ORNL-3144.txt

 

 

o

 

 

~ ORNL
MASTER COPY
ORNL-3144 %C/\ F

C-84 — Reactors—Special Feat
Aircraft Reactors

AEC RESEARCH AND DEVELOPMENT REPORT M=-3679 (24th ed.)

 

 

 

 

 

   

AIRCRAFT NUCLEAR PROPULSIONPROJECT

cay
ke

b EAT
ad

g

rn. J

SEMIANNUAL PROGRESS REPORT

AR
[_‘.

Eili 8-15-77

 

Y

FOR PERIOD ENDING APRIL 30, 1961

 

 

i w} n
6 . -
L . I ; "
. baro f&
7 ;
e \r\ \
" If;;_‘l .
b (.\f.,
$2 3 ]
ey 5
I‘E\‘: \‘I\ »-;,h
Tmmed b A =
e .- %
. et % i
L . T
¢ .."‘S‘“" \ =
e b
'AETWM’HI‘:E:S‘ N o T

 

 

oac

 

 

- B
.p : n
SR
S OAK RIDGE NATIONAL LABORATORY
= fl operated by
S UNION CARBIDE CORPORATION
o ' for the
s U.S. ATOMIC ENERGY COMMISSION
@y 3

 

sND L
3

 

 
 

 

 

 

 

"LEGAL NOTICE

This report was prepared as an account of Government sponsored work. Neither the United States,
“nor the Commission, nor any person 0ct|ng on behalf of the Commlssnon'. . . ‘
"A. Makes any warranty or representutuon, ‘expressed or |mpl|ed wnth respect to the occuracy,
. completeness, or usefulness ‘of the information contained in thls report, or ‘that the use of
any |nformat|on, upporatus method or process_dlsclosed in. th|5 report moy not mfrmge :
prwutely owned rights; ar L - . ‘ '
B. Assumes any liabilities with respect to the use of, or for domoges resulting from tl"le use of“
' any mformcltlon apparatus, method, or process dnselosed in this report. ‘
. As used in the above, ‘person actang ‘on behalf of ‘the Comm|ssuon" includes. any. employee or:
contractor of the CommISslon, or employee of such contractor, to the extent thut such employee -
or contractor of the Commlsslon, or employee of such _controcter prepares “disseminates, or
provides access to, c|ny mformutlon pursuant to l'ns employment or contract wn?h ‘the Comm|ssnon,'
or his employment wnth such contractor.

 

 

 

 
-
&
~

ORI — 3B/ S

(Report Number)

ANP CLASSIFICATION REVIEW FORM 5732;

Report Title:

 

 

Author:

 

Please indicate in square below classification determination

 

/ / Classification confirmed as

in accordance with CG-RAN-4 topic

 

 

/ / Classification downgraded to

”

in accordance with CG-RAN-4 topic

 

/ / Declassification 1s contingent on (Explain)

 

 

 

/ iZ/ Contains no Restricted Data nor other Defense Information

and may be declassified in accordance with CG-RAN-4 Guide

Topic S, 3,/ s 2—,# 6
! 6K gey”

A e £ 1972

C/ISignature Pate

 
OBNL-31H4 ERRATA

p. 10

(1) In equation for ¢7(E,2), "dr" is-replaced by dr."

p. 141

(1) In equations flm?¢n(E,2) and Sn(E;fi), "Sn(E,fiU"is replaced by "S'(E,&)."

! 3
(_/: (ZiiH -2 (£") —g-(fl——))dg ] )

IiH

(2) "éxp {

is replaced by
2 | | |
- 91
nex'p [ f(ZilH g'g;?—l - Z‘.r(g') > dfl':|.l"
S LiH

p. 147

e l n
2nDsiny’

" 1 "

(1) TIn the equation, Tl replaced by

p. 148

(1) TIn Table 12.13, "P(E,8,D)" is replaced by "P(E,6,D)."
ORNL-3144

C-84 — Reactors—Special Features
of Aircraft Reactors
M-3679 (24th ed.)

This document consists of 182 pages.
Copy (7? of 231 copies. Series A.

Contract No. W-7405-eng-26

ATRCRAFT NUCLEAR PROPULSION PROJECT
SEMIANNUAL PROGRESS REPORT Q
for Period Ending April 30, 1961

otaff \*
Oak Ridge National Laboratory “ \3?
?' T
S

Date Issued &wfigl
JUL 11 1961

i, .
T .
& e . .
-
NS e :
TR e e
Voeo T e

OAK RIDGE NATIONAL LABORATORY
Oak Ridge, Tennessee
operated by .
UNION CARBIDE CORPORATION
for the

U. S. ATOMIC ENERGY COMMISSION

 
3

 

 

 
FOREWORD

The ORNL-ANP program primarily provides research and development
support in reactor materials, shielding, and reactor engineering to
organizations engaged in the development of air-cooled and liquid-metal-
cooled reactors for aircraft propulsion. Most of the work described
here is basic to or in direct support of investigations at Pratt & Whitney
Aircraft Division, United Aireraft Corporation, and General Electric

Company, Aircraft Nuclear Propulsion Department.

i, 41

 
1

CONTENTS

QUMM AR Y ottt i it e ettt st ettt aneeneseeesotesaoeessesananansennaes xix
PART 1. MATERIALS RESEARCH AND ENGINEERING

1. MATERIALS COMPATIBILITY vttt ittt et vosnnsnenenssnnoenonnens 3
Oxidation of Columbium Alloys at Low Pressures ...........

Corrosion Studies for Determifiing the Mechanism by Which

Lithium Attacks Columbium .......c.iti it iineenienenss S
. Effect of CGrain Orientation .......eeeeeeeenneerennennn 9
Effect Oof Time ..o i ittt e ittt it te i et e nnnnnnennanas e 9
. Effect of Lithium Purity .......citiieiiiininrnnnnssns 12
Cofrosion Studies on Cb—1% Zr Alloy in Lithium ........... 13
Effect of Oxygen Additions to Cb—1% Zr Alloy on Its
Room-Temperature Tensile Properties .............00... 13
Effect of Heat Treatment of Welds .......... ... ..., 14
Dissimilar-Metal Mass-Transfer Studies in Cb—1% Zr Alloy-
NaK-Type 316 Stainless Steel Systems ........cciivnienesn 15
Effect of Carbon and Nitrogen Pickup on Tensile
P OpEr Lie S i it e it i s e e e e i5
Effect of Stress on the Rate of Carbon and Nitrogen
PiCKUD  tvi i i e e e e 18
Vapor Pressure of NaK (43.7 wt % X) in the Temperature
Range 1520 1o 1832 % vttt it it ittt teteneenneneennnnnns 20
Potassium Compatibility Studies .....cciiiiieteenenennnas 21
Refluxing Potassium Capsule Tests ........cciiiivnnn. 21
| Boiling-Potassium Loop Tests ...... ... 23
. Purification of Potassium Metal: Methods of Reducing the
OxyEen Contbent ..ttt ettt ettt tee et 24
Determination of Oxygen in Potassium .........0cciiienens 24
2. AGING STUDIES OF COLUMBIUM-BASE ALIOYS ....ieiiiivinnnen.. 28
Aging of Wrought Material ......c.ccieiriieenenennnnennnnns 28
Aging of Welded Material .......ccviuieieinniennrerarannnns 31
Internal Friction of Cb—1% Zr ALIOY i eevvrvreeraeennonann 36
3. MECHANICAL PROPERTIES INVESTIGATIONS o veveevnvernennnnn. 42

Columbium Alloy Tube-Burst TeSTS +.vvv it rrvrenreeranencens 42
vi

 

ALLOY PREPARATION v i ittt et e s eenesoenasnonsnsssassssssnsanss
Melting of High-Purity Columbium and Columbium-Zirconium
Alloys ..eeee... Gt e et et e e et es et e ettt
Addition of  Nitrogen to Columbium and Columbium-Zirconium
AlLOYS it e e it e et eres e
Electron-Beam Melting of Columbium-Hafnium Alloys .........
FABRICATION STUDIES 4 ivieunsvenesesnesnonsnescssosnanasasss
Braziné of Columbium .....ccviieenniennns Ceereesaesaas e
Fabrication of Clad Tubing .vvveernrennenss ee et e e eeenaan
BERYLLIUM OXIDE RESEARCH 4t veeviururnenineroonnreacesaanans
Preparation of Refractory Oxides from Molten Salts ........
Preparation and Calcination of Beryllium Oxalate -
Monohydrate ........ciiiiiiiiiiiiii i it i .
Purification of Beryllium Compounds by Solvent Extraction ..
Phase Relationships in BeO-Metal'Oxide Systems  ............
The BeO-Cal System ...ttt iineteteneenernneeneeannanns
The BeO0-MgO, Be0O-CeOp, BeO-Zr0O, Systems .........c.0....
ENGINEERING AND HEAT-TRANSFER STUDIES .....¢.iveierrnnennnns
Boiling-Potassium Heat-Transfer Experiment ................
Thermal Conductivity of Lithium .......ccciiiiiiienennn.
RADIATION EFFECTS v i it ie it tesenenesennsnnsascannansenens
Radiation Effects on Columbium-Zirconium Alloy ............
Radiation Effects on Stainless Steel and Inconel ..........
Beryllium Oxide Irradiation Studies ......civiiriniinnnnns '

PART 2. SHIELDING RESEARCH

DEVELOPMENT OF REACTORS FOR SHIELDING RESEARCH ..........

Tower Shielding Reactor IT (TSR=II) .+.'vvvirruvrnneenn. Ceeee

Reactor Mechanical System ...ttt it i e i

Reactor Contrqls .....................................

Flow Measurements and Fuel Plate Temperatures ..........

Nuclear Measurements ereeies i e e e e e e s e s ee e

 

 

46

46

© 48

50
51
51

54,

56
56

5'7
58
59
60
61
67
67
72
77
77
81
86

93
93
93
95
95
97

o

&
&}

R ——

DEVELOPMENT OF RADIATION DETECTION EQUIPMENT ............. . 99
Gamfia-Ray.SpectrOSCOPy ..................................... 99
The Model IV Gamma-Ray Spectrometer .........civevviuvens 99
Unscrambling of Continuous Scintillation Spectra ........ 100
NeUtron SpPECHIOSCODY v vt v vt e tneeenenenneeesenneancessennns 113
Use-of Silicon Surface-Barrier Counters in Fast-Neutron
Detection and SpectroSCOPY v it i e iintonttnnnereoenennss 113
A TFast-Pulse Integrator for Dosimetry . .......iiiiiieennnnn. 121
BASIC SHIEIDING STUDIES . eteennvuernneennoresnanoannnnonnaens 123
Energy and Angular Distributions of Neutrons Emerging from
Planar Surfaces of Diffusing Media ...........ciiiiiiaenn. 123
Experimental Verification of a Geometrical Shielding
Transformation . .....ii ittt teninernenoonesnsonsonsas 127
APPLIED SHIEIDING ittt et ioeasntntataenenensaenanasosnnnas 129
Preanalysis of Pratt & Whitney Divided-Shield Experiment
B TOF ittt ittt ittt i s e e e e 129
Radiation SOUTCEes .. .iiii i it ininiinnrnnrns che e 131
Attenuation and Transport ProCessSes .....vieervrersasannas 131
Future Calculations .......c....n ST P 139
Mathematical Description of Calculation ......coveivueenen 139
Preliminary Results of the Pratt & Whitney Divided-Shield
Experiment at TOF ...ttt ittt ittt iistnstarsarneeensnenas 148

S A vii
A e wes eleweg T e vt . TR

   

v

 

DI R

[N

©

o

)

2

et A

 

 
s)

 

ANP PROJECT SEMIANNUAL PROGRESS REPORT

SUMMARY

Part 1. Materials Research and Engineering

 

1. Materials Compatibility

 

The oxidation rates of several commercial columbium alloys and numer-
cus experimental alloys were measured in low-pressure oxygen and in low-
pressure air at temperatures up to 1200°C. In most cases, it was observed
that alloying had a detrimental effect on the oxidationlresistance, which
is the reverse of the effect that is observed when the alloys are tested
at atmospheric pressure.

Products from the corrosion of columbium by lithium were shown to
occur as a transgranular precipitate on certain crystallographic planes.
The depth to which the corrosion product is found was determined to be
a function of grain orientation. Although the depth of attack was found
to be independent of exposure time between 1 and 500 hr, oxygen was con-
tinually leached from the columbium by the lithium and the specimens were
weakened by this depletion. The addition of 2 wt % Li,0 to lithium re-
duced the depth of attack in oxygen-contaminated columbium to below that
resulting from exposure to pure lithium. This result is in line with
other evidence that lithium reacts with the oxygen contamination in the
columbium and that the driving force for lithium to produce lithium oxide
is of consequence in the corrosion process. An addition of LisN to the
lithium did not reduce the amount of corrosion.

Additions of up to 1500 ppm of oxygen were made at 1830°F to Cb—1%
Zr alloy. The oxygen contamination incrzased the room-temperature tensile
strength and decreased the ductility of the alloy. When the alloy was
exposed to lithium in this condition it was corroded, and a loss of ten-
sile strength occurred. A high-temperature heat treatment in vacuum at
2910°F, however, nullified the strengthening effect of oxygen and made

the alloy again corrosion resistant.
 

The resistance of Cb—1% Zr alloy welds to corrosion by lithium was
also shown to be strongly influenced by heat treatments. Oxygen additions
at 1830°F to the alloy welds made them very susceptible to corrosion by
lithium, but the effect of additions made at 2190°F was very small. Heat
treatment at 2370°F completely eliminated corrosion by lithium for oxygen
concentrations up to 2500 ppm in Cb—1% Zr welds. All the specimens pre-
pared by welding oxygen-contaminated Cb—1% Zr alloy and testing in the
as-welded condition demonstrated a high susceptibility to corrosion. It
was concluded that the corrosion resistance of the Cb—1% Zr alloy welds
is strongly influenced by the .mode and/or distribution of the oxygen. »

It was found that CbC and Cb,yN layers approximately 1 mil in thick-
ness formed on the surface of columbium or Cb—1% Zr alloy‘speciméns when = s
they were exposed to NaK in a type 316 stainless steel container during
isothermal tests at 1700°F for 500 hr. It has also been determined that
the carbon remains mostly in the surface layers, while the nitrogen com-
pletely penetrates the specimen. This transfer of carbon and nitrogen
results in an increase in the tensile strength and a reduction in ductility
of Cb—1% Zr. When the brittle surface layers were removed in recent ex-
reriments, these same effects on mechanical properties were still observed.
Since the transfer of carbon was limited to the surface layers, it was
concluded that nitrogen is the pfincipal element responsible for the ob-
served changes in the mechanical properties. In tests on unalloyed colum-
bium, stresses up to 70% of yield strength did not increase the rate of
pickup of either carbon or nitrogen, thus a major rate-controlling step
in the transfer process appears to be the one which supplies carbon and
nitrogen to the alloy surface.

The vapor pressure of NaK was measured in the temperature range 1520
to,l830°F. The data were compared with those obtained by calculation
using Raoult'’s law and empirical equations for vapor pressuré as a func-
tion of temperature for the pure species.

Type 310 stainless steel and Inconel were tested in refluxing po-
tassium systems at 1600°F for 1000 hr. TFabrication was completed and
the tests were begun on boiling-potassium loop systems constructed of .

Inconel and Haynes Alloy No. 25. These systems will be operated for.

 

 
i

(»

i

ulRNE.,

3000 hr at a boiler temperature of 1600°F. Two batches (10 and 90 kg)
of commerclal potassium contalning approximately 300 ppm oxygen were puri-
fied by hot-gettering and cold-trapping treatments. The oxygen content

was lowered to approximately 25 ppm by these treatments.

2. Aging Studies of Columbium-Base Alloys

 

Evidence was accumulated which indicated that the aging reactions
which occur in Cb—1% Zr alloy are due principally to the oxygén in the
alloy. It was observed that the tendency to undergo the aging reaction
increased as the annealing temperature was increased in tests of speci-
mens with the same oxygen content and that only alloys of intermediate-
oxygen-content aged when annealed at temperatures of 1600 and 1800°C.
Increasing the carbon and nitrogen impurities in the alloy did not cause
the alloy to respond to aging under the conditions studied.

The use of a Rockwell-B hardness test to evaluate the aging be-
havior of Cb—1% Zr welds was investigated. It was found that the nu-
merical values of hardness are of dubious value as absolute indicators
of aging behavior, since some points associated with brittle and border-
line behavior have lower hardness values than points associated with
ductile behavior. The hardness test, however, has value as an indicator
of trends and provides useful information to assist in the analysis of
the aging behavior.

Various postweld annealing treatments were studied as possible meth-.
ods of circumventing the aging reaction. Annealing from 1 to 3 hr in
the temperature range of 1900 to 2200°F is known to be effective in pre-
venting aging. This annealing treatment caused a heavy precipitate to
appear as a network throughout the matrix, the distribution of which pre-
vents it from acting as a strengthening agent. The results indicate that
even with high oxygen contents in the welds, appropriate postweld anneal-
ing can prevent aging.

An apparatus to measure internal friction has been built and placed
in operation. This eguipment will.be used to study the role of oxygen
in aging of columbium-base alloys. Preliminary measurements were obtained

with columbium and the Cb—1% Zr alloy specimens.

—— 3
 

3. Mechanical Properties Investigations

 

Tube-burst data on Cb—1% Zr alloy specimens were obtained at 1800°F
to provide control data for evaluation of results from in-pile experi-
ments. The data are summarized and compared with those of other investi-

gators.

4. Alloy Preparation

 

Methods were studied for the preparation of columbium and columbium
alloys with closely controlled compoéitions of both the alloying addi-
tions and impurity elements. In this study the eleétron beam, inert-gas-
shielded tungsten arc (noficonsumable-electrode arc), and consumable-
electrode vacuum arc melting processes have been investigated. Emphasis
has been on the first two of these melting processes.

Electron-beam melting was demonstrated to be very effective for pre-
paring columbium of high purity with respect to interstitial impurities,
oxygen, nitrogen, and carbon. However, the addition of alloying elementé
more volatile than columbium, such as zirconium, was difficult to control.

Studies were made to determine the capabilities of the arc-melting
furnaces with respect to sizes of heats of columbium that can be melted
and to the purity which can be maintained in the metal. Preliminary cali-
bration curves were established for the controlled addition of nitrogen
to columbium by arc melting under various partial pressures of nitrogen

and argon.

5. "'TFabrication Studies

 

Experiments showed that 12 refractory-metal-base brazing alloys
which have been developed readily wet and flow on colgmbium;to-columbium
T-Jjoints. No visual evidence of fillet cracking was seen on these joints.
Five of these alloys also appeared promising for joining columbium to
stainless steel, but Jjoint designs should be utilized which minimize the
effects of thermal expansion coefficient differences.

Columbium tubing was clad with type 304 stainless steel or type 446
stainless steel by co-drawing tubes at foom temperafure and annmealing at

about 1000°C for 2 hr to obtain a bond. Copper was used at the interface

xii

 

 
 

of the composite to serve as a bonding metal as well as a diffusion bar-
rier to inhibit the formation of a brittle reaction layer. Preliminary
evaluation of these composites indicated that producing metallurgical

bonding of the cladding and the tube by cold forming and subsequent heat

treatment is not promising.

6. Beryllium Oxide Research

 

The steam precipitation of beryllium oxide from melts containing
lithium and beryllium fluorides is being developed as a direct route to
the formation of crystalline material without the necessity of a cal-
cination step. Tests at 800°C made with helium as the carrier gas for
the steam gave products which were well crystallized but were contami-
nated with corrosion products from the metal container. Additional tests
were in progress in which hydrogen is being used as the carrier gas to
suppress the corrosion of the container.

Two approaches were studied in efforts to obtain crystals of U0,
coated uniformly with BeO: 1in one, preformed crystals of UO, of less
than 20 p in diameter were suspended in a fluoride melt during the stéam
precipitation of BeO; in the other, consecutive precipitations of U0,
and BeO from the same melt were attempted, with the initial precipita-.
tion being very rapid to obtain small particles of UO, and the subse-
quent precipitation being very slow to obtain uniform coating with BeO.

Hot filtration of an aqueous slurry of beryllium oxide monohydrate,
at a temperature above that at which the trihydrate can form, gave a
product almost completely free from traces of the trihydrate. Calcina-
tion of the product gave a BeO product of higher purity than that usually
obtained from the trihydrate.

Exceptionally pure beryllium hydroxide was produced, first, in a
l-g quantity and, later, in a 30-g batch, by the solvent extraction process
previously under development. The material was considerably better than
that available for use as analytical standards, and additional quantities
are being prepared for use as improved standards. The only reportable
impurity in the 30-g batch, as determined by spectrographic techniques,

was 5 ppm maghesium.

xiii

 
&

Investigations of phase relationships in BeO-metal oxide systems
were continued. Data were obtained for the binary systems Be0O-CaO, BeO-
MgO, BeO-Ce0,, and BeO-Zr0O,. The effects of cooling rates on the forma-
tion of the intermediate compound Caj;Be30s; in the Be0-Ca0 system were in-
vestigated, and the information developed was used as a guide for the
successful growth of single crystals of the phase. The eutectic tempera-
ture and composition were determined to be 1860 * 10°C and 69 * 2 mole %
BeO in the BeO-Mg0 system; 1890 + 20°C and 63 * 3 mole % BeO in the BeO-
Ce0, system; and 2045 # 10°C and 58.7 * 2 mole % BeO in the Be0-Zr0,

system.

7. Engineering and Heat-Transfer Studies

 

Construction of an apparatus for studying forced-flow boiling with
potassium has been completed. A temporary boiler section was installed
so that cleanup and the determination of the operational characteristics
could be accomplished while awaiting completion of the final boiler as-
sembly. Preliminary heat-transfer data were obtained for potassium in
the laminar-flow regime; the results are in reasonable agreement with
data on mercury and lead-bismuth eutectic in the same flow range.

Measurements of the thermal conductivity of molten lithium were con-
tinued using a modified version of the axial heat flow apparatus. The
apparatus changes allowed better definition of the heat flow and tempera-
ture distributions and resulted in some improvement in experimental pre-
cision. The most recent set of measurements show less variation with
temperature than did the earlier results; thus, the conductivity varied
from about 27 Btu/hr-ft-°F at 750°F to 29 Btu/hr-ft-°F at 1500°F in con-
trast with a variation from 20 to 40 Btu/hr.ft.°F for the previous meas-
urements over the same temperature range. The results are in good agree-

ment with published results.

8. Radiation Effects

 

In-pile tube-burst tests at 1800°F were conducted on specimens of
Cb—1% Zr alloy in the poolside facility of the ORR. The rather high

stresses used in the experiment caused all specimens to rupture within

xiv s‘:“"“uflllllllllll

 
 

60 hr., The results indicated that the in-pile rupture strength was about
10% less than the out-of-pile strength. Longer times to rupture and higher
radiation doses will be obtained with the use of lower stresses in future
experiments. Additional in-pile tube-burst experiments were completed on
Inconel and stainless steel.

Serious physical damage to BeO which had received fast-neutron dos-
ages of the order of 1021 neutrons/cmz'in the ETR was revealed during
postirradiation examination of test specimens. The damaged specimens con-
"stituted a portion of the 57 BeO pellets 1 in. in length and 0.4 to 0.8
in., in diameter that were contained in 16 capsules distributed among five
separate irradiation assemblies. A comprehensive range of exposures was
achieved at a significant variety of temperatures. Damage observed in
the BeO that received the higher dosages varied from minute cracks to
gross fracture and disintegration into powder.

The BeO has been recovered from the capsules of all five assemblies.
The disassembly and inspection of the last three assemblies was carried
out in the Battelle Memorial Institute hot cell facility. Visual in-
spection, macrophotography, and'physical dimension measurements of the
specimens have been completed. Metallographic examination was completed
for the specimens used in the early experiments, and exploratory gas
analyses and x-ray diffraction studies have been performed. A major por-
tion of the thermal flux dosimetry analysis was completed, and corre-
sponding fast flux determinations are under way. A survey of the tem-
perature data and composition changes of the thermal-barrier gas is being
made to evaluate thermal conductivity changes observed during the progress

of the irradiations.

Part 2. Shielding Research

 

9. Development of Reactors for Shielding Research

 

Upon completion of the critical experiments with the TSR-II, the
reactor was disassembled, all the temporary eguipment which had been in-

stalled for the critical experiments was removed, and the reactor was

S xv
reassembled for operation. As a result of the shakedown runs of the com-
plete system which continued through December 1960, some components had

to be modified before the reactor was placed in operation in January and
subsequently operated at a 100-kw power level. Minor design changes are

contemplated to provide more reliable operation.

10. Development of Reactors for Shielding Research

 

Gamma-Ray Spectroscopy. During shakedown tests with the BSF model

 

IV gamma-ray spectrometer it became apparent that the spectrometer must
be positioned at least 5 ft from the reactor surface to reduce the gamma-
ray background to an acceptable level. As a result, measurements of the
radiations leéving the surface of the reactor will have to be made by
attaching a 5-ft-long air-filled cone to the front of the spectrometer
shield. The large composite NaI(Tl) crystal obtained for use with the
spectrometer has been tested further and appears to be satisfactory.

The finite limit imposed upon the output of a gamma-ray scintilla-
tion spectrometer by the number of channels available in a multichannel
analyzer prevents exact determination of a continuous gamma-ray spectrum.
This difficulty is equal in importance to the better-known problems of-
fered by statistical distributions of ‘observed counts and the lack of
exéct knowledge of spectrometer responses. The problem is currently
being studied by arbitrarily assuming an exact knowledge of the number
of cdunts produced at given energies and an exact knowledge of a spec-
trometer response in order to determine how closely a contimious spec-
trum can be reproduced. Single parameters that are dependent upon the
spectrum, such as gamma-ray dose, can be written as linear combinations
of spectrometer response functions if the cross sections governing these
parameters are indeed exact combinations of a finite number of response
functions. This is not usually true, but the assumption of "nonnegativity,"
that is, that the particle spectrum is everywhere equal to or greater than
zero and that the energy range lies between zero and infinity, permits
reasonable estimates to be made. Upper and lower bounds can be determined
in this fashion for wvarious parameters. A method borrowed from the mathe-

matics of economics and scheduling, called "linear programing', for which

xvi

 

 
 

computer codes already exist, can be used to attack the problem of setting
the narrowest possible upper and lower bounds for a given parameter. For
multiple-parameter problems, Jjoint bounds may be set if the correlations
are known. Approximate methods for determining the "smallest allowed re-
gion" have been found, based on considération of the allowed regions as
rectangles of given length, width, and center coordinates, or an irregular
bodies with given centroids and moments of inertia.

Neutron Spectroscopy. The possible use of a neutron-sensitive semi-

 

conductor detector for neutron spectroscopy has been investigated further
with some resulting modifications in the design. From the tests conducted
thus far it is concluded that the full width at half maximum is relatively
constant, of the order of 300 kev, regardless of the incident neutron
energy. The presence of gamma rays will produce a broadening of the neu-
tron peak, but measurements have been made in gamma-ray fields up to

~200 r/hr without appreciably altering the response. Further, the device
produces a pulse height that is a linear function of incident neutron
energy. The results of calculations indicate that the foreground-to-
background ratio will be about 5 to 1 when the counters are used to meas-}
ure a fission spectrum, and a preliminary experiment with the TSR-II has
substantiated this prediction. The effects of further variations in the
design are being studied; specifically, the effects of increasing the

area of the counters and substituting Li® metal for Li®F as the neutron-
sensitive material.

A Fast-Pulse Integrator for Dosimetry. When the A-1A amplifier used

 

during dosimetry measurements at the Lid Tank Facility was replaced with
a DD-2 amplifier which had a known dead time of 2.6 usec, it became de-
sirable to develop an integrator which had a resolving time within the
same interval so that counting loss corrections could be calculated. As
a result, a new integrator with a total resolution of 2.4 psec was placed
in operation early this year. It consists of two sections, each having
four discriminator stages coupled together as a true binary scale of 16.

The output pulses from both sections are fed to a single decade scaler.

xvii

 
11. . Basic Shielding Studies

 

Bnergy and Angular Distributions of Neutrons Emerging from Planar
Surfaces of Diffusing Media. The investigation of the energy and angular

 

distributions of neutrons emerging from planar surfacesgs of diffusion
media has now included NDA moments method calculations of the energy spec-
tra of neutrons from a point source in an infinite medium of LiH for dis-
tances from the source of 11.5, 23, and 34 g/cm2 and for the four source
~energies 16.3, 1.48, 0.12, and 0.01 Mev. The results indicate that low-
energy eguilibrium is reached at all thicknesses for all input energies
employed, which 1s in agreement with earlier Monte Carlo calculations.
In addition, measurements have been made at the BSF of the angular dis-
tributions of 4.95-ev, l.44-ev, and subcadmium neutrons leaking from a
4=in., -thick LiH slab, the results of which appear to agree with calcula-
tions performed with the NDA NIOBE Code. It was originally planned that
energy spectral measurements would be made with a neutron chopper facility
to be built at the BSF, but the withdrawal of project support has cancelled
the construction of the facility. Some experimental data which can be
compared with computed energy spectra for LiH will be obtained, however,
from a series of measurements at the Linear Accelerator Facility of General
Atomic, San Diego, in early May 1961.
Experimental Verification of a Geometrical Shielding Transformation.
The Lid Tank Shielding Facility is currently involved in a program
to verify a geometrical transformation concept that the axial dose rate
from a large source plate in a homogeneous medium can be inferred from
the dose rates from a small disk source. During the program the source
plate size is being varied by inserting cadmium irises of several different
diameters between the source plate and the incident beam of neutrons from
the ORNL Graphite Reactor. Thus far, only the measurements with a 7-in.-

diam iris have been completed.

12. Applied Shielding

 

The preanalysis for the Pratt & Whitney divided-shield experiment
at the TSF has been completed. The shield design consists of a highly

asymmetric reactor shield surrounding the TSR-II and separated from the

xviii #: sgf

.

 
I

TSF compartmentalized cylindrical crew shield by approximately 64 ft.
The reactor shield incorporates two advanced shielding materials; that
is, lithium hydride as the neutron shield and depleted uranium as the
gamma-ray shadow shield. The radiation sources which were considered
important and calculable included fission neutrons from the core, prompt-
fission and fission-product decay gamma rays from the core and shadow
shield, and capture gamma rays from the reactor and reactor shield. The
attenuation and transport_processes were divided into three categories:
determination of the spectra of radiation leaking from the reactor shield;
the subsequent scattering of this radiation in air; and the attenuation
by the crew compartment of radiation both unscattered and scattered in
air. '

Experimentation with the Pratt & Whitney divided shield mockup at
the TSF is under way, and some preliminary results are available. Dur-
ing the course of the experiments the optimization of the reactor shield
is being checked by use of a water-filled "patch" tank constructed so
that it can be moved around the shield in the horizontal midplane. Also,
the relative effectiveness of water, transformer oil, and a mixture of
transformer o0il and an organic boron compound as the crew compartment
shield material is being investigated. Most of the measurements com-
pleted to date consist in measurements of the dose rates and fluxes in
air around the reactor shield, although some measurements have also been
made inside the crew shield. Additional in-air and crew compartment
measurements are still to be made, as well as a number of spectral meas-

urements.

S xix
A s
- mr TR R TR R Sy
ST S30E ;&‘,:?fii’ -

iR

.

*
eI R

-

i

 
PART 1. MATERTALS RESEARCH AND ENGINEERING
1. MATERTALS COMPATTIBILITY

Oxidation of Columbium Alloys at Low Pressures

H. Inouye

The reactivity of columbium with oxygen places severe restrictions
on the use of columbium at high temperatures because of scaling and em-
brittlement by internal oxidation. Through alloying, it has been possible
to reduce the reaction rates as much as a factor of 100 at atmospheric
pressure. Furthermore, the reaction rates of unalloyed columbium can be
reduced by a factor of 400 by reducing the oxygen pressure to 1 X 10~% mm
Hg in the temperature range 850 to 1200°C.1*2 Studies are being made to
determine whether alloying will have the same effect at low oxygen pres-
sures as at atmospheric pressure.

In the previous progress report,3 the general effect of alloying
additions on the reaction rates was reported for exposure ét 1000°C to
oxygen at a pressure of 5 X 104 mm Hg and at 1200°C to air at a pressure
of 5 x 10™% mm Hg. This study has been extended to include rate studies
at 850°C, as well as screening tests on additional compositions at 1000
and 1200°C.

In general, it has been found that when the alloying element forms
an oxide more stable than the oxides of columbium, low concentrations of
the element reduce the reaction rate in low-pressure oxygen below that
for unalloyed columbium. However, as the concentration of the alloying
element is increased the reaction rate also increases. This behavior is
shown in Fig. 1.1 for zirconium additions and in Fig. 1.2 for aluminum
additions. ©Similar results have also been observed for additions of
cerium, beryllium, hafnium, and titanium. The difference between the
oxidation rate of high-purity columbium and that of commercial-purity
columbium is about that observed for low levels of the alloying elements
(Fig. 1.3). |

 

1ANP Semiann. Prog. Rep. April 30, 1960, ORNL-2942, pp. 3—7.
2H. Inouye, The Scaling of Columbium in Air, ORNL-1565, Aug. 29,. 1956,
JANP Semiann. Prog. Rep. Oct. 31, 1960, ORNL-3029, pp. 36.
UNCL ASSIFIED

ORNL~-LR-DWG 57251R
5.6

1000°C

4.8 4

0, PRESSURE: 5x10” *mm Hg

?2'40 5% Zr
>
TC 1400 min

X
=
< HIGH- PURITY Cb
"f 2.4
T
% 1070 Zr
2 16

0.8

 

0
o 100 200 300 400

TIME (min)

Fig. 1.1. Oxidation Rates of Columbium-Zirconium Alloys at 1000°C
in Low-Pressure Oxygen.

UNCLASSIFIED
ORNL-LR-DWG 57252R

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3.2 !
1000°C ‘
58 0, PRESSURE: 5x10~% mm Hg
_6_,£LCb

2.4 o

+ /
- L~ 2.5% Al
o ek
g //"‘——
~. 20 L Y L
o
E /fi/d ]
z 16 P '
(] 1.5 o]o Al

£\t
o 1.2 _"fl___fl_,.—— 0.5% Al
L
flé.——-—f"’" *
0.8 - ®
0-4 {
0
0 100 200 300 400

TIME {min}

Fig. 1.2. Oxidation Rates of Columbium-Aluminum Alloys at 1000°C
in Low-Pressure Oxygen.

 
UNCLASSIFIED
ORNL-LR-DWG 57250R

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3.0 ’
t000°C
-4
0, PRESSURE:5x10 mmHg
—".-—.
o—*
/
o
£ 2.0 -
-a_ // ELECTRON-BEAM-MELTED Cb
E ®
=z o/ .
S 0'/
LO—"
/
-
°
. { }dww
w /. \
2 10 . :
. 79 r
o

¢ 7—45”0”64i\\‘ARC—CAST Cb

/ (COMMERCIAL)

®

2 |
o

 

 

 

 

 

O 100 200 300 400
TIME {min)

Fig. 1.3. Oxidation Rates of Unalloyed Columbium at 1000°C in Low-
Pressure Oxygen.

Inasmuch as the oxidation characteristics of an alloy are dependent
to a large extent on the type of oxide formed on the metal, it was ex-

pected that a very high concentration of alloying material might prove
beneficial, since the probability of forming a different oxide would be
greatér. However, as shown in Figs. 1.4 and 1.5, the higher levels of
elemental additions proved to be detrimental to the oxidation resistance
for the specific test conditions used.

Tests of another class of alloying elements in columbium showed
little, if any, change in the oxidation rate as the amount of alloying
element was increased. The elements added were tin, molybdenum, tungsten,
and palladium., In these tests the added element formed oxides less stable
than the oxides of columbium. The rate curves for tin additions are shown
in Fig. 1.6.

The oxidation rates of columbium in low-pressure air at 1200°C were
greatly different from the oxidation rates in okxygen, as shown in Fig. 1.7.
In oxygen, the reaction rates fiere linear from the beginning of the test;

however, in air, an incubation period of about 6 hr was required before a
UNCLASSIFIED
ORNL-LR-DWG 57249R

 

 

850°C
0, PRESSURE: 1x10™% mm Hg

3.0 —
2.5 ‘0'4’31:10% Mo-10% Ti- BAL.Cb —
c/

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a
E /
[
~
£ 2.0 F-48:15% W-5% Mo- 0.7 % Zr - BAL. Cb—
3 A-—fiL-_—--AflA--‘}'—'A"
- e AT B =
[ %—
e ot
I 1.5 o
o
o
=
1.0 / .
/ “/ F-82:33% Ta-0.7% Zr—BAL.Cb
/ <
/ —
0.5 "
4 “—i‘_—..
/ ._..--'
! et UNALLOYED Cb
—.
.-l—
0 Se—oT® t I
0 100 200 300 400

TIME (min)}

Fig. 1l.4. Oxidation Rates of Commercial Columbium Alloys at 850°C
in Low-Pressure Oxygen.

measurable amount of oxidation occurred. Furthermore, the oxidation
rates for equivalent oxygen pressures were significantly lower when the
oxygen contained nitrogen (from air). Numerous investigations have shown
that the nitrogen contamination of columbium oxidized in air is slight,
and it appears that the observed difference might be due to a "poisoning"
of the columbium surface or the formation of a nitride layer. _

Since nitrogen has an important effect on the oxidation rate in low-
pressure air, it would be expected that elements which form nitrides in a
nitrogen-containing oxygen atmosphere would show different behavior in a
nitrogen-free oxygen atmosphere. In tests of this hypothesis, it was
found, as shown in Fig. 1.8, that increasing the zirconium content reduced
the oxidation rate in a nitrogen-containing atmosphere; whereas, in a
nitrogen~free atmosphere the reverse effect was observed.

This study has shown that alloying additions do not reduce the oxida-

tion rates of columbium at low pressures in the same proportion as at

 
UNCLASSIFIED

ORNL-LR-DWG 57253R

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.
1000°C
) —~4
i 0, PRESSURE: 5x10° " mm Hg /’U/
5 /
— /
£ 70% Cb-25%Ti—5% Al ALLOY
{
E 4 //
=
= o
<L
. © v
§ / UNALLOYED Cb
2 1
//
0
0 100 200 300 400
N TIME (min)
Fig. 1.5. Oxidation Rates of Cb—25% Ti—5% Al Alloy at 1000°C in Low-

Pressure Oxygen.

2.5
2.0
N
£
o
~
o
E .5
=
<
o
£ 40
o
o
z
0.5
0
Fig. l1l.6.

Pressure Oxygen.

UNCLASSIFIED

ORNL-LR-DWG 58614

 

e

®

 

 

 

 

 

0, PRESSURE: 5 x 10" mmHg

 

 

 

1000°C

 

 

 

 

 

 

 

 

 

 

 

100

200

TIME (min)

300

400

Oxidation Rates of Columbium-Tin Alloys at 1000°C in Low-
UNCLASSIFIED
ORNL-LR-DWG 57247R

 

 

 

 

 

 

 

 

 

 

 

 

 

4q
. l i i o
1200°C, ELECTRON - BEAM— MELTED Cb
m 0, PRESSURE: 5x10™% mmHg
/ A O, PRESSURE: 1% 1079 mmHg
O /’ ® AIR PRESSURE: 5x10~%mm Hg
S ./ A (R, 1107 mmHg)
g / A -
ot /] ®-
= A ./
= 2= # —
<I A ¢
o I /
— A ®
T / ./
D A Ve
W /
s q|n_A o’
I
‘ A /

/ .

‘ -

! ®

A

0 o—0=
0 20 40 60
TIME (hr)

Fig. 1.7. Oxidation Rates of Unalloyed Electron-Beam-Melted Columbium
at 1200°C in Low-Pressure Oxygen and Air.

UNCLASSIFIED
ORNL-LR-DWG 58615

 

 

 

1.2 /‘g
1200°C

AIR PRESSURE: 5x10°% mm Hg /
1.0 /

0.8 3% 2r /

o b4
0.6

N

 

 

 

 

WEIGHT GAIN (mg/cm?)

 

o

5
N

N\

 

 

 

 

 

 

 

 

 

 

 

5‘702!’
,/’/ UNALLQOYED Cb
0.2 //
o / .
0 10 20 30
. TIME (hr)

Fig. 1.8. Oxidation Rates of Columbium-Zirconium Alloys at 1200°C
in Low-Pressure Air,

 

n
atmospheric pressures. In fact, alloying was found to be detrimental to
the oxidation resistance in most cases. Thus, for highly alloyed columbium
containing elements which form more stable oxides than the columbium
oxides, a greater degree of atmospheric purity appears to be necessary to

prevent contamination.

Corrosion Studies For Determining the Mechanism by
Which Lithium Attacks Columbium -

 

 

J. R. DiStefano

Effect of Grain Orientation’

 

It has been shown that corrosion of columbium by lithium is related
to the oxygen concentration of the columbium.? In order to study this
phenomenon, oxygen is added to columbium at 1830°F at OXygen pressures on
the order of 1 X 10™* mm Hg or lower. This is followed by homogenization
(heating for 2 hr at 2370 or 2910°F under vacuum) to obtain a more uniform
distribution of the oxygen. Polycrystalline specimens prepared in this
manner are subsequently exposed to lithium in static test systems at
1500°F. The attack observed on specimens containing 500 to 1500 ppm oxy-
gen is in the form of a transgranular precipitate. The attack varies
from grain to grain and is crystallographic in nature, as‘shown in Fig.
1.9. An effort has been made to determine whether an oxide phase is
present preferentially along these planes prior to exposure to lithium by
examining pretest specimens with the electron microééope; however, thus far
no correlation has been obtained. Studies with columbium single crystals
of known orientation are presently being conducted in order to determine

the habit plane of the corrosion product.

Effect of Time

It has been determined metallographically that the depth of attack

of columbium by lithium is not a function of time during exposure periods

 

4E. E. Hoffman, The Effects of Oxygen and Nitrogen on the Corrosion
Resistance of Columbium to Lithium at Elevated Temperatures, ORNL-2675,
Jan. 16, 1959,
 

 

e

UNCLASSIFIED
Y-33449

-
P Ry

 

\

Fig. 1.9. Effect of Grain Orientation on Depth of Attack of
Columbium by Lithium. Etchant: HF-H>804-HNO3-H20 (11-11-28-50 vol %).
500X (Confidential with caption)

of 1 to 500 hr.’ The specimens continue to lose weight, however, as a
function of time. This can be related to the oxygen being removed from
the specimen. The weight-loss data and the oxygen-loss data are compared
in Fig. 1.10. Hardness measurements were made from one surface to the
other along the cross section of each specimen, and these data are plotted
in Fig. 1.11. These results indicate that although intergranular attack
occurs very rapidly and does not increase with time, an oxygen gradient
is established across the specimen, and leaching of oxXygen by the lithium
continues until a minimum level is reached. This level was approximately
250 ppm for the conditions of this test. The columbium is considerably
softened and therefore weakened by this depletion in oxygen, but there is
no change in the metallographic appearance of the specimen.

The 250- and 500-hr test specimens showed slightly less attack than

the specimens tested shorter times. ©Since the depth of attack in

 

5ANP Semiasnn. Prog. Rep. Oct. 31, 1960, ORNL-3029, p. 9.

10

 
UNCLASSIFIED
ORNL-LR-DWG 57001R
—-1000

WEIGHT LOSS

—-800

OXYGEN LCSS

-600

OXYGEN LEVELS OF Cb SPECIMENS
—-400 PRICR TO EXPOSURE = {000 -1500 ppm
SPECIMEN THICKNESS = 0.043in.

LOSS (ppm)

-200

 

1 2 5 10 20 50 100 200 500
TIME OF EXPOSURE (hr)

Fig. 1,10. Weight Loss and Oxygen Loss of Columbium Specimens as a
Function of Time of Exposure to Lithium at 1500°F.

UNCLASSIFIED
ORNL-LR-DWG 57002

 

170

1hr

160 /f\

150
. ® 2hr \
, At
o
130
o}
N
3hr N
120 T‘\
0 !
/e
1o

 

 

 

 

 

 

 

'HARDNESS,DPH (200g LOAD)

 

 

 

  

 

 

 

 

 

 

 

 

 

100 \
/ 16 hr \
90 / ; \
/ 20 \ \
5 L] g
T
0.040in. -
70 ‘ :
EDGE OF CENTER EDGE OF
SPECIMEN SPECIMEN

Fig. 1.11. BHardness Profiles Across Columbium Specimens Exposed to
Lithium as a Function of the Time of Exposure.

11
polycrystalline specimens varies from grain to grain, the determination of
the effect of time as a test variable could best be evaluated in specimens
cut from a single crystal. ©Opecimens of this type will be exposed to
lithium for 1, 100, and 1000 hr.

Effect of Lithium Purity

 

From previous work® it appears that oxygen or nitrogen in lithium
does not tend to promote corrosion of columbium if the oxygen concentra-
tion of the columbium is low. In the earlier experiments there was no
indiéation as to what the effect might be i1f the oxygen concentration of
the columbium were high. Columbium specimens to which approximately
1000 ppm O, had been added were therefore exposed 100 hr at 1500°F to Li,
Ii + 2 wt % LisN, and Ii + 2 wt % Li,O.

It was found that the columbium specimens exposed to lithium and the
Ii + 2 wt % LisN were completely attacked, but that the columbium specimen
exposed to Li + 2 wt % Li,O was attacked to a depth of only 1 mil. If
the driving force for corrosion is the tendency for lithium to react with

the oxygen in the columbium to form lithium oxide, then this driving force

would be reduced if the oxygen content of the lithium were high. The re-
sults of this test tend to support this supposition. Further tests will
be conducted in an attempt to substantiate this conclusion.

Several possible mechanisms could explain the results of the columbium-
lithium corrosion studies which have been conducted thus far. One mecha-
nism would require lithium diffusion into the columbium and the precipita-
tion of lithium oxide or a complex lithium-columbium-oxygen compound as a
corrosion product. The fact that corrosion occurs so rapidly and does
not progress with time {(as determined by metallographic studies) does not
favor a diffusion mechanism. An alternate mechanism assumes that the
lithium reacts chemically with columbium oxide at the surface of the
specimen which might be present either before exposure to lithium or might

precipitate during the early stages of the test. As was mentioned earlier,

 

6ANP Semiann. Prog. Rep. March 31, 1959, ORNL-2711, p. 11.

12

 
no evidence has been found metallographically to indicate the presence of

columbium oxide before testing.
Corrosion Studies on Cb—1% Zr Alloy in Lithium
J. R. DiStefano

Effect of Oxygen Additions to Cb—l% Zr Alloy on Its Room-Temperature
Tensile Properties

 

A series of Cb—1% Zr alloy sheet tensile specimens were oxidized at
1830°F at an oxygen pressure of 5 X 1072 mm Hg. Following oxidation, one
group of specimens was exposed to lithium for 100 hr at 1500°F Withofit
further heat treatment. A second group was homogenized at 2910°F, and a
third group was homogenized at 2910°F and then exposed to lithium at 1500°F
for 100 hr. All these specimens and a group which received no further
treatment after oxidation were tensile tested at room temperature. The
results of>these tests are plot%ed in Fig. 1.12. The tensile strengths
of the specimens tested in the "as-added" condition increased with in-
creasing oxygen concentration and the ductility decreased (0% elongation

for those specimens containing 1000 ppm 0, or greater). When oxidation

 

3 SRR
(x11500) ORNL-LR~DWG S5TOOO0R
| ] |
OXYGEN ADDED AT 1830°:‘>_’__ ——
‘//

 

L=

LT
100 /{///

 

 

 

OXYGEN ADDED AT 1830°F AND

 

 

 

 

 

 

 

TENSILE STRENGTH (psi)

 

 

 

_J/’ /§XPOSEDT0 Li AT 1500°F FOR 100 hr
./ | _
50 ~au OXYGEN ADDED AT 1830°F
//HEAT TREATED AT 2910°F
,E; L _ o W T !
\ L il |

 

“~OXYGEN ADDED AT {830°F
I[HEAT TREATED AT 2910 °F AND
EXPOSED O Li AT 1500°F FOR 100 fr

0 500 1000 1500 2000 2500
0, CONCENTRATION (ppm)

 

 

 

 

 

 

 

Fig. 1.12. The Effect of Oxygen Additions, Heat Treatment, and Ex-
posure to Lithium on the Room-Temperature Tensile Strength of Cv—1% Zr
AMloy.

13
was followed by a 2910°F heat treatment, the tensile strength remained

constant with increasing oxygen concentration, and there was no loss in

ductility. Exposure to lithium following the oxidation treatment resulted
in attack of the Cb—1% Zr alloy. The depth of attack increased as a func-
tion of increasing oxygen concentration. The tensile strengths of these
specimens decreased as a result of corrosion testing but femained above
those of the heat-treated specimens. The corroded specimens were also
more brittle. If the specimens with oxygen additions were heat treated

at 2910°F before exposure to lithium, the tensile properties were found

to be essentially the same as those prior to exposure, and no attack was
observed.

It is apparent from these results that the form and/or distribution of
oxygen in the Cb—1% Zr alloy has a marked effect on both the strength and
the corrosion resistance of this materisl. In order to predict the cor-
rosion resistance of the alloy, it is necessary that both the oxygen con-

tent and the thermal history of the material be known.

Effect of Heat Treatment of Welds

 

In order to establish the conditions under which corrosion of high-
oxygen-content Cb—1% Zr alloy welds by lithium will occur, eight specimens
which had reeeivedsv®rious pretest treatments were tested in lithium at
1500°F for 100 hr. The results of this test are summarized in Table 1.1.

It may be seen from the table that the pretest treatment determined

whether corrosion did or did not occur. When welding was the final treat-
ment before exposure to lithium (weld Nos. 3 and 7), almost complete attack
of the Cb—l% Zr alloy occurred. Also, as previously o'bserved,7 when oxygen
was added to the welded alloy at 1830°F and no additional heat treatment
was given, the Cb—1% Zr alloy was susceptible to corrosion by lithium.
When the oxygen addition temperature was raised to 2190°F, the attack fol-
lowing exposure to lithium was 2 mils or less and was found in only a few
areas. Heat treating at 2370°F as a final treatment completely eliminated
the attack. It appears that heat treating Cb—1% Zr alloy in the range

2190 to 2370°F stabilizes the oxygen and no corrosion occurs.

 

7ANP Semiann. Prog. Rep. Oct. 31, 1959, ORNL-2840, pp. 30-32.

14

 
Table 1.1. Effect of Oxygen-Contamination Temperature and
Subsequent Heat Treatment on the Corrosion Resistance
of Cb—1% Zr Alloy Welds to Lithium

 

 

weld® Histor;yb Prior to Testing in Lithium 0, Added Attack
No. for 100 hr at 1500°F (ppm) (mils)
Welded; O, added at 1830°F 1700 19
Welded; O, added at 1830°F; heat-treated 1700 0
2 hr at 2370°F
3 O, added at 1830°F; welded 2500 25
O, added at 1830°F; welded; heat-treated 2500 0
2 hr at 2370°F
Welded; O, added at 2190°F 1300
6 Welded; O, added at 2190°F; heat-treated . 1300
2 hr at 2370°F
7 O, added at 2190°F; welded 1800 25
8 '02 added at 2190°F; welded; heat-treated 1800 0

2 hr at 2370°F

 

aInert—gas—shielded tungsten~arc welds in 0.060-in. sheet.

bStarting material was annealed for 2 hr at 2910°F prior to any
subsequent treatment.

Specimens that were corroded had greater hardness than the uncorroded
specimens both before and after the test in lithium. This indicates that
the oxygen distribution which makes the alloy susceptible to corrosion
also strengthens it appreciably. Further studies will be made to deter-
mine whether there is a correlation between Cb—1% Zr alloys which have
been observed to undergo an aging reaction under certain conditions and

Cb—1% Zr alloys which are subject to corrosion by lithium.
Dissimilar-Metal Mass~Transfer Studies in Cb—1% Zr
Alloy-NaK-Type 316 Stainless Steel Systems

J. R. DiStefano, E. E. Hoffman

Effect of Carbon and Nitrogen Pickup on Tensile Properties

It was observed?® previously that CbC and Cb,N layers approximately
0.001 in. thick formed on the surfaces of Cb—1% Zr alloy specimens exposed,

 

8ANP Semiann. Prog. Rep. Oct. 31, 1960, ORNL-3029, p. 13.

15

 
to NaK in a type 316 stainless steel container at 1700°F for 500 hr.

In these tests there was a type 316 stainless steel to Cb—1% Zr alloy
surface area ratio of 10:1. It was also found that when these surface
layers were removed the nitrogen concentration of the specimens remained
high but the carbon concentration was only slightly higher than that of
the starting material.® | |

In order to evaluate the effects of the layers on the mechanical
properties of the alloy, several static tests were conducted under the
conditions deécribed above on alloy specimens annealed at 2190 and 2910°F.
One set of specimens was electropolished to remove the surface layers
that formed during the test. These specimens and specimens with the
layers intact were then tensile tested at room temperature and at 1700°F.
Control specimens were heat treated 500 hr in argon and were also tensile
tested at room temperature and 1700°F. The results of these tests are
summarized in Table 1.2._’

The interpretation of the behavior of the specimens that were annealed
at 2910°F is complicated by the aging that occurred during the test period.
The Specimens annealed at 2190°F did not age during the test period and
therefore gave more reliable information regarding the effect of the carbon
and nitrogen increase. Results of chemical analyses of the Cb—1% Zr
alloy specimens annealed at 2190 and 2910°F are similar in that both indi-
cate a high nitrogen_concentration after removal of the surface layers.
The changes observed in tensile properties of specimens 11, 12, and 13
and of specimens 18, 19, 20, and 21 illustrate the strengthening effect
of the high nitrogen concentration. After removal of the layers, the
tensile strength remains high. These data are consistent with the chemi-
cal analyses for nitrogen.

An interesting effect was observed when the control specimens were
heat treated in a type 316 stainless steel container. The Cb—1% Zr speci-
mens picked up carbon and nitrogen, apparently from the stainless steel,
although an inert-gas environment of argon separated the two metals. An
increase in the tensile strength and a loss in ductility simil@r to those

for the specimens exposed in the NaK system were observed. When the

16

 
specimens annealed at 2190°F were heat treated in a columbium container,

however, no change in tensile properties was found.

Table 1.2. Effect of Brittle Layers on Tensile Properties of Cb—1% Zr Alloy

 

 

Tensile , Elongaticn Chemical Analysis
Specimen Specimen History Test gi:ziéih in 2 in. (ppm)
No. Temperature . Gage —_—_—
(°F) (psi) (%) Ne G 0, H
Specimens Annealed at 2910°F
1 Annealed 70 38 700 14.0 85 120 92 14
2 Annealed and heat treated® 70 79 000P 0 710 560 150 5
3 Annealed and heat treated® 70 65 000P 0 Not analyzed
4 Annealed and exposed® 70 74 000 0 1000 300 170 7
5 Annealed, exposed, and electro- 70 83 000 0 700 120 1S90 19
polished
6 Annealed 1700 25 000 10.5 85 120 92 14
Anmnealed and exposed 1700 €9 800 0 Not analyzed
Annealed, exposed, and electro- 1700 58 300 0 - Not analyzed
polished
Specimens Annealed at 2190°F
9 Annealed 70 37 100 23.5 91 50 100 10
10 Annealed and heat treated® 70 69 500 14.0 830 960 250 19
11 Annealed and heat treated® 70 38 800 26.0 Not analyzed
12 Annealed and exposed® 70 88 500 13.5 920 1730 250 16
13 Annealed, exposed, and electro- 70 79 999 10.0 550 120 270 20
polished
14 Annealed 1700 31 000 13.5 Not analyzed
15 Annealed 1700 32 000 14.5 Not analyzed
16 Annealed and heat treated® 1700 42 000 8.5 Not analyzed
17 Annealed and heat treated?® 1700 39 500 11.5 Not analyzed
18 Annealed and heat treated® 1700 28 200 19.5 Not analyzed
19 Annealed and heat treated® 1700 27 900 18,0 Not analyzed
20 Annealed and exposed ‘ 1700 48 500 6.5 Not analyzed
21 Ammealed and exposed 1700 4'7 500 6.5 Not analyzed
22 Annealed, exposed, and electro- 1700 41 500 4.0 Not analyzed
polished
23 Annealed, exposed, and electro- 1700 44 500 4.5 Not analyzed
polished

 

®Heat treated 500 hr at 1700°F in type 316 stainless steel container with an argon atmos-
phere. :

b . . . -
Specimen aged during heat treatment and picked up nitrogen and carbon from type 316 stain-
less steel container.

CExpose¢f§OO hr at 1700°F in (Cb—1% Zr)-NaK-type 316 stainless steel system; surface area
. ratio of stainless steel to Cb—1% Zr alloy was 10:1.

dElectropolished to remove brittle surface layers; approximately 2 mils removed.

“Heat treated 500 hr at 1700°F in columbium container with an argon atmosphere.

17
Effect of Stress on the Rate of Carbon and Nitrogen Pickup

In order to determine what effect, if any, stress might have on the
rate of pickup of carbon and nitrogen by columbium in a columbium—INaK-
type 316 stainless steel system, columbium specimens were tested in
stressed and unstressed conditions in the same system. Columbium was
used in these tests rather than Cb—1% Zr alloy because wrought-alloy ma-
terial was not available in a form suitable for the fabrication of the
desired specimens. On the basis of earlier tests? it was assumed that
the mechanisms of carbon and nitrogen transfer to the alloy would not be
appreciably different from those for transfer to columbium. Two 10-in.
columbium rods were machined so that each had a 0.4-in.-diam section and
a 0.3-in.-diam section of equal length. One of these specimens was placed
in tension during the exposure while fhe other was suspended immediately
ad jacent to it and was under no load. The test conditions and the results
of chemical analyses of turnings taken from the specimens after test are
summarized in Table 1.3.

These data indicate that for the conditions of this test stress was
not a factor in the rate of pickup of either carbon or nitrogen. Both the
stressed and unstressed specimens picked up substantial amounts of carbon
and nitrogen, and the stressed specimen actually showed less increase in
these elements than the unstressed specimen. Although the nitrogen con-
centration is higher at the surface, some diffusion into the specimen
occurs, but the carbon increase is almost completely confined to the sur-
face layers.

Since no effect of stress was observed, it would appear either that
stress does not affect the diffusion rate of nitrogen and carbon in the
columbium or thét the limiting step in the transfer process at this tem-
perature is the rate at which carbon and nitrogen are supplied to the
columbium surface. That is, the diffusion rate of carbon and nitrogen in
the type 316 stainless steel and/or the solubility of carbon and nitrogen
in the NaK could be the major rate-controlling steps.

 

ANP Semiann. Prog. Rep. April 30, 1960, ORNL~2942, p. 1l.

18

 
Table 1.,3. Chemical Analyses of Columbium Before and After
Exposure to NaK in a Type 316 Stainless Steel
Container for 500 hr at 1700°F

 

Rod Depth of Turnings CPemical Analyses

 

 

Diameter Exposure Condition® From Surface (ppm)
(in.) ‘ (mils) N, o
0.563 As received, not exposed (b) 93 250
0.4 Stressed to 1700 psi 0~5 2000 1600
0.4 Stressed to 1700 psi 5-10 520 380
0.4 Stressed to 1700 psi 10-15 400 290
0.4 Stressed to 1700 psi 1520 420 280
0.3 Stressed to 3000 psi 0-5¢ 2400 1700
0.4 Unstressed 0—5 2500 2300
0.4 Unstressed 510 650 480
0.4 Unstressed 1015 600 250
0.4 Unstressed 15-20 640 420
0.3 Unstressed 05 2900 2500
0.3 Unstressed =10 1100 660
0.3 Unstressed 1015 680 480
0.3 Unstressed 1520 650 420

 

a’I'ype 316 stainless steel to columbium surface area ratio: 10:1.
bBulk analyses; turnings on as-received rod presently being made.
cSam.ples lost for remaining analyses.

10,11 o1 this dissimilar-metal

These and previous isothermal tests
system have shown that nitrogen is the principal element that increases
the strength and reduces the ductility of Cb—1% Zr alloy in the system
Cb~1% Zr alloy-NaK-type 316 stainless steel., Transfer of carbon is limited
to a surface layer which has little or no effect on the mechanical proper-
ties. Stress does not increase the rate of pickup of either carbon or
nitrogen. The effect of the surface area ratio has not been measured in
any detail; but, in a single test conducted with the columbium surface
aresa greater than the type 316 stainless steel surface area, a CbC layer
was found on the éurface of the type 316 stainless steel.

No effort has thus far been made to determine the effect of specimen

thickness, the effect of flowing NaK, the effect of thermal gradients, or

 

10ANP Semiann. Prog. Rep. Oct. 30, 1960, ORNL-3029, p. 13.
11ANP Semiann. Prog. Rep. April 30, 1960, ORNL-2942, p. 9.

19
the effect of reducing the nitrogen concentration of the stainless steel.
These variables will be considered in future tests.

With the data obtained thus far, 1t cannot be stated categorically
that these materials are not compatible under the conditions described,
but only that considerable changes in the mechanical properties of the

Cb—1% Zr alloy can occur under the conditions of these tests.

Vapor Pressure of NaK (43.7 wt % K) in the
Temperature Range 1520 to 1832°F

J. R. DiStefano

The design of the apparatus used in the dissimilar metal tests de=-
scribed above was such that it was necessary to know the vapor pressure
of the NaK in order to specify accurately the stress on the columbium
specimen. Experimental determinations of NaK vapor pressure were there-
fore made in the temperature range 1520 to 1832°F. For these measurements,
the NaK was heated in an evacuated stainless steel container until equi-
librium was reached, and the pressure measured by a strain-gage type of
pressure transmitter (Taylor Instrument Company's Transmitter Model 706
TN1103) was recorded.t? The container was well insulated to avoid re-
fluxing in the system. Above 1830°F, large pressure changes were con-
tinually observed, indicating that the system was in a nonequilibrium
state; hence, the measurements were not continued above this temperature.

The data obtained from this experiment, along with the data calcu-
lated from Raoult's law and empirical equations for the vapor pressure

13 are given in Table 1.4.

of pure Na and K as a function of temperature,
The measured vapor pressure at the highest temperature (1273°K, 1830°F)
was found to be much less than the calculated value, possibly as a re-

sult of dimerization.

 

12y, R. Miller, High Temperature Pressure Transmitter Evaluation,
pPp. 29-32, ORNL-2483, May 16, 1958.

13¢, B. Jackson (ed.), Liquid-Metals Handbook Sodium-NaK Supplement,
Atomic Energy Commission, Department of Navy, Washington, D. C. (1955),
p. 38.

20

 
Table 1.4. Vapor Pressure of NaK
(56.3 Na—43.7 K, wt %)

 

 

 

Temperature Pressure (mm Hg)

°K -~ °F Calculated Measured
1100 1520 ' 758 97
1123 1562 024 1163
1173 1652 1388 1587
1223 1742 2014 _ 2016
1273 1832 2850 2420

 

Potassium Compatibility Studies

 

D. H. Jansen, E, E, Hoffman

Refluxing Potassgsium Capsule Tests

 

The refluxing tests described previously14 are continuing in an ef-
fort to evaluate the relative corrosion resistance of iron-, nickel-,
cobalt~, and columbium-~base alloys in contact with refluxing potassium at
temperatures in the range from 1600 to 2000°F. Sleeve-type inserts of the
test materials have been used to line the walls of each test capsule.
Weight changes observed on specimens from the condenser and boller regions
of type 316 stainless steel {(iron-base) and a Haynes alloy No. 25 (cobalt-
base) refluxing test system have been reported.15 Recent results of
1000-hr tests on Inconel and type 310 stainless steel are given in Fig. 1.13.

The weight losses from dissolution in the condenser (vapor) region
and the weight gains from deposition in the boiler (liquid) region were
larger for type 310 stainless steel than for Inconel. This result is
surprising in view of the fact that type 310 stainless steel has exhibited
more resistance to mass transfer in pumped sodium systems than Inconel.l®

The difference in corrosion resistance observed in these refluxing systems

 

14ANP Semiann. Prog. Rep. Oct. 31, 1960, ORNL-3029, pp. 14~15.
19ANP Semiann. Prog. Rep. Oct. 31, 1960, ORNL-3029, p. 16.
16 ANP Semiann. Prog. Rep. Dec. 31, 1957, ORNL-2440, p. 147.

\

21
may have been influenced by the differences in condenser region tempera-
tures for the two tests (Fig. 1.13). The lower condenser temperature
(1535°F) in the Inconel test, as compared with the condenser temperature

.of 1595°F for the type 310 stainless steel, suggests that less refluxing
occurred in the Inconel test. Such a variation in test conditions would
“tend to explain the ”appérent” superior corrosion resistance of the Inconel.

Additional tests on these alloys at higher temperatures should indicate

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

FLAT CAP
SPECIMEN
UNCLASSIFIED
ORNL-LR-DWG 53497R
R N R
NN
\/ SN %
1535°F UNCONEL)\/‘ §
[N
1595°F (TYPE Q/ N
310 STAINLESS STEEL)igz \ |__TYPE 310 STAINLESS STEEL
e N
N — INCONEL
\ \ _
N N
N N
VRN
V1A
£ N N
= : N : - b
@ N S
g N N
S VIR
N N |
N
VI
N N
N N
Vi1 |
Vi A
Y N '
V) N
i;»‘— _4\§ \
\% Z3
NA 1IN
- \
\1/ /§ —_— .
N
N N . .
N N
A ,~§§
N N
£ N 1N - -
0 N 1 EN - .
N N
- N BN
o NZER AN |
3 VA
3 N 181 ‘ /
%f ] f\ . /
§/ 7 f% /
1IN
of §/ [ /%
1600°F N A T
) v N
iy — |
g -10 0 10 20 30 40 50
THERMOCOUPLE WELL —Y] |/ 2
‘ WEIGHT CHANGE (mg/inS)

 

~MOVABLE THERMOCQUPLE

Fig. 1.13. Test System Configuration and Weight-Change Results for
Inconel and Type 310 Stainless Steel Specimens Tested in Refluxing Potas-
sium for 1000 hr at a Boiler Temperature of 1600°F (871°C).

22

 
more conclusively which of these alloy systems (iron- or nickel-base) has

the best corrosion resistance to potassium in refluxing systems.

Boiling-Potassium Loop Tests

An Inconel and a Haynes alloy No. 25 boiling-potassium loop have been
fabricated. These loops are presently operating at a boiler temperature
of 1600°F. The tests are scheduled for 3000 hr of operation at this tem-
perature. ©Several modifications were made in these loops relative to the
design previously described for the type 316 stainless steel test,17 as
described below. |

Machined, sleeve~type inserts similar to those used in the refluxing
capsule testst? were used to.line the total length (6 ft) of the cold leg
in order to obtain weight-change data, as well as metallographic data, in
the regions where dissolution and deposition have been observed in similar
previous tests on type 316 stainless steel.

Two modifications have been made in the design of the boilers of the
Inconel and Haynes alloy No. 25 systems. The vapor region of the boiler
(2-in. sched.-40 pipe) has been lengthened by 12 in. to minimize the pos-
sibility of liquid carryover to the condenser. The desired condition is
that 1600°F saturated potassium vapor be delivered to the condenser. An
additional boiler modification has been the incorporation of a sodium
jacket contained in 4~in. sched.-40 pipe around the full length (3 ft) of
the potassium boiler. The purpose of this jacket is to provide more uni-

form heating to the potassium in the boiler, which should, in turn, tend

to suppress the pressure and temperature instabilities observed in previous
boiling-potassium loop tests.

Another Haymes alloy No. 25 loop has been designed for operation at
1800°F. 1t is designed to permit quantitative weight-change determinations
on the pipe wall in both the condenser and liquid regions of the cold leg.

A Cb—1% Zr alloy loop for circulating boiling potassium at 2000°F is
also being constructed. This loop will be operated in an inert-atmosphere

chamber.:

 

17ANP Semiann. Prog. Rep. Oct. 31, 1960, ORNL-3029, p. 19.

23
Purification of Potassium Metal: Methods of
Reducing the Oxygen Content

 

 

A, P. Litman, E. E, Hoffman

A program for determining the most effective and practical method of
removing oxygen from potassium has been initiated. Hot gettering, cold
trapping, low-temperature filtration, and vacuum distillation are methods
that will be investigated.
With past purification experience with lithium as a basis, several
batches (10 and 90 kg) of commercial potassium (Mines Safety Appliances) .
were purified by hot gettering and subsequent cold trapping to produce a
relatively pure material for use in compatibility and heat-transfer experi- | -
ments. The results of the two purification experiments were quite similar,
and therefore only the results for the 90-kg batch purification are given
in Fig. 1.14. | |
Oxygen determinations were made by the n-butyl bromide method on
samples of potassium taken at 12 intervals during the 1350-hr purification
experiment. The results indicated that the bulk of the oxygen was re-
moved during gettering with & kg of titanium at 1200°F in the first 250 hr
of the test. Only aAslight additional oxide reduction occurred during the
subsequent 1100-hr cold-trapping treatment. The oxygen content of the
as-recelved potassium was 270 ppm; 250 hr of gettering reduced this to

30 ppm; and 1100 hr of cold trapping yielded a product containing 25 ppm.

Determination of Oxygen in Potassium

 

J. C. White, G. Goldberg

The mercury amalgamation technique and the butyl bromide method 8
are being evaluated to ascertain the most accurate and reproducible method
for determining the oxygen content of potassium. The potassium in ques-

tion is of high purity, containing oxygen in the order of less than 50 ppm.

 

187, C. White, W. J. Ross, and Robert Rowan, Jr., Anal. Chem. 26,
210 (1954).

24

 
GAS

TYPE 347 STAINLESS STEEL TANK
(12 in. OD)

 

 

 

TITANIUM LINER

4-in. HOT AND COLD TRAP

  

 

 

 

 

 

UNCLASSIFIED
ORNL-LR-DWG 40851 R

DISCHARGE

PRESSURE
GAUGE

 

Xy

in

 

 

e —
A

 

 

 

 

 

TITANIUM SPONGE
TITANIUM SCREEN

Oxygen Content**

 

 

 

 

Process Description Teml:;:mms 1;;:; (ppm)
’ Individual Analyses Average
As Received 246, 268, 293 269
- Hot Gottarad at 650°C 650 50 26, 36, 38 33
Using 8.4-kg Ti Sponge 150 18, 50 36
. 650 250 20, 20, 20 20
100 33, 40, 65 46
250 32, 38, 40 v
Cold Trapped* 400 49, 5% 50
Using B.4-kg Ti Sponge 80 550 20, 25, 50 32
45 700 19, 27, 28, 47, 57 3
100 18,27, 35 27
Cold Trapped* 200 16, 42 29
Using 8,4-kg Ti Sponge 35 300 14,19, 20 18
30 400 18, 33 28

 

*Melting Point K = 64°C

**Analytical Method: n-butyl bramide (Ref, ORMNL- 1286}

Fig. l.1l4. Potassium Purification Experiment.

3

25
Pepkowitz and Juddl® first used the amalgamation technique for the
determination of éxygen in sodium. The method involves the repeated ex-
traction of the metallic sodium with mercury; the mercury-insoluble sodium
monoxide floats on the amalgam. When the extraction is completé, the
sodium monoxide is dissolved in water and titrated with standard acid.
With potassium, as with sodium, the assumption is made that the oxide is
the only mercury-insoluble impurity that reacts with water to form the
base. The amalgamation technigue is therefore also applicable to potas-
sium metal analysis.

The original amalgamation apparatus has been simplified, as shown in
Fig. 1.15. The Jamesbury ball valve was modified so that the potassium
sample could be transferred from an inert-atmosphere dry-box directly to
the reaction vessel. The valve acts also as the reaction-vessel cap. The
helium used is of high purity. In addition, an oxXygen getter and a liquid-
nitrogen cold trap are used in series with the helium line.

In operation, the reaction vessel is first evacuated and flamed to
remove all moisture. Mercury, 20 to 30 em’, is admitted slowly to the
reaction vessel, after which the system is pressurized with helium at
2 psia. The reaction vessel side-arm stopcock is closed and the Jamesbury
valve is opened, allowing the potassium sample to drop into the mercury.
When the potassium has reacted with the mercury, the resulting amalgam is
drained, leaving approximately 1/2 in. of mercury above the lower stopcock
to keep from losing the floating oxide. The reaction vessel is alter-
nately evacuated, filled, pressurized, shaken, and drained until a water
extraction of the last draining is neutral to phenolpthalein. The remain-
ing mercury and the insoluble potassium oxide are then washed from the
reaction vessel with water and the resulting base is titrated with standard
acid. |

A few samples of potassium were analyzed by the amalgamation method,
but the results were inconclusive. Sampling techniques are being investi-

gated whereby similar samples of potassium can be analyzed by both the

 

9L, P. Pepkowitz and W. C. Judd, Anal. Chem. 22, 1283 (1950).

26

 
amalgamation and butyl bromide methods to compare the sensitivities of

both methods.

UNCLASSIFIED
ORNL—LR-DWG 58616

M—/
._/\/\//
STAINLESS STEEL
SAMPLE RECEIVER
MERCURY
RESERVOIR

 

 

 

3/4-in. JAMESBURY
VALVE
HELIUM

 

;)i ¢ 0

. __
STAINLESS STEEL
TAPERED JOINT
TYGON TUBING
VACUUM CONNECTION

PYREX PIPE
REACTION VESSEL

 

 

 

 

 

 

      

 

 

T'ig. 1.15. Alkali Metal Amalgamation Apparatus.

27
2. AGING STUDIES OF COLUMBIUM-BASE ALLOYS

Aging of Wrought Material

 

D. 0. Hobson

Data were presented in the previous report1 which indicated that
oxygen in the Cb-1% Zr alloy inhibited the aging reaction. It was postu-
lated that the solution and precipitation of zirconium carbides and/or
nitrides were responsible for the aging reaction. In contrast, daté ob-
tained more recently indicate that oxygen is the chief, and probably the
sole, cause of the aging reaction under the conditions being studied.
Thirteen heats of Cb—1% Zr alloy having oxygen contents ranging from 24
to 950 ppm have been examined. In addition, one heat contaminated to
oxygen levels of 470 to 1200 ppm was examined. All specimens were aged
following a solution anneal at 1600°C. With the exception of three heats
containing only small quantities of oxygen (24 to 50 ppm), a definite cor-
relation was found between the oxygen content and the ageability of the
alloy. For a standardized anneal at 1600°C followed by an aging treat-
ment at 927°C, six heats having up to 260 ppm oxygen showed an aging re-
action. The remaining seven heats (320-950 ppm oxygen) showed no aging
reaction.

This behavior indicates that the solubility of the precipitating phase
was affected by oxygen additions. To further investigate these effects,
various heats were annealed at temperatures ranging from 1200 to 2000°C.

Heat S8FW, which contained 260 ppm oxygen and which had shown aging
after an anneal at 1600°C, was annealed at 1800°C and aged under identical
conditions. As shown in Fig. 2.1, the aging reaction was accelerated.
This behavior is similar to.that of an alloy in which one or both anneal-
ing temperatures is under the solvus line of the precipitating phase, with
the higher annealing temperature resulting in more solution of this phase.

Heat S16EC, containing approximately 950 ppm oxygen, was annealed at
1600, 1800, and 2000°C and aged at 927°C. As shown in Fig. 2.2, the speci-

mens annealed at 1600°C showed no aging response. The specimens annealed

 

1ANP Semiann. Prog. Rep. Oct. 31, 1960, ORNL-3029, pp. 24—25.

28

 
e

UNCLASSIFIED
ORNL-LR-DWG 56B803A

 

{(x 103) | |
© ANNEALED 2hr AT {800°C
90| ® ANNEALED 2hr AT 1600°C

 

 

80— ULTIMATE TENSILE STRENGTH ~— (8} (8)
YIELD STRENGTH
- . P

 

 

/

 

 

Y

 

60 (8)

 

TENSILE AND YIELD STRENGTH {psi)

 

v '/

o

 

\

 

 

 

 

 

 

 

 

 

 

25 50 100

200 500 750

AGING TIME AT 8i5°C (hr)

Fig., 2.1. Results of Tensile Tests of Specimens of Colunibium-
Zirconium Alloy Heat S8FW That Were Annealed 2 hr at 1600 and 1800°C and

Aged at 815°C.

at 1800°C showed some variations in hardness that might possibly be due

to aging. The specimens annealed at 2000°C showed a definite aging peak
after approximately 50 hr at 927°C. (The oxygen content of heat S16EC

UNCLASSIFIED
75 ORNL-LR-DWG 56B05R

A ANNEALED 2hr AT 1600°C
® ANNEALED 2hr AT 1800°C
© ANNEALED 2hr AT 2000°C

o
wn

60

HARDNESS, Rznr

&l
w

o
O

 

45
0

16 25 50 100 200 500
AGING TIME AT 927°C (hr)

Fig. 2.2. Results of Tensile
Tests of Specimens of Columbilum-
Zirconium Alloy Heat S16EC That Were
Annealed 2 hr at 1600, 1800, and
2000°C and Aged at 927°C.

was reduced from approximately
950 to approximately 850 ppm
during annealing at 2000°C.)
Specimens of heat S24WC,
which contained 120 ppm oxygen
and which aged after annealing at
1600°C, were contaminated with
oxygen in amounts ranging fromv
470 to 1200 ppm and then annealed
at 1600 and 1800°C and aged. None
of the higher oxygen content speci-
mens annealed at 1600°C showed
any aging response other than a
drop in hardness. However, after

annealing at 1800°C, the specimens

29

 
with oxygen contents from 470 to 700 ppm showed definite increases in

hardness, as indicated in Fig. 2.3.

UNCLASSIFIED
ORNL-LR-DWG 56806R

 

70

0, CONTENT
65 {ppm)

______-0- 470
530

;;;;;32580____

700

 

 

\

 

 

 

 

 

HARDNESS, R 37
wn o
Q o
#
/
!
!
»
1

 

 

 

 

 

 

 

 

 

 

- 1200
IF.\ _______--"__.—-—-O"
\ ————|e .
45 1 et T $< 840
-~ T~-a
‘t4=:,r”" 800
40
0 25 50 75 100 125 150

AGING TIME AT 927°C (hr)

Fig. 2.3. Bffect of Oxygen Con-
tent on the Aging of Specimens of
Columbium-Zirconium Alloy Heat S24WC-4

That Were Annealed 2 hr at 1800°C and
Aged at 927°C.

UNCLASSIFIED
ORNL- I_.R- DWG 56807

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2000
1900 ,/
AGING OCCURRED ”/,,/’
S 1800 . . ol =//./ A h
i
T 1700 =
= -~
#1600 > o4 -k A A
% ’,,/’//'
b _
1500 1
g
3 NO AGING OCCURRED
uw 1400 &
=
=z
<I
1300
1200 A :
0 200 400 . 600 800 1000 1200

OXYGEN CONTENT (ppm)

Fig. 2.4. Summary of Effect
of Oxygen Content on Aging Reaction
of Specimens of Several Heats of
Columbium-Zirconium Alloy.

30

A summary of the effect of
the oxygen content of the alloy
on the aging reaction is presented
in Fig. 2.4. It may be seen that
as the oxygen content is increased
the annealing temperature must
also be increased in order for
aging to occur. Because of the
present lack of information con-
cerning the ternary system Cb-Zr-0, » -
it would be presumptuous to at-
tempt a quantitative assessment
of the aging reaction in the Cb—1%
Zr alloy. However, it is believed
that ZrO, or a metastable, co-
herent transition phase which
later forms ZrO,; is responsible
for the property changes in the
alloy with heat treatment. Zir-
conium dioxide has been identi-
fied in the Cb—1% Zr and similar
alloys by three independent groups:
ORNL, Pratt & Whitney Aircraft -
Division (CANEL),? and Westinghouse

Research La.boratories.3 -

 

2J. R. Stewart, W. Liberman,
and G. H. Rowe, Recovery and Re-
crystallization of Cb—1% Zr Alloy,
Columbium Metallurgical Symposium,

AIME, Lake George, N. Y., June 1960.

- 2R. T. Begley, W. T. Platte,
A. IT. Lewis, and R. L. Ammon, De-
velopment of Niobium-Base Alloys,
WADC Tech. Rep. 57—344%, Part V.

1)

 
Studies conducted on high-purity Cb—1% Zr heats containing varying
amounts of carbon up to 370 ppm showed no aging at 927°C after annealing
at 1600°C. Preliminary tests with specimens contaminated with 200 to 300
ppm nitrogen similarly showed no aging under these conditions. It is pos-
sible that other conditions of temperature, quench rate, etc., would show

that aging could also be caused by carbon and/or nitrogen.

Aging of Welded Material

 

E. A. PFranco-Ferreira

At thé suggestion of Pratt & Whitney personnel, a study was made of
the use of Rockwell-B hardness data for evaluating the aging behavior of
welds in Cb—1% Zr alloy specimens. Previously reported4’5 bend-test data
for tungsten-arc welds were re-evaluated and correlated with the corre-
sponding Rockwell-B hardness data. At least five hardness readings were
taken directly on the surface of each weld, and it was found that for any
given weld there was little scatter in the data. Therefore the arith-
metical average of the readings for a weld was taken as the hardness value.
For purposes of comparison, the curves of hardness vs aging time at 1500,
1600, 1700, and 1800°F for Cb—l% Zr alloy heat XM-339 (120 ppm 0,, 170
ppm N, and 190 ppm C) were plotted on the same scale as the previous
bend~-test results. These curves are shown in Fig. 2.5, which includes
hardness curves for welds made in vacuum with the electron-beam process.
A comparison of the over-all shapes of the hardness curves with those of
the corresponding bend curves shows a good correlation in general trend.
The electron-beam weld hardness curves exhibit greater variability than
the tungsten-arc weld hardness curves, and the electron-beam welds are
generally of lower hardness. The lower hardness is thought to be associ-
ated with the somewhat higher weld purity.

A closer study of the numerical values of hardness indicates that
these values cannot be cofrelated with bend-test data. It may be seen

that some points associated with brittle and borderline (ductile with

 

“ANP Semiann. Prog. Rep. Oct. 31, 1960, ORNL-3029, pp. 28-36.
°ANP Semiann, Prog. Rep. April 30, 1960, ORNL-2942, pp. 24—36.

31
 

ce

" BEND-TEST BEHAVIOR

LETTERS IN PARENTHESES INDICATE BEND TEST RESULTS FCR ELECTRON-BEAM WELDS:

 

 

UNCLASSIFIED
ORNL-LR-DWG 59006

(D)Y=DUCTILE, (D/C)=DUCTILE WITH CRACKING, (B}=BRITTLE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

DUCTILE SPECIMENS AGED AT 1500°F SPECIMENS AGED AT 1600°C
TUNGSTEN-ARC WELDS
BRITTLE
TUNGSTEN-ARC WELDS
100 -
TUNGSTEN-ARC WELDS
x TUNGSTEN-ARC WELDS
mg 90
|
0=z
Eg 80 (0)——(B)
Sy
g5 ELECTRON-BEAM WELDS
& 70
T ELECTRON-BEAM WELDS (5)
v 60 :
S bucTILE l | | L. W [ |
% TUNGSTEN-ARC WELDS TUNGSTEN-ARC WELDS
w
@ / SPECIMENS AGED AT 1800°C
= SPECIMENS AGED AT1700°C
wl
 BRITTLE
(]
=
wl
m
100 w [ [ [
x TUNGSTEN-ARC WELDS
j% * »—-—"""_'_-‘
we 80 \ TUNGSTEN-ARC WELDS
o)
¥ Ll (D) \ /‘\
Q2 /
e (D) (D) (D)/\(D) T~ \\
a 70 HD) — o
;:[ N O \\
R
- ©) ELECT| ON- EI;EAM \jw—:wsl () (o)
0 5 10 25 50 100 200 300 400 10 25 50 100 200 300 400
AGING TIME (hr) AGING TIME {hr)
Fig. 2.5. Comparison of Bend Test and Rockwell-B Hardness Data for Welds in Cb—1% Zr Alloy Heat

XM- '239 That Were Aged at 1500, 1600, 1700, and 1800°C for Varlous Lengths of Time.
cracking) behavior have lower hardness values than points associated with
ductile behavior. It is, therefore, impossible to predict from knowledge
of the hardness of an aged weld whether or not it will be brittle. The
hardness test, however, has value as an indicator of trends and as an
additional analytical tool.

An extensive study has been made of the use of various postweld an-
nealing treatments as a means of circumventing the aging reaction. The
material used in this study was heat No. S4KW-1, which contained 160 ppm
O, 180 ppm N,, 130 ppm C, 6 ppm Hy, and 1.07% * 0.007 Zr. The response
to aging of the heat was verified by aging a weld for 100 hr at 1500°F;
the aged weld was brittle. Welds were then made under identical condi-
tions and annealed at 1700, 1800, 1900, 2000, 2100, and 2200°F for 1, 2,
3, 4, and 5 hr. Rockwell-B hardness and bend tests were performed on
each weld after annealing. A graphical presentation of these results is
given in Fig. 2.6. After annealing, all welds were aged for 100 hr at

UNCLASSIFIED
ORNL-LR-DWG 56809

 

100 - I !

AS WELDED AND ANNEALED

[ ‘ {D/C}
(D) (0/C) (D} ANNEALED AT 1800°F

 

 

 

 

 

ANNEALED AT 1900°F
ANNEALED AT {1700°F

 

 

 

ANNEALED AT 2100°F

ROCKWELL B HARDNESS NUMBER

 

 

ANNEALED AT 2000°F

 

ANNEALED AT 2200°F
50 —— NOTE: LETTERS IN PARENTHESES INDICATE BEND D
TEST RESULTS AS FOLLOWS:

{D)-DUCTILE

(D/C)-DUCTILE WITH CRACKING
(B)-BRITTLE

 

 

 

40 |
0 1 2 3 4 5

 

ANNEALING TIME (hr)

Fig. 2.6. Influence of Annealing Time at Different Temperatures on
Rockwell-B Hardness of Annealed Cb—1% Zr Alloy Welds.

33
1500°F. The Rockwell-B hardness and bend-test results for these speci-
mens are shown in graphical form in Fig. 2.7.

A study of the hardness curves of Figs. 2.6 and 2.7 indicates that
all the annealing treatments in the 1900 to 2200°F temperature range are
effective in preventing aging under these test conditions. However, an-
nealing at 2000 to 2200°F produces a significant reduction in hardness
values compared with annealing at 1900°F. Optimum annealing times should
range from 3 hr at the lower temperature to 1 hr at the higher temperature.

A metallographic survey was made of the welds, and a composite dis-
play of the results is presented in Fig. 2.8. The microstructures of
welds annealed for 2 hr at temperatures ranging from 1700 through 2200°F
are compared in Fig. 2.8 with the microstructures of the same welds which

were aged for 100 hr at 1500°F after the annealing cycles. In addition,

UNCLASSIFIED
QORNL-LR-DWG 56808
100

 

| | l I

(B) AS WELDED, ANNEALED AND (B)
AGED

(8) | {B) ANNEALED AT {800°F,
| (8) AGED 100 hr AT 1500°F

 

(B)
90

 

 

80

 

ANNEAI_ED AT {90Q0°F,
AGED §QOhr AT {500°F

ANNEALED AT 1700°F,
AGED 100hr AT 1500°F

N7 {D) {D)  ANNEALED AT 2000°F,
D) ) | AGED 100hr AT {S00°F

 

 

70

 

) ANNEALED AT 2400Q°F,
AGED 10Qhr AT 1500°F
(D)

(D) D) |

 

ROCKWELL B HARDNESS NUMBER

60"

 

) (D)

D)
50 L
(D) NOTE: LETTERS IN PARENTHESES INDICATE
BEND TEST RESULTS AS FOLLOWS:
{D)-DUCTILE

{D/C)-DUCTILE WITH CRACKING
(B)-BRITTLE l

I

0 1 2 3 4 5
ANNEALING TiME (hr}

 

 

{0} ANNEALED AT 2200°F,
AGED 100hr AT 1500°F

 

 

 

 

 

40

 

Fig. 2.7. Rockwell-B Hardness Curves for Annealed-and-Aged Welds.
A1l welds aged at 1500°F for 100 hr.

34

 
 

the as-welded and as-welded-and-aged microstructures are shown for com-

parison.

UNCLASSIFIED
PHOTO 53268

¥ 39331 AS WELDED

 

Y 39342 AS AGED

¥ 39620 ANNEALED ¥ 39333 ANNEALED Y 39335 ANNEALED

  

Y 39622 ANNEALED AGED Y 39443 ANNEALED AGED Y 39441 ANNEALED AGED
{700°F IBOO°F 1900°F

¥ 39337 ANNEALED Y 39339 ANNEALED Y 32341 ANNEALED

=

  

¥ 39435 ANNEALED AGED ¥ 39437 ANNEALED AGED Y 39435 ANNEALED AGED
2000°F 2100°F 2200°F

Fig. 2.8. Microstructures of Cb—1% Zr Welds Annealed at Various Tem-
peratures for 2 hr and Aged 100 hr at 1500°F After Annealing.

 

 

 
It may be seen that as the annealing temperature is raised (more ef-}
fective annealing) a heavy precipitate appears as a network throughout
the matrix. The amount of this precipitate increases with temperature,
Further, there is very little change in appearance after aging. It is
postulated that aging no longer occurs after annealing because the pre-
cipitate has been caused to assume a form and distribution which prevents
it from acting as a strengthening and embrittling agent. This type of
behavior is quite evident when the as-aged (brittle) microstructure is
compared with the annealed-and-aged (ductile) microstructures of specimens
which have been annealed from 1900 through 2200°F. Thus far, x-~ray dif-
fraction attempts to identify the precipitaté have been unsuccessful.
Microprobe analyzer equipment is currently being utilized in an attempt
to determine the nature of the precipitate.

The effect of preannealing treatments on welds containing various
amounts of oxygen was studied also. Test samples of Cb—1% Zr alloy heat
XM-339 to which various amounts of oxygen had been added were given a
homogenization treatment in vacuum for 2 hr at 1832°F. Welds were then
made in the contaminated material, and a chemical analysis was run on
each weld. ZFach weld was annealed for 3 hr at temperatures ranging from
1900 through 2200°F, and Rockwell-B hardness and bend tests were run on
each annealed weld. The welds annealed at 2000°F were then aged at 1500°F
for 160 hr and tested after aging.

The results of this study, presented in Table 2.1, indicate that
even with high-oxygen-content welds aging can be prevented by annealing
under the proper conditions. Annealing at a temperature as low as 2000°F
for 3 hr is highly effective in preventing aging. Metallographic examina-

tion of these welds is under way.

Internal Friction of Cb—1% Zr Alloy

 

R. L. Stephenson, H. E. McCoy

The existence of an aging effect in Cb—1% Zr alloys has been demon-

strated, and there is strong evidence that oxygen is involved in the

36
LE

Table 2.1. Results of Annealing and Aging Tests on Oxygen-Contaminated

Welds in Cb—1% Zr Alloy®

 

Mechanical Properties

 

Weld Impurities

 

 

 

{ppm) TAnnealtng As Welded and After Annealing at
em%f;? wre As Annealed 2000°F and Aging
0, N, c
Bend Test RB Hardness Bend Test RB Hardness

820 200 170 4 As welded Brittle 100+ Ductile 86
1900 Brittle 100
2000 Ductile 88
2100 Ductile 75
2200 Ductile 60

1200 170 170 5 As Welded Brittle 100+ Ductile 88
1200 Brittile 100
2000 Ductile 86
2100 Ductile 80
2200 Ductile 68

1700 180 150 6 As welded Brittle 100+ Ductile o1
1900 Brittle 100
2000 Ductile 92
2100 Ductile 79
2200 Ductile 67

4200 180 150 11 As welded Brittle 100+ Ductile 95
1900 Brittle 100
2000 Ductile 95

with cracks

2100 Ductile 89
2200 Ductile 64

 

Hy, 0.87% +

#Parent material from Heat XM-339 containing 120 ppm Oz, 170 ppm N,, 190 ppm C, 1 ppm
0.05% Zr.

A1 annealing done in vacuum for 3 hr.
UNCLASSIFIED

ORNL-LR-DWG 58617 reaction (see preceding section,
o : this chapter).® Previous experi-
TOP FLANGE ™ ments have shown that internal
friction is particularly sensi-
PmlWSE\\‘{j tive to the amount of solute

actually in solid solution.” Thus,
internal friction measurements
should provide an excellent tech-
——— SPECIMEN ~nique for investigating the role

of oxygen in the aging of this

 

columbium alloy. Accordingly, an

<5 internal friction apparatus cap-
able of operating in a controlled
environment at elevated tempera-
tures has been constructed.

The apparatus utilizes a
torsion pendulum, as shown in
Qffl | Fig. 2.9. The wire specimén be -

&4é§fi25§§§§{fl' “ comes the supporting member and

the frequency is determined by

 

the adjustable weights on the
- f LADJUSTABLE
MIRROR——={Y) PENDULUM BOB pendulum bob. Oscillation is

olL initiated by the electromagnets

 

and the mirror provides for meas-

 

o urement of the amplitude by an
Fig. 2.9. Apparatus for Studying optical lever arrangement.
Internal Friction. | The oil serves to damp lat-
eral vibrations. The pendulum is surrounded by an evacuated tube which
is in turn surrounded by a split or "clamshell" furnace. The entire as-

sembly is shown in Fig. 2.10.

 

5D. 0. Hobson; A Preliminary Study of the Aging Behavior of Wrought
Columbium—1% Zirconium Alloys, ORNL-2995, Jan. 6, 196l.

7A. S. Nowick, Prog. in Metal Physics, Vol. 4, p. 37, Pergamon Press,
N. Y. (1953).

38

 
PHOTC 53213

0
i
L
@
0
<1
&
O
=
=

 

A

Sl

 

of Internal Friction Apparatus.

Photograph

2« 1()s

Fig

39

 

 
A

UNCLASSIFIED
ORNL-LR-DWG 58618

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.12
a TEST NO. 1047
® TEST NO. 1004 —540ppm O3>
010 —o TEST NO. 1041 —707 ppm Op ]
A TEST NO. 1009 -718 ppm O,
i_
Z
ul
= 0.08
i
o
o \l
o
. A
= 0.06 ;
=
T 0
<L ®
3
a  0.04
60' -
/\ *
0.02 4 \Q\fl \ -
0 - A T R ®
50 100 150 200 250 300
TEMPERATURE (°C)
Fig. 2.11. Logarithmic Decrement (®) vs Temperature for Columbium

Containing Various Amounts of Oxygen.

UNCLASSIFIED
ORNL-LR-DWG 58619

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.30 o

0.25
.—
=z
i
> 0.20
[0
Q)
L
0
o
S 045
z 3
o
<I
O
o
- 010
20

@
0.05 by
0 - —
0 50 100 150 200 250 300 350 400
TEMPERATURE (°C)
Fig. 2.12. Logarithmic Decrement (&) vs Temperature for Cb—l% Zr

Alloy Containing Interstitial Contaminants.

40
Preliminary experiments aimed at correlating peak height with soliu-
tion content have been performed. All specimens are annealed at 3500°F
prior to measurement in order to place the oxygen in solution. Data for
columbium containing various amounts of oxygen are given in Fig. 2.11.
Qualitative correlation with oxygen content is evident. The internal
friction spectrum between room temperature and 350°C of a Cb—1% Zr alloy
annealed at 3500°F is shown in Fig. 2.12. This spectrum is in agreement
with that of other investigators.8 These results are preliminary, and
improvements have been made in both temperature measurement and specimen

preparation which should give quantitative results in the immediate future.

 

8R. E. Cleary et al., CANEL, personal communication.

41
3. MECHANICAL PROPERTTES INVESTIGATIONS

Columbium Alloy Tube-Burst Tests

 

R. L. Stephenson, R. W. Swindeman

Data have been obtained on the tube-burst strength of Cb—1% Zr alloys
at 1800°F. The data are summarized and compared with the data of other

1,2 in Fig. 3.1.

investigators
The flat slope of .the stress-rupture curve suggests that a slip
mechanism is operative in deformation and failure. This conclusion is
-strengthened by the ductile appearance of the fractured specimen and the
transgrénular failure observed in metallographic examination. A typical

ruptured specimen and its microstructure are shown in Fig. 3.2.

 

lpiscussion of the Structural Alloy Properties Relative to PWAR-11C
Reactor Design, CNIM=-2487, Aug. 31, 1960.

~_°R. T. Begley et al., WADC~-TR-57-344, Part II.

UNCLASSIFIED

a CRNL-LR-DWG 58620
4x10

104
-~ 8
o
a2
a6
w
@ PRESENT WORK
n TUBE-BURST STRENGTH
4 Cb=1% 2Zr

ZZZ777 UNIAXIAL CREEP
Cb-1% Zr IN LITHIUM (REF. 1)

———— UNIAXIAL CREEP
ELECTRON-BEAM—~MELTED Cb (REF. 2)

——e— UNIAXIAL CREEP
POWDER METALLURGY Cb (REF. 2)

 

103
0.4 1 10 100 1000 . 10,000
TIME TO RUPTURE (hr) -

Fig. 3.1. Stress-Rupture Characteristics of Cb—1% Zr Alloys at
1800°F as Obtained at ORNL by Other Investigators.

42

 
 

UNCLASSIFIED
Y 38907

 

 

- . UNCLASSIFIED

 

g ‘ _ . Y 39226 oo

 

 

 

 

» .007
» -" # = 3 N

e .008

 

 

§ <eg = 010

 

~ 011

 

s - . \ 012

 

 

 

 

 

 

Fig. 3.2. Ruptured Cb—1% Zr Tube-Burst Specimen and Microstructure at
Fracture.

43

 
UNCLASSIFIED
PHOTO 53563

 

 

 

 

 

 

 

 

 

 

 

 

 

  

 

I ' | l VYT
! J l :L gy
OAK RIDGE NATIONAL LABORATORY

 

Fig. 3.3. Heating Element of High-Temperature Tube-Burst Test
Furnace.
UNCLASSIFIED
ORNL-LR-DWG 58621

~=——-FPRESSURE TUBE

WATER JACKET. '
?[——fi |
— jb

 

 

 

 

 

 

 

 

’////(L,FURNACE
|— RADIATION SHIELD
SPECIMEN—T l
é
VACUUM

PORT

Fig. 3.4. Schematic Drawing of
Tube-Burst Test Apparatus.

In the course of these ex-
periments it became apparent that
a8 new apparatus was needed for
tests at temperatures above 2000°F
in order to increase the relia-
bility of the results and to re-
duce specimen contaminagtion. A
tantalum-wire furnace, shown in
Fig. 3.3, surrounded by radiation
shields and a water-cooled evacu-
ated chamber was found to be the
most promising design. A drawing
of this arrangement is shown in
Iig. 3.4. Prototypes of this ap-
paratus have been operated at tem-

peratures above 2800°F. In addi-

tion to the high temperature, this furnace offers the advantages of fast

response, modest power requirements, and relatively low replacement cost.

Four apparatus of this type are now under construction.

45
4. ALTQY PREPARATION

Melting of High-Purity Columbium and
Columbium-Zirconium Alloys

 

T. Hikido, D. T. Bourgette

In previous attempts to prepare Cb—l% Zr alloy samples by electron-
beam melting, 1.5 wt % Zr was charged with approximately 350 g of colum-
bium.? The alloy was melted, rolled into strip, sheared, and remelted to
improve homogeneity. The zirconium content of the final product ranged
from 0.57 to 0.88 wt %, with an average of 0.71 wt % for four melts.
Three additional heats of approximately 350 g each were then prepared,
using a similar melting schedule, but an initial charge of 2 wt % Zr was
used in order to increase the final composition to the 1 wt % desired.

The results of chemical analyses of these melts are presented in Table 4.1.

Table 4.1. Results of Chemical Analyses
of Electron-Beam-Melted Columbium-
Zirconium Alloys

 

 

 

7r Content Impurity Analysis (ppm)
Melt No. (W‘t %)
02 N, C H,
215 0.34 64 15 10 13
216 0.19 82 20 10 16
217 0.21 38 11 10 14

 

_ The data of Table 4.l indicate satisfactory purification with respect
to the interstitial impurities but unexpectedly low zirconium content.

The problem of obtaining the desired alloy composition is a consequence of
the wide difference in vapor pressure between columbium and zirconium and

the difficulty of achieving identical pressure and temperature conditions

Trom melt to melt in the electron-beam furnace. It was decided therefore

 

1ANP Semiann. Prog. Rep. Oct. 31, 1960, ORNL-3029, pp. 41—42.

46

 
to purify the columbium and zirconium starting material by electron-beam
melting and then to arc-melt the alloy under an inert-gas atmosphere in
the hope that the alloy content could be controlled more closely without
undue sacrifice of purity.

Tests were made to determine the purity levels which could be main-
tained in the small button furnace and the intermediate-size arc furnace
and the maximum quantity of columbium that could be melted in a single
heat. Establishment of this limiting heat size was necessary to permit
planning of comprehensive test programs in which such aspects as corrosion,
weldability, aging, and mechanical properties would be correlated. It is
desirable to perform all tests on a single heat of material to eliminate
the material variable.

A series of heats of columbium, ranging in size from 300 to over
1500 g, was melted under argon in the intermediate-size arc furnace. The
starting material was in the form of as-reduced roundels, approximately
1/4 in. in diameter by 1/2 to 3/4 in. long. The heats up to 800 g in
size were melted with little difficulty. However, as the quantity of
columbium was increased above 800 g, the high arc densities required to
make the melts resulted in melting of the tungsten electrode tip, re-
peated cracking of the protective cobalt glass, and burning of the rubber
O-rings. Redesign of the electrode has permitted sufficlient cooling to
prevent melting of the tungsten tip. However, the maxXimum heat size 1is
limited to approximately 1000 g by inadequate cooling of the furnace
tank. The tank would have to be completely rebuilt to alleviate the over-
heating problem. With heats greater than 1500 g, certain areas of the
stainless steel furnace tank became overheated to the point of actually
melting. '

The oxygen, nitrogen, carbon, and hydrogen analyses of columbium
heats melted under various conditions by electron~beam and nonconsumable-
electrode arc melting are summarized in Table 4.2. These initial melting
studies indicate that only the oxygen content is increased during inert-
gas arc melting of either low-purity, as-reduced columbium or electron-

beam-purified material. Increases of oxygen content on the order of

47
Table 4.2. Results of Chemical Analyses
of Electron-Beam-Melted and
Arc-Melted Columbium

 

Impurity Analysis (ppm)
Melting Process

 

 

0, Ny c o,
As received 1700 390 300 12
Arc® 2200 390 330 11
ArcP 2100 370 300 7
Electron beam 42 25 30 8
Electron beam and arc® 112 24 31 9

 

“Nonconsumable-electrode arc melt in intermediate-
size furnace under high-purity argon (O, < 2 ppm) pres-
sure of 605 mm Hg. '

bNonconsumable-electrode arc melt in small button -
furnace under high-purity argon pressure of 300 mm Hg.

“Nonconsumable~electrode arc melt of previously
electron-beam~-melted columbium in intermediate-size
furnace.

70 ppm, as observed in arc melting the previously electron-beam~-melted
columbium, should be acceptable in the preparation of material for most
research programs.

Eighteen apfiroximately 350~g heats of columbium were prepared by
electron~beam melting for use in the preparation (by arc melting) of
columbium=zirconium alloys containing zirconium in the rangé of 0.2 to
1.2 wt %. Chemical analyses of three randomly selected  samples from the
18 heats indicated average impurity contents of 40 ppm 0,, 29 ppm Né,
and 23 ppm C., The analyses of the arc-melted alloys prepared from this
melting stock are presented in Table 4.3.

Addition of Nitrogen to Columbium and
Columbium=~Zirconium Alloys

D. T. Bourgette

In order to study systematically the effects of nitrogen on the age

hardening of columbium and columbium-zirconium alloys it is necessary to

48

 
Table 4.3. Results of Chemical Analyses
of Arc-Melted Columbium-Zirconium Alloys

 

 

 

 

Zirconium Impurity Analysis
(wt %) ~ (ppm)
Nominal Actual Ho 0Os No
0.2 0.16 22 140 79
0.4 0.24 38 290 96
0.6 0.36 41 360 %%
0.8 0.51 40 380 71
1.0 0.66 25 210 83
1.2 0.7 30 250 73

 

develop techniques for masking additions closely controlled in both quantity
and homogeneity. Accordingly, a series of 50~-g buttons of electron-beam-
melted columbium was arc melted under argon with nitrogen added over a
range of partial pressures from 25 to 25 000 p Hg. The results of chemi-

cal analyses of these buttons are given in Table 4.4.

Table 4.4. Results of Chemical Analyses of Columbium Arc
Melted Under Mixtures of Nitrogen and Argon

 

Nitrogen Partial Impurity Analysis

 

 

Sample Code Pressure (ppm)
(1) ‘ 02 N B
25-1X 25 150 64 11
50-1X 50 160 100 9
100-1X 100 81 130 8
200-1X 200 : 79 150 8
400-1X 400 73 230 6
700-1X 700 110 420 5
1000-1X 1 000 120 590 5
5 mm~1X 5 000 180 1500 2
25 mm~1X 25 000 140 1800 2

 

From the analytical results it is apparent that the nitrogen con-

centration in columbium increases with an increase in nitrogen partial

49
. pressures, as was expected. It is also observed that the hydrogen con-
centration decreases with an increase in nitrogen concentration and nitro-
gen partial pressure. An explanation of this phenomenon is not immediately
apparent. The increase in oxygen concentration was satisfactorily low,
ranging from 33 to 140 ppm (assumed avefage starting composition in the

electron-beam-melted columbium was 40 ppm).

Electron-Beam Melting of Columbium-Hafnium Alloys
T. Hikido

Columbium~hafnium alloys were electron-beam melted to study their
melting characteristics and to provide material for the study of aging in
these alloys. A melting schedule similar to that reported above for the
columbium-zirconium alloys was followed. The columbium and hafnium were
electron~beam melted separately, charged to the furnace in the desired
proportions, and alloyed. These alloy buttons, approximately 350 g each,
were then rolled into strip, sheared, and remelted to achieve homogeneity.
The resultant buttons were then rolled to 0.040-in. strip and analyzed.

The results of the chemical analyses are given in Table 4.5.

Table 4.5. Results of Chemical Analyses of
Columbium-Hafnium Alloys

 

 

 

 

Hafnium Impurity Analysis
Melt No. (vt #) (ppm)
Charge Final Oz N, C
256 5.0 2.82 37 17 30
268 2.0 1.10 17 13 20
273 1.0 0.54 52 6 20

 

50

 
5. FABRICATION STUDIES

Brazing of Columbium

 

C. W. Fox

The investigation of brazing alloYs for joining refractory metals
has been under way for several years, and the results of the preliminary
phases of this program have been reported.l:2 It is evident that alloys
containing titanium and zirconium as major constituents are potentially
useful. Approximately 100 different binary, ternary, and quaternary
alloy compositions have been formulated and their melting points deter-
mined, and the general wetting characteristics of many of these alldys
on titanium, zirconium, molybdenum, and columbium have been evaluated.

In view of the interest in columbium and its alloys for a variety
of high-temperature reactor applications, this study has been.re-émphasized
with specific direction toward the brazing of columbium. It is expected
that the brazing results on columbium will be applicable to dilute colum-
bium alloys. Since the fabrication.of test components, and in some cases
the construction of reactor assemblies, may require the Jjoining of colum-
bium alloys to stainless steels, brazed joints of columbium and stainless
steel have also been evaluated.

Since the brazing of large subassemblies would be extremely diffi-
cult at temperatures above 1300°C, only those alloys with flow points be-
low this temperature were considered for further evaluation. Twelve
brazing alloys have been tested on both columbium-to-columbium and colum-
bium-to-stainless steel T-joints 3 in. long. All brazing was performed
in a 3-in, -diam muffle furnace in vacuum. Bach joint was held at the
brazing temperature approximately 10 min. The general visual observations
made in these flowability tests are given in Table 5.1.

Typical brazed T-joints are shown in Fig. 5.1. Joints 1, 2, and 3

are columbium-to-columbium specimens brazed with 67% Zr—29% V—4% Fe,

1R. G. Gilliland and G. M. Slaughter, ANP Semiann. Prog, Rep. Sept. 30,
1958, ORNL-2599, p. 29.

2ANP Semiann. Prog. Rep. March 31, 1959, ORNL-2711, pp. 17-19.

51
 

 

Table 5.1.

to-Stainless Steel T-Joints

Results of Brazing Studies on Columbium-to-Columbium and Columbium-

 

Flow Characteristics of Brazing Alloy

 

 

= Flow
Alloy Composition Point
(wt %) (°c) Columbium-to-Columbium  Columbium-to-Stainless
Joint Steel Joint

67 Zr—29 V-4 TFe 1300 Excellent flow Extreme stainless steel
dilution, poor flow

60 Zr—25 V=15 Cb 1280 Excellent flow Severe stainless steel
dilution, fair flow

48 Z2r—48 Ti— Be 1050 Excellent flow Good flow

63 Ti-27 Fe—10 Mo 1250 Excellent flow Good flow

63 Ti—27 Fe=10 V 1280 Excellent flow Severe stainless steel
dilution, very pcor
flow

68 Ti—28 V-4 Be 1250 Excellent flow Poor flow, joint sepa-
rated on cooling

45 Ti—40 Zr-15 Fe 1050 Excellent flow Good flow

75 Zr-19 Cb—6 Be 1050 Excellent flow Good flow

80 Zr-17 Fe—3 Be 1000 Excellent flow Very poor flow, joint
separated on cooling

46 Ti—46 Zr—4 V=4 Be 1000 Excellent flow Very poor flow

95 Zr—5 Be 1000 Excellent flow Good flow

62 Ti—26 Fe—8 Mo Zr 1250 Excellent flow Fair flow, Jjoint sepa-

rated on cooling

 

 

Cb TC Cb

UNCLASSIFIED
PHOTO 52821

Cb TO STAINLESS STEEL

Fig. 5.1. Typical Columbium-to-Columbium end Columbium-to-Stainless

Steel T-Joints Brazed with Various Developmental Brazing Alloys.

flow along the entire 3 in. of joint is evident.

o2

Good
 

80% Zr—17% Fe—3% Be, and 46% Ti—46% Zr—% V—4% Be alloys, respectively.
Joints 4, 5, and 6 are columbium-to-stainless steel specimens brazed with
95% Zr—5% Be, 48% Ti—48% Zr—% Be, and 45% Ti—0% Zr—15% Fe alloys, re-
spectively. The brazing alloy flowed completely along the 3-in. length
on all the joints, and no visual evidences of fillet cracking were seen.
One of the primary problems associated with making dissimilar-metal joints
results from the difference in the coefficients of thermal expansion.

The typical moderate distortion of the columbium-to-stainless steel joint
is illustrated in Fig. 5.2. In spite of this condition, it is felt that
such a joining technique mey be of value in instances where joints can be
designed to compensate for the differences in expansion of the individual

components.

UNCLASSIFIED
Y-39380

 

Fig. 5.2. Columbium-to-Stainless Steel Brazed Joint Showing Distor-
tion Resulting from Difference in Thermal-Expansion Coefficients of the
Two Materials.

53

 

 
Fabrication of Clad Tubing

 

T. K. Roche

Studies are being made to determine the feasibility of cladding colum-
bium and Cb—1% Zr alloy tubing for protection against oxidation. The in-
centive for these studies is the possible application of such a composite
in 1iquid metal-to-air radiators at temperatures in excess of 1600°F.

Past work has shown that columbium can be clad with conventional oxidation-
resistant alloy;, such as type 310 stainless steel and Inconel, by hot-
rolling techniques. A measurable reaction occurs between columbium and
these alloys at 1500°F, the temperature investigated, with the formation
of brittle intermetallic compounds. Copper was found to be a fairly ef-
fective barrier material between columbium and iron-base alloys, although
a slight reaction takes place at the columbium-copper interface, with

some indication of brittleness. Extensive diffusion occurs between copper
and nickel-base alloys at 1500°F.

Recently an effort has been made to clad columbium tubing with stain-
less steel by conventional cold-drawing techniques followed by a vacuum-
annealing treatment to promote bonding. In one case, type 446 stainless
steel was clad to the outside of columbium tubing after first plating the
inside of the type 446 stainless steel tube and the outside of the colum-
bium tube with a thin layer of copper. The tubing was cold drawn to a
final size of 0.455 in. o.d. by 0.040 in. wall, approximately 0.039 in.
of type 446 stainless steel on 0.001 in. of copper on 0.030 in. of colum-
bium. In the second case, type 304 stainless steel was clad to the out-
side of columbium with a 0.010-in. copper barrier between. This thickness
of copper included both a thin copper plate applied to the outside of the
columbium tube and a preplaced copper tube between the plated columbium
and the type 304 stainless steel cladding. The composite was cold drawn
to a final size of 0.350 in. o.d. by 0.050 in. wall, approximately 0.020
in. of type 304 stainless steel on 0.010 in. of copper on 0.020 in. of
columbium. Both lots of tubing were vacuum heat treated for 2 hr at

1850°F after drawing.

54
Metallographic examination of the type 446 stainless steel, copper
plate, columbium composite showed evidence of a thin reaction layer on
the columbium side of the interface and rather poor metallurgical bonding.
Heat treatment of samples of this tubing for 500 hr at 1500, 1650, 1800,
and 1900°F resulted in growth of the reaction layer with increasing tem-
perature. The growth of the reaction layer was accompanied by the forma-
tion of diffusion voids on the type 446 stainless steel side of the in-
terface, Fractures present in the reaction layer indicated its inherent
brittleness, It was apparent that the 0.001-in.-thick copper plate was
ineffective in preventing diffusion of the type 446 stainless steel to
the columbium side of the interface.

Metallographic examination of the type 304 stainless steel, copper,
columbium composite also showed generally poor bonding at the interface,
with scome separation between the plated copper layer and the preplaced
copper tube., Good bonding was obtained between the type 304 stainless
steel and the copper. OSamples of this composite were also heat treated
for 500 hr at 1500, 1650, 1800, and 1900°F. A reaction at the copper-
to-columbium interface occurred but to a lesser extent than in the type
446 stainless steel, copper plate, columbium composite samples. It was
further noted that copper diffused into the type 304 stainless steel.

No apparent embrittlement was associated with these reactions.

On the basis of these studies it appears that the difference in
thermal expansion between the stainless steels and columbium prevents
good bonding of composites by prepared cold drawing and subsequent heat
treatment. This difficulty should be minimized by a hot-working opera-

tion, such as hot extrusion, of tube blanks prior to redrawing.

55
6. BERYLLTUM OXIDE RESEARCH

Preparation of Refractory Oxides from Molten Salts

 

R. E. Moore, J. H. Shaffer

A method of preparing refractory oxides from molten-salt media is
being developed as a process for obtaining high-purity, single-crystal
material at relatively low temperatures. The results of previously re-
ported experimen.ts1 showed that the reactions of water vapor with BeF,
or UF; dissolved in a molten-salt solvent at 800°C would yield single
crystals of BeO and UO,. When both BeF,; and UF, were dissolved in a
molten-salt solvent, their reactions with water vapor, in some instances,
resulted in UO, crystals that were clad with BeO. These unique materials
are of fundamental interest in evaluating these oxides for reactor appli-
cations.

As part of a program to obtain these materials, three preparations
of BeO (~500 g) were made by the steam precipitation process. The aver-
age diameters of the single crystals, measured petrographically, were 10
to 20 u, but a few crystals were as large as 100 p. Spectrographic analy-
ses of the best batch of prepared Be0O gave the following results:

Impurity PPm Tmpurity PPm
Al 10 Pb <10
B 50 Si 50
Ba <5 Sn <10
Ca <20 Ti 100
Cd <l Vv 5
Co . 2 Cr 30
Cu 5 Fe 30
K <25 Li 50

. Mg <10 Mn 15
Mo <10 Ni 200
Na <25 -

Analyses of the other BeO preparations revealed higher concentrations of

solvent and container materials.

 

1ANP Semiann. Prog. Rep. Oct. 31, 1960, ORNL-3029, p. 54.

56
In an attempt to improve the quality of BeO prepared by this process,
modifications to the experimental procedure have been made. In order to
reduce extraneous contamination, special emphasis has been placed on ob-
taining a high-purity fluoride mixture LiF-BeF, (75-25 mole %) as the
starting material. A second modification has been the use of hydrogen
in place of helium as the carrier gas for water vapor. This practice
should prove effective in reducing impurity levels attributed to corrosion
of the reaction vessel. In addition, the molten-fluoride solvent will be
filtered from the oxide product in a preliminary recovery operation.
Small-scale comparative experiments using these revised technigues are
in progress.

Two approaches are being used in the study of BeO-clad crystals of
U0y, both of which depend on the use of U0, particles less than 20 u in
diameter. Observations of previously produced material indicated that
particles larger than this were not effectively coated. In the first ap-
proach, crystals of U0, of carefully controlled particle size are added
to the fluoride reaction mixture prior to the steam precipitation of BeO.
In the second approach, consecutive precipitations of U0, and BeO are
being made from an initial mixture containing UF,; in LiF-BeF,; (75-25 mole
%).’ Observations of previous UQO,; preparations indicated that smaller
particles were produced by more rapid reactions; therefore, a rapid in-
itial rafie of steam precipitation of the U0, is being followed by a slower

rate of steam precipitation to cause the formation of a BeQ coating.

Preparation and Calcination of Beryllium Oxalate Monohydrate

 

B. J. Sturm

Additional BeC304+H50 for conversion to sinterable BeO was obtained
by boiling a saturated aqueous solution, as described previously.2 By
filtering the hot solution through a Buchner funnel maintained at the

same temperature as that of the solution, a batch of 700 g of precipitate

 

2ANP Semiann. Prog. Rep. April 30, 1960, ORNL-2942, pp. 45-46.

57
was collected with almost no formation of the trihydrate. The yield of
BeO upon calcining was 22%, exactly that calculated for the monohydrate.
As the monohydrate had less water of crystallization than the usual tri-
hydrate, it occupied less volume and adsorbed less "mother liquor." The
final BeO was consequently cofisiderably purer than material made from tri-

hydrate.

Purification of Beryllium Compounds by Solvent Extraction

 

R. BE. Moore

In evaluating BeO for special reactor applications, the effect of
impurities on such properties as sintering characteristics and irradia-
tion behavior camnot be assessed because of the nonavailability of ultra-
pure beryllium compounds. A process for purifying beryllium hydroxide
by a solvent-extraction technique is currently being developed for this
purpose. By this method beryllium is extracted from an aqueous phase
contéining nitric acid and ethylenediaminetetraacetic acid (EDTA) into
an organic phase of acetylacetone in carbon tetrachloride. The impurities
present in the beryllium hydroxide starting materials are complexed with
the EDTA; while beryllium is extracted as the acetylacetonate,
Be(CH3COCHCOCH3) . Beryllium is recovered by a subsequent back extraction
with a strong acidic solution. Various separation facfiors for this pro-
cess have been previously reported.3

Initially attempts were made to back extract with a saturated oxalic
acid solution to recover the beryllium as oxalate. This operation proved
to be somewhat cumbersome because of the relatively large quantity of
oxalic acid that was required. In addition, unfavorable spectrographic
results suggested that a simplification of the back-extraction process
was desirable. Accordingly, nitric acid was tested for the back extrac-
tion of beryllium from the organic phase. This step was followed by the

precipitation of beryllium hydroxide upon addition of ammonium hydroxide.

 

3ANP Semiann. Prog. Rep. Oct. 31, 1960, ORNL-3029, p. 56.

58

 
If beryllium oxalate is the desired end product, an additional reaction
of the precipitated beryllium hydroxide with a near-stoichiometric quantity
of oxalic acid can be conveniently accomplished.

In the very first series of gram-scale preparations of beryllium hy-
droxide by this revised technigue the potential capabilities of the ex-
traction process were demonstrated. Although the starting material was
reported to contain 725-ppm metallic and 2000-ppm nonmetallic impurities,
spectrographic analysis of the final product indicated only the following
impurity concentrations. The reportable impurities consisted of 5 ppm Mg,
10 ppm Si, 5 ppm Sn, 5 ppm V, and 5 ppm B. The elements present in quan-
tities below spectrographic standards were Mo, <50 ppm; Zn, <25 ppm; Sr,
Cr, Ni, and Ti, <20 ppm; Al, Ba, Ca, Fe, K, Li, and Na, <10 ppm; Co, Cu,
Mn, and Pb, <5 ppm; and Ag, <0.5 ppm. No nonmetallic impurities were
detected.

In a second larger scale operation, approximately 30 g was prepared
in a one-stage extraction operation. The only impurity which could defi-
nitely be detected by spectrographic analysis of the final product was
5 ppm of magnesium.

The success of this extraction process for obtaining ultrapure be-
ryllium hydroxide has focussed attention on the need for better spectro-
graphic standard mixtures. While these standards have been adequate up
to the present for determining the impurity level of current commerical
and laboratory preparations of beryllium compounds, the future develop-
ment of this liquid-liquid extraction process for obtaining ultrapure
beryllium hydroxide will be facilitated by more accurate spectrographic
analyses. To this end, the initial production of 100-g quantities of
beryllium hydroxide will be used for improving spectrographic standards

for analyzing beryllium compounds,

Phase Relationships in Be(O-Metal Oxide Systems

 

Investigations of phase relatibnships in BeO-metal oxide systems

have continued.* TIn considering BeO as a nuclear reactor material,

 

“ANP Semiann. Prog. Rep. Oct. 31, 1960, ORNL-3029, pp. 57-61.

59
information is needed on reactions of BeO with the oxides which are added
to improve sinterability, the oxides which are present as coatings for
corrosion resistance and fission-product retention, and the oxides which
are added to improve fuel retention. Data concerning the systems BeO-

Ca0, Be0-Mg0O, Be0-CeO,, and BeO-ZrO, have been obtained recently.

The Be0-Ca0 System (R. A. Potter)

 

The effects of heat treatment and cooling rates in dry argon on the
formation and decomposition of the intermediate phase CasBe3z0s in the
Be0O-Ca0 system were investigated. A study of the time-temperature de-
pendence of its formation led to the production of single crystals of
this phase.

Information developed from cooling-rate data showed the producté of
Be0-Ca0 mixtures, including the eutectic composition of 60 mole % BeO, .
cooled from super-solidus temperatures to be rate dependent. The prob-

able reactions were:

Eutectic liquid — BeO + Ca0 |, (1)
Eutectic liquid — BeQ + Ca0 — CasBe30s (2)
Eutectic liquid — Ca,Be30s . (3)

Reaction (1) was carried to completion in melts cooled at 20°C/min over
the range 1440 to 1300°C. The BeO and Ca0 were manifest in a eutectic
structure, as shown in Fig. 6.1. Reaction (2) did not go to completion
at a cooling rate of 180°C/min from 1440 to 1100°C; however, three-phase
structures containing BeO, Ca0, and CasBe30s5, shown in Fig. 6.2, were ob-
tained. Reaction (3) resulted at cooling rates of approximately l60°C/sec
over the range 1500 to 850°C. |
Although cooling data indicated that the intermediate compound had
a temperature range of stability, efforts to establish the range weré
unsuccessful. The compound was not detected in specimens prepared by

heating mixtures of Be0O and Ca0 to 1200 and 1300°C for 100 hr, and its

60
UNCLASSIFIED
PHOTO 54280

 

E . : . e r @ . l}
| (o) I n (6) {

& . ” e

Fig. 6.1. Two-Phase Structures Containing the Be0-Ca0 Eutectic and
the Terminal Component, BeO. (a) BeO-Ca0 (80-20 mole %). (b) Be0-Ca0d
(92-8 mole %). Transmitted light, lower Nicol in. 165X.

formation was not observed in a mixture of BeQ and Ca0 seeded with a small
addition of the compound and heated to 1300°C for 20 hr.

Single crystals of CayBe30s were successfully grown by maintaining

close control of the reaction

Eutectic liquid — Ca,Be30s5

A 60/40 mole ratio mixture of BeO and Ca0 was melted on a platinum re-
sistance strip and visually observed while undergoing cooling. At the
moment of observed crystal growth in the melt, heat was reintroduced to
the system and crystal growth was allowed to proceed. The resulting iso-
lated single crystals are being examined.

The BeO-Mg0, BeO-CeOs, Be0-Zr0, Systems (A. T. Chapman)

 

A technigue has been developed for determining the eutectic tempera-
ture and composition of binary systems which require wvery high tempera-
tures to produce melting. The success of this technique, which is ap-

plicable for simple eutectic configurations such as those found in many

61

 

 
 

UNCLASSIFIED
Y-37820

 

 

 

Fig. 6.2. Three-Phase Structure Obtained in Be0-Cal0 (60-40 mole %)
Specimen Cooled at 180°C/min from 1440 to 1100°C. The tentative identi-
Tication of the phases are: white, Cal; gray, CazBe30s5; fine structure,
Be0O-Ca0 Eutectic.

BeO systems, depends upon the very fluid nature of BeO-rich melts. Pre-
vious work demonstrated that small quantities of BeO-rich melts could
"drain" out of a thin wafer-shaped specimen and be absorbed into a porous
BeO body. The porous BeO acts as a collector for the liquid and also
draws the liquid away from the wafer-collector interface, effectively pre-
venting these pieces from bonding together. The reason the BeO pellet
collector remains porous during heating in air, even at very high tempera-
tures, was discussed by Aitken in a recent paper.5

The eutectic temperature was established by placing the specimen,

normaelly containing 85 mole % or more Be0O, on the porous collector and

 

’E. A, Aitken, Initial Sintering Kinetics of Beryllium Oxide, J. Am.
Ceram. Soc., 43(12), 1960.

62
heating in air to various temperatures. The lowest temperature at which
an appreciable amount of liquid formed and "drained" into the collector
was noted. The loss of weight of the wafer closely approximated the gain

of the collector and, as long as the amount of liquid is reasonably large,

for example, 50 to 150 mg‘loss from a wafer weighing 400 to 600 mg, and
is collected in a 6- to 8-g BeO pellet, the sample and collector can be
readily separated.

The eutectic composition was established by simple material balance
considerations. The starting wafer composition, the composition of the
fired wafers (determined by chemical analysis), and the weight loss of
the specimen were used to determine the composition of the liquid lost
to the BeO collector. The BeO-MgO, BeO-CeO;, and BeO-ZrO, systems have
been investigated by employing this technique. ' '

Be0O-Mg0O. Four different BeO-MgO mixtures were heated on an iridium
strip heater to temperatures above the solidus and quenched. This tech-
nigue, which has been employed to produce BeO-Y,03, BeO-CaO, BeO-Sro;'and
BeO-Las03 compounds, gave no indication of any interaction, except melt-
ing, between the BeO and MgO. Based on these data and the inability to

obtain subsolidus reactions,®

it is concluded, contrary to a previous
report, ! that BeO and MgO do not form a compound.

The compositions of the liquids which "drained" out of BeO-MgO speci-
mens are presented in Table 6.1. Aé a result of these and the previously
reported tests, it is concluded that (l), as mentioned above, BeO and
MgO do not react to form compounds or extensive solid solutions, (2) the
phase diagram has a simple eutectic configuration, although the possibility
of very limited solubility of either phase has not been precluded, and (3)
the eutectic composition is 69 * 2 mole % BeO and the eutectic temperature
is 1860 = 10°C.

Be0-CeO,. The CeOp used for the determination of the eutectic tem-

perature which was reported previously6 contained a large concentration

of iImpurities. Re-evaluation of this system using high-purity CeO, raised

 

6ANP Semiann. Prog. Rep. Oct. 31, 1960, ORNL-3029, pp. 60—6l.
ANP Semiann. Prog. Rep. Arpil 30, 1960, ORNL-2942, p. 53.

63
Table 6.1. Compositions of Liquids Drained from
BeO-MgO Wafers

 

 

Starting Composition of
Composition Maximum Firing Lost Liquid
(mole %) Temperature (mole %)
— (°C) —
MgO BeO MgO BeO
10 20 1930 28.5 71.5

6 9 1900 31.7 68.3

10 20 1900 32.0 68.0

 

the temperature of first melting from the erroneously reported value of
1775 + 20°C to 1890 # 20°C. The compositions of liquid absorbed in the
BeO collector at temperatures above 1890°C are presented in Table 6.2.

The compositions containing the 10 mole % CeQO, produced the most
liquid and, hence, the most reliable data. Although there is considerable
uncertainty in these figures, a good estimate of the eutectic composition
is 63 * 3 mole % BeO. Since there was no evidence of compoufid formation
or extensive solid solution during this investigation, it is concluded
that BeO-Ce0p is essentially a simple eutectic system. The first melting
occurs at 1890 * 20°C, and the liquid contains 63 * 3 mole % BeO.

Table 6.2. Compositions of Liquids Drained from
Be0O-CeO, Wafers

 

 

 

Starting Composition of
Composition Maximum Firing Lost Liguid
(mole %) Tem%er?ture (mole %)

- °C
Ce0s BeO , Ce0,p BeO
10 90 1955 37.4 62.6
10 ‘ 90 1970 39.8 60.2
5 95 1970 31.4 68. 6
10 90 _ 1960 . 37.2 62.8
5 95 1960 29.0 71.0
10 20 ' 1960 34.9 65.1

 

64

 
Be0-Zr0,. The "porous" collector method was employed to investigate
the phase relationship in the Be0-Zr0O,; system. The composition of the
liquid which drained from BeO-Zr0O, specimens was determined for a wide
range of starting compositions. These compositions which represent the
eutectic composition for the BeO-ZrO, system were in good agreement and,
thus, offer experimental evidence that the "porous" collector technique
is a valid and reliable method for studying high-temperature phase equi-
libria, at least for BeO systems.

After it was experimentally established that at temperatures above
2045°C a significant quantity of liquid was formed in the BeO-ZrQO, speci-
mens, the compositions of the liquid which "drained" out of seven samples

were determined. The results are presented in Table 6.3,

Table 6.3. Compositions of Ligquids Drained from
Be0-Zr0O, Wafers

 

 

 

 

Starting Composition of
Composition Maximum Firing Lost Liquid
(mole %) Tem?er?ture (mole %)
°C
Zr0sy BeO Zr0, BeO
10 20 2070 42,8 57.2
10 20 2065 39.0 61.0
5 95 2070 , 41.3 58.7
83.90 16.10 2065 41.5 58.5
89. 24 10.76 2060 41,7 58.3
83.90 16,10 2065 44,0 56.0
89. 24 10.76 2065 38.6 6l.4

 

The question of BeO solubility in ZrO, was not resolved completely.
A mixture of 89.24 mole % ZrO, and 10.76 mole % BeO was heated to a tem-
perature just below the solidus and examined optically and by x-ray dif-
fraction. Both BeO and ZrO, could be seen optically, but it was impossible
to detect the BeO by x-ray diffraction because of the high'concentration

of ZI'02 .

65
Based on these results, it is concluded that the Be0-Zr0O, system has
a simple eutectic configuration, although the possibility of very limited
solubility of BeO in ZrO, cannot be precluded. The first melting occurs

at 2045 * 10°C, and the liquid contains 58.7 * 2 mole % BeO.

66

 
7. ENGINEERING AND HEAT-TRANSFER STUDIES

Boiling~-Potassium Heat-Transfer Experiment

 

A. I. Krakoviak, H. W. Hoffman

Fabrication and construction of the system for studying potassium
under forced-~flow boiling conditions has been com.pleted.l’2 Details of
the assembled uninstrumented and uninsulated apparatus are shown in the
photographs of Figs. 7.1 and 7.2. A temporary test section (a l-in.-o.d.,
0.065-in,-wall, 78-in.-long, type 347 stainless steel tube) has been in-
stalled so that cleanup can be accomplished and experience gained in loop
operation while awaiting completion of the experimental boiler.! This
interim boller is heated by a set of 3-in.-0.d. clamshell heaters rated
at 4 kw/ft. Although the maximum attainsble flux (6 X 10* Btu/hr:ft?) is
well below burnout estimates, it is anticipated that high-exit qualities
will be achieved at liguid flows below 1/4 gpm. Tube~surface tempera-
tures are obtained by a number of Pt vs. Pt~10% Rh thermocouples attached
to the boiler wall and by an optical pyrometer.

The system was cleaned by the circulation of 1200°F potassium at a
1/4;gpm flow rate for 200 hr. Flow and temperature instrumentation were
calibrated during this pericd. It was found that the electromagnetic
flowmeter indicated a rate 6% lower than that measured by liquid-rise
rates using the J-tube liquid-level transducers in both the liquid and
vapor accumulators. An effort to discriminate as to which of the two
measurements is correct is continuing; at the moment it is suspected that
the error is in the electromagnetic flowmeter since some uncertainty
exists concerning the field strength of the magnet.

Preliminary heat-transfer data have been obtained for potassium in
the laminar-flow regime at an average fluid temperature of 950°F. The

results are shown in Fig. 7.3 and compared with the data of Johnson,

 

1ANP Semiann. Prog. Rep. April 30, 1960, ORNL-2942, pp. 64—71.
2 NP Semiann. Prog. Rep. Oct. 31, 1960, ORNL-3029, pp. 62—64.

67
 

      

 
 
 
 
  

UNC LASSIFIED‘
PHOTO 36333

A~
/

o,

g%-,flwr EaA
gy j

b?
LIQUID-VAPOR 1
B

b
/

f

      

&
i .
%

\ SEPARATOR I A

-.' _.E |I I a " \ ..I-.-'I — SURGE TANK

® BOILER
SECTION

-— ‘ — AIR-COOLED »

CONDENSER
ir a-._ :_1'
= i\

Fig. 7.1l. Over-All View of Boiling-Potassium Appsratus Before Instru-

mentation and Insulation.

68
UNCLASSIFIED
PHOTO 36332

— PRESSURE - - o — PRESSUREf
ELEMENT , ELEMENT

LIQUID-FRACTION
ACCUMULATOR :

ELECTROMAGNETIC 18
FLOWMETER £

|, PUMP BY-PASS
7 LINE

 

Fig. 7.2. View of Lower Portion of Boiling-Potassium Apparatus.

69

 

 
UNCLASSIFIED
ORNL—LR— DWG 58622

 

10

8

_ 0.4
Ny, = 0.625 N,
w ©
o
|
=
Q
o
= 4
|_..
=
L
o)
o
=2
=
3
< 2
© Hg AND Pb-Bi EUTECTIC (REF 3)
® K (THIS INVESTIGATION) ‘
{

20 50 100 200 500

Njgg » PECLET MODULUS

Fig. 7.3. Preliminary Data on Heat Transfer with Potassium.

Hartnett, and Clabaugh?® for both mercury and lead-bismuth eutectic in
laminar and transition flow. The agreement is not unreasonable.
Following the laminar-flow experiment (during which instrument cali-
bration was accomplished), the noncondensible cover gas (argon) was re-
moved by evacuation, and potassium boiling at approximately 50% vapor
quality was instituted. The pertinent data obtained in the subseqfient

boiling tests are presented in Table 7.1.

The inlet liquid flow rate ranged from 77 to 98 lb/hr (0.11 to 0.14
ft/sec). Liquid temperatures at the inlet were held at 1130 to 1160°F as
a safety precaution in these initial studies; the saturation temperature
was controlled by varying the exposure of the condenser coils. Large~
scale fluctuations in the boiler tube-wall temperatures, the boiler inlet

and exit pressures, and the boiler inlet flow rate were observed. These

 

*H. A. Johnson, J. P. Hartnett, W. J. Clabaugh, Heat Transfer to
Lead-Bismuth and Mercury in Laminar and Transition Pipe Flow, Inst. Eng.
Res., Univ. Calif. (Aug. 1953).

70
Table 7.1. Data Obtained with Boiling Potassium
in Heat Transfer Apparatus

 

 

Rurnl Boiler Exit Boller Heat Heat Vapor Boiler Exit
N ‘Temperature Flux Balance Quality Velocity
° (°F) (Btw/hr.ft?) (%) (%) (ft/sec)
1 1376 35 100 77 53 107
2 1377 40 400 88 53 122
3 1399 43 000 93 57 130
4 1383 41 900 92 49 127

 

excursions occurred at an average frequency of 3 cycles/min; typically,
the boiler wall temperature rose slowly (over about 20 sec) to a value
about 130°F above the saturation temperature and then fell very rapidly
to the initial temperature. Concurrently, the boiler inlet pressure
climbed from 12 to 27 psia; and the liquid flow dropped from 0.25 to
0.20 gpm. This behavior suggests that the boiling is in the slug-flow
regime. The pressure, temperature, and flow recorders are being modified
so that a more detailed examination of these oscillations can be made., A
30% reduction in the flow (to 63 1b/hr) at a flux of 43 000 Btu/hr-ft? re-
sulted in a boiler-tube temperature rise to 1900°F in approximately 70
sec; safety circuits turned off the boiler power at this point. |
Developmental studies of a high-flux boiler (capability of 5 X 10°
Btu/hr-ftz) have been continued. A satisfactory technique for brazing
the copper disks to the boiler tube has been obtained. Two four-disk
subassemblies were successfully brazed at 1030°C using Nicrobraze 50
alloy preplaced in an 1/8— X l/8—in. circumflerential groove on both the
inner and outer surfaces of each OFHC copper disk. All abutting copper
and stainless steel surfaces were precoated with a 0.001l5-in.-thick
electroless-nickel plate. Diametral clearances after plating were ‘of
the order of 0.002 to 0.004 in., To facilitate outgassing during the
brazing cycle, three equally spaced slots (0.109 in. wide and 4 in. long)
were milled into the outer jacket at the location of the heat-barrier

gaps between the copper disks.

71
Ultrasonic examination of the brazes after a test run indicated good
bonds at both the inner (boiling tube-to-copper) and outer (jacket-to-
copper) surfaces. Some difficulty has been experienced in making the
closure welds for the outgassing slots; cracking in these welds has been
attributed to- contamination by the braze material. Some modification in
the fabrication procedure to eliminate this defect is contemplated. Con-

struction of the final boiler assembly is proceeding.

Thermal Conductivity of Lithium

Jd. W. Cooke

Measurements of the thermal conductivity of molten lithium (99.98
wt % Li) were continued using an improved version of the previously
described” axial heat flow comparison apparatus with compensating guard
heating. Preliminary results are shown in Fig. 7.4.

Changes were made in the apparatus in order to improve the opera-
tional performance and increase the precision and accuracy of the measure-
ments (see Fig. 7.5). An upper heat meter was added, and a coaxial guard»
cylinder fias placed around the test piece. This guard cylinder was con-
structed in three sections, with the upper and lower portions being of
type 347 stainless steel to match the heat meters. The central region was
nickel to approximate the conductivity of the lithium section. The length
of each section was the same as that of the corresponding region of the
test piece, and the three sections were brazed to form a continuous unit.
All thérmocouples, in both the test piece and guard cylinder, were arc-
welded to the bottom of accurately spaced thermowells. Both the main and
guafd heaters were redesigned to give more uniform heating. Highly polished
COpper plates were brazed to the bottoms of both the test piece and the
guard cylinder to minimize the contact resistance to the heat sink.

‘The increased precision in the measurements as the result of these
modifications is apparent in the reduced scatter of the new data (Fig.

7.%4) as compared with the previous results. >Further, the improvements

 

“ANP Semiann. Prog. Rep. Oct. 31, 1960, ORNL-3029, pp. 64—69.

72

 
UNCLASSIFIED
ORNL -LR-DWG 58623

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

45 |
KUTATELADZE ef ai(REF.7) ' A
40 | = =— CANEL (REF. 10) A
------ WEBBER efa/ (REF.9) A
® MILLER AND EWING {REF.5)
g A ORNL (PREVIOUS)
T A ORNL (CURRENT)
x 35
> MELTING
@ TEMPETATURE A
—
g
- | -
-
5 30 | - ——
o - A A -
] ‘
Q -~
4 | T el ”"- A
a | sl
E 25 1 ) -..T'
uIJ | //4 ‘~~‘
— r ‘\‘
\5
20 | A L
o | |
200 400 600 800 1000 1200 1400 1600 1800

TEMPERATURE (°F)

Fig. 7.4. Summary of Data on the Thermal Conductivity of Lithium.

have permitted a closer check on the accuracy of the measurements. The
magnitude of the heat flow in the lithium sample can be better established
through a comparison of the thermal fluxes in the upper and lower heat
meters; the maximum deviation for the present investigation was less than
3%. An "in place" relative calibration of the thermocouples can be made
at periodic intervals at temperatures up to 1200°F. One such check made
following a measurement at 1490°F gave an average deviation of 0.45°F at
1200°F between nine thermocouples in the test piece. The better knowledge
of the thermocouple locations in the guard cylinder allows closer esti-
mation of the radial heat loss; in the current measurements this was found
to be less than 1% of the axial heat flow in the sample region. By fixing
the locations of the thermocouples in the thermowells more positively, it |

has been possible to establish the interfacial temperatures wifih'greater

73
UNCLASSIFIED
ORNL-LR-DWG 58255

 

GUARD
HEATERS

{4
2n.TYP

|

L0

 

|

|

 

 

 

 

)

<

 

 

 

 

 

 

Y7 7 7 7

77T 77

 

 

|

o

 

ml

 

 

L

 

V0727773

NS W 7

\

 

 

L

 

 

 

S\

 

 

 

 

ZS)

-

 

I %

L — EXPANSION TANK

| Al,05 REFRACTORY
HEATER SUPPORT

L MAIN HEATER
L— TIE-DOWN ROD

L— UPPER GUARD RING
UPPER HEAT METER

 

i

-

o

S

 

 

 

A A 7 A NS Y NP 7 7 7]

X7z

7

 

 

 

 

7

 

 

 

 

 

 

 

 

A
[

 

N N

 

7

P

THERMOCOUPLE
LOCATION PLANES

SAMPLE CONTAINER

|- SAMPLE GUARD RING

LOWER HEAT METER
/

/ LOWER GUARD RING

ISOTHERMAL COPPER
" COOLING PAD

/SINK HEATER PLATE

SINK COOLER PLATE

 

 

 

 

 

NN N N NN

NN NN

=

 

7

 

 

 

 

Z

 

 

Fig. 7.5. Modified Axial-Heat-Flow Thermal-Conductivity Apparatus.

T4

 
UNCLASSIFIED

ORNL-LR-DWG 58624

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1100 \\\
@
®
1050 \\
® ® PRIMARY THERMOCOUPLES
A SECONDARY THERMOCOUPLES
1000
~ 950 -
e \/A’_‘tNTERFACE_i'g F
L
" \
D
< o
5 900 A
a
= \\i
wl
- \
\.\
850
\/A’mTERFACE =0.4°F
UPPER HEAT METER LITHIUM SAMPLE LOWER HEAT METER
. (TYPE 347 —— -— (TYPE 347 —=f
STAINLESS STEEL) \ STAINLESS'STEEL)
800 N
N
750 \
Q\\
0 {2 3 4 5 6 7 8 9
AXIAL POSITION (in.)
Fig. 7.6. Typical Axial Temperature Profile, Lithium Thermal-

Conductivity Study.

 

75
certainty. A typical axial temperature profile, shown in Fig. 7.6, indi-

cates that the Interfacial resistances between the sample and the heat

meters were quite small. This agrees well with other studies with alkaline

liquid metals.”*® |
The results of the recent measurements with lithium are given in

Fig. 7.4 in comparison with earlier data and with the results of other

investigators. It may be seen that the new data agree to within an average

deviation of +1.4% with values reported by Kutateladze et al.? obtained by

a method-of successive stationary states. (The results of Kutateladze

et al. for potassium, sodium, and mercury compare closely with the data

8 for the same metals.) In contrast with the

reported by Ewing and Miller
data for sodium and potassium, and earlier results for lithium (Webber

et al.?), the more recent data on the lithium conductivity show a positive
dependence on temperature, that is, the conductivity increases with in-
creasing temperature. For the present data, this variation is not as
pronounced as with the earlier ORNL measurements or with the values calcu-
lated at CANEL'? from electrical-conductivity measurements. While final
énaiysis of these results with respect to establishing the experimental

error is incomplete, it is expected that further corrections will have

only a second-order effect.

 

R. R. Miller and C. T. Ewing, Naval Research Leboratories, Washington,
D. C., private communication.

6¢. T. Ewing, J. A. Grand, and R. R. Miller, The Thermal Conductivity
of Sodium and Potassium, Naval Research Laboratory Report 3835 (August
1951 ).

7S. S. Kutateladze et al., Liquid-Metal Heat Transfer, pp. 2-3, Atomic
Press, Moscow, 1958, translated by Consultants Bureau, New York, 1959.

8C. T. Ewing and R. R. Miller, Thermal Conductivities of Mercury and
Sodium-Potassium Alloys, J. Phys. Chem. 59, 524 (1955).

H. A. Webber et al., Determination of the Thermal Conductivity of
Molten Lithium, Trans. Am. Soc. Mech. Engrs. 77, 97 (1955).

10yyclear Propulsion Program, Engineering Progress Report, Jan. 1,
1960 — Mar. 31, 1960, Pratt & Whitney Aircraft, CANEL Operation, Report
PWAC-601, p. 92 (Secret)

76

 
8. RADIATION EFFECTS

Radiation Effects on Columbium-Zirconium Alloy

N. E. Hinkle, J. C. Zukas, J. W. Woods

Two experimental assemblies containing tube~burst specimens of Cb—1%
Zr alloy were irradiated in the poolside facility of the ORR. The tube-
burst specimens in these assemblies were tested at 1800°F in a helium
atmosphere. Because of early failure of the heaters, the specimens in
the first experiment were not ruptured; Data on gamma heating rates were
obtained, however, and were used to affect successful operation of the
- second experiment.

The second experimental assembly is shown in'Fig. 8.1 before the
final enclosure was installed. This assembly contains eight specimens
and two gettering trains. The specimens and associated pafts are shown
in Fig. 8.2, before and after joining. Brazing and welding of tube-burst
test assemblies are performed in the vacuum chamber of an electron-beam
welding machine in order to eliminate contamination. The windings of the
three-section furnaces shown in Fig. &.1 are of Nichrome V wire sheathed
in stainless steel with MgO insulation. The insulated furnaces are
wrapped with tantalum foil to obtain additional mass (for gamma heating)
and thermal radiation shielding. | | . ,

In order to obtain reliable stress-rupture data for Cb—1% Zr alloy,
it is very important that the tests be performed in = helium.atnosphere
of the highest purity. Therefore a helium purification system was pro-
vided. All organic materials were eliminated from the test system by
using ceramic insulation on the thermocouple and power leads from the
in-pile assembly to the out~of-pile junction box, and the experimental
assembly was baked out at about 275°F in a vacuum. The assembly was then
purged and evacuated a number of times and finally filled with helium.
After insertion in the reactor, the experimenfal assembly was connected
to a helium supply pénel by means of quick-disconnect fittings which es-

sentially eliminated the possibility of admitting any air to the system.

77
84

UNCLASSIFIED
PHOTO B2727

 

 

Pig. 8.1.

Experimental Assembly for the Second Cb—1% Zr Alloy Stress-Rupture Experiment.
 

A

 

UNCLASSIFIED
PHOTO 3542435

Fig. 8.2. Columbium—=Zirconium Alloy Tube-Burst Specimen and Associated Test Capsule Parts. The

furnace shown is one of two types used in the first columbium experiment.
form, platinum alloy windings, and Al,03 insulation.

It consists of a tantalum

 
The helium supply panel includes helium purification trains of charcoal
at liquid nitrogen temperature and calcium at 600°C. The two end furnaces
of the assembly (see Fig. 8.1) contain zirconium foil for scavenging the
oxygen and nitrogen that may be released as the various materials outgas
during the experiment. The specimens were tested at a rather high stress
in the second experiment, and all ruptured within 60 hr. The data for
these specimens and the comparative out-of-pile data from tests in vacuum
are shown in Table 8.1 and in the graph of Fig. 8.3. The in-pile rupture
strength was about 10% less than the out-of-pile strength. |

It is planned to test two of the specimens in the next assembly at

1800°F to obtain longer times to rupture and to obtain data on the

Table 8.1. Results of Stress-Rupture Tests
of Cb—1% Zr Alloy Tubing at 1800°F

 

 

. Irradiation at - Time to
Sp;gj..men %;IS??S ' Rupture Rupture
(Mwhr) | (hr)
(a) 31 000 23
19-6 28 000 © 100 1.5
19-1 28 000 1900 3.8P
(2) 27 500 149¢
19-5 26 000 150 3.5
19-3 26 000 900 28
() 25 000 - 100
19-7 24 0002 1100 35
19-2 24 000 1700 55
18-7 23 000 5500 >160€
19-8 22 500 870 27
19-4 22 500 1400 45
(2) 22 500 6950

 

a'Ou.‘t-of-pile test data.

bSpecim.en stressed at about 1750°F for 57 hr before
increasing the temperature to 1800°F.

| cSpecim.en leaked and was repressurized intermittently
during the test.

dThe calculated minimum stress during the test was
19 000 psi.

eSpecim.en did not rupture during the test.

80
UNCLASSIFIED
ORNL-LR-DWG 55994R

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

50,000

40,000
_ — 2 TH
7 30,000 = TP = e Al L ~ [ H
n :_ m e L Lol -.l—"--.._ '_:E
fi = o IN-PILE, EXP. 1 [il° I \\"‘""“t 1
& 20,000 e IN-PILE, EXP 2 |
5 A 57 hr AT 1750°F

& OUT-OF-PILE
L
0 ) 10 100 1000

TIME TO RUPTURE (hr)

Fig. 8.3. Stress Versus Time to Rupture for Cb—1% Zr Alloy (Pratt
& Whitney Material) at 1800°F in Helium. The horizontal arrow denotes
"not ruptured.'" The vertical arrow denotes a leaky specimen.

remaining specimens at about 2000°F. This experiment will utilize the
last of the material provided by Pratt & Whitney. A second batch of Cb—1%
Z2r alloy, purchased by ORNL, 1s available for future experiments.

Thé first experimental assembly was examined to determine the cause
of the furnace failures and to observe the condition of the speéimens with
respect to impurity contamination. It was found that excessive thermal
expansion had bent the main structural element and obviously damaged the
furnaces and the furnace leads. Possible, slight, surface contamination
was observed and will be investigated in more detail.

The initial postirradiation examination of the specimens used in the
second experiment has shown that six of the specimens exhibited fractures
similar to out-of-pile fractures. These six specimens were split open
with a longitudinal crack. The specimen surfaces appear quite bright,
indicating very little contamination. It is interesting to note that
neither the thermocouples nor the furnaces were damaged during the blow-

out type of rupture, although the furnaces were bent out of their normal

shape.

Radiation Effects on Stainless Steel and Inconel

N. E. Hinkle, J., C. Zukas, J. W. Woods

An experimental assembly was irradiated that contained ten tube-

burst specimens of type 304 stainless steel. The steel specimens were

81
tested at 1500 and 1600°F in air. The results of these tests and compara-
tive out-of-pile test data are given in Table 8.2. Results of earlier
tests indicated that the time to rupture of this material was reduced Dby
a factor of 2 by irradiation at 1500°F, and these data continue to show
the same effect. Sufficient out-of-pile test data at 1600°F are not yet
available for comparison with the in-pile test data.

In a previous report,1 it was suggested that the presence of boron

(specifically Blo) might be responsible for the reduced rupture strength

 

lANP Semiann. Prog. Rep. March 31, 1959, ORNL-2711, p. 60.

Table 8.2. Results of Stress~Rupture Tests of Type 304
Stainless Steel Tubing at 1500 and 1600°F in Air

 

 

Test Specimen Stpess Irradiation Dose Time to
Temperature T No (psi) at Rupture Rupture
(°F) ' P (Mwhr) (hr)
1500 310 6300 5172
318 6300 %87%
537 6300 | 183
319 5800 6912
239 5250 10252
73 5250 1134
536 4700 857
240 4200 24,502
72 4200 1670
17-3 4000 36 900 g51p
17-10 3500 50 080 >1600¢
17-1 3000 : 50 080 >1600¢
17-9 3000 50 080 >1600¢
1600 17-8 5000 3 750 49
360 3700 - 265
17-5 3000 13 500 414
17-6 2600 27 100 870
17-4 2200 43 900 1430
17-17 2200 47 400 >1505¢

 

“Specimens 6 in. long; remaining specimens 2 1/2 in. long.
Specimen received 11 500 Mwhr of irradiation prior to stressing.

cSpecim.en did not rupture.

82

 
of Inconel because of the formation of helium and lithium at the grain
boundaries during neutron bombardment. In order to invéstigate this idea,
the Laboratory purchased six heats of Inconel containing various amounts

of boron (natural and isotopic).? Testing of these materials has begun
with irradiation of two specimens of each of five of the heats. These
tests are being conducted at 1500°F in an air atmosphere. The experimental‘
assenmblies were constructed the same as previous assemblies for testing
standard heats of Inconel. The test.procedures were described previously.3
The experimental assembly presently being irradiated is shown in Fig. 8.4.
The specimen test-parameters, the data obtained thus far, and the avail-

able comparatifie out-of-pile test data are listed in Table 8.3, For

 

2ANP Semiann. Prog. Rep. Oct. 31, 1960, ORNL-3029, p. 75.
3ANP Semiann. Prog. Rep. March 31, 1958, ORNL-2517, p. 58.

Table 8.3. Rupture Characteristics of Inconel Tubing Containing
Controlled Amounts of Boron in Tests at 1500°F in Air

 

 

Heat Boron Content Specimen Stress Irradiation at Time to
No (ppm) No (psi) Rupture Rupture
’ ; ’ (Mwhr) (hr)
0B <10 (a) 6000 155
(a) 5000 320
28-6 5000 4 250 140
(a) 4000 1250
28-2 2000 14 250 475
i (a) 3000 1950
108 100 28=-3 6000 .1 620 52
(a) 5000 580
28-8 5000 1 800 58
(a) 4000 1750
2B 20 (a) 5000 115
(a) 4000 880
(a) 4000 4200
4B11P 40 (a) 5000 105
28-9 4500 5 650 186
(a) 4000 250
28-5 3200 18 600 620
{(a) 3000 980
6B11P 0 28-4 5500 1 750 56
(a) 5000 610
28-10 4500 4 050 133
(a) 3800 1500
6B10° 60 (a) 5000 195
28-1 4500 2 450 80
(a) 4000 255
28-7 3500 6 930 229

 

aOu_t-of-pile test specimen.
PBoron enriched to 98.5% B,

®Boron enriched to 95% B1O,

83
 

78

 

UNCLASSIFIED
PHOTO 53278

'
2
Oax WIDGE NATIONAL LABORATORY

 

Fig. 8.4. Experimental Assembly for a Typical In-Pile Stainless Steel or Inconel Tube-Burst Ex-
periment. The "Y" shaped tubing at the bottom left is part of the air-cooling manifold for the speci-
men cooling system.
comparison,

Inconel.

from the new data.

Figs. 8.5 and 8.6 summarize the data on standard heats of

No conclusions regarding the effect of boron can yet be drawn

- A second experimental assembly is under construction,

and it is hoped that with this additional data some conclusions can be

drawn.

8000 — T T T T T T

UNCLASSIFIED
ORNL-LR-DWG 55999R

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

— *
L IRRADIATION TIME AND {) FRACTION-
6000 o ! oF g'° éfiRNUP : .
. e " . :
o-hfi‘ (ph——8—® .04
_. 4000 100 (7057~ o~ ¢ .
3 550 (30) T ::::::::::
< 1000 (55 S T
§ o IN-PILE TEST |
@ s IRRADIATED AT TEST TEMPERATURE \
fy 2000 PRIOR TO STRESSING
» STRESSED FOR 48hr PRIOR TO NEUTRON |
BOMBARDMENT . .
s IRRADIATED AT TEST TEMPERATURE, POSTIRRADIATION TESTED e °
¢ OUT-OF-PILE TEST | | |4J | |‘ [ 1
1000 P _
30 50 100 200 500 1000 2000 3000 -

TIME TO RUPTURE (hr)

Fig. 8.5. Stress Versus Time to Rupture for Inconel (INCO Heat No.
NX 8962) at 1500°F in Air.

STRESS {psi)

Fig.
fication,

UNCLASSIFIED
ORNL-LR-DWG 56000R
8000

6000

4000

o IN-PILE TEST
® QUT-OF-PILE TEST

2000

 

1000 . -
10 20 50 100 - 200 © 500 1000

TIME TO RUPTURE (hr)

8.6. Stress Versus Time to Rupture for Inconel (CX-900 Spec1-
INCO Heat No, NX 5757) at 1500°F in Air. ‘

85
Beryllium Oxide Irradiation Studies

 

R. P. Shields, J. E. Lee, Jr.

Serious physical damage to BeO which had received fast-neutron dosages
of the order of 102! neutrons/cm? in the ETR was revealed during post-

irradiation examination of test specimens. The damaged specimens constitued

a portion of the 57 BeO pellets, 1 in. in length and 0.4 to 0.8 in. in
diameter, which were contained in 16 capsules distributed among five
separate irradiation assemblies of the ORNL-41 test series. A compre-
hensive rangé of exposures was achieved at a significant variety of tem-
peratures.

The BeO has been recovered from the capsules 6f all five assemblies.
The disassembly and inspection df the last three assemblies was carried
out in the BMI hot cell facility. Visual inspection, macrophotography,
and physical dimension measurements of the specimens have been completed.
Metallographic examination is complete for the specimens used in the early
experiments and exploratory gas analyses and x-ray diffraction studies have
been performed. A major portion of the thermal flux dosimetry analysis is
completed, and corresponding fast flux determinations are under way. A
survey of the temperature‘data'and thermal barrier gas composition changes
is being made to evaluate thermal conductivity changes observed during the
progress of the irradiations.

Damage observed in the BeO that received the higher dosages varied
from minute cracks to gross fracture and disintegration into powder. The
first cracks were found to appear after an estimated irradiation dose of
5 to 8 X lOzo'neutrons/cm? (>1 Mev) in the temperature range‘of 700 to
1000°C, as shown by Fig. 8.7, which is a photograph of sample 3-52. The
BeO samples disintegrated to powder or to easily crushed material when
irradiated at 120°C to a dose of approximately 1 X 1021 neutrons/cm2 (>1
Mev). A sample of the disintegrated material found in capsule 5-7'ié
shown in Fig. 8.8. The capsule cladding increased in diameter up to 3%,
and, as shown, there was fracture or disintegration of the BeO. One cap-

sule burst its cladding after an irradiation dose of 10! neutrons/cm?

86

 
 

 

UNCLASSIFIED
RMG {7281

RMG {7282

 

RMG 17283

Fig. 8.7. Cracked BeO Pellet 3-52 That Was Ir-
radiated to a Fast-Neutronh Dose of ~8 x 1029
Neutrons/cm? in the Temperature Range 700 to 1000°C.
5%

 

UNCLASSIFIED
PHOTO 54274

 

Fig. 8.8. Disintegrated and Frac-
tured Material from BeQO Irradiation Cap-
sule 5-7 After a Fast-Neutron Dose of
1 X 10?r Neutrons/cm? at 120°C. 4X

 
(>1 Mev) at a temperature of 120°C. The partial view near one end of
capsule 5-7 shown in Fig. 8.9 indicates the type of rupture which occurred
in this unit. A summary of the specimen conditions and irradiation tem-
peratures and an estimate of neutron flux are given in Fig. 8.10.
Examinations and analyses of the specimens are continuing at both
ORNL and BMI in order to gain as much quantitative information as possible
regarding the specific nature of the observed damage and to determine the
contributions of dosage and temperature. New information will be forth-
coming after irradiation of a sixth experimental assembly now at the ETR.

This unit contains 1.18-in.-diam specimens 3 in. long that are both

UNCLASSIFIED
PHOTO 654275

 

Fig. 8.9. Ruptured Cladding of Capsule 5-7. 4X

88

 
unclad and encapsulated. This experimental assembly is scheduled for
removal from the reactor toward the end of the current fiscal year.
The integrity of the BeO irradiated in these tests has been demon-

strated to be less than had been anticipated. A need for further and

’

intensive study is indicated.

UNCLASSIFIED
ORNL-LR-DWG 56040A

EXPERIMENT NUMBER

 

 

NO. |
458°C \
(0) \

 

 

 

NO.3 [

 

 

 

(0)

 

 

 

 

 

 

 

 

 

 

 

(0) SOUND SPECIMENS _
(1) FIRST CRACKS NO. 6 [
(2) GENERAL FRACTURES 950°C
(3) POWDER PRESENT (2) @ l

 

 

 

 

 

 

 

 

 

 

 

 

NO.7 [ NO.7 [
120°C |2 140°C |
0 L (3) B
a1 1783 2230 1343 10032 O 2.0 40 60 x10'4
EXPOSURE (Mwd) FAST FLUX (>1Mev)

203x10% 88x10% 1.4x10% 5.6x10% 4.95x10%
IRRADIATION TIME (sec)” :

Fig. 8.10. BeO Irradiation Conditions of ORNL-41 Series Experi-
ments.
PART 2. SHIELDING RESEARCH
9. DEVELOPMENT OF REACTORS FOR SHIELDING RESEARCH

Tower Shielding Reactor II (TSR-II)

 

L. B. Holland

Upon completion of the previously described critical experiments,1'2

the TSR-II was disassembled, all the temporary equipment which had been
installed for the critical experiments was removed, and the reactor was
reassembled for operation. As a result of the shakedown runs of the com-
plete system which continued through December 1960, some components had
to be modified before the reactor was placed in operation in January and
subsequently operated at a 100-kw power level. Minor design changes are
still contemplated to provide more reliable operation. These changes are

described below along with some results obtained during the shakedown runs.

Reactor Mechanical System

 

The changes in the control mechanism housing assembly were described
previously,2 and a spare assembly is now being fabricated according to
this design. The control mechanisms for the spare assembly are identical
with those presently in the reactor; however, closer tolerances were held
in the fabrication of the new mechanisms and more uniformity has been ob-
served in the release and travel of the control rod under scram conditions
on the test stand. The first mechanisms were placed in the reactor in
November 1960 and have been operating in a satisfactory manner without
attention.

In an effort to help maintain a clean system, a 20 000-gal holdup
tank has been added to the reactor system for demineralized water storage.
Several filters have also been added at crucial points, and it is hoped
that these filters, combined with a completely closed water system, will

prevent the buildup of crud in the control mechanism lines.

 

LANP Semiann. Prog. Rep. April 30, 1960, ORNL-2942, pp. 112-14.
2ANP Semiann. Prog. Rep. Oct. 31, 1960, ORNL-3029, p. 82.

93
Two sections of 6-in. hose supplying cooling water to the reactor
have failed under full-flow and full-pressure conditions. The failures
were slow leaks which would not have Jjeopardized the reactor safety after
high-power operation. The vendor has examined the damage and has recom-
mended an improved flange design on replacement hose which will be de-
livered in May. For the present the reactor is operating at reduced flow
with temporary hoses in place.

An operational error caused the collapse of two ionization chamber
wells. The cQoling water pumps were started with the exit line from the
reactor closed. This resulted in overpressurization in the core and failure
of the chamber well., BSince the failure caused a pressure change to at-
mospheric pressure, the chambers were not damaged. The chanbers have been
temporarily insfalled in removable wells in the annular region and are op-
erating satisfactorily. A tag-out procedure has been initiated to prevent
a recurrence of this type of failure. During a scheduled shutdown in June,
the three original chamber wells in the inner region will be replaced with
removable heavy-walled wells.

As previously described,? the TSR-II is suspended from a support
platform so that the reactor can be rotated together with any shield for
beam or shield mapping measurements. At first, considerable difficulty
was encountered in rotating the reactor and shield. A hydraulic torque
booster (which follows the motion of a positioning motor) is used to rotate
the reactor. Originally the hydraulic pump which supplies power to the

torque booster was remotely located, but it had to be mounted on the reac-
.tor platform before it would operate correctly. Electrical cables to the
reactor chambers and drive motors were too stiff to permit 360-deg rota-
tion of the reactor. These cables have been stripped of the heavy insula-
tion, and lighter insulation has been used to permit full 360-deg rotation
for complete mapping of the leakage radiation from the Pratt & Whitney-
designed shield (see chap. 12).

 

JANP Semiann. Prog. Rep. Sept. 30, 1958, ORNL-2599, p. 186.

9%

 
Reactor Controls

 

For the most part, the difficulties in the reactor control system
during the shakedown runs were concerned with minor adjustments or cable
connector troubles. The notable exceptibn is the seat switch system
which, because of its complexity, could not be kept in proper adjustment.
The present system senses a pressure change when a control plate moves a
leaf spring away from an orifice through which a dc pump is forcing a small
flow of water. The system is further complicated because the pressure
switch requires an air reference which equals the core water pressure at
all flow rates and reactor elevations.

Since the orifice and leaf spring part of the circult appear to op-
erate satisfactorily, a flow-type sensing device has been designed and a
prototype model tested. In the prototype modél the flow through the
sensing device when the orifice is open lifts a éonducting ball from a
set of contacts and thereby breaks a low-voltage dec circuit. The only
auxiliaries required will be a dc¢ pump to force water through the orifice.
This system will be independent of both the water flow rate throfigh the’

reactor and the reactor elevation.

Flow Measurements and Fuel Plate Temperatures

 

Measurements were made of the full flow through the reactor and of
the flow distribution in the annular fuel eleménts under'fullelow and
half-flow conditions. It was determined that the maxXimum flow that could
be achieved with all the baffle plates in place in the reactor was 810 gpm
with a pressure drop of 31 psi across the core. The reduction in total
flow from the design flow of 1000 gpm required new studies of both the
inner and annular regions.

Since the optimum flow distribution in both the upper and lower fuel
elements in the inner region under all flow conditions is nearly uniform,
the reduction in total flow has no appreciable effect on the baffle design.
Efforts to reduce the pressure drop through the upper fuel baffle plate
met with no success. A long series of experimental measurements have
provided a baffle design which provides uniform flow distribution. The

baffle plate is being fabricated and will be tested during the month of May.

95
The water velocity outside the U237 .1o0aded spherical cover plates on
the control mechanism housing varies with position; therefore, a calcula-
tion has been made to determine whether the heat transfer coefficient at
the point of lowest water velocity is sufficient to remove the heat gener-
ated when the heat is removed from the external surface only. These calcu-
lations were based on a uniform flow distribution in the upper and lower
fuel elements for the measured total flow through the reactor and the cal-
culated heat generation in the fuel cover plates after several hours at 5
Mw. For a heat generation rate of 101 000 Btu/hr.ft?, the surface tempera-
ture of the cover plates at the point of lowest water flow was calculated :
to be 217.5°F, which is well below the saturation temperature of 259°F for .
a. core pressure of 20 psig. _

For the annular region it was necessary to measure the flow distribu-
tion through the fuel elements and compare the measurements with the calcu-
lated value of the flow required in each channel to keep the water below
the saturation temperature for the minimum operating éore pressure of 20
psig.’ The baffle plate design for the annular elements was determined by
flow measurements on a single plastic mockup of an annular fuel element
which had flat rather than curved plates. The final determination of the
flow distribution through the fuel plates of the annular fuel elements
was made with all the fuel elements in place in the reactor. For full
flow and half flow conditions, a salt solution was injécted into the system
ahead of two conductivity probes which were separated by a known distance.
Measurements of the time for the solution to pass between the probes gave
a value ofrthe flow in each channel, Measurements were made in every other
fuel element and in enough channels in these elements to determine the
flow distribution as a function of channel number. The results corroborate
the single element measurement and show that the flow is sufficient to keep
the plate surface temperature in all channels below 225°F. The core pres-
sure should always exceed 20 psi so that local boiling should not occur

below 259°F. Data of Bernath® show that film boiling will not start unless

 

“Chem. Eng. Prog. Symposium, Series No. 30, Vol. 56, pp. 95-116,
1960. | -

26

 
this value is exceeded by 60°F, which provides a safety margin of 94°F.
There is, however, one channel in the annular elements where the velocity
dropped below the value reguired, but the flow velocity in the adjacent
channels is much higher than required, so the over-all effect is considered
acceptable.

The rather severe winter weather provided a good check of the freeze-
up protection of the water system. For the most part the automatic de-
vices protected the system adequately. Some wire heatefs and insulation
were added where either the water velocity stays low or stagnates because

of instrumentation.

Nuclear Measurements

 

The reactor was reassembled with the spherical cover plates on the
control mechanism housing. The cover plates are loaded with 160 g of
U235 (these plates were erroneously referred to as 116-g plates previ-
ously5). Using these fuel plates limits the excess loading of the reac-
tor to 0.97% £k/k when the reactor is water reflected and at 20°C. (The
present operating limit is 1% Ak/k.)

Attempts to measure the change in reactivity with loss of water
reflection during the critical experiments showed a change of ~0.08%
Ak/k.6 This was not a clean experiment, however, because the reactor was
supported in the empty lead-water beam shield, and approximately l/2-in.
water and some steel were present outside the pressure vessel. The reac-
tor can now be operated in air with no water outside the pressure vessel.
The excess loading drops 0.39% Ak/k to 0.58% Ak/k when the reactor is re-
moved from the pool. As far as excess loading is concerned, the lithium
hydride-uranium shield which was designed by Pratt & Whitney (see chap.
12) is the same as a water reflector.

The absolute power calibration of the TSR-II has not hbeen made, but

it is proposed that the measurement be made by comparing the heat output

 

SANP Semiann. Prog. Rep. Oct. 31, 1960, ORNL-3029, p. 86.
6Tbid.

o7
of the reactor with that of a 40-kfi sufimersible heater. The heater can
be closely coupled to the reactor, and the temperature rise as a'function
of time of the small amount of water in the pressure vessel can be com-
pared for the reactor and the heater. The eiectrical input to the heater
can be measured quite accurately, and corrections can be minimized by op-
erating over a small temperature range where heat input due to pumping
equals heat loss from the system.

Experiments have been run to determine the effect of the regulating
rod position on the reactor neutron and gamma-ray leakage. A change in
leakage of 12.7% was observed for the full regulating rod travel. This
change is due to the location of the ionization chambers with respect to
the regulating rod. As a result of this variatioh it is necessary to keep
the regulating rod position fixed. Thus far this has been accomplished
by holding the system temperature constant.

It is planned to examine the effect on leakage of the shim rod posi-
tion. This will be accomplished by operating the reactor over a tempera-
ture range with the regulating rod kept fixed.

Measurements have been made of the linearity of the reactor power-
1evel-cdntrolling instruments. For these runs, three BF3; counters were
used in addition to the reactor instruments. Preliminary runs show varia-
tions of 10% in the linearity. OStudies are under way to improve the power
level instrument so that such large corrections will not be regquired for

all measurements.

98

 
10. DEVELOPMENT OF RADIATION DETECTION EQUIPMENT

Gamma-Ray Spectroscopy

 

The Model IV Gamma-Ray Spectrometer (G. T. Chapman)

 

Considerable effort and time has been spent in investigating the ef~
fectiveness of the Model IV gamma-ray spectrometer shield and collimator.t
A study of the backgrounds found with the spectrometer in a reactor radia-
tion field has indicated that it is necessary to operate the spectrometer
at distances no closer than 5 ft from the reactor to reduce the gamma-ray
background within the shield. (Background is used here to mean spec-

- trometer response with the lead collimator in the housing closed by a
rotatable lead filler.) Consequently, a 5-ft-long air-filled cone has
been attached to the front of the spectrometer shield to allow measure-
ments of radiation leaving the surface of the reactor. It has also been
shown that.the use of such a cone considerably reduces the number of de-
graded gamma rays due to scattering in the lead collimator of the spec-
trometer housing. The tests for optimizing the cone-collimator system
have not been completed.

The use of an air-filled cone permits the utilization of water as
.additional shielding between the reactor and the spectrometer, and there-
fore essentially all the remaining background measured in the spectrometer
is due to radiocactivity in the components of the shield or crystal. Figure
10.1 shows the background measured under various conditions, with the
energies of the gamma rays giving rise to the predominant peaks indicated,
although the sources of these gamma rays have not been satisfactorily
identified. It is evident that the background below about 4 Mev origi-
nates from long-lived isotopes within the shield of the spectrometer, but
the source of the high-energy gamma ray at about 6.87 Mev is as yet un-
determined.

The large composite NaIl{T1) crystall obtained for use with the spec-

trometer has been tested further. This composite crystal consists of two

 

1ANP Semiann. Prog. Rep. Oct. 31, 1960, ORNL-3029, p. 118.

99
UNCLASSIFIED
2-01-058-0-591

BACKGROUND ~ 16 hr AFTER LAST

10°

COUNTS PER CHANNEL PER SECOND

EXPOSURE TO REACTOR
SPECTROMETER UNDER WATER,

COLLIMATCGR CLOSED

BACKGROUND DURING
REACTCR OPERATION

SPECTROMETER UNDER
COLLIMATOR CLOSED

&

BACKGROUND ~ 4 hr
AFTER LAST
EXPOSURE TO REACTOR
SPECTROMETER N AIR,
COLLIMATOR CLOSED

BACKGROUND ~ 4hr AFTER LAST
EXPOSURE TO REACTOR
1072 SPECTROMETER IN AR,
COLLIMATOR OPEN

REACTOR POWER: 2 watts

3073
0 20 40 60 BO 100

CHANNEL NUMBER (RELATIVE PULSE HEIGHT)

Fig. 10.1.

ated with the Model IV Gamma-Ray

Spectrometer.

100

 

Background Associ-

crystals, one 9 in. in diameter and
5 in. long and the other 9 in. in
diameter and 7 in. long, optically

glued together. Based on previous

measurements and calculations,? a
l-in.-diam, 2-in.-deep well was
drilled into the.crystal to reduce
the "tail" of the pulse-height dis-
tributions. Unlike the previous

crystal with a conical end, this

crystal has shown a uniform response

to gamma-ray energies as great as -
6.1 Mev.

response of the crystal to colli-

Figure 10.2 shows the

mated gamma rays with energies rang-
ing from 0.511 to 2.754 Mev. As
anticipated, the well has simpli-
fied the Compton distribution in
the "tail" of the higher energy dis-

tributions.?

The crystal has been
in almost constant operation for
approximately six months without
noticeable deterioration of the

glued interface.

Unscrambling of Continuous Scintil-
lation Spectra (W. R. Burrus)

 

The problem of "unserambling"
gamma-ray spectrometer data was in-
troduced previously.2’% Difficulty
in unscrambling results from the
fact that while the actual spectra

of interest are a superposition of

 

2Ibid., pp. 102-8.

 
a continuous part and a nearly monoenergetic part, spectrometers always

yield a finite set of numbers as a result of a measurement. In the simple

 

*W. R. Burrus, Unscrambling Scintillation Spectrometer Data, IRE
Trans. Nuclear Seci., NS-7(2-3), pp. 102-11, 1960.

“ANP Semiann. Prog. Rep. Oct. 31, 1960, ORNL-3029, p. 121.

UNCLASSIFIED
2-01-058-0-590

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

100
A\
0.511 Mev T
90
0.662 Mev
80
o Na’* DECAY
& Na?? DECAY
® Cs'>7 pECAY
o
)
3]
Y
o
-
L
=
3
T 50
o
a
LJ
¢ |
2 4 )
E 40
>
O
o
30
7 1.277 Mev 1.368 Mev
. \L £
2.754 Mev
10 £ aj
'éf X&\kn Oggeoap
0 ¥ AOoYcRpnam 0o o
0 10 20 30 40 50 60 70 80 90

CHANNEL NUMBER (RELATIVE PULSE HEIGHT)

Fig. 10.2. Response of the Large Composite NaI(T1l) Crystal to Col-
limated Gamma Rays.

101
threshold foil "spectrometer," the set of numbers is just the counting
rate of the foils after a certain amount of waiting time. For the NaI(T1l)
scintillation spectrometer, the numbers are usually the number of counts

in the channels of a multichannel analyzer.

In order to understand the results of a spectrometer experiment, it
is necessary to obtain the actual spectra from the measured set of values;
that is, the measured set of values must be unscrambled. The solution to
the unscrambling problem has proved unusually difficult and haé been a
great stumbling block to otherwise straightforward experiments. The solu-
tion is complicated by the fact that the number of counts measured actually
has a statistical distribution, and the average cannot be determined
“exactly by any experiment. In many of the previous attempts at unscramb-
ling, it was found that the results were extremely sensitive to very small
statistical fluctuations in the input data. The problem is further com-
plicated by the difficulty of measuring the response of the spectrometer.

| The statistical problem has been well recognized, however, even if
not completely'resolved, but another basic problem has recently become
apparent. The second problem is introduced by the finite nature of the
spectrometer output. Since the desired radiation spectrum is continuous,
it cannot be determined exactly by a finite number of points, thus with
actual instruments which yield c,;, c2, ... c with n seldom larger than
a few hundred, an uncertainty is present. This effect has usually been
completely ighored, yet it may actually be of the same order of importance
as the statistical uncertainty. Thus in this report, it is assumed that
the exact average number of counts ¢y, ¢, ... ¢ is known and that the
response of the spectrometer is known exactly. An effort is then made to
see how closely the continuous spectrum g(e) can be determined. It is
hoped that this treatment of the deterministic part of the unscrambling |
problem will lay the ground work for a realistic and rigorous treatment

of the complete probiem.

Mathematical Formulation. The response of a scintillation spectrometer

is related to the unknown spectrum by the equation

102

 
c, = J; K (e)gle)ae ,  J=1,2, .om (1)

where
c, = average number of counts obtained in the jth channel of the
J spectrometer,

~K.(e) = response function of the spectrometer, that is, the average
J fraction of incident particles of incident energy e which
produce a count in channel j,

g(e) = continuous spectrum, that is, g(e) de is the average number of
particles between e and e + de.

The gualitative appearance of the Kj(e) functions for a 3- by 3-in.
single-crystal NaI(Tl) gamma-ray scintillation spectrometer with 50 chan-

nels is shown in Fig. 10.3.

Bounds of a Single Parameter. If the'quantities C1; Ca, «v. C from

 

a scintillation spectrometer are to be used to determine a certain parameter
which depends upon the spectrum, for example, the tissue dose, the dose

can be expressed in terms of g(e) by the equation

o= J o) ate) ae (2)

UNCLASSIFIED
2-01-058-0-584

 

 

 

 

 

 

 

 

 

 

 

A
[\ === ——
/ 7 —

0 0.2 0.4 0.6 0.8 1.0
ENERGY

 

Fig. 10.3. The Response Functions K;(e) for a Few Illustrative Chan-
nels of a 50-Channel NaI(Tl) Gamma-Ray Scintillation Spectrometer.

103
where

p = tissue dose (in suitable units),

w(e)

dose absorption cross section, that is, the dose due to a single
particle of energy e.
For gamma radiation, w(e) has the well-known form consisting of contri-
butions from the photoelectric effect, Compton effect, and pair produc-
tion. The cross section is relatively smooth, with the exception of a few
sharp "edges" in the lower kev energy range corresponding to photoelectric
shell effects. |

Now if it is desired to be cautious and to insist on using no informa-
tion about g(e) other than that obtained from the scintillation spectrome-
ter, the dose D can be obtained from ¢y, c3, ... c if the dose absorption
cross section w(e) can be expressed as a linear combination of the response

functions:

n
wie) = géi U Kj(e) . (3)

Then

J;°° i u, Kj(e) g(e) de

p= J wie) ale) ae

n

- 2w LR Ge) ale)

.

- (4)

1
.t
<
o

However, if the cross section is not an exact combination of n response
functions, it is not possible to find the dose D at all. To understand
this, consider that a linear combination of response functions can e found
which nearly (but not quite) matches the dose absorption cross section.

For example, it may be impossible to match w(e) perfectly in the neighbor-

hood of an "edge." Thus the combination of response functions will be a

104

 
little high in certain places and a little low in others. It is then pos-
sible to have a component in the gamma-ray spectrum which gives no counts
(in any channel) but which does contribute to the dose. In other fields
such components have sometimes been referred to as "invisible components."5
Mathematically, if any discrepancy exists, then a possible component of
the spectrum may be found which is invisible [is orthogonal to all the
Kj(e)] but which contributes to the dose [is not orthogonal to w(e)].

Thus, to continue the cautious point of view of not making any as-
sumptions about g(e), it is usually not possible to determine a simple
quantity such as dose. This, of course, is a very pessimistic viewpoint,
which presupposes that several invisible components are present in the
spectra which are of such large magnitude that estimates of dose are com~
pletely ruined. Of course, this probably is not true, and the artificial
assumption that large invisible components are not present could be made.

But a much less artificial assumption can be made which will give a
reasonable estimate of dose. It is that

ge) 20 , 0 (5)

IN
M
IN
8

This "nonnegativity assumption" is physically justified in particle spectra,
since a negative number of particles has no physical significance and its
introduction into the formulation easily removes the difficulty of in-
visiblé components.

Although w(e) cannot be matched exactly with a combination of re-
sponée functions, it is perhaps possible to find one combination which is
always lower than w(e) and another combination which is always larger, as

shown in Fig. 10.4. Then, it is possible to write

o <1;‘r <nu+
ggi vy Kile) < wle) < géi 5 Ksle) | (6)

 

R. N. Bracewell and J. A. Roberts, Aerial Smoothing in Radio As-
tronomy, Australian J. Phys. 7, pp. 61540 (1954).

105
UNCLASSIFIED
2-01-058-0-585

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ENERGY

Fig. 10.4. TIllustration of Lower and Upper Bounds for a Cross Section
Formed from Linear Combinations of Response Functions.

- where
u. = coefficients of combination on Eq. 3 which give a lower bound to
J W(e))
u; = coefficients of combination which give an upper bound to w(e).

Now, because g(e) is everywhere a nonnegative function,

v N 0
f D ujj' Kj(e) gle) de = f w(e) g(e) de
J=1 °

0

©° 1

< J. > ut K. .(e) gle) ae (7)
o & 3T
or

n noo
L C,.<D< . C. 8
;Ei uJ 3 <P ;éi uJ cJ s (8)

and thus upper and lower bounds for D have been found, whereas it was
impossible to obtain an estimate (except in 3peciai cases) without the
nonnegativity cohdition.

Now if it is desired to obtain the narrowest possible bounds for the
dose D by using the above scheme, the following two mathematical problems

must be solved:

106

 
n .
Problem I. Find numbers ui, uf, ... wf such that ), u'5 c, takes on it

minimum value consistent with w(e) < ug Kj(e). =1

n
Problem IT. Find numbers uy, uz, ... u, such that 2: ug'c_ takes on

its maximum value consistent with w(e) B-ug Kj(e). J=1

Linear Programing. If Kj(e) is replaced with a table of values tabu-

 

lated at e = e3, €3, ... € then problems I and IT are identical to a much
studied problem in economics and scheduling. This problem, stated in terms
of its economic interpretation, is:

1. "Goods" can be bought from different factories ¥y, Fp, ... F
but an entire production run must be purchased, even if it entails buying
some "goods" which the customer does not want.

2. The n factories charge ¢y, C3, ... c, for a unit of production,
respectively.

3. Each unit contains the following number of '"goods:"

Factory 1 Factory 2 ce Factory n
Good 1 K11 Koy “ue . Km
Good 2 Kio Koo .o Kno
Good m Kim Kom : Knm

Now, the customer wishes to purchase enough goods to satisfy his re-
quirements for w3 of Good 1, wy of Good 2, ... and v of Good m for the
least possible cost, and the problem of finding the number of units to
buy from each factory'to obtain the minimum cost is identical to that of
determining the linear combination of response functions which give the
lower possible upper bound. Similarly, the problem of finding the largest
possible lower bound is equivalent to another "factory-goods" problem.
Problems of this sort, in which a linear function is to be minimized (of
maximized) subject to linear equation inequalities, are known as '"linear

programing” problems. Standard techniques exist for solving this class

107
of problems and have been coded for most digital computers. For example,
a book by Gass® summarizes the available codes.

Bounds on Multiple Parameters. If a set of functions Wi(e), Wz(e),l

 

...‘wm(e) is chosen instead of just one function w(e), then, similarly,

upper and lower bounds for

p, = fmw.(e) gle) de , ' i=1,2, ...m , (9)

can be determined provided a linear combination of response functions can
be found for each.Wi(e) which bracket it from below and from above, as in
Fig. 10.4 for w(e).

These parameters, P1, P2y .- pm, might represent quantities of the
same type as the example for w(e) — thus Wi(e), Wale), ... Wm(e) could be
m different dose cross sections for several materials of interest. But,

"of greater importance, Wi(e) can be any conceptual function chosen by the
experimenter. For example, if the experimenter wishes to describe a con-
tinuous spectrum g(e) by a histogram where the height of the histogram
P1, P2s; ... Pp represents the number of particles between ey and ep, ep

and e, etc., then the appropriate Wi(e) would be given by

1l
=

. . < ex<
Wi(e) if e, sese

. (10)
=0 otherwise.

As another ekample, it has been proposed that one method of reducing
the statistical problem in unscrambling spectra would be to determine

parameters of a smoothed spectrum of the form®

B ==,r0po Si(e) g(e) de , i=1,2, ...m , (11)

where Si(e) is a properly chosen 'smoothing function." In this case, the

appropriate Wi(e) are actually the smoothing functions.

 

®3. I. Gass, Linear Programming, McGraw-Hill, 1958.

108

 
But, unfortunately, as Wi(e) becomes narrower, it becomes harder to
bracket from above and below by a combination of response functions, and
the attainable bounds become poorer and poorer. Thus a practical com-
promise must be made between the desire for a detailed representation of
the spectrum and the desire for close bounds on the resulting parameters.

Figure 10.5 shows several dif-
UNCLASSIFIED
2-01-058-0-586

 

ferent types of Wi(e) and quali-
tatively indicates the types of
bounds which might be obtained
with functions of the type il-
lustrated in Fig. 10.3.

 

 

 

 

 

 

  

 

~ 1A o ™~
Y kf' Joint Bounds. 1In the
Fig. 10.5. Upper and Lower ' multiple-parameter problem, upper
Boundg for Different Shapes of Wi(e) and lower bounds can be obtained
Functions.

for each parameter Pi, Dz, +-« Pm
which represent the extreme values that these parameters could have con-
sistent with the given values of c¢3, c2, ... ¢, and with the nonnégafifiity
condition g(e) = 0. But it may happen that not all the parameters can
simultaneously attain their extreme values. This situation is similar to
one in multivariate statistics where one variable is correlated with
another. When one is large, the other also tends to be large (or small),
depending on the sign of the correlation coefficient, but in the presént
problem all the parameters are exactly given. These Jjoint bounds will Dbe
clarified by giving a simple two-parameter example.

Example of Two-Parameter Problem. It 1s assumed that

 

cL =2 = J; Ki(e) gle) de

and
2 =6=J Kale) gle) ae

are given and that it is desired to find the allowed values of

109
Py = JE Wi(e) g(e) de

and

P2 J;‘ Wao(e) g(e) ae

of (cy,c2) and consistent with the nonnega-

consistent with the wvalues
The four functions are given graphically in Fig. 10.6.

tivity condition.

UNCLASSIFIED
2-01-058-0-587

Wyle) -

  

0 { 2 3 4
ENERGY ENERGY

Fig. 10.6. The Functions Kj(e), Ka(e), Wi(e), and Wp(e) for a Simple
Two-Parameter Problem.

Following the previous plan, combinations of Kij(e) and K,(e) are

shown in Fig. 10.7 that bracket Wi(e) and Wa(e) from above and below.
these cases, the linear programing problem of finding the best bounds can
The same figure (Fig. 10.7) shows

In

be easily determined by inspection.
the upper and lower bounds for the combinations

Wi(e) + Wo(e)

and
Wy (e) + Wa(e)

Substituting the coefficients of combination ug and ug shown in

Fig. 10.6 into Eq. 8 gives

110

 
UNCLASSIFIED
2-01-058-0-588

  

= >
¥ b
ENERGY ——tam ENERGY ——w

® ®
& £
+ +
= >
¥ 5

M

  

ENERGY —— ENERGY —w

Fig. 10.7. Combinations of Kj(e)
and K, (e) Which Bracket Wi(e), Wa(e),
Wi(e) + Wa(e), and 3Wi(e) + Wo(e) from
Above and Below.

UNCLASSIFIED
2-01-058-0-589

BEST INDEPENDENT UPPER

o

RECTANGULAR HULL

 

o 2 4 6 8 10 12 14
P

Fig. 10.8. Allowed Region of
Parameter Space for p; and ps.

8Spl+p2-<.l4
l6$3pl+p2-<_l8

These four inegualities determine
the allowed values of p; and-pg.
The allowed region is indicated in
Fig. 10.8 by the shaded area.
Solution of the Deterministic
Problem. Even though the input
data for the deterministic problem
were exact, it may be seen that
the results for p1, P2, ... P
cannot be specified exactly. The
most precise solution is a com-
plete description of the allowed
region (as in Fig. 10.8), but, in
more complex multidimensional
cases, a complete description is
too cumbersome for practical ap-
plications., One possibility would
be just to give for a result the
coordinates of a point near the
center of the allowed region and
to give the extreme values of each
coordinate. This technique, how-
ever, would-yield far less infor-

mation than would be desired. For

example, it may be seen from Fig. 10.8 that the combination of 3p1 + pp can

be determined with much greater accuracy than either p; or ps separately,

owing to the shape of the region.

Thus an adequate description must con-

tain some information about the shape of the allowed region.

111
Two possible descriptions are the following:

1. GCive the center and the principal axes (length and direction) of the
smallest rectangular region which completely encloses the allowed
region. _

2. Give the centroid and the moments (including cross moments) of in-
ertia of the allowed region, considering that it is a solid body with
unit density.

In terms of these descriptions, formulas can be developed which give
the upper and lower bounds on any combination of the pi's by considering
the geometrical properties of the description.

| Although the determination of the smallest allowed region or of the
smallest rectangular "hull" which enclosed the allowed region is a straight-
forward mathematical problem, no satisfactory computational method for
solving it has yet been found. Approximatermethods exist which give the
approximately correct principal axes as the solution to an eigenvector
problem. The details of these methods will be reported later.

Conclusions. It has been shown that definite upper and lower bounds

can be put upon parameters of a continuous spectrum of the type

pi=.':wi(e) gle) de , i=1,2, ... m ,

by the use of a nonnegativity condition. The size and shape of the allowed
region for the pi's depends upon how well the'Wi(e) functions can be
bracketed between some linear combinations of the Kj(e) functions of the
spectrometer. Approximate descriptions of the allowed region are suitable
for practical applications and allow one to put an approximate bound on

any linear combination of the parameters. The methods which have been dis-
cussed could serve as a point of departure for the analysis of the statisti-
cal part of the problem and will be iImportant in the development of improved
methods for analyzing data obtained from crude spectrometers (such as thres-
hold detectors) where the uncertainty due to the meager availability of
suitable cross sections is more important than statistical errors in the
data.

112

 
Neutron Spectroscopy

Use of Silicon Surface-Barrier Counters in Fast-Neutron Detection and
Spectroscopy (T. A. Love, R. B. Murray, H. A. Todd, J. J. Manning)

The possible use of a neutron-sensitive semiconductor detector for
neutron spectroscopy was discussed previously.7'8 Basically, the spec~-
trometer consists of a thin layer of Li°F (~150 g/cmz) between two silicon-
gold surface-barrier counters, each of which is seated in a fluorothene
sheet. Neutrons are detected by observing the & + T pair resulting from
the Li®(n,a)T reaction.

The details of the counter construction were given previously,8 But
some modifications have since been introduced. A schemétic diagram of
the current design is shown in Fig. 10.9. The separation between the two

silicon counters is now effected by a 0.002-in.-thick ring of Teflon, and

 

"ANP Semiann. Prog. Rep. April 30, 1960, ORNL-2942, p. 117.
8ANP Semiann. Prog. Rep. Oct. 31, 1960, ORNL-3029, p. 128.

UNCLASSIFIED
ORNL-LR-DWG 57789

SILICON: Bmm x {4 mm x 0,381 mm

  
  
   
    
 

 

GOLD: 50 ug/ém®

 

 

Li®F: 150 pug/cmt

 

 

 

 

 

 

 

 

25 ALUMINUM CAN

TOP DIODE ASSEMBLY
245 ALUMINUM MOUNT

  

0002 in. THICK —a==

TEFLON SPACER

FLUOROTHENE DISK
/7 TEFLON TERMINAL

AQUADAG

 

 

 

 

 

 

0004 -in~dia '
GOLD WIRE /4/
. //'q
SILVER v 5
PAINT S GOLD: 50 ug/cm

 

 

o 1
APPROXIMATELY { INCH
BOTTOM DIODE ASSEMBLY

Fig. 10.9. Schematic Diagram of Sandwich-Type Silicon-Gold Surface-
Barrier Counter. :

113
a small wire is used to make the electrical contact to the gold surface.
The wire is cemented to the gold with a tiny drop of Aquadag and then
grounded. A reverse bias of 100 v is applied through a resistor of
~500 000 ohms, and the signal is fed through a coupling capacitor to a
low-noise high-gain preamplifier.

A block diagram of the electronic system for the sandwich counter
is shown in Fig. 10.10. The signal from each detector is fed into a pre-
amplifier, which, in turn, feeds a DD-2 double delay line amplifier with
a O.7-psec clipping time. The output of the amplifier is fed into a dis-
criminator circuit biased to reject all pulses which represent less than

1l.5-Mev energy deposited in the diode.

UNCLASSIFIED
ORNL -LR-DWG 57795

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

DD-
* )l PREAMP » e DISCRIMINATOR
AMPLIFIER
500K iK
_ HANN
. /LA F 1100 v MIXER DD-2 SIGNAL | MULTIC EL
5 AU CIRCUIT AMPLIFIER ANALYZER
AN —=Si
COINCIDENCE { |{GATE
. S00K 1K CIRCUIT

 

 

 

 

 

 

 

. pl; PREAMP * 0b-2 DISCRIMINATOR 1:|

AMPLIFIER

 

 

 

 

 

 

 

 

 

 

 

Fig. 10.10. Block Diagram of Electronic System for Sandwich-Type
Silicon-Gold Surface-Barrier Counter.

The discriminator setting of 1.5 Mev was chosen on the basis of the
response of a single detector exposed to thermal neutrons with no dis-
criminator setting. For incident thermal neutrons, the triton and alpha
peaks are clearly defined, as shown in Fig. 10.1l, and most of the counts
below 1.5 Mev are attributable to gamma rays. About 5% of the pulses be-
low this energy are from reaction products which have lost several hundred
kev of energy. Thus a discriminator setting of 1.5 Mev essentially re-
Jjects all the gamma-ray background with little loss of the counts due to

the reaction products. A gamma-ray background with the discriminator

114

 
UNGCLASSIFIED
ORNL-LR-DWG 57794

2400 600

! |
ALPHA PEAK ¥ 45 % OF TOTAL
ABOVE CHANNEL 25

 

GAMMA BACKGROUND,

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2000 ¥~ NOISE, ETC 500
fl,m ITONS

1600 400
T T
= o
= o
8 ¢ 2
S 1200 300 S
2 »
c c
-

800 % 200

ALPHAS
1.5 Mev 1
400 ' 100
& 9
0 900, v %‘e "=;= 0
0 10 20 30 40 50 60

CHANNEL NUMBER

Fig. 10.,11. Pulse-Height Response of Single Surface-Barrier Detector
Exposed to Thermal Neutrons (No Discrimination).

setting of 1.5 Mev has the same effect as noise; that is, the peak from
monoenergetic neutrons will be broadened.

The loss of counts as a result of the reaction products will be much
greater for higher energy neutrons, perhaps as much as 30% for 10- to 15-
Mev incident neutrons, owing to the forward peaking of the reaction prod-
ucts, Calculations are under way for determining the extent of the loss
as a function of energy.

The coincidence circuit of Fig. 10.10 requires that the pulses occur
within 107 sec of each other, and the output bf this circuit is used to
gate the multichannel analyzer. Thus, with this circuit, Lis(n,a)T events
in which the alpha particle and triton deposit at least 1.5 Mev in sepa-
rate counters are recorded, but those events that deposit less than this
amount in either counter are eliminagted.

The output of the preamplifiers is also. fed into a mixer circuit.

The output of the mixer circuit is amplified by a third DD-2 amplifier

115
whose output is recorded in the multichannel analyzer. The input of each
preamplifier also has provisions for accepting a pulse from a multichannel
pulse generator. This arrangement allows each amplifier and preamplifier
to be checked separately for linearity and the analyzer to be checked for
both linearity and zero position. The pulse generator also allows a
secondary check for stability.

Experimental Data and Discussion. The response of the semiconductor

 

spectfometer t0 monoenergetic neutrons up to energies of 3.5 Mev and also
to 14.7-Mev neutrons was reported previously.7’8 Further studies of the
response to monoenergetic neutrons in the energy region from 3.5 to 8.2
Mev and to groups of monoenergetic neutrons from the Be®(d,n)Bl? reaction
have now been made.

Background effects from (n, charged particle) reactions appeared at
all neutron energies above the threshold energy (approximately 2 Mev) of
Si(n, charged particle) reactions. Pulses from fhese charged particles

were recorded only in the neutron pulse-height spectrum for incident

neutrons above about 5 Mev, since the reactions are endothermic. 1In order
to determine the background, a pair of silicon diodes without Li®F was
constructed, and pulse-height spectra for diodes with and without Li®F
were compared for monoenergetic neutrons at several energies. Later modi-
fications now allow gains to be adjusted more accurately for matching both
sets of diodes by use of a pulse generator.

A comparison of the foreground and background for 5.73-Mev incident
neutrons from the D(d,n)He? reaction is shown in Fig. 10.12. The peak at
a pulse height corresponding to neutrons of 2.19-Mev energy is due to carbon
contamination of the target and comes from the Clz(d,n)N13 reaction.
Figure 10.13 shows a foreground-background comparison for neutrons of 8.13
Mev. |

 The pulse-height spectrum for groups of neutrons from the Be?(d,n)BLO
reaction (deuteron energy of 1.85 Mev) is shown in Fig. 10.1l4a. This is
to be compared with the spectrum obtained for the same reaction by the

time-of-flight method® (see Fig. 10.14b). 1In Fig. 10.14 the numbers in

 

Private communication from J. H. Neiler et al., ORNL Physics Division.

116
UNCLASSIFIED
ORNL-LR-DWG 57792

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

900 T i
® FOREGROUND |
© BACKGROUND GAINS NOT NORMALIZED
800 THERMALS ’
+ 4.6 Mev C'z(dn} ni3 i
2.19- Mev NEUTRONS D (d,n)H2
+ 4.6 Mev 5.73- Mev NEUTRONS
+ 4.6 Mev
700 fi H
600
4 ®
@ 500
<
o
£z
O
@ ¢
c [ ]
S 400
®
p
*lo
300 3
¢ .:
P J 9
200 A
. o 3
I .
100 - / » \.
: b L ||'/
® /';E o‘d ® \*.0”
':O .ou .-.. .. '
O &, l
0 o 90,9
0 20 30 40 50 60 70 80

CHANNEL NUMBER

Fig. 10.12. Pulse-Height Response of Sandwich-Type Surface-Barrier
Counter Exposed to 5.73-Mev Neutrons.

parentheses are the values reported for neutrons from this reaction. As
expected, the resolution of the sandwich counters at the lower energies
is not good enough to resolve the 0.96-Mev group from the 1.33-Mev group,
but at the higher energies it is quite good. The data in Fig. 10.14
represent measured pulse-height spectra, and the energy dependence of the
neutron detection efficiency has not been taken into account in either
case. The data shown in Fig. 10.15 demonstrate that the device gives a

puise height which is linear over the range examined (0 to 14.7 Mev).

117
UNCLASSIFIED
6 ORNL— LR—DWG 57790

GAINS NOT NORMALIZED

4.62-Mev

NEUTRONS
A C12(d, 7) N‘.3
1

counts /channel

# 8.13-Mev
NEUTRONS
D(d,n)He3

BACKGROUND

 

0 20 40 60 80 100
‘ CHANNEL NUMBER

Fig. 10.13. 'Pulse-Height Response of Sandwich-Type

Ssurface-Barrier
Counter Exposed to 8.13-Mev Neutrons.

118
l_l
|_l

UNCLASSIFIED

. ORNL—LR-DWG 57793
103

£, =1.85 Mev
Lab. ANGLE =0 deg

4q
10 THERMALS _L

S NUMBERS IN PARENTHESES
ARE REPORTED ENERGIES

UNCLASSIFIED
2—Q1—077—-NS—11—13—-55—1

100 p.h.u. = 75 musec
|
( ) {5.39)
{0.986)
0.95 (G.Og)
4.00)

103

counts /channel

{4.39)
440

(2.55)
2.'|f0

counts /microcoulomb

Beg(d,n)s‘o—fd =1.85 Mev
LAB ANGLE 20°—5-meter FLIGHT PATH

 

10 20 30 40 50 &0 70 80 300 400 500 600 700 800 900 1000 100
CHANNEL NUMBER PULSE HEIGHT UNITS

Fig. 10.14. (a) Pulse-Height Response of Sandwich-Type Surface-Barrier Counter to Neutrons from
the Be?(d,n)B'® Reaction. (b) Pulse-Height Response of a Time-of-Flight Spectrometer to Neutrons
from the Be®(d,n)B'0 Reaction.
UNCLASSIFIED
ORNL-LR-DWG 57791

 

 

 

 

 

140 | -
T(d’ fl) JES— -
i
120 - T, n) —
//////, 18.3
100

 

i
/'fh—E!e9 (@, 7) 13.36 Mev
o | | |

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

= »
5 80 A B (0 ) 12. 42 Mev
E ,//, | |
? ////'\\Bé(mnMQ25Mw
|
é 80 : .‘/ |
/////"“-cm(d,n)7.77 Mev
40 ] .
' /*(j
/ THERMAL 4.6 Mev
20 —
0 |
0 2 a4 6 8 10 12 14 6 18 20

NEUTRON ENERGY (Mev)+ 4.6 Mev

Fig. 10.15. Pulse-Height Response of Sandwich-Type Surface-Barrier
Counter as a Function of Neutron Energy Plus 4.6 Mev.

Conclusion. From the data of the figures represented here and those
published previously, it is concluded that the full width at half maximum
is relatively constant, of the order of 300 kev, regardless of incident
neutron energy. The presence of gamma rays will produce a broadening of
the neutron peak, but measurements have been made in gamma-ray fields up
to approximately 200 r/hr without appreciably altering the response.
Further, the device produces & pulse height that is a linear function of
incident neutron enefgy.

The background data for monoenergetic neutrons with energies of 4,
5.73, 7.40, 8.13, 13.7, and 14.7 Mev have served as a basis for an esti-
mate of the background to be expected when the counters are used to
measure a fission spectrum. Calculations predict that the foreground-to-
background ratio will be about 5 to 1. A preliminary experiment was per-
formed with neutrons from the Tower Shielding Reactor IT to get an esti-
mate of the foreground and background, and the results indicated that the
‘background was approximately 20% of the foreground, which is in agreement
with calculation. A measurement of the reactor spectrum will be made in

the near future.

120
Further Developments. In an attempt to improve the counting effi-

 

ciency of these counters, the effects of several variations in the counter
design are being investigated. Specifically, a sandwich counter with an
area approximately four times the area of the counters used in the above
experiments is now under construction. In addition, the feasibility of
substituting 1i® metal for Li®F as the neutron-sensitive material is being
studied. Favorable results in these experiments will permit the use of
about 20 to 50 times as much Li® as is presently used, which, assuming
normal incidence to the plane of the Li® layer, will increase the counting
efficiency of the counters from about 107 to about 10-° for neutrons in
the Mév region and will result in a greater foreground-to-background ratio
for a given spectrum. ©Since the foreground-to-background ratio is deter-
mined by the relative number of higher energy neutrons present, this in-

crease in Li® will allow a much "harder" spectrum to be examined.

A Fast-Pulse Integrator for Dosimetry

 

P. T. Perdue

The integrator which was used for dosimetry at the Lid Tank Facility
for several years is composed of an eight-channel discriminator whose out-
put is fed to the first four binaries of two modified Q-1010 scalers. The
5-psec resolving time of this integrator was considered sufficient when
used with the A-1A amplifier, since the resolving time of the amplifier
was of this order, but with the advent of the DD-2 amplifier and its known
2.6-usec dead time it became desirable to decrease the resolving time of
the integrator to within the same interval so that counting loss corrections
could be calculated. Development of such an integrator10 was thus initiated
by Lid Tank Facility personnel. The new integrator, which has a total
resolution of 2.4 usec, was placed in operation early this year. It con-
sists of two sections, each having four discriminator stages coupled to-
gether as a true binary séaie of 16. The output pulses from both sections

are fed to a single decade scaler.

 

10p, T, Perdue et al., Fast-Pulse Integrator for Dosimetry, ORNL re-
port (to be published).

121
Increasing the speed in the integration process necessitated the
development of several additional ideas. For example, to minimize the
counting losses in the binary stages under the various "off-on" combina-
tions, the slope of the input pulse from the DD-2 is being used to deter-
mine the time interval between successive discriminator responses. A
special pulse shaper is being used between binary stages to maintain a
maximum pulse width of 0.2 usec at the input of the next stage. The width
of the output pulse from the integrator sections to the decade scaler has
been reduced to 0.3 upsec through the design of a shaper circuit, giving a
resolution of approximately 1.0 usec in the scaler. Cathode interface
problems, experienced in the Schmidt-trigger discriminators of each of the
stages, have been eliminated through the use of a 6922 double triode,
where both halves of the tube are in steady-state operation within the grid
base.

Operation of the instrument with various fast-neutron detectors has
indicated that the statistical spread in the count rate from the integrator
output is virtually the same as the maximum spread on the total number of

pulses fed into the integrator.

122

 
11. BASIC SHIELDING STUDIES

Energy and Angular Distributions of Neutrons Emerging
from Planar Surfaces of Diffusing Media

 

V. V. Verbinski

The initiation of a program for obtaining basic information on the
energy and angular distributions of neutrons in the electron-volt region
throughout various shielding materials was reported previously, along
with the results of a preliminary experiment for investigating the angular
distribution of neutrons emerging from poisoned hydrogenous slabs.l As
originally conceived, the over-all purpose of the program was to determine
the applicability of several techniques for calculating neutron energies
and angular distributions within high-performance materials, specifically
LiH. The evaluation of the calculational techniques was to be aided by
experimental measurements made with a proposed neutron chopper.2 Insofar
as possible, both the calculated and measured data were to be obtained as
energy spectra of neutrons at several points inside the shield, energy
spectra of neutrons leaking from the surface of the shield, and angular
distributions of the leakage neutrons.

Recently, however, budgetary support for construction of the neutron
chopper facility was withdrawn. In order to obtain other experimental
data which could be compared with the calculations, plans were immediately
initiated to perform a series of measurements at the Linear Accelerator
Facility of General Atomic, San Diego, in early May 1961.

-Three calculational methods are being utilized: +the NDA NIOBE (direct
numerical integration of the Boltzmann equation) code, the moments method,
and the ORNL O5R code. Where possible, the source energy and LiH geometry
are similar so that direct comparisons of calculations can be made. The
NIOBE code is set up to handle only spherical geometry, the moments method

only infinite geometry, and O5R code any geometry. Placing a source at

1ANP Semiann. Prog. Rep. Oct. 31, 1960, ORNL-3029, p. 179.
2Ibid., p. 142.

123
a reasonable distance from the shield will allow comparisons of the spec-
tral and angular distributions of leakage neutrons from slabs of LiH cal-
culated with the NIOBE and O5R codes and will also allow comparisons of
all three methdds for internal spectra at several depths in very thick
slabs.

Thus far, energy spectra of neutrons from a point source in an infi-
nite medium of LiH have been calculated by the moments method for distances
from the source of 11.5, 23, and 34 g/cm? and for four source energies |
of 16.3, 1.48, 0.12, and 0.01 Mev. These preliminary spectra, normalized
at 1 volt in Fig. 11.1, show that low-energy equilibrium is reached at all
thicknesses for all input energies employed. This equilibrium spectrum is
in agreement with earlier Monte Carlo calculations made at NDA. One re-
sult of the Monte Carlo calculation, labeled "Fission,"” is for a fission
input spectrum and a source distance of 26 cm. It can be seen that the
spectral shape of this curve is practically identical with the curve for
1.48-Mev neutrons at 23 g/cm?.

Some NIOBE calculations? of the angular distributions of neutrons
leaking from LiH are also available, although they were not made specifi-
cally for this program. Here, of course, a spherical geometry was used,
but it is not expected that this will significantly affect a comparison
with slab-geometry experiments. The angular distributions of leakage cur-
rent, J(u), calculated by the NIOBE code for 5.51-, 1.58-, and 0.037-ev
neutrons have the form 1 + Ay, where p = cos8@ and 6 is the angle from

perpendicular; that is,

J(u) < 1 + 3,25u for 5.51-ev neutrons
J(u) 1l + 2.64u for 1.58-ev neutrons

J(u) o« 1l + 1,02y for 0.037-ev neutrons

 

3Herbert Goldstein, Nuclear Development Corporation of America,
private communication.

124

 
UNCLASSIFIED
2-01-058—-0-592

SOURCE
ENERGY

16.3 Mev
1.48 Mey

104
SOURCE

DISTANCE =11.5 g/cm?
0.12 Mev

0.01Mev

 

____________ FISSION
(MONTE CARLO)

103 16.3 Mev
- 1.48 Mev
N
-9-
Wy
Ll
>
=
<
|
o 7
FISSION |,/
102
16.3 Mev
1.48 Mev
10!
349/cm2
0.12 Mev
100
0.0t 0.4 1 10 100 1000 10,000 100,000

NEUTRON ENERGY (ev)

Fig. 11.1. Moments Method Calculations of Neutron Spectra Times
Energy at Various Distances from Monoenergetic Point Sources in an Infinite
Medium of LiH. Spectrum at each depth normalized to 1.48-Mev curve at
1 ev. "Fission" curve from earlier Monte Carlo calculations.

125
The NIOBE calculations are to be compared with Bulk-Shielding Fa-
cility measurements of the angular distributions of neutrons leaking from
a 4-in.-thick LiH slab. The experimental arrangement was the same as
that described previously,1 where neutrons from the Bulk Shielding Reac-
tor were incident on the shielding sample after traveling through an air-
filled collimator. On the opposite side of the shield sample was posi-
tioned an air chamber from which air-filled tubes extended at various
angles from the normal up to 66 deg. Detector foils were placed at the

far end of each tube. These measurements gave the‘following results:

I+

J(u) <1 + (2.00 * 0.25)u for 4.95-ev neutrons

J(p) <1 + (1.75 + 0,25)u for l.44-ev neutrons

+ 1.4

J(u) o<1 + (0.9_0.4

Ju for subcadmium neutrons
In the subcadmium measurements, instabilities were encountered in the
counting circuitry, and the measurements are therefore being repeated.

The linear dependence of J(p) on p is to be expected when the flux
varies linearly near the surface of the LiH, and the flattening of J(u)
with lower energies is due to a decrease in extrapolation length with
decrease in energy. In contrast, the angular distribution in a non-
moderating shield with high l/v absorption peaks strongly in the forward
direction as the neutron energy decreases. Therefore, the slowing down
in a hydrogenous material is seen to be of major importance in deter-
mining the angular distribution of leakage neutrons.

As mentioned above, measurements of leakage spectra and of spectra
within LiH slabs were to have been conducted with a neutron.chopper fa-
cility planned for the BSF. 1In conjunctiqn with this effort, a program
was developed for the IBM-7090 computer to calcplate the fransmission
function and burst shape of the mechanical chopper. In addition, several
calculational techniques were developed for taking into account edge leak-

age, divergence of the neutron beam, and off-axial effects of the outer

126

 
slits. Also, numerical techniques were evolved which resulted in large
economies in computing time. A report is in preparation outlining these
endeavors.

The series of measurements to be conducted at General Atomic, San
Diego, will begin in early May 1961. A preliminary measurement conducted
at GA had adequately demonstrated the feasibility of the series of planned
measurements. Although the Linac spectrum is not identical to a reactor
spectrum, the NDA calculations in Fig. 1l.1 indicate that the results are
not very sensitive to the value of source energy for the high-energy more

penetrating neutrons.

Experimental Verification of a Geometrical Shielding Transformation

 

L. Jung J. M. Miller
F. J. Muckenthaler J. Grant

The Lid Tank Shielding Facility is currently being utilized to verify
a geometrical transformation concept that the axial dose rate from a large
source plate in a hOngeneous medium can be inferred from the dose rates

from a small disk source. Expressed mathematically,

2
D(z,a) =D (%,—Ji-) + D z? + E;-, 21+ cee T
vn v
(1) a? a
+ D zZ< + y —

where
a = radius of source plate,

7z = distance from source plate along a line perpendicular to center of
source plate,

=
I§

integer (a/A/f = radius of the small disk source).

127
During this investigation the source plate size will be varied by in-
serting cadmium irises of sevéral different'diameters between the source
plate and the incident beam of neutrons from the ORNL Graphite Reactor.
Cadmium was selected for this purpose because it provides less scattering
of thermal neutrons near the edge of the iris, thus giving less error in
describing the source area. The use of cadmium gives rise to inaccuracies
in the fast-neutron and gamma-ray dose rate measurements in the media- be-
cause of cadmium capture gamma rays, however, and verification of the
theory must therefore depend only on the thermal-neutron flux measurements.

Thus far, only the measurements with a 7-in.-diam iris have been completed.

128

 
12. APPLIED SHIELDING

Preanalysis of Pratt & Whitney Divided-Shield Experiment at TSF

 

S. K. Penny, D. K. Trubey

The preanalysis for the Pratt & Whitney divided-shield experiment
at the Tower Shielding Facility has been completed. This preanalysis, |
which predicts the neutron and gamma-ray dose rates to be measured in
the experiment, was performed to serve three purposes. First, the calcu-
lations were intended to guide the experimentalists in setting up the
measurements; second, a comparison of the calculated and experimental re-
sults will indicate how well the theoretical methods at hand can be used
to predict measurements; and third, the disagreements will lead to im-
provements in both the theoretical methods and the experimental methods
so that they can be more directly correlated.

The Pratt & Whitney divided-shield design consists of a highly asym-
metric reactor shield surrounding the Tower Shielding Reactor II that is
separated from the TSF compartmentalized cylindrical crew shield by ap-
proximately 64 ft. The reactor shield, a sketch of which is shown in the
following section of this chapter, incorporates two advanced shielding ma-
terials, lithium hydride as the outer neutron shield and depleted uranium
as the gamma-ray shadow shield. The design dimensions for the reactor
shield are given in detail in Table 12.1, and those for the crew compart-
ment are given in Table 12.2.

For a complex shield such as the Pratt & Whitney divided shield,
every conceivable source of radiation must be considered and carefully
weighed before a decision can be made as to its importance. Also, every
theoretical method considered must be examined as to its applicability,
avallability of calculaticnal methods, and practicality before it is used
in the calculations. Thus, calculations for a complex shield system rep-
resent a fairly large undertaking, and the final results contain many
nuances of the above considerations. TFor reasons of convenience and sim-
plicity, the housing for the reactor chambers, etc., positioned above the

reactor core was ignored in the calculations, and the reactor shield was

129
 

 

Table 12.1. Gamma-Ray Shield Thicknesses and Outer Radii of Pratt & Whitney Reactor Shield
Nominal dimensions: inner spherical radius of shield 19.364 in.
thickness of inner stainless steel 0.125 in.
can
thickness of LiH "cooling layer"” 6.430 in,
annulus
thickness of gamma-ray shield stain- 0.375 in.
less steel support can
thickness of forward steel bulkhead’ 0.25 in.
thickness of rear steel bulkhead 1.00 in.
v Nominal Gamma-Ray  Outer Radius v Nominal Gamma-Ray  Quter Radius
Polar Angle Shield Thickness® of Shield® Polar Angle Shield Thickness® of ShieldP
(deg) (in.) (in.) (deg) (in.) (in.)
0 3.68 54 .27 100 0 39.15
5 3.65 54,05 105 0 38.75
10 3.55 53.770 110 0 38.35
15 3.38 53.05 115 0 37.88
20 3.19 52.45 120 0 37.55
25 2.85 51.50 125 0 37.15
30 2.50 50. 56 130 0 36.75
35 2.09 49,75 135 0 36.15
40 1.60 48,95 140 0 35.65
45 1.04 47.95 145 0 35.00
50 0.68 47.25 150 0 34.45
55 0.35 46.31 155 0 33.56
60 0.05 45,55 160 0 32.95
65 0 b4ér .65 165 0 32.28
70 0 43.85 170 0 31.85
75 0 42,85 175 0 31.50
80 0 42,05 180 0 31.45
85 0 41.16
90 0 40.35
95 0 39.65

 

 

thick stainless steel support can.

130

aDimensions do not include 0.125-in.-thick outer stainless steel can or 0.375-in.-

bThese dimensions are nominal to the inner surface of the outer stainless steel can
and include allowance for 0.25 in. of foam plastic to take up gaps and allow for expansion
with temperature.

Table 12.2. Dimensions of Cylindrical

Crew Compartment

Inside diameter = 36.0 in.
Inside length = 72.0 in.

 

 

Material Region Thickness
(in.)
b Sides and front 0.20
H;0 Sides and front 8.00
Pb Rear 3.27
H-0 Rear 18.00

 

 
considered to be azimuthally symmetrical about the reactor—crew compart-

ment axis.

Radiation Sources

 

The radiation sources which could conceivably be important are (1)
fission neutrons from the core and from the uranium shadow shield, (2)
prompt-fission and fission-product decay gamma rays from the core and
shadow shield, (3) capture and inelastic scattering gamma rays produced
in the reactor and reactor shield, (4) capture and inelastic scattering
gamma rays produced in air, and (5) capture and inelastic scattering
gamma rays produced in the crew compartment. The fission neutrons in
the shadow shield were neglected because this source was quite small. |
A1l inelastic scattering gamma rays were also neglected because no suit-
able method was available for treating this source. In addition, the
capture gamma rays in the ailr and in the crew compartment were neglected
because no adequate calculational method was available for computing the
distributions of thermal and epithermal neutrons in these regions. These
sources are expected to be small since the low-energy neutrons are markedly

suppressed by lithium hydride.

Attenuation and Transport Processes

 

The attenuation and transport processes used in the calculation can
be divided into three categories: (1) the radiation emergenf from the
reactor—reactor shield system, hereinafter called the leakage spectrum,
(2) the subsequent scattering of this radiation in air, and (3) the at-
tenuation by the crew compartment shield of radiation which is both un-
scattered and scattered in air. The calculations for each category will
be discussed separately, and the assumptions and approximations that were
involved in each case will be given.

Leakage Spectrum. The reactor shield leakage spectra for both neu-

 

trons and gamma rays were calculated on the basis of "moments method"
datal for lithium hydride. The data were used as a point-to-point at-
tenuation kernel from a source point in the reactor—reactor shield system

to a point in air. No attenuation or buildup in air was included in this

131
part of the calculation. The assumption that the energy spectrum at a
point in air can be characterized for both gamma rays and neutrons by
data for lithium hydride is Jjustified on the following bases. Monte Carlo
calculations indicate that the emergent energy spectrum for gamma rays

3 is characteristic of the final material if

incident on infinite slabs?s
that material is at least one mean free path thick. These conditions are
satisfied in the reactor shield, since lithium hydride is the final ma-
terial (outer layer) and its thickness is relatively large. The magnitude
of the gamma-ray energy spectrum depends, of course, on the actual total'
number of mean free paths thickness. In the case of neutrons, the fact
that fast-neutron removal theory4 works fairly well makes possible the

use of the emergent energy spectrum characteristic of the last hydrogenous
material (that is, lithium hydride), since the neutron slowing down is due
primarily to hydrogen. Here the magnitude of the energy spectrum depends
on the total equivalent thickness of the last hydrogenous material, where
the equivalent thickness is calculated on a hydrogen density basis. Cor-
rections for norhydrogenous materials are effected through the use of re-
moval cross sections.

Since the "moments method" data were obtained for an infinite medium,
their use in penetration calculations for finite media overestimates the
energy spectrum, particularly at the lower energies. Furthermore, using
the method of a point-to-point kernel leaves the angular distribution of
the radiation completely open to question. In this calculation it was as-

sumed that all the radiation is moving along the line of sight from the

 

1H. Goldstein and J. E. Wilkins, Jr., Calculations of the Penetration
of Gamma Rays, NY0-3075, June 24, 1954; D. D. Babb, J. N. Keller, and E.
McCray, Curve Fits of Gamma-Ray Differential-Energy Spectra, NARF-59-36T
(MR-N-251), Nov. 1, 1959; H. Goldstein, Some Recent Calculations on the
Penetration of Fission Neutrons in LiH, NDA-42, Aug. 7, 1957.

21,, A. Bowman and D. K. Trubey, Stratified Slab Gamma Ray Dose-Rate
Buildup Factors for Lead and Water Shields, ORNL CF-58-1-41, Jan. 16,
1958.

| 3H. Goldstein, Fundamental Aspects of Reactor Shielding, p. 223,
Addison-Wesley, Reading, Mass., 1959.

“Tbid, p. 326.

132

 
source point to the field point. The actual case is that radiation emerges
from all points on the shield surface with an angular distribution which
depends on where the surface point is located. The errors involved in
this assumption can be qualitatively evaluated.

The neutron source for the leakage calculation is the number of fis-
sion neutrons produced per unit volume per unit reactor power with epi-
thermal fissions neglected. The gamma-ray source is the number of gamma
rays produced per unit volume per unit reactor power of a given energy
arising from neutron absorptions, which include fission and thermal-neutron
radiative capture. In ordef to obtain these sources, the thermal-neutron
flux throughout the reactor—reactor shield system must be known. The flux
slope in the core and immediately outside the core was obtained by calcu-
lations with the GM-GNU reactor code’ for the case of a spherical reactor
surrounded by 20 cm of water. It was extended by joining it with the
measured BSR thermal-neutron flux at distances greater than 10 em from
the reactor boundary. It was then normalized to a total power of 1 watt.

Since the medium surrounding the TSR-II in the Pratt & Whitney design
is lithium hydride instead of water, it was necessary to convert the flux
for water to that for lithium hydride. This can be done approximately by
assuming that most of the slowing down of neutrons to thermal energies is
effected by hydrogen and that the effect on the slowing down by non-
hydrogenous materials is taken into account by the use of removal cross
~ sections. If leakage is neglected, the number of neutrons arriving at
thermal energies by hydrogen collisions in corresponding volume elements
of the two materials is equal under these assumptions.

The corresponding volume elements are found through the transforma-

tion from medium 1 to medium 2 by

PrT1 = P22

SGNU-II, A Multigroup One-Dimensional Diffusion Program, General
Motors Corp. Report 101, Nov. 12, 1957,

133
and thus

Pi1
AV, = dV; -5 >
P2

where
pi = hydrogen density in medium i,
r, = spacial coordinate in medium i,
dVi = differential volume element in medium i.

The expression for the number of neutrons slowing down to thermal energies
(and consequently being absorbed) in a differential volume element in a
pure hydrogen slowing-down medium is

1) (2

(
o r 2, 'r
e © ’ Egl)¢(l)(r1) dv, = e ° Z£2)¢(2)(r2) dva ,

 

where

Zii) = macroscopic removal cross section for nonhydrogenous ma-

terials in medium i,
Zéi) = macroscopic thermal-neutron absorption cross section in
' medium i,
_¢(i)(ri) = thermal-neutron flux in medium i at rj’
so that, finally, -
(1) 3
¢(2) n = E—?;)— (0_2) ¢(l)(I‘1) exp (Ei,l) *Zi.z)pl/pz) ry .
P2 &Y P1

_ For the sake of simplicity, it was assumed that the thermal-neutron
flux‘had only the radial dependence that was calculated with the reactor
code (and extended by BSR data) and modified to correspond to lithium
hydride, although the reactor—reactor shield system is actually asymmetri-
cal in shape., This would probably not be too much in error except close

to the outer boundary were it not for the presence of the shadow shield.

134

 
Since this assumption ignores the presénce of the shadow shield, however,
and consequently the strong depression of the flux it causes, the number
of gamma rays born in the shadow-shield region is overestimated. (This
was borne out by further calculations with the GM-GNU code performed by
W. E. Kinney of ORNL after the major part of this preanalysis was com-
pleted.) However, even with this overestimate the contribution from the
shadow shield is small.

In calculating the thermal-neutron flux, various regions of the TSR-
IT were homogenized for the sake of convenience and also to-save time in
the calculation of the gamma-ray leakage spectrum. Four regions were
used: the central control plate housing, the core annulus, the lead layer,
-and the remainder, which was mostly aluminum and water. The thermal-
neutron flux was not averaged over these regions. Errors may be intro-
duced by this homogenization, but they are probably small. By far the
largest contribution to the gamma-ray leakage spectrum was that of the
capture gamma rays born in the last homogenized region of the TSR-IT,

The results of these calculations were used to predict dose rates
in air and in the crew compartment from radiation unscattered in air, that
is, the direct-beam dose rates. Direct-beam dose rates in air 50 ft from
the reactor are given in Tables 12.3 and 12.4 and those in the crew com-
partment are given in Table 12.5. The results were also used to obtain
an equivalent point source to be used in the air-scattering calculations.
The equivalent point-source angular distribution is found by calculating
the number current spectrum on the surface of a very large sphere at a
given angle and multiplying the results by the radius squared.

Air Scattering. The angular and energy distributions of neutrons
which are emitted from a monodirectional and monoenergetic point source
and subsequently scattered in air were estimated by using data from the
Convair D-35 Monte Carlo code.® This code incorporates inelastic as well
as elastic scattering, and the distances from the source point to the

field point range from 10 to 100 ft. It also estimates the dose rate in

 

M. B. Wells, Monte Carlo Calculations of Fast Neutron Scattering in
Air, NARF-60-8T, Vols. 1 and 2 (FZK-9-147), May 13, 1960.

135
air from the scattered neutrons; hence at this point the estimate of the

dose rate 1n alr has been completed.

Table 12.3. Calculated Neutron Dose Rates for a
Point in Air 50 ft from the Reactor

 

Neutron Dose Rates (rep/w-hr)

 

 

6
a
(deg) Direct Scattered Total
0 5.91 x 107° 2.53 x 10™7 2.59 x 1077
5 6.17 X 107° 2.53 x 107 2.59 x 1077
15 8.29 x 107° 2.56 x 10~7 2.64 % 10°7
30 2.00 x 1078 2.66 x 10°7 2.86 x 107
60 1,17 x 1077 3.19 x 10-7 4.36 x 107
90 5.68 x 10™7 4.55 x 10~7 1.02 x 10-
120 1.60 x 10~ 7.34 % 10-7 2.33 x 10-°
150 5.03 x 1076 1.22 x 10~ 6.25 % 1076
180 1.12 x 10°° 1.70 x 10-° 1.29 x 10~?

 

% = polar angle, the polar axis being the axis of
metry through the reactor and crew compartment.

sym-

The positive

direction of the polar axis is from the reactor to the crew

compartment.

- Table 12.4. Calculated Gamma-Ray Dose Rates for a
Point in Air 50 ft from the Reactor

 

Gamma-Ray Dose Rates (r/w-hr)

 

 

6
: a,
(deg) Direct Scattered Total
0 1.42 x 107° 9.36 x 10-° 1.08 X 10°°
5 3.60 x 1076 9.48 x 1076 1.31 X 1077
15 1.26 x 10-° 1.01 x 10~° 2.27 x 107°
30 5,32 x 10°° 1.25 x 1077 6.57 x 107
60 2.91 x 10°% 2.%44 x 10-° 3,15 x 1074
90 3,91 x 1074 3,40 x 107° 4,25 x 1074
120 5.17 X 1074 4.37 x 1075 5.61 X 10~%
150 6.38 x 1074 5,28 x 1072 6.91 x 10~%
180 6.98 x 10™% 5.70 x 1077 7.55 X 10~%

 

136

%Defined in Table 12.3.

 
Table 12.5. Calculated Neutron and Gamma-Ray Dose
Rates for a Position Inside the Crew Compartment

 

 

Neutron Dose Gamma-Ray Dose
Rate Rate
(rep/w-hr) (r/w-hr)
Scattered radiation 1.55 x 107° 7.9 % 107°
Direct beam 9.93 x 10-11 1.05 x 1078
Total 1.65 x 107° 1.84 x 1078

 

Monte Carlo calculations yielding energy and angular distributions
and dose rates for gamma rays scattered in air are also available,7 but
because the energy and angular distribution data, which fill four large
volumes, are not in a form that can be readily used with computers only
the dose-rate data could be used directly. For this preanalysis it was
necessary to obtain the energy and angular distributions of gamma rays
emitted from a point source by using the approximation of single scatter-
ing with no attenuation on either leg in air. This is probably not ter-
ribly wrong,8 except for the region of the front of the crew compartment,
where it is certainly an underestimate.

Air-scattered neutron and gamma-ray dose rates calculated by these
procedures are given in Tables 12.3 and 12.4, respectively, for a point
in air 50 ft from the reactor.

Crew Compartment Attenuation. The neutron attenuation in the crew

 

compartment was calculated by use of the Monte Carlo ABCD Code.® This
code assumes the geometry of a cylindrical crew compartment, takes into
account inelastic as well as elastic scattering, and assumes that there

is a magnetic tape containing the angular and energy distribution of

7R. E. Lynch et al., A Monte Carlo Calculation of Air Scattered Gamma
Rays, ORNL-2292, Vols, 1-5, Sept. 10, 1958.

8D. K. Trubey, The Single-Scattering Approximation to the Solution
of the Gamma-Ray Air-Scattering Problem, ORNL-2998, Jan. 20, 196l.

?Herbert Steinberg, Monte Carlo Code for Penetration of Crew Com-
partment-IT, TRG-211-3-FR. '

137
neutrons scattered in air (that is, it uses the output from the Convair
D-35 code referred to above). This angular and energy distribution is
assumed to be uniformly distributed on the outside surface of the crew
compartment in such a way that if the thicknesses on the crew compart-
ment approach zero, the dose rate in the crew compartment approaches that
in air. The code only allows one medium, making it necessary to replace
the lead on the rear of the crew compartment with an equivalent water

thickness, prescribed as follows: 10

ZPb

tHZO = th —;; XH20 = 0.58 th 5
where
t = thickness,
Zib = removal cross section of lead = 0.116 cm™ 1,
KH o= relaxation length of water = 10 cmn.
2

The small lead thickness on the side was ignored.

The attenuation of gamma rays by the crew compartment was computed
with the Monte Carlo TRIGR-P Codell for slabs. As mentioned in the pre-
ceding discussion, the energies and angles of incidence upon the slabs
representing thicknesses on the crew compartment were determined by the
single-scattering approximation. The results were integrated 6ver the
proper range of angles of incidence, and the dose rates were combined by
weighting with twice the fractional area of the particular surface of the
crew compartment., This gives the corfect limit as the crew éompartment
thicknesses approach zero. However, the weighting should have been done
at the time of integration over the angles of incidence and it should have
been with the fractional projected areas onto planes normal to the in-
cident directions. The neutron and gamma-ray dose rates inside the crew

compartment after scattering in air are given in Table 12.5.

 

107, vy, Blosser, ORNL, private commnication.

llMonte Carlo Calculations of Gamme Ray Penetration, TRG 125-FR-II
(WADC 59-771).

138

 
Future Calculations

 

As pointed out in the preceding steps, several improvements are needed
for calculations of this type. For example, the moments-method type of
calculation of the leakage spectrum should be replaced By some finite-
medium calculation which also predicts angular distributions and can handle
multiregion shields. It is doubtful that this can be done in the near
future, however. If would also be desirable to be able to predict the
thermal-neutron flux throughout the entire reactor—reactor shield system,
as well as in air and in the crew compartment. In addition, some method
for handling inelastic scattering gamma rays should be devised. This 1is
possible, but it would have to be done in conjunction with the neutron
slowing-down problem. Other aids would be the availability of air-
scattered gamma-ray data in a usable form, which would probably be best
done by recalculation, and the availability of crercompartment penetra-
tion codes which take into account the crew-compartment geometry and allow

for multiregion shields. The latter problem is not insurmountable.

 

Mathematical Description of Calculation

Consider a point isotropic source in a finite medium which is sur-
rounded by a vacuum. This source gives rise to radiation leaking from
all points on the surface of the medium. If we approximate the flux,
not the current, on the surface by that in an infinite medium, then we
are overestimating the flux. If we further assume that the radiation
leaving the surface is directed along the line of sight from the source
to the point on the surface, then the flux is overestimated in that di-
rection and of course completely neglected in other directions. There-
fore, under these assumptions, if we are interested in the flux at some
point in the vacuum, the error in the estimated flux depends on hdw im-
portant the contribution is from other parts of the surface which have
radiation directed toward the point in the vacuum. For the system in
question, the resulting leakage spectrum at the point directly to the
rear of the reactor shield is certainly overestimated. The trend of the
error is nebulous for the point directly to the front of the reactor

shield, since the obvious errors tend to compensate.

139
This analysis also holds for the leékage spectrum headed toward the
detector, if the surrcunding medium is air, but the contribution to the
flux from radiation leaving the shield surface and subsequently scatter-
ing in air must be considered. Here again, the error is nebulous, since
it depends on both\the importance of the angular distribution (including
the magnitude) of the radiation leaving the surface and the probability
of scattering in air and reaching the field point.

The estimate of the flux in air in accordance with the assumptions
resulting from gamma rays produced in the reactor—reactor shield volume

and not scattering in air is given by

~ ¢(r) : :
_¢7(E:2) = zl'{; fdl” F:“F E Za(_r_)fyl(Eo) dEq X

Z
X [zmrz?M (Ef u(Eo,27) dz’; Eo):, )

0

and the equivalent point source of gamma rays leaking from the reactor

shield is

§7(8,0) = = far ¢(x) T s [N (B ao x

Z
X [szM (Ef w(Eo, 27) at’; Eo)] ,
0 ,

where

¢7(E,D) = flux of gamma rays of energy E per unit energy
range at the spacial position D in air,

n
~
—
=
o
~
il

number of gamma rays per sec per unit energy range
per steradian emitted from the equivalent point
source in the direction given by the unit vector g,

r = spacial position of a point in the reactor—reactor
shield volume,

140

 
dr = differential volume element at r,
= thermal-neutron flux at r,

= thermal-neutron macroscopic absorption cross sec-
tion (including fission) of the ith nuclear species
at r, '

71(E0) = number of gamma rays emitted with energy Eg per
unit energy range resulting from a thermal-neutron
absorption, 12

I = distance from the source at r to the shield surface
measured along the direction of the unit vector
D<:f£ in the case of the flux in air and along the

unit vector { in the case of the equivalent point
source,

1’ = distance from the point socurce to another point
within the shield measured in the same manner as I,

IJ(EO) l/)

gamma-ray macroscopic absorption cross section at
energy Eg and at the spacial position determined

; by 17,

.l; W(Bo, 17) a1’

I

total number of mean free paths at energy Eg for
the distance I,

4r1?M(E,X;BEg) = 4m1° times the flux of gamma rays of energy E given
by the moments-method solution for a point isotropic
source emitting gamma rays of energy Eg at a dis-
tance X in mean free paths at energy Eg in an in-

finite medium of lithium hydride.

These same quantities for neutrons are

 

 

. r .
(,bn(E’.]_).) = E de—I—D—_-—lz sz(_I;) [47TZ§N(E, le)] X
b fLiH (17)
X exp f (Z -5 (1) B )dl/];
’ [ * t LiH
sn(E,-'fi) = Z]:T—T: fd; ¢(r)vip(r) [4miZN(E,1))] X
’ LiH / -p(l’) /
o[£ (47500 £22) ],

 

128, Troubetzkoy and H. Goldstein, A Compilation of Information on

Gamma—Ray Spectra Resultlng from Thermal Neutron Capture, ORNL-2904,
May 17, 1960.

141
where
l

f dl/ D(Z/))
€ o PLin
v = number of fission neutrons per fission,

l

 

I

Zf(r) = macroscopic thermal-neutron fission cross section at r,

Prif = hydrogen density in lithium hydride,

0(1’) = hydrogen density at spacial position determined by 1/,
ZLlH = macroscopic removal cross section for lithium in
t lithium hydride,

Zr(l’) macroscopic removal cross section for nonhydrogenous

material at spacial position determined by 17,

4vl: times the flux of neutrons of energy E given by
the moments-method solution for a point isotropic
fission source at a given distance ! 1n an infinite
medium of lithium hydride. ©

2
4WleN(E,le)

The reactor—reactor shield is composed of spherical or, at worst,
azimuthally symmetrical layers of materials. The problem of computing
the lengths along the directions E<:f£ or 5 across regions is done.as
follows. The radius of the layers is a function of the angle 6 that the
radius makes with the axis of symmetry; i.e.,

R = radius vector to the intersection of a line in the
direction § (from the source point) and the shield
layer in guestion,

R=10Q+r,

R(8) cos 6 = 1 cos V¥ + Z,

R2(6) = 12 + r2 + 21(r - Q),
@ = angle between symmetry axis and the direction fi,

= angle between symmetry axis and the direction fi,

 

V
Z = projection of r on the symmetry axis,
l=—(£-§) i\/(£-5)2+R2—r2
= 0.

If the layer is spherical, the problem is trivial. For the layers
which are not spherical, an iteration technique is employed. Tables of
R versus cos 6 and R versus R cos 6 are generated. An initial guess of
the value of cos 6 1s made and hence an initial R is obtained. A value

of I is then computed with this estimate of R by the quadratic formula,

142

 
4)

and an estimate of R cos € is made with the resulting value of [ by

Rcos @ =171cos ¥V +72 .

Hence a new R is obtained for the next iteration. The test for convergence

 

 

 

is
I,
i+
1 - 7 < some small number,
i
where i denotes the order of iteration. When this is satisfied, Zi+l is
accepted.

This scheme works very well as long as R varies slowly compared with
1, which is the situation for practically all cases. Fortunately, the
cases where the scheme does not work well are few in number and also un-
important. The neutron flux spectrum and dose rates on the shield sur-
face are presented in Tables 12.6 and 12.7, ahd equivalent point-source
spectra are presented in Tables 12.8 and 12.9. The dose rate in air at
spacial position D from radiation leaving the reactof—reactor shield system

and scattering in air is given by

D_(D) = fae faf s(E,0) T(E,8-D,D) |,

K3
P
I
»
Oy
S
2
i

dose rate in air due to a monodirectional monoenergetic
point source,

0p!

P

=

@
Il

equivalent point source spectrum,

o
Il

unit vector in the direction of D.
The dose rate in the crew compartment due to radiation leaving the reactor—
reactor shield system, scattering in air, and penetrating the crew compart-

ment is
Dg(D) = Sfam St s(,8) p(g,8,0) |,

143
Table 12.6. Neutron Leakage Spectrum at Shield Surface

 

Neutron Flux (27 neutrons/w.sec-cm?.Mev)

 

 

 

 

 

 

 

 

 

 

8
(deg)®
E =1.1 Mev E = 2.7 Mev E = 4 Mev E = 6 Mev E = & Mev E = 10.9 Mev
0 6.96 X 1072  3.11 X 1072 2.48 x 1072  2.02 x 1072 1.45 x 1072 7.79 x 10~
20 1.21 X 1071 5.37 X 1072  4.29 x 1072 3,42 x 1072 2.43 x 1072 1.27 x 10~2
30 2.27 x 10°Y  9.85 x 1072  7.90 x 10-2 6.18 x 10”% 4.31 x 1072 2,20 x 102
40 4,90 x 107 2,10 x 107t 1.67 x 107} 1.28 x 107! 8.84 x 1072 4.38 x 1072
50 9.78 x 1071 4.14 x 107!' 3.27 x 107! 2.46 x 107! 1.65x 1071 7.98 x 1072
60 1.87 x 10° 7.81 x 10°Y  6.11 x 107Y  4.49 x 107! 2,94 x 10! 1.37 x 107!
70 4,24 x 100 1.7 x 10° 1.33 x 10° 9.46 x 107t 5,91 x 107* 2.60 x 107%
80 9,78 x 10° 3.92 x 10° 2.88 x 109 1.93 x 10° 1.13 x 10°  4.60 x 1071
90 1.87 x 10t 7.37 x 10° 5.28 x 109 3.36 x 10° 1.88 x 10° 7.4 x 107t
100 3.44 x 10t 1.34 x 10° 9.30 x 10° 5.62 x 10° 3.01 x 10° 1.07 x 10°
120 7.00 x 10% 2.67 x 10% 1.81 x 10t 1.02 x 10t 5.17 x 109 1.70 x 10°
140 1.63 x 102 6.13 x 101 3.95 x 10t 2.04 x 10t 9,78 x 10° 2.90 x 10°
160 6.12 X 102 2.26 X 102 1.33 x 102 6.06 x 10t 2.64 x 10t 6.62 x 10°
180 7.57 x 102 2.78 x 102 1.6l x 102 7.23 x 101 3,10 x 10! 7.62 x 10°
%Defined in Table 12.3.
Table 12.7. Neutron Dose Rates at Shield Surface
e Dose Rate e Dose Rate
(eg)? (rep/w-hr) (deg) (rep/w-hr)
0 6.92 x 107 80 6.63 X 1077
20 1.17 x 1078 90 1.20 x 10-%
30 2.10 x 1076 100 2.08 x 1074
40 4,36 x 1076 120 4.04 x 1074
50 8.29 x 1076 140 9.07 X 10-%
60 1.52 x 10°? 160 3.28 x 1073
70 3.17 x 10°° 180 4,02 x 1073
aDefined in Table 12.3.
Table 12.8. Equivalent Point Source Neutron Spectra
Neutron Current (27 neutrons/w:sec-Mev-Steradian)
a
(deg) E=1.1Mev E = 2.7 Mev E = 4 Mev B =6 Mev E = 8 Mev E = 10.9 Mev
0 1.43 x 103 6.40 x 102 5.11 X 102 4,14 x 102 2.97 x 102 1.60 x 102
5 1.42 x 103 6.36 x 102 5.08 x 102 4,12 x 10% 2.96 x 102 1.60 x 102
15 1.87 x 103 8.29 x 1072 6.61 x 102 5,31 x 102 3.78 x 102 2.00 x 102
30 4,10 x 103 1.79 x 103 1.43 x 103 1.13 x 102 7.90 x 102 4,07 x 102
60 2.85 x 10% 1.18 x 10% 9,23 x 10° 6.69 x 103 4.33 x 103 1.99 x 10?
90 2.12 x 107 8.29 x 104 5.90 x 10% 3.69 x 10% 2.04 x 10% 7.60 x 103
120 6.64 X 107 2,53 x 10° 1.70 x 10° 9.38 x 10% 4.75 x 104 1.53 x 10%
150 2.10 x 108 7.80 x 10° 4,75 x 10° 2.28 x 10° 1.04 x 107 2.82 x 104
180 4.90 x 10° 1.80 x 10° 1.04 x 10% 4,67 X 10° 1.99 x 10° 4.90 x 10%

 

aDefined in Table 12.3.

144

 
Table 12.9. Equivalent Point Source Gamma-Ray Spectra

 

Gamma-Ray Current (2 photons/w?sec-Mev-Steradian)

 

 

6
(deg)® - '
E = 0.5 Mev E=2.0Mev E=4.0Mev E =6.0Mev E = 8.0 Mev E = 10.0 Mev
0 3.25 x 106 1.01 x 106 2.94 x 10° 2.89 x 10° 2.25 x 10° 9.93 x 103
5 4.45 x 106 2.48 x 108 7.02 x 10° 8.13 x 10° 7.19 x 107 5,32 X 10%
15 2.30 x 107 7.23 x 108 2.24 x 10°® 2.65 x 106 2.88 x 108 1.73x 10°
30 1.52 x 108 2.79 x 107 8.44 x 106 9,54 x 108 1.18 x 107 7.47 x 103
60 1.78 x 10° 1.41 x 108 3.56 x 107 3.16 x 107 4.00 x 107 2.44 x 1068
a0 2.20 x 10° 2.13 x 108 5.11 x 107  4.28 x 107 5.50 x 107 3.35 x 106
120 3.23 x 10° 2.66 x 108 6.19 x 107 5.18 x 107 6.70 x 107 4.07 x 106
150 4.39 x 10° 3.17 x 108 7.03 x 107 5.72 x 107 7.39 x 107 4. b4l x 108
180 4.95 x 107 3.44 x 108 7.46 x 107 5.94 x 107 7.66 x 107 4.58 x 108

 

®Pefined in Table 12.3.

where P(E,fi,D) is the dose rate in the crew compartment due to radiation
leaving a monodirectional monoenergetic point source, subsequently scatter-
ing in air, and then penetrating the crew compartment at a distance D away
from the source.

For neutron dose calculations, values of P(E,fi,D) are obtalined directly
by using the ABCD code and D-35 code for neutrons. For gamma rays it is
computed on the basis of single scattering in air and the penetration of

“infinite slabs using the TRIGR-P code. The procedure is as follows:

P(E,,0) = .=~ [ap(5,8,0,8) + Ap(m,8,0,0) + ap(e,d,0,0)]

total

where

P(E,,D,A) = dose rate behind an infinite slab with constituents and
orientation the same as area A on the crew compartment
from radiation leaving a monoenergetic monocdirectional
point source, subsequently scattering in air, and then
Penetrating the crew compartment which is at a distance
D away from the source,

Atotal = total crew compartment area,

145
AS = side crew compartment area,
AR = rear crew compartment area,
AF = front crew compartment area.

Since the equivalent point source is azimuthally symmetrical, we use

A A 1 2”- ~
P(E,0-D,D,4) = == [ d¢ P(E,q,D,4)

=V 2T /
1 do A
“mmemy J e L w9 B s (R

X I[ES, cos (K,B,(b)] H[COS (K-)B}(b)] s

where

A A

cos ¥ = (<D

2

cos (A,B,¢) = cosine of the angle between the inward unit normel and
the direction of the scattered radiation,

B = angle between axis of symmetry and the direction of
the singly scattered radiation,

¢ = azimuthal angle about the symmetry axis,

n = electron density of air,
Qg = Klein-Nishina differential scattering cross section per .
s electron,
Y+ = scattering angle,

A = unit vector denoting inward unit normal to the area A,
E. = scattered energy given by Compton formula,

I(E,cos n) = dose rate behind an infinite slab due to an incident
‘ current of photons having an energy E and a direction
given by the cosine of the angle 1 with the slab normal,

1, x>0
H(x) = {Oj x < 0.

A better estimate of P(E,Q(@,D) would have been:

146

 
" or do )
S dp fo dp n — (E,¥+8) X

P(E,8-D,0) = 2 2—1@,‘@
j <m0 a0

}\j Ccos (jij,f3,(b) II[§<35 (;;j1f3a‘bil

EiAi cos (fii,6,¢) H[cos (fii,B,q’))]

X

 

IEES,COS (KJ,B,¢)] ,

*

where i and j denote different areas on the crew compartment.
. The functions T(E,(:D,D) and P(E,Q:D,D) are presented in Tables 12.10
through 12.13.

Table 12.10. The Function T(E,8,D) for Neutrons and a Distance D of 64 ft

 

T(E,0,D) (ergs/g.hr.neutron/sec)

 

 

 

 

 

 

0
(deg)® _ -
E =1.1 Mev E =2.7 Mev E = 4 Mev E = 6 Mev E = 8 Mev E = 10.9 Mev
5 5.3 % 10710 1.72 x 10710 2,56 x 10710 2,57 x 10710 2.88 x 10719 2.96 x 10-10
15 6.13 x 10°11  5.30 x 10°11  7.96 x 10711 8.44 x 1071 7.48 x 10711 6.81 x 107!}
30 5.93 x 10-11  2.84 x 1011 3.85 x 10711 2,89 x 10711 2,73 x 10711 2.34 x 10711
60 130 x 107} 1.06 x 10" 1,54 x 107}r  1.09 x 107*Y  9.40 x 107} g.25 x 107}2
90 74l x 10712 6.08 x 10712 1.07 x 1071 6.6l x 1071? 5,12 x 10712 4.85 x 10712
120 652 x 10-12  4.06 x 10~12  8.32 x 10°12  4.20 x 10712 3,69 x 1072 3.04 x 10712
150 6.36 x 10°12  3.64 x 10712 7.63 x 10712 3.03 x 1012 3.26 x 1071%  2.83 x 10-12
180 863 % 10712 2.7 x 10712 7.32 x 10712 2.99 x 10-%2  2.44 X 10712 2.71 x 107*2
®hefined in Table 12.3.
* Table 12.11. The Function P(E,6,D) for Neutrons and a Distance of 64 ft
* P(E,6,D) (ergs/g-hr+neutron/sec)
(deg)? ~ ~ ~ ~
E = 1.1 Mev E = 2.7 Mev F = 4 Mev F = 6 Mev E = & Mev E = 10.9 Mev
5 5.08 x 10°1%  2.11 x 10712 3.57 x 10°12 1,01 x 1071} 7.6l x 107*%  1.30 x 10711
15 1.00 x 10-3%  8.28 x 10713 1.82 x 10712 2.35 x 1072  2.62 x 1071%  4.02 x 10712
30 6.78 x 10715 2.65 x 10713  1.15 x 107*%2 1,19 x 107}*  1.85 x 10°12  1.61 x 10712
60 3.08 x 1075 7.19 x 10714 3,92 x 107} 5.03 x 107}* 9.13x 107!  6.01 x 10743
90 Z.57 x 10°Y5  3.27 x 10-Y% 3,13 x 107} 2.17 x 107}? 2.4l x 10713 1,98 x 107%3
120 539 x 1016 7.91 x 10715 1.07 x 10°!2  7.96 x 107'4 1,80 x 10717  5.34 x 10-14
150 3.7 x 10716 1.70 x 1071%  5.0L x 10°** 3.0l x 107'%  1.58 x 10°1%  1.72 x 1071%
180 3.62 x 10-16  2.77 x 10715 7.44 x 107'% 2,10 x 107*%  1.66 X 1014 5,00 x 10714

 

aDefined in Table 12.3.

147
Table 12.12. The Function T(E,8,D) for Gamma Rays and a Distance D of 50 ft

 

T(E,0,D) (r/hr-photon/sec)

 

 

a

(deg) E = 0.5 Mev E = 2 Mev E = 4 Mev E = 6 Mev E = 8 Mev E = 10 Mev
1 940 x 10713 1.50 x 10712 1.80 x 10°*% 1,91 x 1077%  1.96 x 1012 2,03 x 10712

5 500 x 10-13  4.60 x 10713  5.50 x 10712  4.60 x 10712 4.50 x 10713 4.50 x 10712
15 300 x 1071%  5.50 x 10°%%  5.30 x 10-Y%  4.47 x 1071%  3.96 x 10-1%  3.62 x 10714
30 1 35 x 10-¥4  1.30 x 10-Y*  1.00 x 10-*% 8.15 x 10~%%  6.81 x 107} 5.80 x 10-1°
60 3.40 % 10715 2.80 x 1015  1.90 x 10-% 1.37 x 107! 1.09 x 10-15  g.95 x 107%°
90 181 x 10715 1.20 x 10°1%  7.30 x 10-¥6 5,45 x 107} 4.35 x 10-1% 3,58 x 10°16
120 1.14 % 10715 6.70 x 1076 4.05 x 1076 2,92 x 1071®  2.42 x 10-16 2,14 x 10°1°
150 1.0 x 10°15  5.30 x 10°16 3,20 x 107*¢ 2,35 x 107}% 1.93 x 10716 1,67 x 10-1©
180 9.75 x 10~16  4.70 x 10°'6  2.80 x 1076  2.12 x 107'® 1.7 x 10-16 1,42 x 10716

 

®Defined in Table 12.3.

Table 12.13. The Function P(F,6,D) for Gamma Rays and a Distance D of 50 ft

 

D sin 6 P(E,0,D) [{r-cm)/{hr-photon/sec)])

 

 

(deg) E = 0.5 Mev E = 2 Mev E = 4 Mev E = 6 Mev E = 8 Mev E = 10 Mev
5 1.04 x 10-1%4  3.89 x 10712  6.32 x 10712  6.84 x 10712 6.79 x 10-1?  6.56 x 10713
15 713 x 1075  1.50 x 107}3 2,11 x 107'*  2.12 x 10712 2.00 x 10713 1.85 % 107%°
30 4ooh % 10715 3.14 x 10-Y%  4.00 x 1074 3,77 x 107'%  3.41 x 10-14  3.06 x 10714
60 5 46 x 10715 1.71 x 10715  1.50 x 107 1.31 x 10-*® 1.13 X 10-1% 9,80 x 10-13
90 T mo % 10715 6.07 x 10716 3.42 x 10716 2,40 x 1071 1.85 x 10-16  1.51 x 10°'°
120 9.7 % 10716 3.24 x 1016 1.76 x 10716 1.21 x 1076  9.24 x 1017 7.47 x 1077
150 5 56 % 10°16  1.70 x 1076 8.98 x 10717  6.12 x 10717 4.64 X 1017 3.74 x 10717
180 0 0 0 0 0 0

 

®pefined in Table 12.3.

Preliminary Results of the Pratt & Whitney Divided-Shield
' Experiment at TSF

V. R. Cain

The Pratt & Whitney divided-shield mockup experiment at the TSF,
discussed in the preceding Section, is the first such experiment with the
TSR-IT as a source. Some preliminary results are available.

The Pratt & Whitney divided-shield mockup consists of a reactor

shield}3,1% constructed to a design which was optimized insofar as

134, C. Woodsum, Pratt & Whitney Aircraft Shield Mockup Experiment for

the TSR-II, Sixth Semiannual ANP Shielding Information Meeting, NARF-59-14T,
vol. III. |

14ANP Semiann. Prog. Rep. Oct. 31, 1959, ORNL-2840, p. 139,

148

 
possible with the available data and the c¢ylindrical compartmentalized
crew compartment fised in. earlier experiments at the TSF. Both are shofin
suspended. at the TSF in Fig. 12.1, and another view of the reactor shield
is shown in Fig. 12.2. The reactor shield consists of a lithium hydride
neutron shield and a uranium gamma-ray shield, both largely concentrated
on the side facing the c¢rew compartment. Except for the region above the
core required for cooling-water flow and the reactor controls, the lithium
hydride shield surrounds the reactor core. The uranium gamma-réy shield
is constructed in the shape of a lens and is contained within the thickest
section of lithium hydride. The various shield regions can be seen in
Fig; 12.3. Additional dimensional details are given in Table 12.1.

In order to check the optimization of the reactor shield, a water-
filled "patch" tank, also shown in Fig. 12.3, was constructed so that it
could be moved around the shield in the horizontal plane.

The TSF compartmentalized crew shield is shown in cross section in
Fig. 12.4. The shaded areas indicate the tanks that are filled in the -
calculated "optimum" cofifiguration. During the course of the experiment,
these tanks are being filled with Watér, transformer oil,‘or a mixture
of transformer oil and an organic boron compound called "Borester 7"
(C18H3606Bo). Table 12,14 gives a comparison of some of the properties
of these liquids, along with those of a type of borated polyethylene which
might be used in a high-performance shield. The optimum configuration
results in liquid thicknesses of 18 in. on the sides and 19.5 in. on the
front. (The front thickness is intended to be black to the small amount
of radiation expected from this direction.) Not shown is the internal
lead shielding, which consists of 3.27 in. of lead in the rear (next to
tank No. l) and 0.20 in. of lead on the sides in the optimum configura-
tion. Most of these crew shield parameters are to be varied during the
experiment.

The coordinate system used to specify in-air detector positions in
the experimental program is shown in Fig. 12.5. The angles ¢ and 6 des-
ignate the detector position in the vertical and horizontal planes, re-

spectively, and p is the distance from the center of the reactor to the

149
 

0T

i TF kST

S S

(& —F T W ——
BT FrY- ¥ R

!
;
e

N ]

TS

| & ¥ .

 

UNCL ASSIFIED

| PHOTO 53817

3 - -y . .
'. e M TN .

Pratt & Whitney Divided Shield Mockup Suspended at TSF.

 
[ UNCL ASSIFIED
44?'.: PHOTO $3820
Nl

     

| __HAZ4RD | : =T
KEEP OUT = -':-n.- ‘_'- . : """“‘*--..

Fig. 12.2. Pratt & Whitney Shield Around TSR-II.

L5l

 
  

 

 

 

   
 
 

 

 

 

 

 

 

0
‘Y{

EALLELLL X

AR AT IE T T IR,

 
 

 

B L

 

 

31.45

 

 

 

 

 

 

 

 

¥ TOWER SHIELDING REACTOR I

—— 53750 max, gz |
S 2250 MIN pig

P
S5y
O/Ifé\
W

Sp&

 

 

 

 

 

   

 

 

 

 

 

 

 

L ELTLLL L LA LT

 

 

 

 

ol

 

>4\

DEPLETED URANIUM

 

 

ORNL-LR-DWG 58223

MOVABLE WATER-
FILLED PATCH

  

AXIS OF

 

 

 

 

_J

 

    

LITHIUM HYDRIDE

ALL DIMENSIONS ARE IN INCHES

4 0 4 8 12
e — e —
INCHES

 

Fig. 12.3.

152

Cross-Sectional View of Pratt & Whitney Reactor Shield.

SYMMETRY

-

A\l
 

131.60 in. OUTSIDE

 

UNCLASSIFIED
2—01—-056—15—P-490R2

 

e——37.05 in. OUTSIDE ‘—‘”-—7 73,68 in. OUTSIDE——————-‘iZOAS in. OUTSIDE —=

12.0in. (TYP)
{10 10a 9 16 21 26 3

 

 

 

 

 

 

 

 

 

 

 

 

1242 in.—= frea—
g G |
t : ]
z < 4 t
< _~ 36.0in. DIA ~ D U D U
rol oz K
8 et ‘{ 17.74in. Dm—T
g 18.0 in. OD
—0.188in.
1
0.188 in. Q.188 in.
12.36 in; 24.06in. (TYP} 7.96 in.
O.IBBI‘Ln.
250in.% n N 0250,
22T Nosron %,
ALL METAL THICKNESSES ARE 0.125in. (TYP) UNLESS SPECIFICALLY DESIGNATED OTHERWISE
Fig. 12.4. Cross-Sectional View of TSF Compartmentalized Crew Com-
partment. |
UNCLASSIFIED
2-01-060-95
OAK RIDGE GRID NORTH
—J
<1
Q
—
4 09
L)
- |
¢ £ |
////
// /
— /
. f )
" &
= r /- . )
CENTER LINE OF TSF PAD
REACTOR CENTER
Fig. 12.5. Coordinate System Used for TSF Pratt & Whitney Ex-
periment. ' ;

153
Table 12.14. Compositions of Crew Compartment Shielding Materials

 

 

 

Compositicn
Material Relative
ateria Density Hydrogen Carbon Oxygen Boron
(atoms/cm?®)  (atoms/ecm®)  (atoms/cm®)  (wt %)
_ x 1022 % 1022 X 1022
Water 1.00 6.69 3.34
Transformer oil: &6.7 wt % C, 0.87 6.6 3.78
12.7 wt % H
Transformer oil and Borester '7: 0.92 6.2 3.35 0.45 2.9
50 wt % each
Borated polyethylene: 3.8 wt % 0.9 7.8 3.94 3.0

B4C, 96.2 wt & (CH,),

 

detector. When measurements are being made in the crew compartment, the
distance from the reactor center to the rear surface of the crew shield
cavity is 50 ft, with the crew shield centerline on the line determined
by 8 = 0 and ¢ = 90 deg. The distance between the rear lead in the crew
shield and the center of detection of the detector is designated by the
symbol x,

At times the reactor is rotated about its vertical centerline, in
which case the counterclockwise angle through which it turns from its
normal position (i.e., with the thickest point of the shield at 6 = O and
¢ = 90 deg) must be given. The symbol used for this angle is Q.

Thus far, the experimental program is about one-half complete. There-
fore, the description of the program which follows is divided into measure-
ments already completed and p}anned measurements.

Completed Measurements. Most of the measurements completed to date
have been made in air around the reactor shield. At each point the measure-
ments usually included the fast-neutron dose rate, the gamma-ray dose rate,
and the thermal-neutron flux. The detector positions used for these measure-
ments are given in Table 12.15. For each position a complete rotation of
the reactor was made; i.e., O was varied from O to 360 deg. Samples of
these measurements are shown in Figs. 12.6 and 12.7, which represent map-
pings of the fast-neutron and gamma—ray dose rates in the horizontal plane

50 ft from the center of the reactor as a function of .

154

 
Table 12.15, Coordinate Positions
for In-Air Measurements Around
Pratt & Whitney Shield

 

h, Reactor

 

0 ¢ (deg) Altitude
(ft)
58.6 in. 90 100
65.1 in. 67.5 100
65.1 in. 112.5 100
50 ft 90 15
50 ft 90 50
50 ft 20 175
75 ft 45 125
75 ft 20 175
75 Tt 135 175

 

In addition to the in-air measurements at various distances from the
reactor, several points were obtained with counters directly on the surface
of the shield. Also, an altitude traverse from h = 6 ft to h = 175 ft
was run for a detector position specified by p = 65 ft, € =0, and ¢ = 90
deg. For this run, the reactor was at o = O.

Some of the dose-rate measurements inside the crew shield have also
been completed. One altitude traverse was made with three detectors at
the lateral positions shown in Fig. 12.8. For this traverse, the detector
coordinate x was at 36 in. In all other cases, x was varied between 12.5
and 53.5 in., continuously for gamma-ray dose rates and thermal-neutron
fluxes and in about &-in. steps for fast-neutron dose rates. Data of this
type have been taken with the optimum crew compartment configuration using
water, oil, and the borated oil mixture. Data for the borated oil mixture
were also taken with all the front tanks (tanks 27, 28, 29, 30, 32, and
33) drained to determine the relative amount of radiation penetrating
the forward compartments. |

Measurements have also been made inside the crew shield as a function

of the position of the water-filled patch. These were performed with the

155
optimum crew shield configuration, except that water rather than borated

olil was used in the tanks.

UNCLASSIFIED
2- 04 - 056 -XXX-832

¥

-1
TISSUE

o

FAST-NEUTRON DOSE RATE (ergs-g

 

w05

180 200 220 240 260 280 300 320 340 O 20 40 60 80 {00 20 140 160 180
a, REACTOR HORIZONTAL ROTATION ANGLE (deg)

Fig. 12.6. Fast-Neutron Dose Rates 50 ft from Center of Reactor as
a Function of @, TSR-II in Pratt & Whitney shield. '

156

 
L

»

GAMMA-RAY DOSE RATE (ergs-q}'lssu E-hr"-m" )

   

Planned Measurements. Additional in-air mapping measurements are

 

planned, primarily traverses in the angle ¢ with p = 25 ft and with various

UNCLASSIFIED
2-01-056 -XXX -833
Tom

1072

wn

~N

10-3

 

104
180 200 220 240 260 280 300 320 340 O - 20 40 60 80 100 120 140 160 180
a, REACTOR HORIZONTAL ROTATION ANGLE (deg)

Fig. 12.7. Gamma-Ray Dose Rates 50 ft from Center of Reactor as a
Function of &, TSR-II in Pratt & Whitney shield.

157
values of 6., Collimated dose-rate measurements will also be made as a

function of reactor rotation, with the collimator looking only at the

shield surface, to determine scattered and direct components of the in-air

measurements, and some foil measurements directly on the reactor shield

surface will probably be

   
 

 
 
  

/ HOLDERS FOR FND,
ASD, OR BF,

made.

 

 

 
      
     

 

  

CAVITY
CENTERLINE

POSITION C

 

UNCLASSIFIED
2-01-056-XXX~-834

  
    
   
   

 

HOLDER FOR .
PARAFFIN-COVERED \
BF3 COUNTER

   

 

 

     

 

 

 

 

 

 

 

 

 

 

 

TRAVERSING
MECHANISM

MAXIMUM OF
i-in. Pb SIDE
SHIELDING
INSIDE CAVITY

Fig., 12.8. Counter Positions Inside 36-in.-diam Cavity in Crew Com-

partment (Looking Toward

158

Reactor).

 

~d
»

3

 

Gamma-ray and neutron spectral measurements are desired, the most
urgent being the neutron spectrum at the rear of the reactor shield. If
time permits, both neutron and gamma-ray spectra will be taken at several

positions.

TR -~
Reports previously issued in this series are as follows:

160

ORNL-528
ORNL-629
ORNL-768
ORNL~858
ORNL-919
ANP-60
ANP-65
ORNL-1154
ORNL-1170
ORNL=-1227
ORNL~1294
ORNL-1375
ORNL-1439
ORNL-1515
ORNL-1556
ORNL-1609
ORNL-1649
ORNL-1692
ORNL-1729
ORNL-1771
ORNL-1816
ORNL-1864
ORNL-1896
ORNL-1947
ORNL-2012
ORNL-2061
ORNL-2106
ORNL-215%
ORNL-2221
ORNL-2274
ORNL-2340
ORNL-2387
ORNL-2440
ORNL-2517
ORNL-2599
ORNL-2711
ORNL-2840
ORNL-2942
ORNL-302%9

Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Perioad
Period
Period
Perioad
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period .
Period
Perioad
Period
Period
Period
Period

S

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
Ending

November 30, 1949
February 28, 1950
May 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
October 31, 1960

 

L1

b
)

O R -gowumduwH

?flF:?fl?fl§1$1}+§2§1;:Fip*?-Fzgizzgzfiaz:;iz:w g QrrdorEmEHOn g

 

ORNL-~3144

C-84 — Reactors—Special Features
of Aircraft Reactors
M-3679 (24th ed.)

INTERNAL DISTRIBUTION

M. Adamson 44, F. C. Maienschein

W. Allen 45. W. D. Manly

E. Beall 46. A. J. Miller

S. Billington 47. K. Z. Morgan

F. Blankenship 48. F. J. Muckenthaler

P. Blizard 49, M. L. Nelson

L. Boch 50. P. Patriarca

BE. Boyd 51, S. K. Penny

B. Briggs 52. A. M. Perry

D. Callihan 53. P. M. Reyling

E. Center (K-25) 54, H. W. Savage

A, Charpie 55, A. W. Savolainen

E. Clausing 56. E. D. Shipley

S. Cockreham 57. 0. Sisman

B. Cottrell 58. M. J. Skinner

L. Culler 59. G. M. Slaughter

A. Douglas 60. A. H. Snell

L. Fowler 61. I. Spiewak

P. Fraas 62. B. J. Sturm

H. Frye 63. C. D. Susano

J. Gray 64. J. A. Swartout

L. Greenstreet 65. D. B. Trauger

R. Grimes 66. D. K. Trubey

Guth 67. G. M., Watson

0. Harms 68. A. M. Weinberg

Hikido 69. J. C. White

R. Hill 70. E. P. Wigner (consultant)
. E. Hinkle 71, C. E. Winters

E. Hoffman 72-78. Laboratory Records Department

W. Hoffman 79. Laboratory Records, ORNL R. C.

Hollaender 80-83. ORNL — Y-12 Technical Library

B. Holland Document Reference Section

Tnouye 84—86. Central Research Library

H. Jordan

W. Keilholtz

P. Keim

J. Keyes

G. Lafyatis

R. Laing

S. Livingston

N. Lyon

G. MacPherson

E. Maerker

- 161
87.
88-91.
92-93.

9.
95-97.
98—99.

100.
101-102.
103-104%.

105.
106—108.

109.

110.
111-112

113.

114,
115-120.

121.

122.

123.
124~125.

126.

127.

128.

129.

130.

131.

132.
133-134.,

135.

136.

137.

138.

139.
140-147.
148-150.

151.

152.

153.
154~155.

156.

157.

158.

159.

160.

161.

162.

163.

164 .

162

EXTERNAL DISTRIBUTION

AiResearch Manufacturing Company

Air Force Ballistic Missile Division

ATPR, Boeing, Seattle

AFPR, Douglas, Long Beach

AFPR, Douglas, Santa Monica

AFPR, Lockheed, Marietta

AFPR, North American, Downey

Air Force Special Weapons Center

Air Research and Development Command (RDRRA)
Air Technical Intelligence Center

ANP Project Office, Convair, Fort Worth
Albuquerque Operations Office

Argonne National Laboratory

Army Ballistic 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 Laboratory

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, Aiken

Engineer Research and Development Laboratories
General Electric Company (ANPD)

General Electric Company, Richland

General Nuclear Engineering Corporation.
Hartford Aircraft Reactors Area Office

Idaho Test Division (LAROO)

Knolls Atomic Power Laboratory

Lockland Aircraft Reactors Operations Office
Los Alamos Scientific Iaboratory

Marquardt Aircraft Company

Martin Company

National Aeronautics and Space Administration,
National Aeronautics and Space Administration,
National Bureau of Standards

Naval Air Development Center

Naval Air Material Center

 

Cleveland
Washington

 

b
A

P

o

165.
166.
167.
168.
169.
170.
171.
172.
173-174.
175~178.
179.
180.
181—-182.
183,
184,
185.
186.
187.
188.
189-190.

191-193

194~205.
206—~230.
231.

 

Naval Air Turbine Test Station

Naval Research laboratory

New York Operations Office

Nuclear Metals, Inc.

Oak Ridge Operations 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. S. Naval Radiological Defense ILaboratory
University of California, Livermore
Westinghouse Bettis Atomic Power Laboratory
Wright Air Development Division

Technical Information Service Extension
Division of Research and Development, AEC, ORO

AL
e TR

A ihe

S 163
 

SESTRICIED DRI .
This document contains Repssssinsretpeimrpemsegesinsd in the Atomic Energy

Act of 1954, 1Its t disclosure of its contents in any
manner to an unauthor i ‘ wed. '

3