not only for comparison with calculations already performed but also to support the design criteria of the neutron chopper. Because of the current interest in high-performance neutron shields, the experiment has included several measurements on polsoned and unpoisoned hydrogeneous slabs. A review article by Bethel® gives an angular distri- bution of ¢(p) = 1 +/3 p for pure paraffin (or water) calculated by the 184, A. Bethe, Revs. Mod. Phys. 9, 132-33 (1937). 179 diffusion theory approximation, and this has already been verified some- what at thermal energies by Fink.!'®? One goal in this experiment was to determine whether this angular distribution is also valid in hydrogenous shields at other energies and to investigate how it varies with the addition of poison, the thickness of the sample, and the angle of incidence of the input neutrons. If the angular distribution were found to be independent of the angle of incidence, it could be assumed that the Milne problem?® is correctly mocked up. This is of particular interest, since many calculations use.the Milne geometry (semiinfinite medium with the source at ~»). In order to investigate the angular distribution of neutrons in nonslowing-down media, a group of measurements was also made on poisoned and unpoisoned lead slabs. These were compared with calculations which have been made by Case et al?l with the one-velocity transport theory. Description of Experiment The experimental apparatus, as positioned adjacent to the reactor, is shown in Fig. 11.11. Neutrons from the reactor pass through a 10-in.- diam air-filled collimator to the shield sample where, after some diffusion, the attenuated neutron flux emerges into an air chamber. Some of the neutrons pass into the air-filled beam tubes and to the detector foils. Neutrons not captured by the foils mostly strike the end of the air- filled beam-catcher cap. Except for the section adjacent to the shield sample, the air chamber was lined with a 1/4-in. thickness of well-packed BiOC powder to minimize scattering within the chamber. The beam tubes and the beam- catcher caps were also lined with the BiOC. In order to investigate the *9G. A. Fink, Phys. Rev. 50, 738 (1936). 205ee, for example, F. T. Adler and C. Mark, "Milne Problem with Capture," MI'-66, Chalk River. ' 21K, M. Case, F. de Hoffmann, G. Placzek, "Introduction to the Theory of Neutron Diffusion,"” U. S. Govt. Printing Office, Washington 25, D. C., p. 147 (1953). 180 THIS TUBE USED FOR MONITOR FOILS Cd-COVERED .. HALVES ~7 UP —-t—.- | ___‘/ BEAM TuBE)’) PROFILE —FOIL STACK 0.02 in. Cdf NEUTRON BEAM DETECTOR FOIL DETAIL FPig. 11.11. 8T T=4in.OR 6in. | | AIR-FILLED ~ CHAMBER UNCLASSIFIED 2-04-058-567 [rnt— 2'/2 in. &° \ 7 Q° _H__ - - :__—‘________,__————:' > / AIR-FILLED 10n. /X COLLIMATOR | gaN“TUB T NeameLe \R—F\\—\’ED / /’/ / ‘ A e \/ /' ‘ . { POWDER / i /. o & /I/ TUBE-DETAIL WITH 0 BEAM-CATCHER CAP & IN PLACE Apparatus for Measuring Angular Distributions of Neutrons Parallel Slab of Material. ,,,,,,,,,,,,,,,,,,,, S LR S oa e POWDER, TIGHTLY PACKED BLOCKING FOIL DETAIL Diffusing Through a Plé,ne effect of the angle of incidence of the input neutrons, an air-filled adaptor tilted 30 deg with respect to the normal in the horizontal plane was placed between the shield sample and the 10-~in,-diam collimator. The shield sample, the air chamber, and the beam tubes were thereby tilted 30 deg with respect to the input neutrons. The circular detector foils were divided into half-disks as shown in the foil detector detail in Tig. 11.11. One stack of half-disks was cadmium~covered (20-mil thickness), and the other stack was bare. With this arrangement, several difference measurements were possible. With the gold foils, for example, the difference in neutron activation between the B-foil and @he C-foil gave a measure of the gold-resonance neutron flux. The B-~foil was black to-most of the gold-resonance neutrons but only about 3% absorptive to thermal neutrons. And with the indium foils, the differences between the B~ and C~foil activations (called the B-C activity) gave a measure of the indium-resonance flux. The D-C activity gave a measure of the thermal-neutron flux. -Each angular distribution determination (each run) was made with two foil exposures. In one exposure, gold, indium, and cadmium blocking foils were inserted at the air-chamber end of the beam tubes (good geometry), as shown in the blocking-foil detail of Fig. 11.11. This exposure determined the background activity of the detector foils. After removing and counting the detector foils, the exposure was repeated with the gold and indium blocking foils replaced by a copper foil of equal scattering probability. The detector-foil activity in the second exposure thus was the foreground activity. The background B-C gold~foil activity was then subtracted from the foreground B-C'activity to obtain the net activity. | Each group of detector foils was always removed within about 15 min after exposure and counted three times so that erratic counts could be eliminated by "voting out” the bad count. Counting errors were somewhat infrequent. A 20~channel analyzer stored the photopeak counts from a 3= by 3-in. Nal detector. After corrections for foil decay and foil- weight variations were made, the data were normalized and plotted 182 as ¢(u)/¢(1) versus u, where p = cos 6 and 6 is the angle from the normal. Resgults The plots of ¢(n)/¢(l) versus p are presented in Figs. 11.12 and ll.lé. Bach column of graphs represents one run with the exception of the last column of Fig. 11.13 where runs 9 and 10 are combined. In the colums with three graphs, the top graph presents thermal-neutron data, the middle graph indium-resonance (1.44=-ev) data, and the bottom graph gold-resonance (4.9-ev) data. The samples used in runs 1 through 3 (Fig. 11.12) were 6-in.-thick hydrogeneous slabs. with boron poisoning of oy = 0, 4.66, and 72 barns, respectively (oy = poisoning per hydrogen atom at thermal energy). The corresponding poisoning for each energy region is given by the C value shown on each graph, where C = ZS/(Za + Zs). The angular distributions for all these runs are strikingly similar and seem to follow the ¢(n) = 1 +./3 p angular dependence (shown by solid lines in the graphs) for a nonpoisoned hydrogenous medium given by Bethe.'® In run 4 the 6-in.-thick slab used in run 3 was reduced to 4 in., and the angular distribution was found to be identical to that obtained in run 3, within the accuracy of these measurements. The curved, dashed lines in runs 3 and 4 are from one=velocity calculations by Case et al.?! for the C=values given in each graph. (The thermal data for these runs were statistically meaningless and were therefore deleted; reruns are being made.) Runs 5 through 7 (Fig. 11.13) were identical with runs 1 through 3, exceptrthat the angle of incidence of thé input neutrons waSVBO deg. Again no change in ¢(u) is observed. It is this observed constancy of the angular distribution with the angle of incidence which supports the design of the BSF neutron chopper. If it had been found that'the angular distribution of the emergent neutrons varied with the angle of incidence, it would not be feasible to measure the energy spectra for normal incidence only, as is planned with the chopper, and to measure the energy spectra as a function of angle of incidence would greatly compli- cate the design of the chopper. 183 The samples used in runs 8 through 10 consisted of é-in. thicknesses of powdered lead, unpoisoned in runs 8 and 9 and poisoned with ByC in run 10 (C = 0.17 for thermal neutrons). The powder density was 7.14 g/cm?. In run 8 the input neutrons entered at an angle of 30 deg, while for runs 9 and 10 they were normally incident. It is observed that in lead the angular distribution for the 30-deg input neutrons did not peak at p = 1 and also that the neutrons emerging from the lead samples had a memory of their input direction. Yet in runs 8 and 9 the "good-statistics" thermal-neutron data suggest an approximate ¢(p) = 1 + 2u dependence for angles far from the angle of input of the neutrons, as the one-velocity (C = 1) calculation predicts. The run 10 data superposed on the run 9 data in Fig. 11.13 show that there is indeed more forward peaking for the poisoned lead fhan for pure lead. The dashed curves with these plots represent the one-velocity calcfilations of Case et al.?! for the poisoned sample. A crude fit is attempted at larger angles (small u). These "one-velocity” runs must be repeated using a heavier slab of lead. Conclusions The experimental results for hydrogenous materials agree with the angular distribution of ¢(p) = 1 +\f§ n given by Bethel® for nonpoiscned hydrogenous media and also show that the angular distribution is un- affected by poisoning for the range of poisoning investigated (1 2 C 2 0.75). Furthermore, a comparison of runs 3 and /. shows that ¢() is not affected by varying the sample thickness from 4 to 6 in. This observed constancy with thickness is to be expected once spectral equilibrium is reached, and 1t can be assumed that spectral equilibrium was reached in this material, since Beyster et al.?? have shown that the spectrum is constant in a much more weakly poisoned hydrogenous material (polyethylene) at thick- nesses above 2 in. These experiments indicate then that after the neutron energy spectrum becomes stabilized, the angular distribution of neutrons °2J. R. Beyster, W. M. Lopez, R. E. Nather, and J. L. Wood, "Measure- ment . of Low Energy Neutron Spectra,’ GA-1088 (Nov. 13, 1959). 184 2-01-058-568 ' UNCLASSIFIED 1.2 | C=1.0 C=0.86 fi 1.0 \ l . 1 0.8 N, : ,)/;\ N . \}\ THERMAL =IZ 0.6 ; i\ 8 (ANGLE lf!hom NORMAL ) (~0.25ev) =|® ~ p = cos ,. 0.2 fl ° | 1.2 [ C=1.0 C=0.98 _ . I lc=0.75 : : €=0.75 1.0 ” ¢ | ! ,}//\:\;\ | F | \ 0.8 N N 2NN §, SN PN INDIUM \\ N // \\ \F\I' // \\ N . —_ / \ B / N RESONANCE (T o6 N 2 M 1 < X _ (1.44ev) *I° N \ - ‘ \ N 44 ev N N ‘o i \\\. N 0.4 < N ! N ~ . ~ \\ 1 ~ ~ \ It ~~L 0.2 e T \‘*-\\ | N 0 ‘ 1,2 i‘ c=1.0 €=1.0 [ "|¢=0.84 C=0.84 10 4 i v - :j i\T\ /‘?\ix | / \:$\ ¢ { N . / \ \ 0.8 \i\ I_/ A \$\ ‘ / CN GOLD N \N SINN / N ; v * ‘\ - S \ RESONANCE 2|z o0s ~ AN AN AN N N 4 e \ N \ ~ \ (4.9 ev) \\\\ F ~ 0.4 \\\\ AN N4 ~ N \\\\ \\. . \\ 9 0.2 W o~ 0 ' - . | 08 10 08 06 04 02 0 08 10 08 06 04 02 O 08 10 OB 06 04 02 O 08 10 08 06 04 02 O p , oo ' K ' RUN NO. 1. RUN NO. 2. RUN NOQO: 3. RUN NO. 4. 6in. Hy0, INPUT L 6in. BORIC ACID SOLUTION, INPUT L 6in. METHYL BORATE + METHANOL, INPUT L 4in. METHYL BORATE + METHANOL , INPUT L Fig. 11.12. Angular Distributions of Neutrons Emerging from Poisoned Hydrogenous Slabs with Thermal, Indium Resonance, and Gold Resonance Energies. The solid lines represent the ¢{p) = 1 +4/3 p angular dependence given by Bethe for a non- poisoned hydrogenous medium. The dashed lines give ¢(p) values from one-velocity calculations by Case et al. for the specified C values. Error bars indicate statistical uncertainties only. " . 185 INPUT NEUTRONS INPUT NEUTRONS INPUT NEUTRONS UNCLASSIFIED 1.2 ' 2 4 | (2 2-01-058-569 - . - ' €=1.0 ' €=0.86 ' €=1.0 *C=1.0 , - . 0 ¢ =047 LO/’/) \?\ } 2.0 1.0 ; * 0.8 \‘\\\ ! \\\\\ i it =cos & (ANGLE FROM NORMAL) ' : 08 * THERMAL ‘ “\\ ~ . 1 (A~ = o6 . ™, 1.2 . 0.6 et (~ 0.25 ev) zy& J AN . ’//(°f\\\\ 0.4 AN \\\ 0.8 b 0.4 ¥ S ' N ] cr \\ ‘ / \ \ .\. // . \ 7 0.2 0.4 T 0.2 [ A\ ™~ . . —~_ j J o\\- . © INPUT NEUTRONS ° 0 1.2 i 2.4 : 12 ¢ C=1.0 C=0.98 C=0.75 C=1.0 e C=1.0 2 * 4 1 " o C=0.59 10/{' ANE ~ ' .20 10 1 0.8 \\\\T 4 LN 1.6 i, S T \ l// 1\ I\ 0.8 INDIUM o 9 }\§ 7N o RESONANCE s’: 0.6 N : \ A 2 06 (1.44ev) *° \\ N NN v 0.4 ¥ \\\\ AN N os L, 0.4 G : ey N . -] v ~o T //L‘\i\'\\ 0.2 > - 0.4 '\\\“\\\ 0.2 }/ A k\‘me\\ : '--..__' 7 - E\\§\\{ i \ 0 0 0 T 1.2 2.4 1.2 C=1.0 C=1.0 ' l C=0.84 § c=1.0 ¢C=1.0 1.0 : b ll 2.0 ; 1.0 ? ¢=0.73 /N 0.8 |~ 1 T T\ 7N N, 1 6 0.8 " ¥ . s 7 T \l . . GOLD I W\\ A N T\N ~ . RESONANCE 3|= o6 { ~ ) 1.2 X 0.6 (49 eV) w\ \ \\\ * 5 - 0.4 ' ™. \ o \ e . o A° \ N ~ ~ 0.8 —s —— 0.4 7N e \\\ - \\ ** - | \\ /o § Q\\ - 0-2 \\ 0.4 0.2 ‘U..,._ ; A : --...g__--_.-q-‘- —— O f o o - 08 10 08 06 04 02 O 08 10 08 06 04 02 O 08 10 08 06 04 02 O 08 {10 08 06 04 02 O 08 1.0 08 06 04 02 O m p | p : P p RUN NO. 5. RUN NO. 6. . RUN NO.7. RUN NO. 8. « RUN NO.9, 6in. PURE Pb, INPUT L 6 in. H20 ,INPUT 30 deg 6in. BORIC ACID SOLUTION, INPUT 30 deg Fig. 11.13. Angulér Distributiohs of Neutrons Emerging from Poisoned Hydrogenous and Lead Slabs with Thermal, Indium Resonance, and Gold Reso- nance Energies. The solid lines in runs 5, 6, and 7 represent the ¢(p) = - 1 +./§ p angular dependence given by Bethe for a nonpoisoned hydrogenous 186 6in. METHYL BORATE + METHANOL, t | INPUT 30deg medium. The solid 1 dependence expected for pure lead (C = 1). values from one-velo values. Error bars 6in. PURE Pb {p=7.14g/cc), INPUT 30deg o RUN NO.10, 6in. Pb + B,C, INPUT L ines in runs 8, 9, and 10 represent the ¢(n) = 1 + 2u The dashed lines give ¢(p) city calculations by Case et al. for the specified C indicate statistical uncertainties only. in typical hydrogenous materials, both poisoned and unpoisoned, is con=- stant. This being true, the angular distribution given here can be used in shielding calculations for a poisoned hydrogenous material, and the energy spectrum need be determined for only one thickness of the material, providing the thickness is greater than 2 in. This will greatly simplify the shieldifig problem, since calculations for this energy region are difficult because of crystal structure and molecular binding effects. The experiments with the lead slabs agree with the one~velocity calculations of Case et al.?! qualitatively; i.e., from the superposed plots of runs 9 and 10, the poisoned lead certainly shows stronger forward peaking than the unpoisoned lead. However, the fact that the angular distribution is affected by the angle of incidence of the neutrons shows that the Milne problem was not properly mocked up and precludes further discussion until additional measurements with thicker lead slabs are made, An IBM-704 Program to Determine Angular Distributions of Neutron Histories Generated by the 05R Code There are many problems in the mathematical analysis of the behavior of nuclear reactors which, for one reason or another, are not easily handled by analytical or conventional numerical calculations but which can be treated by the Monte Carlo method. To facilitate such a treat- ment, a general-purpose Monte Carlo reactor code, identified as the Oak - Ridge Random Research Reactor Routine (05R), was developed fbr the IBM="704 computer. It was prepared to perform three general tasks: (1) to calculate specific reactor parameters, up to and including the set of parameters describing an entire reactor; (2) to examine the capabilities of the Monte Carlo method for reactor calculations; and (3) to furnish results suitable for use as comparison standards in evaluating other methods of computation. To facilitate the third task, great care was taken to assure an adequate representation of neutron cross-section data, and, as a | result, cross sections may be specified at as many as 10% energy points, 187 Because the output of the code includes the velocity components of the neutrons after each collision, it is adaptable to the investigation of the angular distribution of neutrons across plane boundaries. For the angular distribution investigation, an additional IBM=-704 code 1is being developed that will process the neutron histories generated by the O5R code. To do this it is assumed that the neutrons enter the end of a cylinder and then diffuse through the length of the cylinder. Any given cylinder is divided infio ten or fewer cross-sectional regions, and the neutrons leaving each region are aivided into 20 equal solid- angle .groups for each of 28 energy groups. The total weight of neutrons in each group and the first five angular moments for each energy group crossing each boundary are recorded. Further, the total weight of neutrons becoming thermal in each region is determined, and the angular distribution of the neutrons emerging from the opposite end of the cylinder is thus determined. Upon completion the code will be used to predict the angular distri- butions of neutrons emerging from hydrogencus slabs (specifically, lithium hydride) in a geometry which will be analogous to the experi- ments reported in the preceding section entitled "Angular Distribution of Neutrons Emerging from Planar Surfaces of Diffusing Media." Development of an IBM-704 Analytical Code for Analysis of Axially Symmetric Reactor Shields A series of codes, each of which can be used for a line-of-sight calculation of a separate component of the radiation from a reactor shield, is being written for the IBM-704 computer. These codes will be appli- cable to the preanalyses of Tower Shielding Facility experiments and also to general shield design. Thus far, three codes have been completed: a code for the primary neutron dose component, one for the primary gamma-~ray dose component, and a third for gamma-ray buildup. In the program for determining the primary neutron component, an axially symmetric reactor shield cbnfiguratiofi is assumed, and the dose rate 1s computed for a detector point on the axis of symmetry. The 188 reactor core is defined either as a homogeneous sphere or as a homogeneous spherical annulus, and the surrounding shield is considered to be a seriés of concentric homogeneous spherical annuli plus an outer shaped region. A single shadow shield‘may be imbedded in the outer region. Its inner surface must be spherical, and its thickness may be a function of polar angle. (A more general arrangement, which will permit three such shadow shields, will be introduced in the program later.) _ The fuel region is divided into annular sections, and the dose rate to any point on the axis of symmetry is calculated on a line-of=-sight basis. The attenuation in the hydrogenous portion of the shield is assumed to be of point kernel form, such as that given for moments- method calculations;?3'2% in practice, the logarithm of the expression 4m?D(r), in which D(r) is the dose rate at a distance r from a point - source of neutrons, is fitted with a polynomial, and the coefficients are fed in as input data., The attenuation in the heavy materials in the shield is based on removal cross=-section data. The attenuation through both the hydrogenous and the heavy materials on the rear of the crew compartment is lumped with that of the reactor shield. Provision is made in the code for a maximum of 15 reactor shield regions. The dimensions of each region, the increment of the angular variable, the angular thickness functions for shaped portions of the shield, and the radial power density function of the active core are all treated as input data. The calculation of distance traversed in each of the regions has been coded as a Fortran subprogram and can be used in other main programs of this study if required. ~ This basic code; although designed to compute fast-neutron dose_ rates, is also used to calculate primary gamma~ray dose rates. Capture gamma rays resulting from thermal-neutron captures within the limits of the TSR-IT reactor tank are included in the primary contribution. 23R, Aronson et al., "Penetration of Neutrons from a Point Isotropic Fission Source in Water," NYO-6267, NDA 15C-42 (Sept. 22, 1954). 2%R. Aronson, J. Certaine, and H. Goldstein, "Penetration of Neutrons from Point Isotropic Monoenergetic Sources in Water," NY0-6269, NDA 15C-60 (Dec. 15, 1954). ' 189 A volume integration over the source region is carried out, and attenua- tion along each line-of-sight path is computed by using the product of an exponential times the builldup. Values of the bulldup factor are taken from data for water that are based on the total number of mean free paths traversed.?? A Fortran subprogram utilizing the dose-rate buildup factors for water?? has been written to provide a polynomial approximation to the bulldup. A fit of the cube root of the buildup as a function of the number of mean free paths was determined, for fixed energy, by using third~-degree polynomials, and the various coefficients were fitted, in turn, by using fourth-degree polynomials. The resulting expressions - give values within 5% of the published experimental data over the entire energy range from 0.255 to 10 Mev and from O to 20 mean free paths. The machiné time required to obtain a single value of the buildup is approxi- mately 8.4 msec. '_ The gamma-ray buildup code and the subprogram for calculation of distance traveled in each shield region have been coded and "debugged." The remainder of the programs described above have been coded and are in the process of "debugging." Calculations of Inelastic Cross Sections for High-Energy Particles Incident on Complex Nuclei One factor that will dictate the payload capacity of future space vehicles is the weight of the radiation shielding needed to protect personnei and equipment. In order to design minimum-weight shielding and therefore to maximize the payload for a given thrust, it is necessary to have available és accurate and detailed information as possible about the penetration of shields by radiation in space. One component of space radiation is protons that appear in solar flares and in abundance in the first Van Allen belt, with a maximum flux 2°H. Goldstein and J. E. Wilkins, "Calculations of the Penetration of Gamma Rays, Final Report," NY0=-3075 (June 30, 1954). 190 2.5ec”™t (ref. for energies greater than 40 Mev of about 104 protons-.-cm” 26). The details of how these protons are stopped in shields by ioniza- tion collisions is well understood; however, the shielding problem is complicated by nuclear events that can take place with the higher energy components of the proton energy sfiectrum, which extends to approximately 700 Mev. One result of an inelastic collision of a high-energy proton with'a complex nucleus is the ejection of one or more nucleons. The neutral charge component of the nucleon, the neutron, is difficult to " stop and may govern the shield size and weight. A detailed knowledge of the total and differential cross sections for the inelastic nuclear events is essential for accurate shielding calculations. Although some of the cross-section data for these reactions have been obtained experimentally, they are insufficient for a compre- hensive shielding calculation. An experimental program for finding all the necessary data would be prohibitively long and expensive. TFor this reason it is necessary to rely on calculations to supply the missing | information, and a program for performing the calculations is under way. In conjunction with the high-energy accelerator shielding effort, a Monte Carlo code is being written to calculate the inelastic cross sections for highe-energy particles incident on complex nuclei. The low=energy portion of the calculation, which will éxtend to approximately 1 Bev of incident energy and will be directly applicable to the space- vehicle shielding effort, has been coded and is being "debugged." A preliminary check of the models and procedures used in the calculation has been made by caldulating some total inelastic cross sections and comparing.them with experimental results. In Tables 11.4 and 11.5, which present a portion of the comparison, ob is the calculated»inelastic Cross section, O, the experimental cross section, og the.geometrical Cross section determined from the model of the nucleus used in the calcu~ lation, and E the incident energy of the particle, The results obtained 265, ¢. Freden and R. S. White, Phys. Rev. Letters 3, 9 (1959). 191 thus far indicate the general validity of the approach, and it is antici- pated that the differential data will compare favorably with experimental data. Table 1l.4. Total Inelastic Cross Sections for Neutron Incident on Various Elements at Various Energies Calcula?ed Experimental Geometrical Energy . Inelastic . . Material . Cross Section, Cross Section, (Mev) Craoss Section, o (mb) o (mb) o, (mb) X g & Al 536 500 + 507 769 Cu 892 910 + 50 a 1138 Pb 1768 1850 = 180 2045 765 A 470 435 + 107 Cu 821 822 = 23b Pb 1700 1923 + 62 8. P. Millburn et al., Phys. Rev. 95, 1268 (1954). bN. E. Booth, G. W. Hutchinson, and B. Ledley, Proc. Phys. Soc. 71 (3), 293 (1958). Table 11.5, Total Inelastic Cross Sections for Protons Incident on Various Elements at Various Energies Calculated Energy CL Tnelastic Experimental Geometrical (Mev) Material . o aection, CroSS Sectiom, Cross Section, o (mb) ’ O (mb) o, (mb) c 185 Al 440 408 769 Cu 782 7467 1136 Pb 1652 1550 204.5 900 Al 469 370 + 297 Cu 812 740 * 52 Po 1670 1660 = 50 8. P. Millburn et al., Phys. Rev. 95, 1268 (1954). Py, E. Booth et al., Proc. Phys. Soc., 70 (3), 209 (1957). 192 o Calculation of the Penetration of High-Energy Particles Through Shields As was stated in the preceding section, there is a necessity for obtaining accurate and detailed information about the penetration of high-energy particles through shields that may be used on space vehicles, In this connection, a code is being developed for calculations of the transport of high-energy particles in matter within a broad energy range. The first phase of the development of the code will cover the lower energy range,:up,to 1 Bev, and the particles considered will be protons, which are found in abundance in space, and secondary neutrons, which arise as a result of inelastic collisions of the primary protons. The energy degradation of the protons by ionization collisions will be treated by using the well-known stopping power formulas, and the nuclear inelastic events will be treated by using the data generated by the calculation mentioned in the preceding section. The transport of the neutrons born in the shield will be treated in a way similar to that for protons, except that no degradation by ionization collisions need be considered. The calculation will be set up to be as general as possible in order to obtain all the data desired for shielding design. Preliminary study indicates that, for simplicity and general usefulness, slab geometry should be the first shield geometry considered. 193 12. APPLIED SHIELDING DEVELOPMENTS Preanslysis of Pratt & Whitney Divided Shield Experiment at TSF An extensive calculational program has been undertaken in order to estimate the neutron and gamma~ray dose rates that will be fieasured in the planned divided~shield experiment for Pratt & Whitney Aircraft Company at the Tower Shielding Facility. As reported previously,l the TSF compartmentalizeq,detector tank will be utilized as a crew com- partment, and the radiation source will be the TSR-II encased in the Pratt & Whitney urahium—lithium hydride shield. The preanalysis program is an attempt to predict the dose rates by using basic principles and | the best calculational tocls presently available. The important gamma-ray sources that are being considered are the prompt-fission and fissilon-product decay gamma rays and those gamma rays which result from thermsl-neutron capture in the reactor shield assembly. Gamma rays resulting from thermal-neutron capture in air and in the crew shield, és well as those resulting from inelastic scattering of neutrons, are neglected. The neutron sources are, of course, the fast neutrons from fission. The contributors to the dose rate in the crew compartment divide generally into two parts: (1) the direct beam, and (2) radiation scattered in air. The direct-beam dose 1s calculated by using a point- to~point kernel and subsequently integrating over the source volume and source energies. The air-scattered radiation is calculated in three‘ parts. First, an energy and angular distribution for a point source equivalent to the reactor-shield assembly leakage is calculated by using a point-to-point kernel. This equivalent point source 1s then used to obtain a scattered energy and angular distribution in the proximity of the crew compartment through the use of air-scattering codes. Finally, the scattered energy and angular distribution 1is LUANP Semiann. Prog. Rep. Oct. 31, 1959," ORNL-2840, p. 139. 194 a attenuated through the crew shield by using appropriate Monte Carlo codes. The p01nt to~point kernel used for the direct-beam neutron calcula-~ tion describes the dose rate in lithlum hydride resultlng Tfrom a point 2 fission source,” and the kernel used for the air-scattering neutron calculation describes the differential number spectrum,in lithium hydride from a point fission source,.? The 1ntegration over source energies has, of course, already been performed and is 1ncluded in the kernels. The attenuation by nonhydrogenous materials is included by the use of removal Cross sections. The additional hydrogen in fhe system, that which is not in lithium hydride, is taken into account by calculating the equiya— lent path through lithium hydride and adding it to the actual lithium hydridé path and then using the sum for the argument in the kernel. This resulting kernel, of course, must be corrected for the lithium which is actualiy not present along the equivalent path. Thé-equivalent lithium. hydride path is obtained simply by the product of the actual path and the ratio of the hydrogen density to that in 1ithium hydride. | The point-to-point kernel used for the direct-beam gamma-ray calculation describes the product of the uncollided flux-and the dose rate'buildup factor® in lead, and the kernel for the air-scattering gamma-ray calculation describes the product of the uncollided flux and the differenfial energy spectrum® in 1lithium hydride plus the uncollided energy flux. The bulldup factor in lead is used for the total number of mean free paths along the line of sight, since lead is the last material and represents a considerable fraction of the total number of mean free paths. ‘ | | The energy and angular distribution of neutrons in the proximity of the crew shield is obtained by using the results of the Convair D-35 °H. Goldstein, "Some Recent Calculations on the Penetration of Fission Neutrons in LiH," NDA-42 (Aug. 7, 1957). JH. Goldstein and J. E. Wilkins, Jr., "Calculations of the Penetration of Gamma Rays,'" NDA-DC-41 or NYO-3075 (June 30, 1954). 195 air-scattering code,4 which accounts for inelastic as well as elastic scattering. The calculation for the gamma rays is effected by use of a simple model which assumes single scattering with no attenuation on either leg, except for geometric attenuation.’ The attenuation of the air-scattered radiation through the crew compartment is obtained for neutrons by using the Monte Carlo ABCD code,6 which also takes into account inelastic as well as elastic scattering. The gamma-ray calculation is performed by using the results of the Monte Carlo TRIGR-P Slab Code.’ The finite size of the crew shield is con- sidered by weighing the contributions from the side and ends of the crew shield by their fraction of the total area on the shield surface. The coding and "debugging" for the calculation is still in progress, and the results will be presented in a later report. LTSF Study of Secondary Gamma-Ray Production in U238_1iH Configurations A study of the shielding properties of a depleted uranium and ‘normal lithium hydride slab combination has been completed for the Pratt & Whitney Aircraft Company. The experiment was designed primarily for examining the secondary gemma-ray production in the shield and, in addition, for obtaining pertinent information regarding the attenuation properties of the shield as a function of thickness and position of the uranium within the shield. The uranium-LiH shield was preceded by a mockup that simulated the materials used in the design of the Pratt & Whitney 1l-c reactor so that the emergent neutron spectrum incident on “M. B. Wells, "Monte Carlo Calculations of Fast Neutron Scattering in Air," Convair-Fort Worth Report FZK-9-147, Vol..I (May 1960). °See "The Single-Scattering Approximation to the Gamma-Ray Air- Scattering Problem," Chap. 11, this report. 6"ANP Semiann. Prog. Rep. Oct. 31, 1959," ORNL-2840, p. 104. 7H. Steinberg, "Monte Carlo Calculations of Gamma Ray Penetration,” TRG-125-FR-II (Nov. 30, 1959). 196 the inner face of the shield would be representative of the spectrum expected from the reactor. The mockup consisted of dry slabs of beryllium, aluminum, boral, iron, and Inconel, Measurements were made in borated water contained in an aluminum tank following the shield. ‘ The initial measurements indicated the usefulness of boral in reducing secondary production in the uranium. Placing'l/2 in. of boral in front of 3 in. of uranium followed by 24 in. of LiH reduced the gamma- ray dose rate behind the shield by nearly 40%. Since the reduction in dose rate was greater than the 20% specified in design criteria, it was felt that it would be advantageous if a 1/2-in. boral sheet preceded all éhield configurations that contained uranium. Replacing the 1/2 in, of -boral with 1 in. of LiH reduced the gamma-ray dose rate nearly 70%. No change was observed in the secondary gamma-ray contribution when 1/2 in, of boral was added on either side of the 1l-in. LiH slab. In addition to the mockup measurements, the neutron and gamma-ray attenuation characteristics of LiH were exafiined through a series of measurements behind O-, 6-, 12-, and 24-in. thicknesses of the material, Fast-neutron and gamma-ray dose-rate measurements and thermal-neutron flux traverses were made perpendicular to the shield, as well as radially. For gamma rays the relaxation length was about 22 cm, and for fast neutrons it was about 4.8 cm. These values do not include any corrections for the aluminum containers or for geometry. | Gamma-ray measurements were also made behind a LiH-uranium shield as an aid in calculation of the secondary gamma-ray production in uranium. Data were obtained for several thicknesses of uranium, varying from 3/4 in. to 4 1/2 in., placed at various positions within the LiH. The results indicate that, after the first 6 in. of LiH, changes in the gamma-ray dose rate are very small when the uranium thickness is only 3/4 in. However, for 4 1/2 in. of uranium, nearly 12 in. of LiH is necessary before the secondary contrifiution becomes negligible. DNeutron measurements were also made behind these configurations, but the statistics on the fast-neutron measurements were poor because of the low count 197 rates. The results are, however, indicative of large changes. TFor instance, the relaxation length in borated water following the shield when no LiH preceded the uranium was different from that with LiH in front, except when the uranium was placed at the rear edge of the shield. With only 1 in. of LiH between the 3 in. of uranium and borated water, the fast-neutron dose rate increased almost a factor of 2 over the dose observed when 12 in. of LiH was present. Thermal-neutron measurements also indicated a large increase under identical conditions. In fact, the thermal-neutron flux intensity was essentially constant for a specified LiH thickness, regardless of the thickness of uranium (up to 4 1/2 in.) if the uranium was at the end of the shield. Further work is planned to investigate this effect. - In another series of measurements the uranium slabs were divided into two groups and, finally, into tfiree groups separated by various thicknesses of LiH. At no time in this series of measurements was there ever less than 12 in, of LiH following the last piece of uranium, and the total LiH thickness was always 24 in. Over-all comparison of these data with the data obtained for a configuration containing a single location of uranium is not feasible at this time. It may be stated, however, that the gamma-ray dose rate behind the split uranium shield 1s greater for some specific configurations and less for others. - Measurements were made of the fast-neutron spectrum at various places in the LiH shield with fission foils and sulphur., The shield congisted of 12 in. of LiH, 3 in. of U, and then an additional 12 in. of LiH. DPositions for measurements were after O, 1, 3, 6, and 12 in. of LiH, the last one being located between the LiH and uranium. The spectrum immediately following the reactor component mockup was obtained for comparison purposes with that predicted from calculations. Gold folls were also placed at these points to measure the thermal-neutron flux., A series of gold-foil mappings were made through the reactor mockup to aid in calculation of the neutron capture gamma rays. 198 LTSF Shielding Measurements for Aerojet Mobile. Power Reactor ML-1. o A serles of experlments was performed at the LTSF to evaluate the shield configuration pr0posed for a moblle power reactor (1dent1f1ed as the MI~1) that is being de51gned by the Aerojet-General Corporation. Thermal-neutron flux, fast-neutron dose, and gamma-ray dose measurements were made behind slab mockups of the proposed shield, and the postshutdown decay characteristics of the mockup were determined. . The configuration which represented the ML-1 reference design con- . sisted of 2 1n, of lead, 1/4 in. of aluminum, 2 in. of Hevimet, 1/4 in. of aluminum, 1/4 in. of Lucite, 1/4 in. of type 347 stainless steel, 1/8 in. of boral, 3 3/4 in. of lead, 1/4 in. of stainless steel, 5 in. of dry amwonium pentaborate (APB), 1/8 in. of stainless steel 1/8 in. of boral, 1/8 in. of aluminum, 25 in. of 7.5 wt % APB solution, and 1/8 in. of aluminum. The measurements weré made in the APB solution. The effect of the boron in the APR solution was to reduce the thermal-neutron flux UNCLASSIFIED 2-01-057-0-482 (x107) * ~ 3.4 -2 - ] \b = - 3.0 \a . ' ~N - = - z =53.0cm . . 2.6 N . ergs.g”! A . ‘\\\\. o 46 86 126 166 206 246 TIME AFTER SHUTDOWN {min) ' Fig. 12.1. Gamma-Ray Dose Rate Behind the Aerojet ML-1 Shield Mockup as a Function of Time after Power Shutdown. 199 ®, by a factor of between 10 and 100, the fast-neutron dose rate by about 10%, and the gamma-ray dose rate by about a factor of 2. A study of the optimum placement of lead within the configuration indicated that at least some of the lead must follow the stainless steel tank containing the 5 in. of APB to suppress the capture gamma-ray production in the tank walls., Since the decay characteristics of this reactor shield are of particular importance, the gamma-ray dose rate beyond the mockup (with plain water substituted for the APB solution) was studied as a function of the time after power shutdown. The results are shown in Fig. 12.1. 200 ‘\ -13. SHIELD DESIGN Optimization of a Reactor Shield The optimum design of a reactor shadow shield has been studied, as discussed previously,l by considering a uniformly thick lead shadow shield and a point scurce of monocenergetic gamma rays, the shield and source being embedded in water. The optimum shadow=-shield half-angle as a function of angle from the source-detector axis was computed on the basis of a simplified analysis of the air-scattered and first-=collision primary gamma-ray contributions. Preliminary consideration has since been given to the design of a shadow shield of optimum shape. As before, only air=scattered gamma, rays and those that had undergone a single collision in water and had scattered toward the detector were considered, and simplifying assump= tions were made to facilitate the calculations. The alr-scattered gamma- ray contribution was formulated on the basis of line~of-sight travel from the source through the shield, taking into account the differential . energy spectra of the scattered component but not the angular redirection of the emergent beam. The first-collision contribution was represented by assuming line-of=-sight attenuation and infinite-medium point~source dose buildup on the second leg of travel through the shield. It was assumed that data for a homogeneous material could be used that were based on the total number of mean free paths traversed. The simplified formulations of the dose-rate components for a monoenergetic source are given below and the geometrical relationships of the parameters are indicated in Fig. 13.1: D,g = J.S(a) 27 sing Ao e [g(a,E) + s f2(a,m) @B glo,2)] (1) 1"ANP Semiann. Prog. Rep. April 30, 1960," ORNL~2942, p. 91. 201 UNCLASSIFIED 2-01-059-578 POINT OF /fif SCATTER — . SOURCE - WATER MEDIUM Fig. 13.1. Geometry for Reactor Shield Optimization. and. ’ inoe? o’ Lt ’ Dpe = Jf 8(a?) 27 sim/ aate T 4 (a’) ar X 1 =X2 X — € B1(E//,x3) , a? where D AS = air-scattered gamma~ray dose rate, D, = first~collision gamma~ray dose rate, FC 202 (2) HiMH2 ta (@) T1,72 angle from source-detector axis to radius vector through source, angular source strength, pata(a) + (pa — pa)ta(a), (B )ty (a/,r) + [pp(B/Y) — wa(E77)] 12(a/,r), linear absorption coefficients in water and lead, respectively; these coefficients are constants of the source energy E, except where dependence on E//, the energy after scatter, is indi= cated, differential energy spectral density of the scattered com=- pgnent of gamma rays emerging from the shield, multiplied by 2 e J _TUnctibn giving air-~scattered dose rate at detector per unit source strength, » ‘macroscopic differential scattering cross section, source~detector separation distance, dose~rate buildup factor, thickness function describing the shape of the water medium (the length of the radius vector from the source to the out-~ side of the shield as a function of ), thickness of lead shield measured along radius vector, total distance in shield and distance through shadow shield, respectively, from point of scatter in direction of detector. The lead shadow shield was assumed to have axial symmetry and to be bounded by concentric spherical surfaces, the inner radius being constant and the outer radius being a step function of «. The effect of changing t,.-by a small amount, At,, over a small range of @, AX, is given by the following expressions: AD AS = =3(Q) 27 sinx Ao {e (b2 — p1) g(o,E) + +J'f(x1,E/) dB/ gla,B’) X - o {e ' [(Bafx) =1} ‘ X (Lo — p1){ Oty (3) . a,xl o 203 and ~p1y()/sine’ AD., = ~f s(a’) 21 simy/ ao’ e b (@f) x Q:I : | i _ 3 .7? X[(%)m*%&"] 3 e Bam)lx X FAY ) 3 (4) where B, is the energy buildup factor. The assumption is made in Eq. 3 that the change in the differential energy spectrum is a second-order effect relative to the change in total energy flux for small changes in shadow shield thickness. In addition, there is a corresponding rate of change in total shield weight given by M = 2m(pe — p1) [rg + ta(@)]? sinx A Oty (5) where p1,p2 = densities of water and lead, respectively, ro = inner radius of lead shadow shield. The expressions on the right~hand sides of FEgs. 3, 4, and 5 can be calculated and plots of p(ty) = AM/(ADAS +-ADFC) versus t, for various values O can be obtalned. The level lines of p(t,) then determine corresponding pairs of values (t,,0) or "optimum shape” shadow shields. It should be noted again that many simplifying assumptions have been made, notably in regard to the evaluation of buildup factors and differential energy spectra and the use of somewhat unrealistic angular distributions of -the buildup component for nonisotropic point sources in a nonuniform finite medium. In addition, the entire procedure is predicated on the condition that the computed shadow shield thicknesses are slowly varying functions of the angular variable. 204 l{i} - Before en optimization scheme such as this one can be applied, it 1s necessary toc have a means of computing the relevant dose-rate com=~ ponents for fixed-geometry shield systems. Accordingly, considerable effort during the period has been devoted to the development of codes to compute these and other dose-rate contributions (see '"Development of an IBM~704 Code for Analysis of Axially Symmetric Reactor Shields," chap. 11, this report). | 205 Reports previously ilssued 206 ORNL-528 ORNL-629 ORNL~768 ORNL-858 CRNL~919 ANP-60 ANP-65 ORNL-1154 ORNL-1170 ORNL=1227 ORNL~1294 ORNL-1375 ORNL~143%9 ORNL~-1515 ORNL-1556 ORNL-1609 ORNL-1649 CRNL-1692 ORNL-1729 CRNL~1771 ORNL~-1816 ORNL-1864 ORNL~-1896 ORNL-1947 CRNL~2012 CRNL-2061 CORNL-2106 ORNL-2157 ORNL-2221 ORNL~2274 CORNL-2340 ORNL~2387 ORNL-2440 ORNL=-2517 ORNL-2599 CRNL-2711 ORNL-2840 ORNL~-2942 Period Period Period Period Period Period Period Period Period Period Period - Period Period Period Period Period Period Period Period Period Period Period Period Period Period Period Period Period Period Period Period Period Period Period Period Period Period Period e in this series are as follows: Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending Ending November 30, 1949 February 28, 1950 Msy 31, 1950 August 31, 1950 December 10, 1950 March 10, 1951 June 10, 1951 September 10, 1951 December 10, 1951 March 10, 1952 June 10, 1952 September 10, 1952 December 10, 1952 March 10, 1953 June 10, 1953 September 10, 1953 December 10, 1953 March 10, 1954 June 10, 1954 September 10, 1954 December 10, 1954 March 10, 1955 June 10, 1955 September 10, 1955 December 10, 1955 March 10, 1956 June 10, 1956 September 10, 1956 December 31, 1956 March 31, 1957 June 30, 1957 September 30, 1957 December 31, 1957 March 31, 1958 September 30, 1958 March 31, 1959 October 31, 1959 April 30, 1960 . . HFOwe-agowUuMwWNH 29. 37, 39. 40, 41. - - . ?U.Eirdtlfj?fltfib‘?_mtdzzkflztfl_z_bdiflf—lD’QL"U*‘-‘:!Z’.’JUW.'ZUOP.':UQPtlj*tjljmc-lfl H W P . J. Keyes G R S Adamson Allen Beall Billington Blankenship Blizard Boch Boyd Briggs Callihan . Center (K-25) . Charpie . Clausing Cockreham Cottrell Culler Douglas Emlet (K-25) Fowler Fraas . Frye Gray Greenstreet - :'JU[.—'.C-I}IthL—'bdbb‘tdm@b_mbtdtdb‘*d*fimng Hikido R, Hill E. Hinkle E. Hoffman W. Hof'fman Hollaender B. Holland Inouye . Jordan . Keilholtz . Keim . Lafyatis . Laing . Llvingston . W e h'v ORNL-3029 C-84 — Reactors—Special Features of Aircraft Reactors M-3679 (24th ed:.) INTERNAL DISTRIBUTION 42. 43. 44 . 45.. 46 . 47. 48, 49. 50, 51, 52, 53, 54 55. 56. 57. 58. 59, 60. 61. 62. 63. 6. 65. 66. 67. 68. 69. 70. 71. 72. 73. - T4, 75-81. 82. 83-86. 8789 CEHGPQUUUAWHPQREORP TP NN RGNS o Wm NGO QEOQ=ZMNm . Luce Lyon MacPherson Maerker Maienschein . Manly Miller Morgan Muckenthaler Murray (Y-12) Nelson Patriarca o= RN Penny Perry Reyling savage savolainen Shipley Sisman . J. Skinner . M. Slaughter H. Snell Spiewak HrWORIRW>UO G . Sturm susano Swartout Trauger Trubey Watson - Weinberg White : Wigner (consultant) Winters ILaboratory Records Department TLaboratory Records, ORNL R. C. ORNL — Y-12 Technical Library Document Reference Section Central Research Library 207 90. 9194, 95-96, 97. 98-100. 101-102. 103, 104-105. 106~107. 108. 109111, 112, 113, 114~115. 116, 117, 118-123., 124, 125, 126. 127-128. 129, 130. 131, 132. 133. 134. 135, 136~137. 138. 139, 140. 141. 142. 143~150. 151-~153. 154, 155, 156. 157-158. 159. 160. 161, 162. 163. 164, 208 EXTERNAL DISTRIBUTION AlResearch Manufacturing Company Air Force Ballistic Missile Division AFPR, Boeing, Seattle AFPR, Douglas, Long Beach AFPR, Douglas, Santa Monica AFPR, Lockheed, Marietta AFPR, North American, Downey Alr Force Speclal Weapons Center Air Research and Development Command (RDRRA) Air Technical Intelligence Center ANP Project Office, Convair, Fort Worth Albugquerque Operations Office Argonne National Laboratory Army Ballistlc Missile Agency Army Rocket and Guided Missile Agency Assistant Secretary of the Air Force, R&D Atomic Energy Commission, Washington Atomics International Battelle Memorial Institute Brookhaven National Iaboratory Bureau of Naval Weapons ' Bureau of Naval Weapons General Representative BUWEPSREP, Aerojet-General, Azusa BUWEPSREP, Chance Vought, Dallas BUWEPSREP, Convair, San Diego BUWEPSREP, Grumman Aircraft, Bethpage BUWEPSREP, Martin, Baltimore Bureau of Yards and Docks Chicago Operations Office Chicago Patent Group Defense Atomic Support Agency, Washington Director of Naval Intelligence duPont Company, Ailken Engineer Research and Development Laboratories General Electric Company (ANPD) General Electric Company, Richland General Nuclear Engineering Corporatlon Hartford Aircraft Reactors Area Office Tdaho Test Division (LAROO) Knolls Atomic Power Laboratory Lockland Aircraft Reactors Operation Office Los Alamos Scilentific Laboratory Marquardt Aircraft Company Martin Company National Aeronautics and Space Administration, Cleveland Naetional Aeronautics and Space Administration, Washington 165. 166. 167. 168. 169, 170. 171. 172. 173. - 174, 175. 176—177. 178-181. 182. 183. 184—185. 186. 187. 188. 189. 190. 191. 192-193. 194—196. 197-208. 209-233, 234, National Bureau of Standards Naval Air Development Center Naval Alr Material Center Naval Alr Turbine Test Station Naval Research Laboratory New York Operations Office Nuclear Metals, Inc. Oak Ridge Operaticns Office Office of Naval Research _ Office of the Chief of Naval Operations Patent Branch, Washington Phillips Petroleum Company (NRTS) Pratt & Whitney Aircraft Division Public Health Service Sandia Corporation School of Aviation Medicine Sylvania-Corning Nuclear Corporation Technical Research Group USAF Headquarters USAF Project RAND U. S. Naval Postgraduate School U. 5. Naval Radiological Defense Laboratory University of California, Livermore Westinghouse Bettis Atomic Power Laboratory Wright Air Development Division . Technical Information Service Extension ) Division of Research and Development, AEC, CRO 209 L, e ? f -