ORNL-2711.txt

o \H‘I
(TR -, £k
- 3 445k 0251079 9

ORNL-2711

C-85 - Reactors-Aircraft Nuclear N
AEC RESEARCH AND DEVELOPMENT REPORT Propulsion Systems
M=3679 (22nd ed., Rev.)
777 2\
rraRy el =
¥ S S
3 S PECIG B R
o ¥ ¥ - oe awey be calfeae 2 ame
> a 5 23 OV STV (Y S—
E% 8 = $30 AIRCRAFT NUCLEAR PROPULSION PROJECT X
== e #
T8 =25 .58 SEMIANNUAL PROGRESS REPORT
TIgI<:t
3 o= 2t= FOR PERIOD ENDING MARCH 31, 1959
B -t : Esi
AT
=>5 -85
Q)= ==
EQ L E 82
zo X 3y E
s E g =335 =
= o
=]
g
S

OAK RIDGE NATIONAL LABORATORY

operated by
UNION CARBIDE CORPORATION
for the
U.S. ATOMIC ENERGY COMMISSION

ORNL-2711
C-85 — Reactors-Aircraft Nuclear
Propulsion Systems
M-36T79 (22nd ed.)
This document consists of/ 7fpages.

Copy 7?01:“ 225 copies. Series A.
Contract No. W-TL0O5-eng-26
ATIRCRAFT NUCLEAR PROPULSION PROJECT

SEMIANNUAL PROGRESS REPORT

For Period Ending March 31, 1959

Date Issued

APR 251959

OAK RIDGE NATIONAL LABORATORY
Oak Ridge, Tennessee
operated by
UNION CARBIDE CORPORATION
for the
U. S. ATOMIC ENERGY COMMISSION

MARTIN MARIETTA ENERGY SYSTEMS LIBRARIES
- |

3 445k 0251079 9

FOREWORD

The ORNL-ANF 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 bpasic to or in direct support of investigations under way at
General Electric Company, Aircraft Nuclear Propulsion Department, and
Pratt & Whitney Aircraft Division, United Aircratt Corporation. This
report is divided into four major parts: 1. Metallurgy, 2. Chemistry

and Radiation Damage, 3. Engineering, and 4. Shielding.

CONTENTS

Sm-‘w—iR L B BN BB B R BN N N R N N N RN B R RN I N L R N BN RN RN BN R N BN TR N BN N N R BN RE N RN B R R B RN BB BN R N N

1.1,

1.2.

1030

1.4,

PART 1. METALLURGY

MAT’ERIALS FABRICATION RESEARCH LB BN B BN BN BN B BN BN B BN BN L BN BN BN B BN CBE BN BN BE BN B BN BN BN AN

Yttrium and Yttrium Hydride Studies eeeecsccccescescscsoscscsn
Yttrium Preparationl ececececcesccccccccassoassscscosassssce
Zone Refining of Yttrium ceeececcccccscvssccccocssnsnncs
Single Crystal Growth by Zone Refining .ceccecesvrcccces
Solid-State Electrolysis as a Means of Purifying
YtErium seveccecocecocecoscasecseosssssossvssossssosnssnse
Deformation Mechanisms of Ytirium seeceeecccccccccccnce
Yttrium Hydriding eceeccecccscececsccooscssccsscsscoccne

Preparation of Columbium AllOYS ceseoscecceccecsccosssssccnss
Effect of Zirconium AAQ1tiONS eesecccscccsccssscscsccsce
Effect of Beryllium AAAitions sseeecccccscsscccncconncsne
Effect of Additions of Rare Earth Metals and Yttrium ...

Studles of Reactions of Columbium with Oxygen and Nitrogen ..
EXtruSion Of Clad Refractory Mewls [ B BN BN B B BN BE BN BN BN BE BN BN BN BN BE BN BE BN BN BN BN BN N |
CORROSION ST{.]DES LA N B BN BN B B BN BN BN BN BN BN BN B BE DR AR L BN BE B BN BN R BN BN B B BN BN R B BL BN IR BN BK BN BN BN BN ]

Effects of Oxygen and Nitrogen on the Corrosion Resistance of
Columbium to Lithium at Elevated Temperatures seeecececsce

Compatibility of Columbium with Molten Lithium Contaminated
with Lithium Nitride or Lithium Oxide Ssscs0essssscesssseacs

Dissimilar Material Mass Transfer Effects on a Columbium—
Sodium=Type 316 Stainless Steel System eeceeescccsccsssscs

Mechanical Properties of Type 316 Stainless Steel After
Exposure to Lithilml [ B BN BN BN BN NN BN B RN RN BN N BE B BN BN BN BN BN BN BN BN BN BN BN BN BN BN BN BN BN BN BN BN BN BN

SOldbility of Lithium Oxide in Lithium esecseesssesesesnsse
Lithium Corrosion Of Cermets P S8 0 B0 EIIROSIOIRIOEIESIOBROESINROEESTSTS
wEI‘DDIG AND BRAZING STUDES ® & 08 &0 5 08 00 800 S0 S OO0 SBESSSE sl

Development of Refractory-Metal-Base Brazing Alloys for
Joining Container Materiasls for Molten Lithium ecececececse

WEIding Studies on Columbium © 08 0000 0000000000000 00000000O0OFTE
MECHANICAL PROPERTIES INVESTIGATIONS eseecescecccsscccscsccns
Testing Procedures 0 0 0 0000080000600 06000600600 000000000080 O0OCRTGTE

Mechanical Properties of Pure Columbium cececececcccsccoccsces
Material L BN BN RN RN BN B B BN B B BN B BN BN B NN RN BN B NN BN BN BN BN BN BN B BN NN BN N BN NN NN NN NN B BE BN BN BN NN BN BN

Recrysta-llization Stlldies LI L BN BN B BN BN BN BN B B BN BN B BN BN BN B BN BN B B BN BN BN B BN )

ix

O N0\ o FLwww w

Ho
H O
1.5.

2.1,

2.2.

2.3.

3.1.

vi

Creep=Rupture TestS seesescescscscsccvsccnssacccccensas
Tensile TesStS eescessccrccoscoscoscossssscsscssssssonssnsnse
Comparison of Results with Those of Other
InvestigAtors eeeeocsoeccscscssscccsscnaccccscssonsce
Fatigue TeStS eeessessessscctssssscscccoscsncsscscsnnsses

CERAMICS RESEARCH L B B B B A B AN B B B I N BN AL BN BN R B DR B BN BN BN BN BN BN BN BE BN BN BN BN BN BN BN BN N

Beryllium OxXide StUJieS eeessccsescccsocscncsccsscsscssscssnnsos
Requisites for Sinterabllity ceccecescceccccscsccccasne
Grain Growth Studies ¢80 000000000000 sResRIBCRIOOIOBOLS

PART 2. CHEEMISTRY AND RADIATION DAMAGE

MArIERIALS CIEMISTRY LB B B B AN L BN BN I B BN B BN BN L BN BN BE BN BN BN BN BN BN BN BN BN BN BN BE B B BN NN BN BN NN )

Preparation of Charge Material for Reduction to Yttrium ...
Phase FEquilibria in the LiF-YF_ System eseccesccccccces
Preparation of YF3 from ¥p03 Jeeeecescrcsresccsncnases
Laboratory-Scale Preraretion of LiF-MgFE-YF3

Mixtures LI B B B BN B BE AN BN B BN BN B BN BB B NE O BN BN B R B B N BN BN BN B B B B B B B RN NN

Production~Scale Operations ceeeesccccscccscsccsccsesce
Preparation of Pure Beryllium OXide eseseccosccescccecccons

Purification of Lithium Metal by Extraction with
MOlten Salts LI R BN B BN BN BN BN BN BN BN BN RN B BN RN BN BU NN BN BN BN BN BN B B BN N I I IE BN I I I I I A B A I N )

ANALYTICAL CHEMISTRY SO P 0 S0 G000 P SOOI N0 CANPSIROOIOSOIOIEIBDROETRIREES
Determination Of Oxygen in LiF-MgFe—YFB 0 0002000000000t
Determination of Oxygen in Yttrium-Magnesium AllOY ecceccacee

Analyses of Yttrium Metal eceececcccccssceccccccscscaccscasns
Determination of OXYZEN ceeevececescscocccscccssnsssnsns
Determination of Nitrogen .ccecceccccscsccssscsccccanans
Determination of Fluorine +seececececssccccscscsscnsoses
Determination of Magnesium .ececescesscescccsanccecsss
Determination of Lithium eeccesccccocccsccsssscsccssasce
Determination of Titanium eseceecccccccsccecososcsccosesse

RADIATION EFFECTS LR B BB B B BN B BN B B I BN N N N I N N N N N NN N NN N NN NN N EEERE R
DiOde Experj_ment LA B B B B B BN B BN BN B B BN N NN BN NN N RN W I I N N I N N AN NN NN NNEN]
Creep and Stress Rupture Tests Under Irradiation .cececececes

Irradiation of High-Temperature Moderator Materials

in ETR LB A NN NN R R RN RN NN NN NN NN NN N N RN

PART 3. ENGINEERING

CODI:POI\ENT DEVELOPIENT AND TESTING L AR BN B BN BN BN B BN BN BN BN BN N BN B Y BN RN B NN WY
Irradiation Test of an Oil-Lubricated Pump Rotary Element ..
Dynamic Seal ResS€arCh sesecscecsccccecssscsssccscssccsccsnces
Thermal Stability Tests of Metal Shells .eeececccccsceacons
ENGINEERING AND HEAT TRANSFER STUDIES eeecescesccssccscccses
Effect of Thermal-Stress Cycling on Structural Materials ..
Molten Lithium Heat Transfer eoeesceccesccesccccsccsccssens
Thermal Properties of Columbium and Lithium ececececcoccces
INSTRUMENTATION AND CONTROLS cceccescescccscccscssscscacass
Data Acquisition System eeeeeccccecsscceccsccosscoscscsssssse
Liquid=Metal=Level TransduCers csecesccssccssessccscsccssccss

Thermocouple Development Studi€S eceeesccccscccsccsscsscase
Thermocouples for Use at High Temperatures cecesececcsses
Chromel-Alumel Thermocouple Life TesStsS sececsssccscease

APPLTIED MECHANICS ceesccccscccoccccccncescssssssscscssscss
Basic Problems in Elasticity eecceccecescscescccccascssssee
Creep Buckling of Shells with Double Curvature ececeeseccecsee
ADVANCED POWER PLANT DESIGN eeseccesacsscccsssccscccsccsce

Vortex Reactor Experj—rflents LA R R NNEENENENENENNENNENNENNNENNNEFEEN]
Separation EXperiments sececcecccscesccscssccssccscsse
I.a-mimr InjeCtion LN B NN E N NNEENNENNNENNNNNENNNNENNNEENENEEEE X

Satellite Power Plant eceeeescceccecscccessscscscocossncoes
Radiator Design Considerations ececececccesccssssscocsces
Turbine Design Considerations ecececescececescsscssscsee
Choice Of CYCle eeeesssecescsceccassscscscsssssccnscas
Rubidium=-Vapor-Cycle Power Plant eeecceeccscccccccsses
Ma jor Development ProblemsS scecscescssssscssscscccssse

PART L4, SHIELDING

SHEIJDDIG TIEORY (I B N NN NN ENEEREN NN NE NN NN NN R R
Response Functions of Sodium Iodide Crystals eecececccccsces

A Monte Carlo Code for Computing Fast-Neutron Dose Rates
Inside a Cylindrical Crew Compartment eccesecesscseccscscccss

Comparison of LTSF Measurements of Thermal-Neutron Fluxes
in Water with Cornpone Reactor Code Calculations eecececess

Prediction of Thermal-Neutron Fluxes in the BSF from

LTSF Data (A B N BN NENEENENENRNENEENENNENNEENNNENNENENENNFENNNENNNENE XK

A Monte Carlo Code for the Calculation of Deep Penetrations
of Gamma Rays P C 0 000 0000 0000000000660 80000600000 GC0CGCRORAGESETS

67
69
72
T2
[
78
19
79
80

80
80
80

83
83
83
88

88
89
92

92
94
99
100

101
107

111
111

111
11k
116

120

vii
Ll'.e.

L.3.

Lok,

hoso

viii

——

Calculation of Bremsstrahlung Dose Rates Which Result from
an Activated-Lithium COOling SYStem ceesssessssessenease s

LID I:EANK SHELDING FACIIJIIIY LN B BB BE B BN BN BN BN BE BN BN B BN BN B BN BN N NN RN A B BN B NN N

Experimental Flux Depression and Other Corrections for
GOld FOilS Exposed in Water 0PI RPN RROOIBRGIOIROETRATRSTS

BULK SHIELDING FACILITY seececocccsoccccsceascccscccsccoans
The GE~BSF Study of Radiation Heating in ShieldS .eeecseases
The BSF Stainless Steel-U0O, Reactor (BSR=II) ecceccccscsnas
The Model IV Gamma«~Ray Spectrometer eseesecscsccscscscccnns
Cross Section for the Li6(n,a)H3 Reaction for 1.2

sE 5.800 Mev OO 8 0 &80 0RO O OB NSO SO B0 OO RNOORNBOOSSSP ORISR

The Energy Spectrum of Prompt Gamma Rays Accompanying the
Themlal FiSSion Of U L L B BN BN B BN BN B BN BN BN B B BN ORK BN BN BN BN BN BN B BN BN NN BN BN IR BN NN N

TWER SHEIJDII‘IG FACILITY ® 0 00 &4 58 S0 800 00O OSSR SBPEH NSNS e

TSF Studies in Support of the GE-ANP Program: Rechecks and
Extensions of the 277 Experiments cecececesscscasccssscses
Recheck of Fast=Neutron Measurements cecseceececsccscccce
Recheck of Gamma-Ray Measurements secececcceccscsccccces
Recheck of Thermal-Neutron Measurements eceecesececccces
Mapping of Radiation Around the Reactor Shield c.eceee
Measurements of Gamma~Ray Spectra eceecececcccsccsccos

TG.JER SHEIIDING REACTOR II o 0 090 00 08 00 0O S S SO S G0 SO S S SS S ERDS
Critical Experiments and Calculations eeeccssccsssccsssccss
Flow Distribution Studies L BB BN BN BN BR BN R BN DR BN BN BN BN BN BN BN BE BN BN NN BN BN NN RN BN BN BN BN BN B ]

Heat Transfer STUudies cececescocccccetesssssscsssncssascas

121
124

124
127
127
128
132

133

136
139

139
140
140
141
141
141

143
143
145
146

ANP PROJECT SEMIANNUAL PROGRESS REPORT
SUMMARY

Part 1. Metallurgy

l.l1. Materials Fabrication

Developmental studies on a pilot-plant scale of methods for the
production of high-purity yttrium metal were continued. The impurities
of major concern in the material produced thus far are oxygen, nitrogen,
and fluorine. Purification of yttrium by zone refining on a laboratory
scale was found to be somewhat more effective during runs carried out
in vacuum than in runs carried out in an inert atmosphere, based on the
removal of impurities by volatilization and on the microstructure.
Single crystals have been grown successfully by careful regulation of
the rate of zone travel,

Numerous alloys of columbium with small amounts of zirconium,
beryllium, and rare earth metals have been prepared to test the effects
of these additions in stabilizing dissolved oxygen in the columbium.
Arc melts were prepared and reduced to strip for use in corrosion tests.
Alloys containing zirconium and beryllium were subject to cracking
during cold rolling, but little difficulty was encountered in rolling
the alloys containing rare earths, No corrosion of these alloys was
observed after exposure for 100 hr in static lithium at l500oF.

The reaction rates of oxygen with columbium at low pressures and

o
at temperatures of 850 and 1000 C are being determined to establish the

limiting pressure permissible for exposure of the columbium. The reaction

is being studied by following the change in weight of a specimen exposed
at selected temperatures and pressures. At oxygen pressures below
% x 1072 m He at 850°C and 5 x 1070 mm Hg at 1000°C, the reaction rate
is negligible., A tentative relation between the reaction rate, time,
and pressure has been determined.

Two attempts to extrude molybdenum clad with type 446 stainless

steel were unsuccessful because of segmentation of the molybdenum.
Extrusion billets of columbium clad with type 446 stainless steel and

columbium and molybdenum clad with aluminum bronze have been prepared.

1.2. Corrosion Studies

Investigations of the mechanism of corrosion of columbium in lithium
have revealed that attack is due primarily to oXygen contamination of
the metal. Columbium oxide at grain boundaries is reduced by lithium
at elevated temperatures to form a two-phase grain-boundary corrosion
product of lithium oxide and columbium; columbium nitride is not attacked.
Corrosion tests were also run in which columbium was exposed to static
lithium to which 1lithium oxide or lithium nitride had been added. These
additions appeared to have no effect on the corrosion attack.

Tests to determine the compatibility of components in a columbiunr
sodium-type 316 stainless steel system indicate that nitrogen and carbon
are leached from the stainless steel and deposited as columbium nitride
and columbium carbide on the surface of the columbium. There was columbium
on the surface of the stainless steel, and the carbon content of the
steel had decreased.

Grain-boundary attack and reduced ductility were observed in type
316 stainless steel specimens following exposure to lithium contaminated
with lithium nitride., Lithium oxide in lithium had no significant
effect on the steel in similar tests.

The solubility of lithium oxide in lithium was determined for the
temperature range 250 to 400°C. The solubility of approximately SO ppm
at 25000 was found to increase rapidly to approximately 640 ppm at MOOOC.

Nickel-bonded TiC-base and cobalt-bonded TiC- and WC-base cermets
in which the binder ranged from 3 to 19% (by weight) were found to have
poor corrosion resistance to molten lithium at lSOOOF in 100-hr exposures.
Tungsten~-bonded and molybdenum-bonded TiC-base cermets are being fabri-

cated for testing.

l.5. Welding and Brazing Studies

A development program is being conducted to find refractory-metal-

base brazing alloys for application in high-temperature reactors using

x -/
lithium as a coolant. The alloys found to be acceptable consist primarily
of zirconium and titanium. Sound joints have been produced on the follow-
ing base metals: zirconium, titanium, molybdenum, and columbium,

The welding of columbium without the use of controlled-atmosphere
chambers is being studied. The inert-gas, tungsten-arc process is being
used on sheet columbium, and the inert-gas-shielded, metal-arc process
is being used on plate columbium. Ductile welds with bright, clean sur-
faces have been produced on both plate and sheet. The corrosion resistance

of the welds to lithium will be determined.

1.4. Mechanical Properties Investigations

Techniques have been developed for tensile tests of columbium in
the temperature range 1800 to 2000°F in an inert atmosphere. Hardness
and chemical analysis data have indicated that the test system being
used is effective in preserving the purity of the columbium.

Creep rupture tests have shown the creep strengths of two heats to
differ substantially. In comparisons of data from several sources,
small variations in residual impurities were found to have a very marked
effect on the tensile and creep properties of the unalloyed metal.

The study of the failure of metals as a result of mechanical or
thermal stress fluctuations was continued. Comparisons of ORNL data with
results from the subcontract work being conducted at Battelle Memorial
Institute and the University of Alabama tend to show that the frequency

effect is a manifestation of an abrupt change in the deformation mechanism.

l.5. Ceramics Research

Techniques are being studied for the laboratory production of a
beryllium oxide powder which can be used in cold pressing or extrusion
processes and sintered to a high density. A method for obtaining pure
beryllium oxide by calcining beryllium oxalate has been tested. Fairly
pure powder produced at several calcining temperatures (600 to 900°C)
was cold pressed and sintered, and it was found that sinterability was
adversely affected as the powder surface area increased. The powder

surface area was largest at the lowest calcining temperature. Bodies
from the powder calcined at 9OOOC were sintered to 9L4% of theoretical
density. Oxides prepared from beryllium oxalate by three different
laboratories were compared on the basis of surface area measurements,
impurities, and fabricated densities. It was found that the sintering
characteristics of the purer materials were markedly inferior.

A defective lattice structure was considered as a possible requisite
for the sinterability of BeO, but material treated to induce lattice
defects showed no change in structure. Means for controlling grain growth

of BeO shapes during sintering and service are being studied.

Part 2. Chemistry and Radiation Damage

2.1. Materials Chemistry

Experiments in the preparation of charge material for reduction to

Yttrium have been continued. A phase diagram for the LiF-YF, system

>
was constructed.

Optimization studies were made of the process for obtaining oxygen-
free L1F--MgF2-YF3

Hydrofluorination for 6 to 8 hr was found to remove all but 150 to 200

mixtures for reduction to yttrium-magnesium alloys.

ppm O2 without increasing metallic impurities.,

An experimental apparatus was assembled for extracting lithium metal
impurities by circulating lithium metal through an extraction chamber
containing the molten LiCl-LiF eutectic mixture. In a single extraction
experiment 2290 ppm oxygen was removed. The lithium metal throughput

was 700 g, Mechanical improvements of the apparatus are needed.

2.2, Analytical Chemistry

Methods of analysis have been developed for samples from the steps

in the production of yttrium. The analyses now being performed include
the determination of oxygen in LiF-MgF2-YF5 and in yttrium-magnesium
alloy and the determination of oxygen, nitrogen, fluorine, magnesium,

lithium, and titanium in yttrium metal.

2.5, Radiation Effects

The IO of a grown-junction, single-crystal silicon diode has been

F i
S

measured as a function of reciprocal temperature before and subsequent
to a series of reactor irradiations. A change in current mechanism was
observed that has not yet been explained.

The effects of irradiation on the creep of metals at high tempera-

tures are being studied. Tube burst tests of Inconel in air in the MIR

and the ORR at a temperature of l5OOOF indicate that irradiation reduced

the time to rupture. The possibility that the irradiation effects in
Inconel are at least partially due to gas pressure in grain-boundary
voids is being investigated. The gas is believed to be helium formed
by thermal-neutron capture in Blo.

Two capsules, one of yttrium hydride and one of beryllium oxide,
were irradiated and subjected to a few thermal cycles for a period of
approximately 64 hr in the ETR while it was operating at a power of 175
Mw. The capsules are being returned to Oak Ridge for examination. The

absence of abrupt changes in temperature during the test indicates that

the specimens did not crack.

Part 5. Engineering

3.1l. Component Development and Testing

The oil-lubricated pump rotary element that had been operated in
a gemma-irradiation facility in the MIR canal was disassembled and
inspected. The only damage found that could be attiributed to radiation
effects was the hardening and embrittlement of the Buna N O-rings that
were exposed to high levels of radiation. Elastomeric O-rings of this
type would not normally be used in high radiation areas. The viscosity
of the circulated oil increased 45% because of the irradiation.

A dynamic seal tester is being developed with which to obtain
information on the basic phenomena involved in the operation of face-
type shaft seals for high-temperature pumps. Measuring devices for use
with the tester are being studied.

Another test was completed in the investigation of the thermal
stability of thin metal shells. Further evidence was obtained that

welded shells can be as structurally sound as all-machined shells,
S

%.2. Engineering and Heat Transfer Studies

Experimental studies ot the effect of thermal-stress cycling on
structural materials were continued with the pulse-pump system. The
effect of exposure time at a constant cyclic frequency of 1.0 c¢ps was
investigated for several inside-wall fiber stresses, In contrast to
previous results at lower cyclic frequencies, the major effect appeared
to be subsurface void formation along grain boundaries and 'cracking-
out" of surface grains. Attack to depths of 0.012 in. was noted in the
heavy-walled regions of the test sections after 214 hr (770,000 cycles)
of exposure. The severity and the concentration of the attack decreased
with decreasing wall thickness (decreasing wall stresses). The data
provide a preliminary tie-in with earlier thermal-~stress-cycling experi-
ments with a pressurized system and indicate that application of relatively
small thermal-stress fluctuations over a sufficiently long period of time
may influence the extent of corrosive attack.

An experimental system with which to determine the local heat-
transfer coefficients for molten lithium has been completed, and operation
will begin in the near future. Measurements of the thermal conductivity
of columbium have been initiated. A longitudinal heat flow device is
being used for these measurements. Similar measurements of molten lithium

are to be made.

3.5, Instrumentation and Controls

The data acquisition system was modified and placed back into
operation. It is now operating with eight recorder inputs that take a
total of 96 individual inputs. Based on the operating experience thus
far, it appears that the system can be used to record data from an
operating system.,

Life tests of liquid-metal-level transducers were continued, and
a study of thermocouples for use at temperatures up to hOOOOF was
initiated. Several thermocouple metals and insulating materials were
tested for compatibility in helium. Life tests of Chromel-Alumel thermo-

couples in sodium at l5OOOF were continued.,

v R
— 4
Py
' ! . 3

<

3.4, Applied Mechanics

Studies were carried out on the elastic behavior of cylindrical
shells and of tapered circular plates, Equations for calculating
stresses, displacements, and rotations were developed, as well as tables
of numerical values for all functions.

In experimental studies of the creep buckling of shells with double
curvature, it has been found that shells with the lower radius-to-
thickness ratios tend to collapse much more symmetrically than do those
with the higher ratios., 1In cases where shells collapse in highly
unsymmetrical shapes, the buckling starts at regions near the supported
edge because of localized bending stresses. There is some indication that
in long-time tests these bending stresses are reduced by creep effects

until they no longer influence the buckled shape.

3.5. Advanced Power Plant Design

In the study of the feasibility of the vortex reactor concept,
experiments were continued on the effect of vortex strength on the
separation of light and heavy gases and on the use of laminar injection
to increase the local Mach number ratios (higher vortex strength).
Quantitative separation experiments were performed which established
the strength of the concentration peak in a He-CBFl6 system as a function
of the bleed ratio and the bleedoff radius. Comparison of the data
with theoretical predictions indicated that the turbulence level of the
system is sufficiently high to cause an effective diffusivity that is
much greater than the molecular diffusivity. The use of laminar injec-
tion to correct this defect is being studied, and preliminary data indicate
higher Mach ratios than observed with turbulent injection.

Design studies of auxiliary power units for satellites were con-
tinued. The effects of the cycle working fluid and operating conditions
on both the radiator specific weight and on the turbine specific weight
were considered, and a preliminary layout for a system employing rubidium

vapor as the working fluid was prepared.
Part 4. Shielding

4,1. Shielding Theory

A calculation has been undertaken to determine the response of
sodium iodide crystals to photons. The calculation will consider the
effects of bremsstrahlung and annihilation radiation in detail, as well
as the effects of certain crystal geometries. Typical bremsstrahlung
spectra resulting from first collisions of incident photons are shown,

A Monte Carlo code has been devised with which fast-neutron dose
rates inside a c¢ylindrical crew compartment can be computed if the flux
incident on the outside of the crew shield is specified. The code has
been designed to accept as input data the output of a Convair code
devised to calculate air-scattered fast-neutron fluxes.

The Cornpone reactor code is being used for calculations of neutron
fluxes in various media in the Lid Tank Shielding Facility. Thus far
only the thermal-neutron fluxes in plain watier have been calculated for
four assumed configurations, three finite configurations and one infinite
configuration. The results for the finite cases agree with the experi-
mental results to within 10% for the first 32 cm from the source.

Predictions of the thermal-neutron fluxes to be expected near the
Bulk Shielding Reactor were made on the basis of Lid Tank Shielding
Facility experimental data transformed to the BSF reactor geometry. The
predicted fluxes were consistently higher than the measured fluxes, by
a factor of 2.73 at a distance of 40 cm and a factor of 1.54 at a
distance of 115 cm. These discrepancies are of the same order of
magnitude as those in earlier calculations in which the fast-neutron
dose rates were compared.

The development of a code to calculate at a point detector the
angular and energy distributions of gamms rays emitted from a monoener-
getic, point isotropic or point monodirectional source embedded in an
infinite homogeneous medium of constant density is continuing. Several
attempts to develop a method which could be used to calculate the response
of a detector at separation distances up to 20 mfp in a three-dimensional

geometry have given negative results. The so-called "conditional™ Monte
Carlo method is now being investigated.

In some of the Pratt & Whitney Aircraft fast-reactor systems which
are under consideration the natural lithium coolant passes directly from
the core to the engine radiators without the use of an intermediate heat
exchanger. Some of the Li7 will capture neutrons as it passes through
the core to form Li8, which, in turn, will decay by beta emission. The
majority of the Ii decays will occur in thinly shielded regions outside
the primary shield, and bremsstrahlung radiation will result from the
slowing down of the electrons. Calculations have been performed to
estimate realistic upper-limit dose-rate contributions from this source
for a typical supersonic aircraft configuration. The results of these
calculations indicate that in such an aircraft the dose from the lithium

coolant is small compared with that received via the primary shield.

4,2, 1id Tank Shielding Facility
An investigation is under way at the LTSF to determine the thermal-

neutron flux depression in a medium caused by the presence of a gold-foil
detector and also to determine other corrections which should be applied
to foil measurements. From exposures of foils of various thicknesses,
the experimental ratioc of the saturated activity per unit mass of a
"zero"” thickness foil exposed in water to that of a commonly used 0.002-
in.-thick foil has been determined to be 1.35 + 0.20. Several calculated
values vary from 1.08 to 1.30.

4.3, Bulk Shielding Facility

The second series of measurements for the GE-BSF study of the pro-
duction of heat in radiation shields is nearing completion. This second
series of measurements became necessary because the sample cases leaked
and because there was unexpected bowing of a reinforced wall of the tank
containing the shielding configurations. It has been reported by GE that
the results obtained thus far from the second series of measurements are
in reasonable agreement with their predicted values of energy absorption.

Fabrication of the grid plate and the fuel elements for the BSF

stainless-steel-clad UO,.-fueled reactor (BSR-II) is nearing completion,

2

—‘;:i

and fabrication of the control system is proceeding as the design permits,
The initial low-power critical experiments are scheduled to be performed
at the BSF in May, and control tests will be performed at the MRTS SPERT-I
Facility soon thereafter., In order to produce a period as severe as 3%
msec for the control tests, it will be necessary to use a "double-ended"
insertion device which will have a lower poison section and an upper fuel
section, Calculalions indicate that with the proposed device it will be
vossible to increase the reactivity as much as 6.5% Sk/k. The control
plates were calculated to be worth 14.1L4% fk/k.

All components with the model IV garmma-ray spectrometer have been
delivered to the BSF with the exception of the lead-lithium housing,
which is being machined. Recent experiments with the large NaI(Tl) crystal
to be used as the detector in the spectrometer have indicated that the
double-peak distribution which occurs when gamma rays are collimated
into the conical end of the crystal not only may be the result of the
location of the interaction but may also be affected by the optics of the
crystal and/or reflector.

Since the interpretation of the data obtained with the recently
developed Li6I(Eu) scintillation spectrometer will depend on a knowledge
of the Li6(n,a)H3 cross section, measurements of this cross section were
made for the energy interval 1.2 < E_ < 8.0 Mev. The results are in good
agreement with those reported by Ribe [Phys. Rev. 103, 741 (1956)] for
the energy region (1.2 to 6.5 Mev). T

Some ma jor parts of the final analysis of the data collected in the
experiment to determine the energy spectrum of prompt gamma rays accompa-
nying the thermal fission of U235 for the energy interval between O.4 and
10 Mev have been completed. When the final spectrum is determined a new
experiment to remeasure the spectrum for the region below 0.8 Mev will

be initiated.

4.4, Tower Shielding Facility

The 2T experiments performed by the General Electric Company at
the TSF two years ago were checked and extended by ORNL personnel. Most

of the experiments consisted in measurements of radiation in the GE crew

xviii Y

compartment both as a function of altitude and as a function of the
shielding configuration on the TSR 64 ft away, the axes of the crew and
reactor shields always being kept in the same horizontal plane. The
fast-neutron dose rates observed in the two series and the corresponding
calculated dose rates were in agreement for all cases in which the 27
(borated water) shield was on the bottom of the reactor; for some cases
in which the 27 shield was removed, the fast-neutron measurements in the
ORNL series agreed to within 15% with the measurements of the GE series
and were as much as 50% higher than the calculated values. All the
gamma-ray dose rates measured in the two series were in agreement, but
the experimental and calculated gamma-ray dose rates differed as much as
a factor of 2. The thermal~neutron fluxes measured in the two series
were also essentially in agreement. The extensions to the experiments
that were included in the ORNL series consisted of mapping the radiation
around the reactor shield and measuring the energy spectra of direct-
beam gamme rays from the reactor shield and of gamma rays in the crew
compartment, The crew compartment spectra showed the usual 10.8-Mev and
2.,25-Mev gamma-ray peaks resulting from neutron capture in Nl5 and
hydrogen, respectively. A series of peaks which were attributed to
neutron capture in the aluminum and iron structure was observed near

7.6 Mev, and there also appeared to be a peak at 0.5 Mev which probably
resulted from both boron capture and annihilation radiation. A series
of peaks below 10 Mev is as yet unexplained. Preliminary analyses of
the direct-beam spectra indicate that the high-energy peaks are probably

caused by neutron captures in the iron structure of the shield.

4,5, Tower Shielding Reactor II

When the fuel elements fabricated for the TSR-IT were assembled for
a water-reflected, room-temperature critical experiment, the core was
found to be subcritical by approximately 0.T7% fk/k even in the absence
of the cadmium control grids. A series of investigations, both calcula-
tional and experimental, was initiated in an attempt to discover why a
core calculated to be above critical by approximately T% was slightly

subcritical and to determine what changes in the design of the reactor

would be necessary in order to use the elements., The cause of the
discrepancy between the calculated and the critical masses is still
undetermined; however, it was found that introducing a material such
as aluminum or air into the internal water reflector region can result
in as much as a 10% increase in reactivity. The changes required to
give the necessary increase in reactivity are being incorporated in
the design of the reactor.

Recent flow distribution and heat transfer studies have shown that
satisfactory flow rates through the annular fuel elements can be achieved;

flow studies for the central fuel elements are still under way.

PART 1. METALLURGY
l.l. MATERTALS FABRICATION RESEARCH

Yttrium and Yttrium Hydride Studies

Yttrium Preraration

Research on the preparation of yttrium is being conducted in the
Metal Processing Laboratory described previously.l The basic equipment
consists of two 50-1b fluoride salt purification rigs, arranged in
parallel, and a 150-lb-capacity (rated in steel) vacuum-induction-
melting furnace., The process under investigation involves the reduction

of a mixture, YF -MgFe-LiF, with lithium to form an yttrium-magnesium

alloy. The magnZsium is removed from the alloy by vacuum distillation
(sublimation), and the yttrium is then consolidated into ingot form by
a suitable melting process. Chemical analyses of the yttrium produced
thus far indicate that oxygen, nitrogen, and fluorine are still the

ma jor impurities of concern, The electron-beam-melting process shows
promise for removing fluorine to the desired level. The oxygen and
nitrogen problem is being attacked by determination of the sources of
contamination in the various processing steps and continued efforts to

improve the purity of the reactants. A few cold rolling tests have

indicated a need for improvement in the ductility of the material.

Zone Refining of Yttrium

Several experiments have been carried out in a study of the zone
refining of yttrium utilizing an adaptation of the procedure developed
for the zone refining of zirconium. The experiments were carried out
in vacuum and in inert atmospheres. In the experiments carried out in
vacuum, considerable amounts of volatile impurities condensed on the
inside of the apparatus. The deposited material was identified by

X-ray analysis to be a combination of YF5, YOF, and Y205. In the

lT. Hikido, ANP Semiann., Prog. Rep. Sept. 30, 1958, ORNL-2599,

P 9

experiments carried out in an inert atmosphere, little or no material
was volatilized from the yttrium.

The microstructures of the specimens refined in inert atmospheres
were similar to the as-cast structure, whereas, in comparison, the
microstructures of the specimens refined in vacuum were free of impurity
inclusions. A full evaluation of the amount of purification that can

be attained with this process is in progress.

Single Crystal Growth by Zone Refining

In the growth of single crystals by the solidification of molten
metal, a single nucleus or seed crystal is sometimes difficult to
obtain. Therefore several nuclei are permitted to form and the number
of crystals is subsequently reduced, since in theory one crystal will
grow in preference to the others because it most closely approaches
the preferred direction of growth. Evidence that this process occurs
in the zone refining of yttrium has been found. It was noted in the
zone refining work that when a moderate rate of zone travel was used,
the zone-refined specimen had large grains that were elongated in the
direction of zone travel. A few single crystals have been grown by

this method.

Solid-State Electrolysis as a Means of Purifying Yttrium

DeBoer and Fast2 showed that when direct current is applied
through a zirconium wire, the oxygen impurity in the zirconium moves
toward the anode. This principle has been used at General Electric
Company, ANPD, for the purification of yttrium with respect to oxygen.
Specimens purified-in this manner had grains which were as large as
the diameter of the specimen and approximately 1 in. long. Apparatus

for solid-state electrolysis experiments at ORNL is now being constructed.

7. H. DeBoer and J. D. Fast, Rec. trav. chim. 59, 161 (1940).

.

Deformation Mechanisms of Yttrium

. The zone-refining and purification methods described above show
promise as means for obtaining the high-purity metal and single crystals
| - needed for studies of the deformation mechanisms of yttrium. Tenta-
tively, the program of investigations will include the determination
of (1) the room-temperature deformation mechanisms, (2) the critical
resolved shear stresses for these mechanisms, (3) the effect of temper-
ature on the critical resolved shear stress and the mechanisms, and
(4) the effect of interstitial compounds on the critical resolved shear
stress for slip at room temperature as a function of oxygen and nitrogen

content.

Yttrium Hydriding

It is desirable to obtain a high-purity hydride of yttrium for a
determination of an accurate set of dissociation pressure curves and
for use as a standard in the determination of the effects of impurities
(as present in the metal or as introduced by the hydrogen) on the
physical properties of the hydride.

High-purity hydrogen is being obtained by the decomposition of
uranium hydride. Decomposition of uranium hydride begins at approxi-
mately 5OOOC, and a precise partial pressure of hydrogen can be main-
tained by close temperature control of the uranium hydride.

Recent hydriding experiments with yttrium containing approximately
1000 ppm oxygen, 200 ppm nitrogen, 1000 ppm fluorine, and 5000 ppm

. titanium to which approximately 1 wt % hydrogen had been added yielded
a hydride that was metallic in appearance. After approximately 2 hr of
exposure to air, the hydride began to turn blue, and, after 24 hr in
air, the specimen was royal blue in color. Diffraction data showed

that YOF (rhombohedral) was present on the surface of the specimen.

Preparation of Columbium Alloys

) Based on the evidence that the corrosion of columbium by lithium

often is related to the presence of oxygen in the "pure" metal, an

effort is being made to improve the corrosion resistance by "neutralizing'
the oxygen through alloying. The elements being considered as alloy addi-
tions are those having oxides with high, negative free energies of forma-
tion, including zirconium, beryllium, cerium, lanthanum, misch meta.l,3
and yttrium. In order to obtain the specimens needed to evaluate the
influence of these elements on the corrosion resistance of columbium in
lithium, 100-g nonconsumable-electrode arc melts of the alloys were
prepared, and attempts were made to reduce the ingots to strip by cold

rolling. The elements and quantities added are listed below:

Alloy Additions Quantities Added
(rt %)

Zirconium 0.5, 1, 3, 5
Beryllium 0.25, 0.5, 1
Cerium 0.25, 0.5, 1, 3, 5
Lanthanum 0.5, 1, 3, 5

Misch metal 0.5, 1, 3, 5
Yttrium 0.5, 1, 3, 5

The results of the fabrication experiments completed thus far are

summarized in the following sections.

Effect of Zirconiwuwm Additions

Additions of as little as 1 wt % zirconium resulted in the formation
of a phase which the existing phase diagram indicates to be eutectoidal
alpha zirconium. There were indications that this phase was also present
in the 0.5% zirconium alloy. The volume of alpha zirconium present
increased with increasing zirconium content and made the alloys increas-
ingly susceptible to cracking upon cold rolling. It is possible that
the presence of alpha zirconium in the microstructures of the lower

zirconium content alloys was the result of nonequilibrium cooling from

5Typical analysis of misch metal: 52% cerium, 2Lk% lanthanum,
18% neodymium, 5% praseodymium, 1% samarium.

LO)Y

the melt. Consequently, the solubility of zirconium in columbium may
be greater than would be inferred from the as-cast microstructures of
these alloys. Corrosion samples of all the zirconium-bearing alloys,

except the 5% zirconium alloy, have been prepared.

Effect of Beryllium Additions

The low solubility of beryllium in columbium results in the pre-
cipitation of a phase in a columbium-beryllium alloy which 1s presumably
a columbium-beryllium intermetallic compound. All the beryllium-
containing alloys were subject to cracking upon cold rolling, with the
cracking tendency becoming progressively worse with increasing beryllium
content. Specimens of an alloy containing nominally 0.25% beryllium

have been prepared for corrosion evaluation.

Effect of Additions of Rare Earth Metals and Yttrium

From the standpoint of theilr oxide stability some of the more
attractive additions which could be made to columbium include cerium,
lanthanum, misch metal, and yttrium. In the melting of these alloys,
there was the tendency for those containing greater than about 1% added
material to have a very irregular as-cast surface. The rare earth and
yttrium additions made to columbium resulted in an incfeasing quantity
of oxide particles in the alloys as the amount of alloying material was
increased. Little or no difficulty was encountered in rolling these
alloys to strip.

Rolled and annealed arc-melted alloys of cerium, beryllium, and
yttrium with columbium showed reductions in oxygen content and a hard-
ness decrease as compared with unalloyed columbium. Zirconium additions
appeared to strengthen the columbium, as well as to reduce the oxygen
content. The results of oxygen analyses and hardness tests of the
various columbium alloys are presented in Table 1.1.1. The chemical
analysis results also indicated that segregation of the alloying
materials was not serious. Samples cut from the ends and middle of
a 6-in. by 1/16-in. rolled and annealed strip of an alloy with a nominal
0.5 wt % zirconium content analyzed 0.63, 0.57, and 0.53 wt % zirconium.

Table 1.1.1l. Oxygen Content and Hardness Values of Various
Arc-Melted Columbium Alloys

Nominal Material Oxygen Content DPH Hardness Value
Composition (ppm) (500-g load)
Pure columbium 690 107
Cb-0.25 wt 9% Ce 150 79
Cb-0.50 wt % Ce 180 93
Cb-0.25 wt % Be 450 90
Cb-0.5 wt % Zr 99
Cb-1l wt % Zr 420 114
Co-3 wt % Zr 450 156
Co-1 wt % Y 68

*
All alloys were made from the same columbium heat.

No attack of any of the alloys listed in Table 1l.l.1l was observed after
a 100-hr exposure to static lithium at 1500°F.

Studies of Reactions of Columbium with Oxygen and Nitrogen

The product of the reaction of columbium and air at high tempera-
tures has been found to be dependent on the pressure of the ambient air.
At pressures of about 1 atm, the principal reaction product is Cb205;
hovever, at very low pressures (lO_6 atm), the reaction involves primarily
the solution of gas in the metal without the formation of a visible
oxide film. Between these extremes is a range of pressures for which
the formation of CbO is observed.

Since the contamination of columbium with small amounts of oxygen
and nitrogen seriously affects its chemical and mechanical properties,
studies have been conducted to determine the reaction rates of columbium
with these gases at very low pressures in order to establish the limiting
gas pressures to which the metal may be exposed at various temperatures

without detectable alteration of its chemical and mechanical properties.

The establishment of these critical pressures at various temperatures
will serve as a gulde in high-temperature operations such as annealing
and mechanical and corrosion testing. In addition, these studies with
columbium metal will establish base-line data for evaluating the contami-
nation resistance of the columbium alloys which are being developed.

The reaction rates of columbium with oxygen and nitrogen are being
determined by following the weight change of a small specimen exposed
to the pure gas at selected temperatures and pressures. The weight
change is determined by noting the deflection of a sensitive quartz
spring from which the sample is suspended. The pressure in the reaction
chamber is regulated by a system of leak valves.

In the tests conducted thus far it has been found that oxygen is
the principal contaminant with which columbium reacts at temperatures
up to 1000°C. At these same temperatures, the reactions between columbium
and nitrogen up to pressures of 2 u are negligible.

At the two temperatures (850 and 1000°C) at which these tests have
been conducted, the effect of the temperature on the reaction rates is
most pronounced at lower pressures; that is, at an oxygen pressure of
2 X 10_5 mm Hg the reaction rate at 1000°C exceeded that at 85000 by a
factor of 5, but at a pressure of 2 x lO-u mm Hg, the reaction rates
differed by only a factor of 1.5.

The reaction rate curves for columbium and oxygen at pressures less
than 2 x lO_u mm Hg have been found to be linear with time, and they
show no tendency to deviate from this relationship even after the
specimens absorb over 1 wt % oxygen. No oxide films are visually
evident under these conditions. At pressures greater than 2 x lO-u mm
Hg, the reaction rate curves become parabolic and CbO films are observed
on the metal.

A tentative relationship between the quantity of oxygen absorbed

per unit area, time, and oxygen pressure at 1000°C has been determined:
q = ktP"

vhere q is the quantity of oxygen absorbed (mg/cmg), t is time of

exposure (min), P is oxygen pressure (p), and n and k are constants.
In this equation, the constant k contains a diffusion coefficient and
a solubility factor. The exponent n has been found to be unity for

2

oxygen pressures up to 1 x 10 7 mu Hg. The exponent has not been
determined at higher pressures, but is expected to obey Sieverts' square-
root law. The values of q for times of 100 min at 1000°C increased from

2 ym Heg to 60 at oxygen pressures of

8 at oxygen pressures of 5 x 10~
5 X lO-lF mm Hg.

Tentative limiting oxygen pressures at which negligible reaction
occurs with columbium have been determined to be 3 x lO'-5 men Hg at 85000
and 5 X 10-6 rm Hg at 1000°C. If the reactions between columbium and
nitrogen are assumed to be negligible, the limiting air pressures become
1.5 x lO‘-lF and 2.5 x lO“5 mm Hg for temperatures of 850 and lOOOOC,
respectively. The factors that affect the permeability, J (or q/t),

appear 1o agree with the requirements of Ficks' first law of diffusion.

Extrusion of Clad Refractory Metals

A billet for the duplex extrusion of a refractory metal clad with
an oxidation~resistant alloy was designed, as described previously,
and the following billets were machined and assembled: +two of type 446
stainless steel-clad molybdenum, two of type 46 stainless steel-clad
columbium, one of aluminum-bronze-clad molybdenum, and two of aluminum-
bronze-clad columbium. An attempt was made to extrude the two type 446
stainless steel-clad molybdenum billets; but, because of an incorrect
eXtrusion temperature or lack of adequate lubrication, the molybdenum
segmented. The extrusion was made at llTSOC with glass and Necrolene
as lubricants. In the future, more adequate lubrication will be

provided.

AHH. Inouye and D. O. Hobson, ANP Semiann. Frog. Rep. Sept. 30, 1958,
ORNL~2599, p k4.

10

1.2. CORROSION STUDIES

Effects of Oxygen and Nitrogen on the Corrosion Resistance
of Columbium to Lithium at Elevated Temperatures

Columbium undergoes intergranular corrosion attack when exposed
in molten lithium at elevated temperatures both in static systems at
constant temperature and in systems in which the molten lithium is
circulated under nonisothermal conditions. The attack has been found
to be particularly severe in and near weld zones.l A systematic investi-
gation of the role of atmospheric contaminants has revealed that the
attack is due primarily to oxygen contamination of the colum.bium.2
Data were obtained which indicated that columbium oxide at the grain-
boundaries was reduced by lithium to form a two-phase grain-boundary
corrosion product of lithium oxide and columbium and that columbium
nitride was not attacked. These observations were found to be consistent
with the relative thermodynamic stabilities of columbium oxide and

columbium nitride in molten lithium.

Compatibility of Columbium with Molten Lithium Contaminated
with Lithium Nitride or Lithium Oxide

Static tests were conducted at l5OOOF for 100 hr on columbium in
contact with lithium contaminated with 5000 ppm nitrogen (as lithium
nitride) or 5000 ppm oxygen (as lithium oxide) to determine the effects
of these impurities on the corrosion resistance of pure columbium. No
effects on the columbium were observed, except a thin film of columbium
nitride on the surface of the specimen exposed to the lithium contaminated

with nitrogen.

1

P 15.

2E. 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. E. Hoffman, ANP Semiann. Prog. Rep. Sept. 30, 1958, ORNL-2599,

11
Dissimilar Material Mass Transfer Effects in a Columbium—
Sodium-Type 316 Stainless Steel System

A columbium specimen and a type 316 stainless steel specimen were
suspended in a type 316 stainiess steel capsule containing sodium at
lYOOOF for 1000 hr. The specimens were held in place in the capsule by
columbium wire. A scaly surface film was found on the columbium specimen
and the columbium wire upon visual examination after the 1000-hr
exposure, and weight-change data indicated that the columbium specimen
had gained 0.11 g (29 mg/in.g) in weight. Metallographic examination
of the colurbium specimen showed two surface layers (total thickness
0.8 mil), and an x-ray examination revealed the layers to be CbC and
CbeN. Hardness measurements showed that both layers were quite hard,
as indicated in Fig. 1.2.1. In addition to the brittle surface layers,
intergranular cracks may be seen to a depth of 2 mils. Chemical analyses
of the exposed columbium wire (0.035 in. dia) showed a nitrogen concen-
tration of %350 ppm compared with 310 ppm before the test and a carbon
concentration of 2150 ppm compared with 140 ppm before the test. The
type 316 stainless steel specimen showed a negligible weight change as
a result of the exposure to sodium; on the other hand, columbium was
detected on the surface of the steel specimen, and the carbon concen-
tration had decreased from 420 ppm to 300 ppm.

A second test in which a columbium specimen and a type 316 stain-
less steel specimen were suspended in sodium in a type 316 stainless
steel capsule at lTOOOF for 900 hr has recently been completed and
preliminary results support the analysis presented above. Metallographic
examination of the columbium specimen again showed two surface layers,
and chemical analysis of the exposed columbium wire showed a carbon
concentration of 2350 ppm that is to be compared with 140 ppm before
the test.

12

SECRES
. Y-28321

COLUMBIUM BEFORE TEST <—i——- COLUMBIUM AFTER TEST

DPH —2073

DPH- 1654

DPH-105

o0
00"
0"

T
INCHES

5667

T
15|O X

Fig. 1.2.1. Metallographic Appearance of Columbium Specimens Before and After Exposure to Sodlum
in a Type 316 Stainless Steel Capsule at 1700°F for 1000 hr. Note the brittle surface layers and the inter-
granular cracks of the specimen exposed to sodium. As polished. (EEEMwith caption)

13

Mechanical Properties of Type 316 Stainless
Steel After Exposure to Lithium

Austenitic stainless steels have shown grain-boundary penetration
of lithium after exposure at elevated temperatures to lithium intention-
ally contaminated with lithium nitride.5 Additional static tests to
determine the effects of lithium nitride and lithium oxide additions
on the corrosiveness of lithium on type 316 stainless steel indicate
that the oxide addition had only a minor effect in comparison with the
effect of the lithium nitride. The grain-boundary penetration which
occurred in the nitride-addition test substantially reduced the room-

temperature ductility of the stainless steel test specimen.

Solubility of Lithium Oxide in Lithium

The solubility of lithium oxide was determined at four temperatures —
250, 300, 350, and MOOOC — with a minimum of seven samples having been
taken at each tem.perature.5 Activation analysis methods were used to
determine the oxygen concentration of the lithium samples. The results
of the analyses indicated a solubility of approximately 90 ppm at 25000
with a rapid increase to approximately 640 ppm at 400°C.

Lithium Corrosion of Cermets

Cermets appear to be more suitable than pure metals, alloys, or
ceramics for use as bearing and valve seat materials for operation in
molten lithium at l2OOOF and higher. Pure metals and alloys tend to

solid-phase bond or gall; ceramics lack ductility and are relatively

3E. E. Hoffman, Corrosion of Materials by Lithium at Elevated
Temperatures, ORNL-2674, p 108 (March 9, 1959).

“Ihid., p 231.
>Ibid., p 233.

G. Leddicotte, Anal. Chem. Ann. Prog. Rep. Dec. 31, 1957, ORNL-2453,
p 30.

14

difficult to fabricate.

Initial evaluations of a few typical cermets were made by exposing
them to molten lithium. A summary of the results of the tests, including
tests on binder metals, is presented in Table 1.2.1. The cermets listed
in the table are typical of the most common types of cermet.

The results given in Table 1.2.1 indicate that the nickel-bonded
TiC-base cermets and the cobalt-bonded TiC- and WC-base cermets do not
have sufficient corrosion resistance to lithium to be considered as
promising bearing or valve seat materials. The attack by lithium appeared
to concentrate in the metal phase of the cermets with the cobalt
exhibiting greater corrosion resistance than the nickel. As expected,
these results had the same trend as those obtained with the pure metals
cobalt and nickel.

The extent of the corrosion of the cermet was proportional to the
exposed area of the metal-binder phase. The corrosion was also affected
by the method of fabrication, as illustrated by the results for similar
cobalt-bonded cermets made by different companies.

It is known that molybdenum has good corrosion resistance to
lithium, and, as the data in Table 1.2.1 indicate, so has tungsten.

In an effort to utilize the corrosion-resistance characteristics and
other favorable properties of these two metals, tungsten-bonded and

molybdenum-bonded TiC-base cermets are being fabricated for testing.

15
Table 1.2.1. Results

of Exposure of Carmets and Carmet-Binder Metals to Lithivm at 1500°F

Specimen Nominal Composition Test " we;gn q\unge D:pth :f
Tested* (wt %) Container AN ° pecnr;an ftac Notes
(mg/in.%) {mils)
Metals

Tungsten 100 Columbium 0.4 0 0 No attack detected

Cobalt 100 Columbium 0.4 -4 0 Slight, uniform attack; metallographic
examinations did nat reveal any
preferential attack

Nickel 100 Columbium 0.4 -2427 Attacked heavily, intergranularly and
transgranularly

TiC-Base Nickel-Bonded Carmets

K150A 10 Ni--10 CbTaTiC,-80 TiC Iron 0.4 ~72 12 Nickel-binder phase attacked; speci-
men did not crack

KI51A 20 Ni-10 CbTaTiC,-70 TiC Iron 0.4 -695 65 Nickel-binder phase attacked; speci-
men cracked during test

K152B 30 Ni-6 CbTaTiC,-64 TiC fron 0.4 ~954 Q0 Nickel-binder phase attacked; speci-
men cracked during test

K162B 23 Ni~3 Mo—6 CbTaTiC,- fran 0.4 -668 40 Nickel-molybdenum-binder phase

64 TiC attacked; specimen cracked during
test
Cobalt-Bonded TiC- and WC-Base Cermets

K138A 19 Co-81 TiC Columbium 2.0 -23 Slight attack an binder phose through-
out l/“-in.-diu specimen

K21 10 Co-90 WTiC, Columbium 2.0 =31 6 Cobalt-binder phose attacked; speci-
men did not crack during test

AA 3 Co-97 WC Columbium 2.0 =2 Specimen (‘/‘ in, dia) cracked during
test

K8 3.5 Co-96.5 WC Columbium 2.0 ~14 A 1-mil-thick reaction loyer on speci-
men

A 6 Co-94 WC Columbium 2.0 -9 Specimen ('/4 in. dia) cracked during
test

GG 12 Co-88 WC Columbium 2.0 =22 1-7 Approximately 75% af attack was

7 mils deep and concentrated in
metal phose

*Materials with the prefix K were manufactured by Kennametsl, Inc., Latrobe, Pa.; ofl ather specimens except the metals were manu-
factured by Adamas Carbide Corp., Kenilworth, New Jersey,

**A, is the surface area of fest specimen, and V is the valume of lithium ot 1500°F (816 °C).

16

1.3. WELDING AND BRAZING STUDIES

Development of Refractory-Metal-Base Brazing Alloys for Joining
Container Materials for Molten Lithium

Brazing alloys based on the refractory metals are presently being
studied for application in high-temperature reactors using lithium as
a coolant. The preliminary work was described previously.l Brazed
specimens consisting of inverted T-joints are now being used in evalua-

tion studies of the alloy systems listed below:

Zr—Be—Fe Zr—TFe—Be Ti~Cr-Be
Zr—-Be-}o Ti—Zr-Be Ti-Zr—V-Be
Zr—V—Be Ti-Zr—V Cb~Ti-Be
Zr—Cr-Be Ti—Zro Fe—Cb—Be
Ti-Fe-Be Ti-Zr—Fe

Each alloy is studied to determine its melting point and its flow,
wetting, and joining characteristics on the base metals columbium,
molybdenum, zirconium, and titanium.

On the basis of flow point, wettability, and Jjoint integrity, several
alloys appear to be promising at the present time as brazing materials
for one or more of the base metals. These alloys, together with their
flow points and brazing characteristics, are described in Table 1l.3.1l.

Corrosion testing of refractory-metal-base brazing alloys in
high-temperature lithium 1s also continuing. The large weight losses
reported previously2 for zirconium-base binary alloys exposed to lithium
in columbium capsules are believed to have been caused by dissimilar-
metal mass transfer of zirconium to the capsule walls. The bath zones

of the columbium capsules were gold colored after the test, and microspark

lR. G. Gilliland and G. M. Slaughter, ANP Semiann. Prog. Rep.

2D. H. Jansen, ANP Semiann. Prog. Rep. Sept. 30, 1958, ORNL-2599,
P 23.

17
8T

Table 1l.5.1.

Alloys for Container Materials for Molten Lithium

summary List of Promising Refractory-Metal-Base Brazing

Brazing Alloy Brazing Brazing Characteristics on Indicated
Composition Tenmperature Base Metal
(wt %) (°c) Good Flow Sound Joint

67 Zr—29 V=i Fe 1300 Cb and Mo Cb and Mo
60 Zr—25 V—-15 Co 1300 Cb and Mo Cb and Mo
95 Zr-5 Be 1000 Cb, Mo, Zr, and Ti Cb, Mo, Z2r, and Ti
70 Zr—20 V=10 Cb 1300 Cb and Mo Cb and Mo
80 Zr-15 Fe—5 Be 1300 Cb, Mo, Zr, and Ti Zr and Ti
48 Zr—u8 Ti—4 Be 1100 Co, Mo, Zr, and Ti Cb, Mo, Zr, and Ti
63 Ti—27 Fe—10 Mo 1180 Cb and Mo Cb
63 Ti—27 Fe—lO0 V 1300 Cb, Mo, Zr, and Ti Cb, Mo. Zr, and Ti
63 Ti—28 V-4 Be 1250 Cb, Mo, Zr, and Ti Cb, Mo, Zr, and Ti
80 zr—-l2 Fe-8 Be 1100 Cb, Mo, Zr, and Ti Cb, Mo, Zr, and Ti
45 Ti—U45 Zr—10 Fe 1200 Cb, Mo, and Zr Cb, Mo, and Zr
T4 Zr—18 Cb-3 Be 1100 Cb, Mo, Zr, and Ti Cb, Mo, Zr, and Ti
b7 Zr-b7 Ti-~3 V-3 Be 1100 Cb, Mo, Zr, and Ti Cb, Mo, Zr, and Ti

spectrographic analysis indicated the presence of zirconium in these
areas.

Zirconium- and titanium~base alloys containing iron or molybdenum
have shown excellent corrosion resistance. The 63% Zr—27% V-10% Cr and
70% Zr-20% V-10% Cb alloys have also been tested and have shown good
corrosion resistance to lithium at l?OOOF in 500~hr static tests.

In view of the promising results obtained in these preliminary
phases of the alloy-development program, preparations are being made
for mechanical tests of the brazed joints. A ftorsion-testing machine

is being constructed to permit testing of the brazed joints in shear.

Welding Studies on Columbium

A program of tesis has been initiated to investigate the feasibility
of welding columbium without the use of a controlled-atmosphere chamber.
The inert-gas tungsten-arc process (direct-current, straight-polarity)
vas used to fusion butt weld 0.040-in. columbium sheet. The weld had
a very bright surface and excellent ductility. Micrchardness traverses
showed no variations in hardness across the welded joint. A high degree
of ductility was also indicated by the absence of cracks in 180 deg
transverse and longitudinal face- and root-bend samples. The optimum

conditions used to produce these welds are listed below:

Welding current 60 amp
Welding speed 8 in./min
Arc length 1/16 in.
Helium cover gas flow rate Q0 cfh
Helium backing ges flow rate 60 cfh

The corrosion resistance to lithium of welds made under these
conditions will be determined. Room~ and elevated-temperature mechanical

property tests will also be conducted.

19

The welding of columbium plate by the inert-gas-shielded, metal-
arc process (direct-current, reverse-polarity) is also being studied.
The equipment being used is shown in Fig. 1.3.1. A trailer shield con-
structed to provide an additional blanket of inert gas to protect the
welded joint from contamination during cooling is shown in Fig. 1.3.2.

Both 1/8-in.- and 1/b-in.-thick plates of columbium have been
investigated to date. Bead-on-plate welds made on material of both
thicknesses were found to have excellent ductility, based on microhardness
traverses. The welds were clean and bright. A single-pass, square~butt
weld was also made on 1/k-in.-thick plate in which approximately 50%
penetration was obtained. Helium, rather than argon, was used as the
cover and backing gas because it permitted a superior bead contour and
better penetration. The columbium filler wire used in the study was
0.035 in. in diameter.

As a result of these favorable tests, a more complete and compre-
hensive study is planned in which optimum conditions for producing
sound, ductile welds on 1/Lk-in.- and 1/2-in.-thick columbium plate will
be determined. In addition, multipass-welding feasibility studies will
be made in which a fusion root pass and subsequent filler metal passes
will be used. Single-pass welding studies will also be conducted for

which larger diameter filler wire will be used.

20

UNCLASSIFIED
PHOTO 44105

Fig. 1.3.1. Equipment for Inert-Gas Shielded Motal-Arc Welding of Columbium Plate. Welding gun, trailer shield, and movable train are at the left

and the voltage and current recorders are at the right. _wi'h caption)

UNCLASSIFIED
Y-26023

Fig. 1.3.2. Trailer Shield Used in Inert-Gas Shielded, Metal-Arc Welding of Columbium Plate. -
with caption)

22
1.4, MECHANICAL PROPERTIES INVESTIGATIONS

It has been necessary to develop equipment that is capable of
v operating at high temperatures and which protects the specimen from
contamination during testing in order to evaluate the mechanical proper-
ties of columbium and columbium-base materials. The tendency of such
materials to react with air makes it necessary that they be protected

while at elevated temperatures.

Testing Procedures

Although several different designs of apparatus have been considered,
a slight modifiication of the standard environmental test equipment has
proved quite adequate for testing in the temperature range 1800 to QOOOOF.
The specimen is enclosed in a section of 2-in. sched.-h0, type 310 stain-
less steel pipe, and the pull rods are sealed at each end of the test
chamber with U~-cup seals. The entire assembly is heated by placing a
standard 2200°F resistance furnace about the chamber. Swaged platinum—
platinum~-10% rhodium thermocouples are used for measuring the temperature
of the specimen. Various materiasls have been used for pull rods, but
molybdenum has proved to be the most satisfactory.

Strain measurements are made during tensile tests with an exten-
someter which attaches directly to the specimen. The extensometer is
brought through U-cup seals at the bottom of the test chamber, and the
- strain is measured by means of a variable differential transformer.

Strain measurements are currently being made during creep tests by a
dial gage. Extensometers for making these measurements are being
developed.

Considerable emphasis has been placed upon obtaining a pure argon
atmosphere for testing. After the specimen has been assembled in the
test apparatus and the system has been checked for leaks, heating is

. begun. The heating rate is adjusted so that the pressure never exceeds
0.2 p. Af'ter the specimen has reached the desired temperature, an

. atmosphere of approximately 10 psig of pure argon is admitted. This

static atmosphere is used throughout the duration of the test. A
tantalum sheath is placed about the gage length of the specimen to getter
any oxygen which may be present.

The hardness and specimen analysis data presented in Table 1l.4.1
demonstrate the effectiveness of this test procedure in preserving the
purity of the columbium. Test 124, which ran for 718 hr at 1800°F, was
the only test which resulted in any marked increase in the specimen
impurity content. Although the nitrogen content of this specimen after
the test was almost double that measured before the test, it may be
noted from Fig. l.4.1 that the depth of penetration of the surface
contaminants appears to be only 2 to 5 mils. A hardness traverse across

the specimen substantiated this observation.

Mechanical Properties of Pure Columbium

Material
Two heats of pure columbium material, heats K-4 and WC-18, were

studied. Analyses of these heats showed the following impurities:

Impurity Content (wt %

Heat K-k Heat WC-18
Carbon 0.02¢ 0.003
Oxygen 0.033-0.040 0.16
Nitrogen 0.041~0.0u7 0.0082
Hydrogen 0.0005 0.0012
Zirconium 0.085 0.00121
Tantalum 0.039 0.0466

Heat K-4, which was supplied by Kennametal Inc., was used primarily for
establishing testing techniques because a limited amount of material

was available. Heat WC-18, the other material studied, was supplied

by the Wah Chang Corporation. The material was received in billets

5 in. in diameter, which were extruded at 1200°C to 1 in. in diameter
and then swaged to 5/8 in. in diameter. The surfaces of the swaged rods

were then cleaned and rod specimens having a gage length of 2 in. and

2k

a2

Table

l.4.1. Results of Impurity Analyses and Hardness Tests of
Pure Columbium (Heat K-L4) Creep Test Specimens

Shoulder Hardness,

*
Test Time of Impurities Found {wt %) DPH Values
No. Analysis Oxygen Nitrogen Hydrogen Before Test After Test
12k Before test 0.0%3 0.041 0.0005 150.5159.0 145.2-161.9
0.040 0.047 0.0005
After test 0.049 0.070 0.0016
at 1800°0F 0.0kLL 0.099 0.0015
146 Before test 0.040 0.0L7 0.0005 150.5-159.0 126.7-135.4
After test 0.041 0.0483 0.0004
at 1800CF
121 Before test 0.040 0.0L7 0.0005 150.5-159.0 147.8-159.0
After test o} .+ 0.051 0.0007
at 1800°F ogdh2 0.051 0.000L
133 Before test 0,040 0.047 0.0005 150.5-159.0 131.0-150.5
After test 0.041 0.048 0.0007
at 2000°F 0.0%5 0.0012

*
Results of duplicate analyses presented if available.
& Al *  UNCLASSIFIED
5 { 3 n ]

Y-28445
) N ¢ TN
\ S = ~
o £ v
NN 4 A «

\
INCHES

Fig. 1.4.1. Surface of a Columbium Creep Specimen from Heat K-4 Which was Tested at 1800°F and 15,000 psi

in Stagnant Argon for 718 hr (Test 124). with caption)

26

a diameter of 0.375 in. were machined from the swaged rods.

Recrystallization Studies

In order to find an annealing treatment that would give a reasonably
uniform grain size in the test specimens, studies were made of the
recrystallization characteristics of columbium for various amounts of
cold work. The results obtained for a series of specimens of heat K-4
are shown in Figs. 1.L4.2 through 1.4.7. Diamond-pyramid hardness values
for the specimens are given in the figure titles. As may be noted, the
hardness decreased as the annealing temperature was increased up to
lBOOOC, and then a slight increase occurred. The increase in hardness
may have been due to a precipitate which forms at l5OOOC. The nature
of the precipitate is indicated in Fig. 1.4.8, and the tendency of the
precipitate to gather in pools in the grain boundaries after long periods
of heating is shown in Fig. 1.4.9. No suitable identification of this
precipitate has been made. On the basis of these studies a pretest
annealing treatment of 2 hr at lBOOOC in vacuum was selected for both
heat K-4 and heat WC-18. L& e

Creep-Rupture Tests

The results of the creep-rupture tests which have been completed
and those which are in progress are summarized in Table 1.L.2. As
indicated, several of the tests were terminated because of pull-rod
failures and, hence, were useful only in checking the effect of the
test environment on the purity of the specimen. It is of particular
interest that the creep strengths of the two heats, K-4 and WC-18,
differ substantially. The effect of pure nitrogen on the creep properties
of columbium is also significant. The creep curves, as presented in
Fig. 1.4.10, indicate that the diffusion of nitrogen into the metal

reduces the deformation rate.

Tensile Tests

The results of tensile tests on columbium specimens are summarized
in Table 1.4.3. Although a tensile test per se was not run on heat K-,

a reasonably good approximation of the yield stress could be made from

B g
bl 4
3 oo
N Coe

. 7. 57 UNCLASSIFIEDS
)

by : o
# -z
/8 o

p z

Bk k

EFE

Fig. 1.4.2. Microstructure of Columbium (Heat K-4) After an 80% Reduction in Area by Extrusion at 1200°C and
a Further 60% Reduction in Area by Cold Swaging to o j-in.-dia Rod. (DPH: 193.0-201.2). Etchant: H,O-HF-
HNO,H,50,. -| with caption)

Fig. 1.4.3. Microstructure of Specimen in Fig. 1.4.2 After Annealing for 1 hr at 1100°C in Vacuum. (DPH:
191.2-196.3). Etchant: H,0-HF-HNO,-+H,S0,. —ivh aption)

28

15" UNCLASSIFIED |
L Y2697

W

oy bx-
S
z

o

SETETR

B

o

o

l

250%

Fig. 1.4.4. Microstructure of Specimen in Fig. 1.4.2 After Annealing for 1 hr at 1200°C in Vacuum. (DPH:
171.4-178.2). Etchant: H,0-HF-HNO4H,S0,,. Qs‘ caption)

" UNCLASSIFIED,
Y-26928 -

Fig. 1.45. Microstructurs of Specimen in Fig. 1.4.2 After Annealing for 1 hr at 1250°C in Vacuum. (DPH:
145.2-168.2). Etchant: H,0-HF-HNO4-H,S0,. m caption)
N G
Y
S P R

INCHE
fedcii

FEEEERER

Sy <!

250X

Fig. 1.4.6. Microstructure of Specimen in Fig. J.4.2 After Annealing for 2 hr at 1250°C in Vacuum. (DPH:
147.8-150.5). Etchant: H,0-HF-HNOy-H,S0,. with caption)

. UNCLASSIFIED.
o ot Y:28414

T
INCHE
i

EF

FEFER

'k

-4
&

9 2s0x [
2 (51

Fig. 1.47. Microstructure of Specimen in Fig. 1.4.2 After Annealing for 2 hr ot 1300°C in Vacuum.
150.5-159.0). Etchant: H,0-HF-HNOH,SO,. (GRS with caption)

30

“ UNCLASSIFIED
s Y-28415

Fig. 1.4.8. Specimen of Fig. 1.4.7 ot a Higher Magnification. Note precipitation. (Bl with coption)

UNCLASSIFIED
> Y-28444
: w
. H
5
4 2
4 % 2002

B

\ ;
- : N, » - 6 =

Fig. 1.4.9. Specimen of Heat K-4 Which was Annealed for 2 hr at 1300°C in Vacuum and Tested in Creep at
15,000 psi and 1800°F in an Environment of Static Argon for 718 hr (Test 124). Etchant: H,O-HF-HNO,-H,SO,.

(- with caption)

31

\N
rno

Table 1.4.2. Summary of Results of Creep Tests on Pure Columbium

Pretest heat treatment: annealed 2 hr at lBOOOC in vacuum
Test atmosphere: static argon, except as indicated

Heat Test Test Test Time to Strain at
Designation No. Stress Temperature Rupture Rupture Comments
(psi) (°F) (hr) (%)
K-4 1254 15,000 1800 718 Pull-rod failed; specimen
showed no strain
*
146 16,000 1800 1610 6.21
145 17,000 1800 0.1l3 Lo.62
121 13,000 1800 <0.10 40
133 11,000 2000 82 Pull-rod failed; specimen
showed ~ 10% strain
*%
WC-18 178 3,500 1800 L26.6 20.65
*H
180 3,500 1800 307 .4 14.8
181 3,500 1850 55.8 62.5
¥
190 3,500 1850 115 10.05 Test environment is

flowing nitrogen

*
Test discontinued before rupture.

Tests still in progress; values presented are the last readings taken.
Table 1.4.3. Summary of Results of Tensile Tests on Pure Columbium

Pretest heat treatment: annealed 2 hr at
1300°C in vacuum

Test atmosphere: static argon

Test temperature: 1800°F

Heat Test Yield Stress, Ultimate Strain at
Designation No. 0.2% Offset Stress Rupture

(psi) (psi) (%)

K-4 121 ~17, 000"

WC-18 T-1 3,370 6190 64
T-2 3,486 6836 62
T-3 7,286 8786 70

*
This value was estimated from a curve obtained by step-loading a
creep test specimen.

a load curve obtained by step-loading a creep test specimen to 18,000
psi. A yield stress of 17,000 psi was estimated from these data at
1800°F. Two tensile tests at 1800°F showed heat WC-18 to have a yield
stress of approximately 3400 psi. Again, the marked difference in the
properties of the two heats was evident. Heat WC-18 also appears to be
more ductile than heat K-k. The type of failur§*€£é€h06curred with
heat WC-18 specimens is shown in Fig. 1.4.11.

The significant effects which small amounts of impurities appear
to have on the mechanical properties of pure columbium prompted further
studies of the effects of impurities. Tensile and creep specimens have
been prepared from the same heat of metal to which a specific amount of
oxygen was added so that the effect of oxygen on the mechanical properties
could be studied. The program of tests has not been completed, but the
tensile curves for a sample from heat WC-18 to which 0.1 wt % oxygen was
added and a sample from heat WC-18 without the oxygen addition are com-
pared in Fig. 1.4.12. As may be seen, the yield point of the oxygen-

contaminated specimen is more than twice that of the purer material.
| R G s

w

UNCLASSIFIED
ORNL-LR-DWG 36386

% RUPTURED 55.8 hr WITH
/ 62.5% STRAIN

48

44 —- -

|
40 - = P

~
®

Y
2

STRAIN (%)

TEST 181

© PURIFIED STATIC ARGON ATMOSPHERE

® PURIFIED FLOWING NITROGEN ATMOSPHERE

E%

o 25 50 75 100 125
TIME (hr)

Fig. 1.4.10. Results of Creep Tests of Pure Columbium (Heat WC-18) at
1850°F and 3500 psi. Test specimen 0.375 in. in diameter with 2-in. gage

length. ”ivh caption)

UNCLASSIFIED
Y-28576

Fig. 1.4.11. Columbium (Heat WC-18) Tensile Specimen Tested to Failure at 1800°F. —wioh

caption)

3k
Neither the ultimate strength nor the ductility was greatly affected by

the oxygen addition.

Comparison of Results with Those of Other Investigators

As mentioned above, the results obtained for the two heats of so-
called "pure" columbium used in this investigation are far from being
in agreement, and therefore comparative data of other investigators were

1-5

examined.” The creep-rupture data that have been obtained at various
laboratories are compared in Fig. 1.4.13 on a Larson-Miller master rupture
curve.6 Analyses of the columbium specimens compared in Fig. 1.4.13

are given in Table 1.4.5, and the test conditions are described in Table
1.4.6.

As may be seen, the data of Fig. 1.4.13 are scattered over such a
range that no suitable correlation with the Larson-Miller parameter can
be obtained, even though the parameter presents a rather broad and
inclusive method for adgusting data points to a straight line. Because
of the logarithmic time function in the parameter, the scatter of these
data is actually greater than is evident at first glance.

Data obtained at ORNL and at Pratt & Whitney for the yield strength
of cqlumbium at 1800°F are compared in Fig. 1.k.1k. Again, the range

of experimental values is quite large.

1

lS. W. Poremba, H. A. Saller, and J. T. Stacy, Initial Investigation
ot Niobium and Niobium-Base Alloys, BMI-1003 (May 23, 1955).

2R. Parkman and O. C. Shephard, Investigation of Materials for Use
in Heat Transfer System Containing Liguid Lead Alloys (Report XII);
Final Report, ORO-45 (June 11, 1951).

3R. L. Orr and D. W. Bainbridge, "The Correlation of High-Temperature
Rupture Data for Niobium," Institute of Engineering Research Report,
University of California, Series 22, Issue 28, July 1, 1953.

uNuclear Propulsion Program Engineering Progress Report October 1,
1958 to December 31, 1958, PWAC-53%4, p 119.

5E. J. Jablonowski, F. R. Shober, and R. F. Dickerson, '"Mechanical
Properties of Niobium-Base Alloys," Progress Relating to Military
Applications During February, 1958, BMI-1257, p 25.

6F. R. Larson and J. Miller, Trans. ASME Tk, 765 (1952).

55
UNCLASSIFIED
ORNL-LR-0OWG 36387R

000 ——1———3——=3__C 1

P R I Rk T 1 ~ PURE GOLUMBIUM
/7 | T~ CONTAMINATED WITH
8000 ‘,L ; =i+ 04 WT % OXYGEN
' RN
7000 | - Iy
| "] —— PURE COLUMBIUM D
I // \< \\
6000
e \
- \
-1/
» 5000 N \\
~ \
0 \ \
4 4000 v
2 \
T \
o \
3000
2000 _—
1000
0
0 10 20 30 40 50 60 70

STRAIN (%)

Fig. 1.4.12. Tensile Curves for Pure Columbium and for Columbium Con-
taminated with 0.1 wt % Oxygen Tested at 1800°F In Argon. Test specimen
0.375 in. in diameter with 2-in. gage length. —wifh caption)

.
ORNL—LR-D!! !!392

1800C
16000
_—— 18000 |
30‘000 OIRNL-L = 394 E
a | a | ' e
[¥1)
o i 12000 - —
20,000 - * — o
. ‘ 45 A ] ©
o 3#
! : o : ~ 10000
X ORNL HEAT A ; g °
O BMI HEAT A ‘ =
= 10000 - & papkMaN aND SHEPHARD T —] 2 8000 -—]
& | & DRR AND BAINBRIDGE  _ 4 ) 2
w L ® HEAT K-4 _ | . _ ©
a & HEAT WC-18 1‘ S
i 3 ) . o 6000 B .
£ soogl & PWAC- o N ¥ £ '
@ 0 Pwac-2 v | o
L § BMI HEATE __ | S o 4000 3 L
SEE TABLE 1.4.6 FOR TEST » 3
| CONDITIONS AND DATA REFERENCES — ¥
T=ABSOLUTE TEMPERATURE {°R) 2000 g
4= RUPTURE TIME {hr) i
2000 . w Y
42 44 46 48 50 52 54 56 o 3

T{20+10¢ 7) x 1072

Fig. 1.4.13. Larson-Miller Plot of Creep Data Fig. 1.4.14. Comparison of Yield Strength Data

for Various Heats of Columbium. for Vartous Heats of Columbium.

36

Table 1.4.5. Analytical Data on Impurities in Columbium Billets

Quantity Found

In ORNL In Heat In Heat In BMI In BMI
Impurity Heat A K-L WC-18 Heat A Heat B
(at. %) (wt %) (vt %) (wt %) (vt %)
Carbon 0.0% 0.026 0.003 0.02
Oxygen 0.00k 0.0%3%0.040  0.16" 0.011-0.022 0.05
Nitrogen 0.0125 0.041-0.047  0.0082°  0.001-0.03
Hydrogen 0.00024  0.0005 0.0012°  0.0003-0.001k
Zirconium 0.085 0.00121
Tantalum 0.039 0.0466 0.16

¥
Vendor's analyses before extrusion.

It is felt that the scatter in the data for the mechanical proper-
ties of "pure” columbium is due to one or more of the following facto.s:
purity of the material, differences in pretest mechanical and thermal
histories, and the purity of the test environment. Although the mechanical
properties of all materials are somewhat sensitive to these factors, it
appears that columbium is considerably more sensitive than other materials,
particularly with respect to chemical composition. Hence, the presently
available data for the mechanical properties of "pure" columbium have
little engineering meaning. Until a columbium alloy is developed which
shows considerably more stability with respect to mechanical properties,
it is felt that a study of the relative effects of various contaminants
on the properties of "pure' columbium would be more informative than the
collection of a large amount of engineering test data on any one heat

of columbium.

Fatigue Tests

Fatigue tests on Inconel at 1 and 0.1 cps are in progress under a
subcontract with Battelle Memorial Institute. Analyses of hysteresis

loops obtained in the tests permit calculation of the plastic strain

57

o4

Table 1.4.6. Data on Test Conditions Under Which Data Plotted
in Fig. 1.4.13 Were Obtained

Laboratory at Which Columbium Specimen Test Test Literature
Test was Conducted Designation in Environment Temperature Reference
Fig. 1.4.13 (°F) in Text
Oak Ridge National ORNL Heat A Argon 1500
Laboratory
Battelle Memorial Institute BMI Heat A Vacuun 1500 1
Stanford University Parkman and Shephard Lead 1500 2
University of California Orr and Bainbridge Helium 1600 3
Helium 1735 3
Oak Ridge National Heat K-4 Argon 1800
Laboratory Heat WC~18 Argon 1850
Pratt & Whitney Aircraft PWAC-l* Lithium 1800 L
PWAC-E** Lithium 1800 L
Battelle Memorial Institute BMI Heat B Vacuum 2300

*
Data presented on Fig. 1.4.13 as an average value for several arc-melted heats.

*%
Data presented on Fig. 1.4.13 as an average value for several electron-beam-melted heats.

induced per cycle. Results obtained at 1300 and lBOOOF show excellent

agreement with ORNL data in the number of cycles to fracture obtained

at 2 cpm using equivalent plastic strains as the comparison factor.

6 .
Thus, the freguency effect noted previously 1is more representative of

an abrupt change in deformation mechaenism than it is of a continuous
variation with frequency.

Thermal fatigue studies in progress under a subcontract with the
University of Alabama are presently concerned with the effect of fre-
quency and also the change in material behavior as a function of the
temperature. Afi analysis of the hysteresis curves is being made to
determine the effect of thermal cycles in which work hardening occurs
in one half the cycle and softening occurs in the other half. No

conclusions can be drawn from the limited data presently available.

59
1.5. CERAMICS RESEARCH

Beryllium Oxide Studies

The properties of beryllium oxide are being investigated in order
to develop a process for producing a beryllium oxide powder of consis-
tently high purity and high sinterability. The requirements of an oxide
powder suitable for production of high-density BeO shapes by cold
pressing or extrusion are being studied. Means will also be sought
for the inhibition of grain growth at elevated temperatures (2750°F).

The initial objective of these studies is to obtain a suitable
and consistent BeO powder for the production of high-density shapes by
cold pressing or extrusion and sintering. Tentatively, it has been
specified that the material is to contain less than 500 ppm of impurities
and that it is to be sinterable to a density of approximately 9u4% of
theoretical. Investigation of methods for preparing the powder are
described in Chapter 2.1 of this report, and the calcining and sintering
studies are described here.

The first approach to the problem was to duplicate and, if possible,
improve upon the product derived from the oxalate according to a process
developed by GE-ANPD and designated GEOM-grade beryllium oxide. A
quantity of GEOM-grade beryllium oxide was obtained from GE~ANPD for
processing to establish a basis for comparison and correlation of data.
Fabrication variables were adjusted until it was determined that the
sintered density of 94% of theoretical attained at GE could be duplicated
with standard l/2-in. pellets. Pellets of the required density were
obtained by prepressing the powder at 15,000 psi, cold forming at a
pressure of 10,000 to 20,000 psi, and sintering at l650oC in a helium
atmosphere in a carbon-tube induction furnace.

Several lots of beryllium oxalate prepared at ORNL as described in
Chapter 2.1 were calcined to the oxide in an oxidizing atmosphere in
quartz or alumina containers. A product with lower aluminum, calcium, and
silicon content was obtained by calcining in quartz. The calcining

temperatures were varied from 900 to 600°C in order to vary the surface

Lo
area of the product. The densities obtained by sintering powder of

various surface areas are given in Table 1.5.1. The results indicate

Table 1.5.1. Surface Area and Sintered Density of Beryllium
Oxide Powder Obtained by Calcining Beryllium Oxalate
at Various Temperatures

Calcination Powder Surface Density, %, Obtained
Temperature Area (mg/g) by Sintering at 1650°¢
(°c) in Helium
900 2k.9 93.5
800 3k 92.4
700 % 65 89.7
600 111 Could not be fabricateq

without laminations

that sinterability is adversely affected as the powder surface area is
increased. Although optimum calcining conditions have not been established,
it has been determined that if the pbrescribed chemical procedures are
followed and the oxalate is calcined at 9OOOC, a product comparable in
appearance, surface area, purity, and sintered density to that prepared

at GE is obtained. Beryllium oxide rowder made at GE-ANPD by the oxalate
process 1s shown in Fig. 1.5.1; that made at ORNL which was of comparable
sinterability is shown in Fig. 1.5.2.

Materials of greater purity than those described above were prepared
at ORNL and at The Beryllium Corp. by using the GEOM process, but when
fabricated and sintered under the same conditions, the sintering charac-
teristics of the purer materials were markedly inferior. Beneficiation
of the powders by mixing with stearic acid did not improve their sintering
characteriéfiics. Since a number of variables in the sintering properties
of these mafiérials have not yet been defined, no conclusions can be based
on these résults, but it appears to be significant that the calcium and

silicon contents are appreciably greater in the highly sinterable

2n

UNCLASSIFIED
YE-5308

Fig. 1.5.1. GEOM-Grade Beryllium Oxide Prepared by GE-ANPD.

22,700X. _wi'h caption)

S UNCLASSIFIED
YE-5220

£

Fig. 1.5.2. GEOM-Grade Beryllium Oxide Prepared by ORNL.

23,200X. NP ith caption)
material, as shown in Table 1.5.2. It is planned to continue the

Table 1.5.2. Sintered Densities of GEOM-Grade Beryllium
Oxide of Varying Purity

Major Impurities

Processor Powder Syrface (ppm) Sintered Density
Area (m“/g) Ca Si (% of theoretical)
GE-ANPD 1Y 170 100 QL
ORNL 2L 230 130 93.5
Beryllium 17 10 Trace (5
Corporation
ORNL 21 70  Faint trace 164

preparation and study of very pure beryllium oxide.

Requisites for Sinterability

Although a satisfactory oxide can be obtained by the oxalate-
calcination process on a laboratory scale, the requisites for a satis-
factory starting material have not yet been established. X-ray spectro-
meter analyses were made of the GE and Beryllium Corp. oxides listed
in Table 1.5.2 in order to determine whether the sinterability of Be0
may be contingent on some type of defective lattice structure within the
crystallites. Since no shifts could be observed in the diffraction
patterns and there were no differences in line thickness, it is presumed
that the unit cell sizes are identical and the particle sizes of the
crystallites are relatively equal.

A thermal treatment was used to induce lattice defects in the
Beryllium Corp. material in an attempt to change its sintering charac-
teristics. The material was cold pressed into 3/8 x 3/8-in. pellets,
heated to 900°C (the original calcination temperature), and quenched
in a bath of liquid nitrogen. Pellets were fabricated from the quenched
oxide by the standard cold pressing and sintering techniques. Density

measurements showed no change in sintering characteristics, and an x~-ray

k3

diffraction examination of the material indicated no change in the

unit cell size.

Grain Growth Studies

Beryllium oxide shapes fabricated from oxides that meet the purity
and density specifications exhibit excessive grain growth both during
the initial sintering operations and during service. Although the
optimum grain size has not been established, crystals greater than 35
to 4O | adversely affect the mechanical properties of the sintered
shapes.

It is planned to introduce trace amounts of additives to the oxide

by coprecipitation in an effort to inhibit or control grain growth.

Ll

-

PART 2. CHEMISTRY AND RADIATION DAMAGE

2.1. MATERIALS CHEMISTRY

Preparation of Charge Material for Reduction to Yttrium

Phase Equilibria in the LiF-YF_, System
-~

A phase diagram for the LiF-YF5 system has been constructed from
the results of thermal-analysis and thermal-gradient quenching experi-
ments. The system contains a single eutectic at 19 mole % YF5’ which
melts at 69700. The single compound observed is LiF-YFB, which melts

incongruently to YF, and liquid at 816°C.

5

The phase diagram differs considerably from data presented by

Dergunov,l who reported a single eutectic at 74hk°C and 19 mole % YFB’

but who found no evidence of the binary compound.

Preparation of YF, from Y295
~
The preparation of YF5 by reaction of solid Y

HF at 1100°F has been studied,-’~ and YF

205 with anhydrous
3 with as little as 900 ppm of

oxygen has been obtained by long treatments with HF. However, since
a subsequent liquid~-phase hydrofluorination of the entire charge of
fluoride salts readily removes large quantities of oxygen, a more

economical treatment time can be used and a cruder product accepted.

When Y, O, is treated with an excess of molten ammonium bifluoride,

23
ammonium hexafluoyttrate is obtained that may be decomposed at 600°C

to form yttrium fluoride. The equations for this simple process are:

6NHuHF2 + Y205-——4>2(NH4)§YF6 + 3HL0

(NHM)BYFé-——e>YF + 3NH, + 3HF

5 5

1

E. P. Dergunov, Akad. Nauk. S.S.S.R. 60, 1185 (1948).

°G. J. Nessle et al., ANP Quar. Prog. Rep. March 31, 1958,
ORNL-2517, p 53.

6. J. Nessle et al., ANP Semiann. Prog. Rep. Sept. 30, 1958,
ORNL-2599, p 65.

Analysis of the product, based on reaction with BrFi, showed only 300
ppm oxygen. Such material is sufficiently pure for direct use in phase
studies.

Crystals of (NHLL)5YF6 produced by this method were examined with
the x-ray diffraction camera, but they were so fine that optical micro-
scopy was not possible. The x-ray diffraction pattern, which has

apparently not been reported in the literature, was obtained.

Laboratory Scale Preparation of LiF-MgF -YF, Mixtures
(= o

Experimental preparations on a 5-1b scale were continued to optimize
the process for obtaining oxygen-free LiF-MgFE-YF3 mixtures for reduction
to yttrium-magnesium alloy. Previous experimentation had shown that
hydrofluorination for 3 hr yielded a fluoride mixture containing about
1000 ppm oxygen. Increases in the hydrofluorination time produced mix-
tures with lower oxygen content, as listed below, where the oxygen values

are averages for at least three batches:

Hours of HF Treatment Oxygen Content of
at 850°C Product (ppm)
5 1100
b Loo
P 300
6 150
8 200

No increase in concentration of metallic impurities resulted from the
increased hydrofluorination time. The average values for metallic
impurities were: 200 ppm Fe, 50 ppm Cr, 50 ppm Ni, <20 ppm Cu, and
<10 ppm S.

The 150 to 200 ppm O

2
may be a consequence of traces of H

remaining after 6 to 8 hr of hydrofluorination
2O in the "anhydrous" HF or in the

hydrogen subsequently used to reduce NiFe, FeF., and CuF,. IZquipment

27 2

is being assembled to permit the use of dilute mixtures of F2 in HF to

ensure the absence of H20 in the HF.

L8

Three runs were made to determine the feasibility of using LiF-
MgFE-Y205 mixtures directly in the molten-salt purification procedure.
An arbitrary hydrofluorination time of 10 hr at 85000 was chosen for

these runs. The LiF-MgFE-YF mixture obtained from this process averaged

>
about 900 ppm oxygen, and there was extensive corrosion of the processing
equipment by the wet HF. It was therefore evident that prior conversion

of the YEO3 to YF5 was the preferred technique.

Production-Scale Operations

One run was made during this report period to convert 364 1lb of
YEOB to YF5' Hydrofluorination at llOOoF for 40 hr yielded 439 1lb of
YF3; this represents a 93% yield. The remainder of the material was
lost to the HF disposal system as fine dust. The oxygen content of the
product was slightly less than 1%, based on the mean of four analyses.
This material is satisfactory for use in the molten-salt hydrofluorina-
tion process.

Twenty-seven 50-1lb batches of LiF-MgF2~YF were prepared. Since

the gas sparging system and lines are identicaf with those on the 5-1b
units, agitation of the melt and efficiency of contact between melt
and vapor are poorer in the larger scale apparatus. Accordingly, the
oxygen content of the product is higher for a given hydrofluorination
time than in the smaller unit. The oxygen content is given below as a
function of treatment time in the 50-1b units; analytical results are
not available for tests in which 18-hr HF treatments were used. The

following oxygen values are averages for at least three batches:

Hours of HF Treatment Oxygen Content of
at 850°C Product (ppm)
L 900
6 650
8 450

Simple equipment is being assembled to stir the melt during the
hydrofluorination and hydrogenation steps. With improved agitation,
it is anticipated that removal of oxygen to the 150- to 200-ppm level

49

should be achieved. It is possible that even lower oXygen values can

be realized if HF-F2 mixtures are employed.

Preparation of Pure Beryllium Oxide

A study of procedures for the preparation of very pure BeO has been
initiated as part of an effort to develor improved beryllia ceramics
for high-temperature moderators and fuel elements (see Chap. 1.5, this
report). The effort to date has been confined to a study of modifications
of the beryllium oxalate process that has been shown by the General
Electric ANP program to yield sinterable BeO.

Beryllium hydroxide obtained from The Beryllium Corp. has been used
as the starting material. This material is dissolved by fusion with
oxalic acid monohydrate and the resulting melt is cooled and dissolved
in the minimum volume of hot water. After filtration to remove impurities
with insoluble oxalates, this solution is cooled to yield crystals of a
beryllium oxalate, This material, whose principal x-ray diffraction
lines are at 4.59, 2.885, and 3.22 R, is probably BeCQOM-BHEO. Drying
this material at 120 to lMOOC yields a phase, with principal lines at
7.08, 3.98, 3.82, and 2.475 R, whose yield of BeO on calcination shows
it to have been BeCeou-HEO.

The beryllium oxide prepared in this laboratory compares favorably
both in purity and in sintering characteristics with a sample obtained
from General Electric. When containers of Teflon or of polypropylene
were substituted for glassware in this process, the calcium content was
lowered from 180 to 70 ppm and the silicon content from 110 to a value
less than the 80 ppm in the spectrographic standards.

Purification of Lithium Metal by Extraction
with Molten Salts

The effective extraction of lithium metal impurities by the molten

LiCl-LiF eutectic mixture was partially demonstrated pJ:'evn‘.ously.LL In

uG. M. Watson and J. H. Shaffer, ANP Semiann. Prog. Rep. Sept. 30,
1958, ORNL-2599, p 67.

50

an attempt to improve this extraction process, an experimental assembly
was constructed to provide continuous extraction by circulating lithium
metal through an extraction chamber in a closed loop. The primary
advantages of this system are that relatively large volumes of lithium
metal may be processed and that contamination of the purified product
through mechanical handling is minimized.

In operation, this laboratory-scale extraction process has presented
various mechanical problems. While these difficulties are attributed
primarily to freezing of the lithium metal in unheated portions of the
lithium transport system — the magnetic flowmeter tube and the electro-
magnetic pump cell — techniques gained from use of the apparatus indicate
that a satisfactory operating procedure can be achieved.

The data obtained from a single extraction experiment are presented

below:
Weight of lithium metal 700 g
Weight of LiCl-LiF in extractor 13500 g
Extraction time 25 min
Lithium flow rate 2250 cm5/min
Alkalinity in salt phase following 0.217 meq/g of salt
extraction
Apparent oxygen removed 2290 ppm

The completion of this experiment was interrupted by freezing of metallic
lithium in the magnetic flowmeter tube. This failure will be avoided

by raising the temperature of the circulating lithium metal.

2.2. ANALYTICAL CHEMISTRY

Methods of analysis have been developed and evaluated specifically
for application to samples of materials and products from the yttrium -
pilot plant. Three types of samples are involved: the salt mixture
LlF—MgFg—YFfi,
the yttrium metal produced from the alloy.

the yttrium-magnesium alloy produced from the salt, and

Determination of Oxygen in LiF-MgF,.-YF

5

The oxide content of the LiF-MgFE,-YF5 samples is of primary interest.
A fluorination methodl in which the metallic oxides present are reacted
with KBrFu to liberate oxygen gas is now used routinely for the determi-
nation of oxygen in these samples. Results obtained for duplicate samples
agree to within 10% in the range 500 to 5000 ppm of oxygen. On the basis
of the analytical data, the salt samples appear to be homogeneous with
respect to oxide contamination and, furthermore, are not significantly
subject to atmospheric contamination. Iron, nickel, chromium, and sulfur

are also determined in the salt by conventional methods.

Determination of Oxygen in Yttrium-Magnesium Alloy

Oxygen contamination is also of highest importance in the alloy.
Attempts were made to use both the vacuum fusion and the inert-gas fusion
(Leco) methods for the oxygen determination, but the high temperatures,
AJEOOOOC, involved in such determinations precluded success. Magnesium
was volatilized and, in the vapor state, was found to "getter" the carbon
monoxide formed by the reaction between the oxide and carbon. The
fluorination technique described above was thus found to be the only
applicable method for the analysis. 1In the fluorination method, magnesium
is converted at h5OOC to its fluoride salt, which, of course, does not

lA. S. Meyer, Jr. and G. Goldberg, ANP Quar. Prog. Rep. Sept. 30,

1957, ORNL-2387, p 150. .

52

react with the oxygen produced by the reaction between KBrF4 and the
metal oxides present. Special care is taken to avoid excessive exposure
of the alloy to the atmosphere, although preliminary experiments have
demonstrated that the alloy does not react with the atmosphere appreci-
ably, even when rather finely divided. The possibility remains, however,
that in the initial exposure of the alloy to the atmosphere a protective
oxide film is formed. Attempts are to be made to sample the molten alloy
and avold atmospheric contact prior to analysis.

Nitrogen and fluorine are also determined in the alloy by the

conventional micro-Kjeldehl and distillation methods, respectively.

Analyses of Yttrium Metal

Determination of Oxygen

Both the inert-gas fusion (Leco) and the vacuum-fusion methods are
being used for the determination of oxygen in yttrium metal. 1In general,
the agreement between the methods is satisfactory (210%). The coeffi-
cient of variation on standard iron samples is about 5% for both methods.
Gross disagreements have been observed from time to time which are
considered attributable to nonhomogeneity of the samples, since agreement
is usually excellent on vacuum-annealed metal samples. Replicate samples
from a particular billet were taken at 0.5-in. intervals and values for
oxygen were established to provide a truly reliable value for the oxygen

concentration.

Determination of Nitrogen

The micro~Kjeldahl and the vacuum-fusion methods are used for the
determination of nitrogen in yttrium metal. These methods are satis-

factory in all respects.

Determination of Fluorine

The Willard-w:i.nter2 distillation method is used to remove the

2H. H. Willard and O. B. Winter, Ind. Eng. Chem., Anal. Ed. 5,
No. 1, 7 (1933). -

fluorides from yttrium metal as hydrofluosilicic acid and a colorimetric
determination of the fluorides is made by the conventional thorium-thoron
procedure. This method is entirely satisfactory for determinations of
fluorides in excess of 50 ppm. Some attention is being given to the
pyrolysis technique for the separation of the fluorides. The results
obtained to date indicate that the fluorides are definitely segregated
by this technigue.

Determination of Magnesium

Flame photometry has been apprlied with limited success to the
determination of magnesium in yttrium metal. The lack of success has
been attributed to two factors: (1) the relative insensitivity of the
emission-band spectra of magnesium, and (2) the high background emission
due to the yttrium matrix. As a consequence, methods of separating
magnesium from yttrium are being investigated. An ilon-exchange method5
was tested in which sulfosalicylic acid was added to a buffered solution
(pH 8) to form an anionic complex with yttrium and a cationic complex
with magnesium. The yitrium complex is then absorbed on an anion-exchange
resin (Amberlite 1RA4O1l) to separate it from the magnesium. Preliminary
results indicated that approximately only 70% of 50 mg of yttrium was
removed by this separation, and, in addition, 40% of 1 mg of magnesium
was coadsorbed. The separation does not therefore appear to be feasible.

A liquid-liquid extraction methodlL for this separation proved some-
what more promising. In this procedure, yttrium in a solution (pH 1,

1 M Cl) was extracted with 0.5 M di-2-ethylhexyl phosphoric acid in
cyclohexane. Magnesium, supposedly, 1s not extracted under these
conditions. In excess of 99% of 50 mg of yttrium was extracted in two

steps; however, as much as 60% of the magnesium was also extracted.

5R. T. Oliver and J. S. Fritz, Ion-Exchange Separation of Metals
by a Single-Pass Method, ISC-1056 (June 1950).

n
J. C. White, Extraction of Metal Jons with Di-2-Ethylhexyl
Phosphoric Acid, ORNL CF 57-2-37 (Feb. 8, 1957).

gLt

Further work is planned to improve the extraction by altering the

conditions so as to decrease the extraction of magnesium.

Determination of Lithium

The flame photometric method was found to be satisfactory for the
determination of lithium in yttrium. The emission band at 670 my is
sufficiently sensitive; that is, approximately 10 ppm of lithium is
detectable.

Determination of Titanium

p)

The thiocyanate extraction method” has been applied successfully
to the determination of titanium in yttrium. This method covers an

extremely wide range of concentration, from 5000 to 5 ppm.

’J. P. Young and J. C. White, Anal. Chem. 31, No. 3, 393 (1959).

22
2.5. RADIATION EFFECTS

Diode Experiment

A grown-junction silicon diode was placed on a Teflon terminal
strip, and its reverse characteristics were measured on two independent
instruments. The sample was approximately 1.0 cm long and 5 mm on a
side.

One of the systems was built around a vibrating-reed electrometer.
The head of the electrometer was modified with specially selected
premium components. The stability and noise were significantly reduced.
The sample reverse current was measured across a decade shunt (lO2 to
1012 ohms) by the electrometer.

The problem with the system was trying to measure reverse current
at constant sample voltage. The sample current was measured at ten
voltages from 1 to 10 mv. A selector switch, which was mechanically
coupled to the input switch of a 1l2-point recorder, provided the desired
sample voltage. The output of the electrometer was measured on the
recorder. At the same time, a read-out slidewire on the recorder
exactly balanced out the voltage drop of the current shunt. Thus for
any voltage drop on the shunt, the sample voltage was raised by exactly
that value which would keep the total voltage on the sample constant.
The system is shown in Fig. 2.3.1. It may be seen that the system has
several servo loops in it; their balancing speed was chosen such that
there was little, if any, interaction between the loops. At extremely
high shunt resistance, however, the RC time constant caused by the
shunt and sample resistance and the capacitance of all the shielding
required lengthy waiting periods for the system to reach a stable
condition.

A servo system was incorporated to automatically measure the zero
drift of the electrometer once every 12 points and to adjust the elec-
trometer to correct for the drift. Thus, the problem of zero drift at
low currents for extended periods has been eliminated. With the system

described in Fig. 2.35.1, it was possible to follow the small reverse

56
voltage current of a diode to lower temperatures than has been possible
up to this time and to monitor this current accurately for extended
reriods.

Another useful instrument in the study of semiconducting barriers
is the function plotter or "x-y" recorder. Such an instrument can
readily plot the current flow through the barrier as a function of the
applied voltage. Because of the nature of the work, such an instrument
should meet some rather strict requirements. Some of these are listed
below:

1. It should be capable of measuring currents over a range of 10
or 11 orders of magnitude. OSuch a range is often covered in studying
the forward and reverse characteristics of barriers as a function of
both temperature and radiation.

2. It should have well-shielded and isolated circuits. Electrical
noise may be picked up by wires to and from the sample, be rectified,
and cause erroneous readings. Such errors may also be caused by ground
loops. Hence, it is desirable to use instruments having floating inputs
in order to have only one ground point in the experiment circuity.

5. It should have high input resistance. It is often desirable
to detect millimicroampere current changes in samples having hundreds
of megohms resistance.

An instrument system has been built which satisfies the above
requirements. It consists primarily of a standard "x-y'" recorder
having floating inputs. Each axis of this recorder is preceded by a
battery-powered, isolated-from-ground, high-input-resistance amplifier.
A forward-and-reverse resistor bank is used with the recorder to permit
forward and reverse characteristics that may differ by a factor of 10
to lO6 to be recorded on the same graph at the turn of a switch. Terminals
for external resistors are also provided. The units are connected through
a plug-board to permit various functions of the amplifiers (which are
in reality resistance, current, and voltage meters moditied for this
application) to be used in conjunction with the recorder.

The instrument system has been shielded to the extent of providing

o7

a copper screen shield between the operator and the instrument to keep
electrostatic change from causing erroneous recordings. The system can
accurately detect and record voltages of less than 1 millivolt to 1000
volts and currents of less than lO-ll to 10 amp or more.

The relative properties of the ditftusion mechanism of conduction
and the charge-generation mechanism of conduction were discussed pre-
viously,l and it was shown that a germanium diode conducts initially
by the diffusion mechanism and then shifts to a charge-generation mecha-
nism because of the shallow levels introduced by irradiation just below
the conduction band.

Since silicon Jjunctions conduct initially by a charge-generation
mechanism, it was not obvious a priori how the changes in IO were
produced. Therefore, the equipment mentioned above was assembled and
a grown-junction, single crystal was obtained. Measurements of IO at
various temperatures showed an activation of conduction associated with
energy levels lying near the middle of the forbidden band. Consequently,
any terms introduced by levels produced by irradiation would show up in
the low-temperature range and would become important at higher temper-
atures as their density increased. The results of a series of exposures
in the ORNL Graphite Reactor are shown in Fig. 2.3%.2.

Plots of IO as a function of the reciprocal of the absolute temper-
ature show two distinct activation energies, neither of which can be
clearly attributed to a charge-generation mechanism. There are two
possible explanations. Haynes and Hornbeck have observed two levels
in silicon at 0.57 ev and -0.79 ev from the conduction band.2 Evidence
of a spectrum of levels introduced by neutron bombardment running from

0.16 ev below the conduction band toward the middle of the forbidden

lJ. C. Pigg, C. C. Robinson, 0. E. Schow, ANP Quar. Prog. Rep.
March 31, 1958, ORNL 2517, p ©60.

2J. R. Haynes and J. A. Hornbeck, "Trapping of Minority Carriers in
Silicon," p 321 in Photoconductivity Conference, ed. by R. G. Breckenridge,
B. R. Russell, and E. E. Hahn, Wiley, New York, 195k.

58

UNCLASSIFIED
ORNL-LR-DWG 36949

VIBRATING—
DIODE DECADE SHUNT HIGH REED | INPUT
SAMPLE
10*2 —10* "2 ghms “ow | ELECTROMETER SELECTOR ==

—| ’—- 0=10mv poeme—  SWITCH

VOLTAGE AUTOMATIC
DIRECT- AUXILIARY L. VIBRATING-REED
SELECTOR RECORDER
CURRENT == £ SLIDEWIRE ELECTROMETER =
SUPPLY 0-10mv ZERO—-DRIFT
1-10 mv
—I CONTROLLER

DIRECT-
CURRENT
SUPPLY

I
I
|
I
I
SWITCH 0-3mv II
I
|
|
|
I
I

Fig. 2.3.1. Dlagram of Circuit for Diode Experiments with Auto-
matic Vibrating-Reed Electrometer Zero-Drift Adjustment.

UNCLASSIFIED

-4 ORML-LR-DWG 36950
10 I [ 1 -
ACTIVATION ENERGY, 0.72 ev !
INTEGRATED FLUX, 4.6 x 10> neutrons 7cmé
| i !
!
10°° # |
L0.75¢ev | '
1.8 x 10! neutrons secm?
|
| ‘ ..._ 5
I I I
a -0 . 365ev
£ 7.3x10!5
8 " 2
) eutrons /cm
(58 N

0 58ev
UNIRRADIATED

103/7 (°x)

Fig. 2.3.2. Change in I, Activation of a Grown-
Junction Silicon Diode as o Result of Neutron Bom-
bardment.

09

band had been reported by Wertheim.5

Such a spectrum could constitute
a low-density, impurity type of band which could provide the small
currents observed. The effective band gap would be essentially that
observed in the first irradiation.

A decrease in the barrier height would also produce such an effect.
This becomes evident in the data after the third irradiation and is
suggested by the second irradiation. Since neither barrier height nor

capacitance was measured, the question 1s not resolved at this time.

Creep and Stress Rupture Tests Under Irradiation

The effects of irradiation on the creep of metals at high temper-
atures are being studied. Two sets of tube-burst, stress-rupture
tests of Inconel have been completed at lBOOoF. The first set of tests
was conducted in HB-3 of the MIR and the second in the poolside facility
of the ORR. There is good agreement of the results from both sets of
tests which show that irradiation (in air) at a stress of 5000 psi
shortens the time to rupture by at least a factor of 4. At lower
stresses (longer times to rupture) a reduced time to rupture is also
observed, but the data are less accurate. The reduced time to rupture
is believed to be due, in part at least, to the presence of boron in
the grain boundaries of the Inconel.

The present series of tests was conducted in air because earlier
testsl‘L attempted in a helium atmosphere had failed when thermocouple
errors were introduced by unknown contaminants (believed to have been
supplied by thermal insulation). The MIR creep apparatus was described

earlier.5

5

G. K. Wertheim, Phys. Rev. 111, No. 6, 1500 (1958).

uJ. C. Wilson et al., ANP Quar. Prog. Rep. Dec. 31, 1957, ORNL-24LLO,
p 207.
p

p 55.

J. C. Wilson et al., ANP Quar. Prog. Rep. Mar. 31, 1958, ORNL-2517,

60

The ORR tests were conducted in the pool of the reactor in front of
fuel element A~5. The geometry is probably better than in HB-3 of the
MTR, and the fast neutron fluxes are comparable (see Table 2.%.1). The

Table 2.%3.1. Effect of Neutron Bombardment on the Time to
Rupture of Inconel Tubing at 1500°F in Air

MIR fast flux (HB-3): >1 Mev, 2 x lO15 neutrons/cmg-sec 5
ORR fast flux (poolside facility): »>1 Mev, ~'5 x 10+3 neutrons/em”-sec

Wall

Specimen Thickness Stress Time to Rupture

Number (mils) (psi) Irradiation (hr)
71 30 3000 MR 784
15 30 3000 MTR > 869
DD 20 3000 ORR L67
L8 30 3000 None »2850
6k 30 4000 MTR 343
65 30 4000 MIR 279
43 30 4000 None 801
Ll 30 4000 None 197
68 30 5000 MTR 9L
76 30 5000 MT'R 96
2-1 30 5000 ORR 91
2~5 30 5000 ORR 100
2-6 30 5000 ORR 57*
35 50 5000 None 473
L1 30 5000 None L87
28 30 5000 None 410
37 50 5000 None heg**

*
Shorter time to rupture possibly due to 20-min temperature excursion
to 1600°F.

Tests conducted with helium inside specimens and air outside. All
other tests had air inside and outside of specimens.

61
great advantage of the ORR 1s that there is much more room for specimens,
access 1s easier, and the cost and time required for building experiments
is a fraction of that for the MTR. The initial test apparatus held only
six specimens, but ten are being built into the second apparatus. The
specimen can is of the same height as the reactor fuel elements (24 in.),
and it is 3 in. wide and 12 in. deep. The specimen design (including
the air-cooling finger to dissipate excess gamma-ray heat) is the same

as that of the specimens used in the MTR apparatus.

The results of the MIR, ORR, and out-of-pile tests in air at lSOOOF
are presented in Table 2.3%.1. Four specimens ruptured between 90 and
100 hr at 5000 psi in MIR and ORR. One ORR specimen ruptured at 37 hr,
but this rupture is believed to have been due to a 20-min temperature
excursion to 1600°F at 30 hr. Out-of-pile test specimens ruptured at
about 450 hr under the same conditions. It may be concluded that a
definite decrease in time to rupture is suffered by Inconel under these
conditions.

For tests at a stress of 3000 psi the data are less conclusive
because the longer operating times (including numerous reactor shutdowns)
increase the chances of equipment errors or failures. In addition,
rupture at longer times in air atmospheres is often slow and difficult
to detect, even upon postirradiation examination.

Metallographic examinations of earlier in-pile specimens tested
in an inert atmosphere had disclosed that the events leading up to
fracture were somewhat different in the irradiated metal.. Two important
observations were made of irradiated Inconel. First, the specimens
failed with much less prior elongation and much less general cavitation
at grain boundaries. Second, groups of spherical cavities were often
found at grain boundaries, but the cavities so formed were not always
at grain boundaries that were more or less normal to the major stress.
In unirradiated Inconel, grain boundary separations are roughly normal
to the direction of the major stress.

Examinations of some of the recent MIR specimens (unetched) have

shown features similar to those observed in the inert-atmosphere tests;

62
in particular, the spherical grain-boundary cavities are prominent, and
they are formed on grain boundaries that are not normal to the direction
of stressing. Further examinations of MTR and ORR specimens will be made
when the latter are removed from the reactor.

As suggested by Cottrell,6 it appears that the growth of cavities
at grain boundaries will occur by the Nabarro-Herring creep process
when the applied tensile stress across a grain boundary is greater than
P = g;z, where p is equivalent pressure tending to close a cavity, y is
the surface energy, and r is the radius of the cavity. For a cavity to
grow, a certain normal (to grain boundary) stress is required, but a
shear strain is necessary for nucleation of the cavity. In the case of
Inconel, cavities were observed to form and grow on grain boundaries on
which there was little or no normal, tensile component. It is therefore
suggested that irradiation effects in Inconel are at least partially due
to gas pressure in grain-boundary voids. The gas is believed to be
helium formed by thermal-neutron capture in Blo. Boron is usually present
in Inconel (on the order of 0.005% or greater) and is likely to be
segregated or precipitated at grain boundaries because of its extremely
low solubility in nickel. Thus the mechanism of fracture in irradiated
Inconel is likely to be different from that in the unirradiated metal
because of the helium pressure exerted in the voids once they are formed.
It is possible that nucleation of the voids may also be affected by helium
production in the grain boundaries. Microexamination will help confirm
the above hypothesis. If the hypothesis is correct, it will aid in
extrapolating and predicting behavior under other conditions.

If the irradiation effects in Inconel are due principally to boron
in the grain boundaries, the results may give no indication at all about
the high-temperature creep of metals not containing boron. But, if

boron is the major cause of the shortened time to rupture, materials to

6A. H. Cottrell, "Theoretical Aspects of Fracture," National

Academy of Sciences — National Research Council, International Seminar
on the Atomic Mechanisms of Fracture, Swampscott, Mass., April 12-1k,
1959 (to be published).

63

be used at high temperatures in much higher fluxes must have proportion-
ately lower boron contents in order to be free of the effects observed
in Inconel. Boron is often present in small quantities in metals, and
intentional boron additions are now made to some stainless steels to

improve the hot-working characteristics.

Irradiation of High-Temperature
Moderator Materials in ETR

Two capsules, one of yttrium hydride and one of beryllium oxide,
were irradiated for a period of approximately 64 hr in the ETR while it
was operating at a power of 175 Mw. Temperature control, which was
accomplished by varying a mixture of helium and argon, functioned as
planned..7 The gas mixture was nearly 100% helium during the first half
of the run because the gamma heat exceeded the expected value of 25 w/g
by about 10%. A gradual buildup of gamma-ray flux made it necessary
toward the end of the run to use 100% helium in the gas annulus. Four
thermal cycles occurred in which the temperature of the specimens was
reduced to less than 7500F. At this time the capsules are enroute to
Oak Ridge for examination. The following temperature conditions existed

during the test:

Outside temperature of yttrium hydride 15550F
Central temperature of yttrium hydride 1921°F
Outside temperature of beryllium oxide lEh?oF
Central temperature of beryllium oxide lj}EoF

The absence of abrupt changes in temperature indicates that the specimens
did not crack during the test.

Two additional test assemblies are being prepared. These assemblies
will contain three specimens of yttrium hydride and three of beryllium
oxide. These specimens are to be tested for a longer time and will be

thermally cycled more times than in the test described above.

{W. E. Browning, R. P. Shields, and J. E. Lee, ANP Semiann. Prog.

6L

PART 3.

ENGINEERING

3.1. COMPONENT DEVELOPMENT AND TESTING

Irradiation Test of an Cil-Lubricated
Pump Rotary Element

Operation of the rotary element of an oil-lubricated pump in a
gamma-irradiation facility in the MIR canal was terminated on
October 15, 1958, after achieving an accumulated dose of lOlO r in the
bearing and seal region of the conventional test bearing housing. The
bulk of the o0il that was circulating received a dose of 2.L x lO8 T.

The viscosity of the circulating oil increased L45% during the test, and
one sample of the oil that leaked from the lower seal showed a viscosity
increase of 169%. These viscosity increases were caused primarily by
radiation, The time of irradiation was 5016 hr, and the total operating
time was 6018 hr.

The test piece was disassembled and inspected, and the seal and
lower bearing were found to be clean and in good condition. No
polymerized oil was found in any of the flow passages. The Buna N O-rings
which had been exposed to high levels of radiation were very hard and
brittle, while hardening of O-rings in other areas was relatively slight.

Elastomeric O-rings would not be used in high radiation areas in

actual pump installations.

Dynamic Seal Research

The development of face-type shaft seals for high-temperature
pumps has been hampered by a lack of information on the basic
phenomena involved in the operation of the seal. At present, the
design of a seal must be carried out by cut-and-try methods based on
limited empirical data.

Parameters to be investigated include the effects of seal diameter,
radial face width, axial (interplate) clearance, relative velocity of
the two plates, and sealant pressure, viscosity, and temperature on the
flow between the seal plates and on the torque exerted on the stationary

seal element. Other elements of the problem involve investigation of the

67

effects of the surface tension of the sealant and of the nature of the
circulation of the oil inside of and through the faces of the seal.
Initial studies have shown that two aspects of the design of the test
equipment require particular attention: first, means must be provided
to hold the axial gap between the seal faces at a preset value, with a
minimum of runout or other variation; and, second, means nust be devised
to measure the axial gap accurately while one of the seal plates is
rotating.

Specifications were written for the design of a dynamic seal
tester, which was to consist basically of two sealing disks (about 3 in.
in diameter) mounted on precision spindles, with the faces of the two
disks separated by a small gap (20 to 500 pin.). The seal faces were to
be optically flat, and the axial and radial runout of the rotating seal
face was not to exceed 10 pin., The rotating and stationary seal faces
were to be parallel within 10 pin; the gap between seal faces was to
be adjustable from 10 puin. to 0.001 in.; and the gap was to be measured
with one seal face rotating at a rate of from O to 3600 rpm.

Several design layouts were made, and it was decided to mount the
spindles in hydrostatic bearings in order to attain greater accuracy
than thought possible with rolling contact bearings. In order to
obtain the desired precision, the critical parts were to be generated to
tolerances in the 10=-iin. range. To determine whether these tolerances
could be met in a reasonable length of time and at an acceptable cost,

a survey was made of more than 20 manufacturers of precision machining
or measuring equipment. Three companies indicated that the parts could
be manufactured to meet the required conditions.

A number of measuring devices based on the use of optical equipment,
electronic optical devices, magnetic reluctance methods, and capacitance
methods were investigated, and it was determined that capacltance probes
offered the greatest promise, particularly since commercial devices were
available for performing capacitance measurements in the required range.
It would be necessary, however, for ORNL to develop the method of
applying the capacitance probes to the measurement of the seal garp.

68

A program for evaluating the capacitance method by measuring small
gaps statically at first, then dynamically in air, and finally in oils
or other seal testing fluids, has been initiated. It has been found that
gaps ranging from 100 to 1000 pin. can be measured under static
conditions to an accuracy of +3 win. The gaps were obtained by com-
bining precision gage blocks, and the minimum gap that could be obtained
with available blocks was 100 pin. Equipment has been set up to per-
form dynamic tests of the capacitance method, and preliminary experi-
ments indicate that 1t should be possible to develop an adequate device
in an acceptable length of time.

The 1nitlal layout of the seal tester has been completed, and
engineering estimates of the cost of fabrication of the high-precision
parts have been received from two firms. A preliminary layout of the
hydraulic supply system for the hydrostatic bearings has been completed
and sent to a consultant on hydrostatic bearing systems for review.
Vibration surveys have been made in Building 9201-3 to locate an area
most favorable for the operation of the seal test equipment. Possible
locations for the tester are being studied and a suitable enclosure is
to be designed. Detailed design of the tester will start as soon as a
method of measuring the seal gap has been proved by bench tests.

Thermal Stability Tests of Metal Shells

A third, thin-shell, Inconel test model was subjected to a
creep-buckling test at 15000F in an atmosphere of helium with an
external pressure of 52 psi. The shell buckled in 199.9 hr as
compared with 275 hr for the first shell and 111 hr for the second shell.
The test shell was similar to shells 1 and 2 described previously,l’2

lJ. C. Amos and R. L. Senn, ANP Quar. Prog. Rep. March 31, 1958,
ORNL 2517, p 69.

2B. L. Greenstreet and R. L. Senn, Thermal Cycling and Buckling
Tests of ART-Outer-Core-Shell Models, ORNL CF 586-5-6 (May 5, 1953).

with the exception that it was fabricated by welding together oversize
sections and reducing the weldment to the final dimensions by machining.
The upper half of the test shell had been subjected to 602 thermal
cycles under conditions similar to those imposed in shells 1 and 2
(described in detail in ref. 2) prior to the creep-buckling test.
Nondestructive dye-penetrant and x-ray inspections disclosed no adverse
effects on the shell that could be attributed to the thermal cycling.
Photographs of shells 2 and 3, after buckling, are compared in
Fig. 3.1.1. The results of these tests indicate that welded fabrication

can produce shells as structurally sound as all-machined shells.

70

UNCLASSIFIED
PHOTO 33765

UNCLASSIFIED
PHOTO 30754

SHELL NO. 2 SHELL NO.3
BUCKLING TIME 111 hours BUCKLING TIME 199.9 hours

Fig. 3.1.1. Photographs Showing Test Shells 2 and 3 After Buckling Under an External Pressure of
52 psi at 1500°F.

L

3.2. ENGINEERING AND HEAT TRANSFER STUDIES

Effect of Thermel-Stress Cycling on Structural Materials

The experimental studies of the effect on Inconel of thermal-
stress cycling in a fused-salt environment (NaF-Zth-UFh, 56~39=5
mole %) were continuedl with the use of a high-frequency pulse-pump
loop. Tests 13 and 1L, a continuation of the series at a cyclic
frequency of 1.0 cps, were completed. The results of these tests are
given in Table 3.2.1. On the basis of the improved temperature-

measuring techniques used in tests 13 and 1L and the similarity of the

Table 3.2.1l. Summary of Results of High~Frequency Thermal-Stress
Cycling of Inconel Pipe

Mean temperature: 1405 + 10°F
Inside diameter of test section: 0.L85 in.

Estimated Measured Wall Calculated
Test Inside Wall Temperature Thickness Stress on Total Number
No. Temperature Amplitude on of Test Inside Wall of Exposure
Amp%itude* Outsige Wall Section Fibers Cycles
("F) (°F) (in.) (psi)
12 +63 +5.5 0.1h7 10, 500 360,000
i18.8 0.091 7,700
13 +6} +6.3 0.147 10, 600 770,000
+16.2 0.091 8, 400
1l +46 +4.0 0.147 7,600 2,204,000
+13.0 0.091 5,700
+23.4 0.060 3,800

¥*
90% efficiency assumed; Jakob's equation used for calculating ¥{;
entrance effects neglected; and flow determined by heat balance.

lJ. J. Keyes, A. I. Krakoviak, and J. E. Mott, ANP Quar. Prog.

Rep. Dec. 31, 1957, ORNL-2240, p 5hi.

T2

operating conditions for these runs to those in test 12, the estimated
inside wall temperature fluctuation and inside wall fiber stress for
test 12 have been revised downward slightly from the previously
reported values.2

The results of tests 12 through 1b constitute a set of data for
studying the effect on the Inconel wall of 1,0-cps temperature fluctua-
tions for varying exposure times. The test sections used in these
runs were fabricated with a two-step axial variation in wall thickness
(as indicated in Table 3.2.1) so as to include in each individusl run
the effects of various inside wall stresses. The specimen used in
test 12 showed no cracking in either region of the test section after
360,000 cycles of exposure corresponding to a maximum stress of 10, 500
psi per l/h cycle. In contrast, the test 13 specimen, which was stressed
at essentially the same stress levels and exposed slightly more than
twice as long as the test 12 specimen, showed heavy intergranular crack-
ing to a depth of 0.006 in. in the upper section and lighter cracking
(0.0015 in.) in the lower section. A photomicrograph showing typical
cracks in the thick-walled region of the test 13 specimen is shown in
Fig. 3.2.1. As was to be expected, there was a marked increase in the
number of cracks in the stressed region created by the transition
between the upper and lower regions of the test section.

In test 1k the maximum stress was lowered to 7600 psi and the
exposure increased to 2,204,000 cycles. Metallographic examination of
the test specimen showed moderate to heavy intergranular attack and sub-
surface void formation that extended in some instances in the upper
thick-walled region to a depth of 0.012 in., as shown in Figs. 3.2.2
and 3.2.3. The pitted appearance of the surface seems to result from
the "cracking-out" of surface grains following corrosive attack along
the grain boundaries. Both the depth and concentration of attack were

noted to decrease with decreasing wall thickness. As in the previous

EJ. J. Keyes and A. I. Krakoviak, ANP Semiann. Prog. Rep.
Sept. 30, 1958, ORNL=-2599, p 83.

7>
UNCLASSIFIEDT
T-16196

Flg. 3.2.1. Effect of High-Frequency Thermal Cycling on Inconel In a Fused-Salt Environment — Test 13.
Photomicrograph of typical intergranular cracks in 0.147-in.-wall section. 250X. (fmmmmewith caption)

Th

. UNCLASSIFIED!
T-16517

Fig. 3.2.2. Effect of High-Frequency Thermal Cycling on Inconel in a Fused-Salt Environment — Test 14.
Photomicrograph of intergranular attack and subsurface void formation observed in 0.147-in.-wall section. 250X.

Tommm with caption)

5

. UNCLASSIFIED'

T-16549

Salt Envirenment — Test 14,

Fig. 3.2.3. Effect of High-Frequency Thermal Cycling on Inconel In o Fu
Photomicrograph of typical cracks and inner surface spalling observed in 0.147-in.-wall section. 250X. (Swmwmty

with caption)

6

test, increased attack was observed at points of transition in wall
thickness. The results of test 1k are of particular interest in that
(1) they provide a preliminary tie-in with earlier thermal-stress
cycling experiments with a pressurized system,3 and (2) they indicate
that application of relatively small thermal-stress fluctuations over
a sufficiently long period of time may influence the extent of

corrosive attack.

Molten Lithium Heat Transfer

Fabrication of an experimental system for the determination of
local heat-transfer coefficients for molten lithium has been completed.
Sufficient lithium to fill the loop is now being purified, and
preliminary clean-up operations will begin shortly. Since adequate high-
temperature corrosion-resistant materials were not available, type 347
stainless steel was used for all system components and piping, and to
avoid excessive corrosion it will be necessary to restrict experimental
operation to temperatures below 1000°F.. However, the data obtained,
when presented in dimensionless form, will be applicable to higher
temperature problems.

The test section is a 3/16-in.-ID, 11/32-in.-wall tube divided
into a hydrodynamic entrance region of 25 x/d length and two heated
sections each having a length of LO x/d. Pairs of diametrically
opposed thermocouples were resistance-welded to the tube wall at
positions along the heated portion of the test section. Since the
temperature differences to be measured are small, extreme care was
taken in the fabrication and installation of the thermocouples. Specially

annealed and calibrated Chromel and Alumel wires were used. Circulation

3H. W. Hoffman and D. P. Gregory, ANP Quar. Prog. Rep. Sept. 10, 1956,
ORNL-2157, p 229; Dec. 31, 1956, ORNL-2221, p 202; Mar. 31, 1057,
ORNL~-227h4, p 85; June 30, 1957, ORNL-2340, p 82; Sept. 30, 1957,

ORNL-2387, p 9l.

7
in the system is provided by a small inert-gas-shielded, oil-sealed,

centrifugal pump. Flow rates will be determined by an electromagnetic "
flowmeter. A gas-vortex heat exchanger was installed as the heat sink.
The entire system will be enclosed in an inert atmosphere (argon) box
to minimize the hazards associated with the use of molten lithium. It
is planned to obtain data for laminar, transition, and turbulent

(to Np, = 50,000) flow,

Thermal Properties of Columbium and Lithium

Measurements of the thermal conductivity of columbium are being
made with the use of a longitudinal heat flow apparatus. The test
specimen (a 6-in.-long, l-in.-dia rod) is contained between cylindrical
heat meters of type 347 stainless steel of equal diameter. Thermocouples
(Located at the axial centerline in radial wells) are spaced 1 in.
apart along the length of the apparatus. Radial guard heating is
provided at discrete intervals by a set of heated plates. An
electrical heat source at the top of the upper heat meter and a water
sink below the lower meter complete the apparatus. The entire assembly
is contained within an inert-atmosphere box to ensure against columbium
oxidation at the higher operating temperatures.

The apparatus for the measurement of the thermal conductivity of
molten lithium is being modified to provide more effective guard .

heating. Concurrently, lithium of sufficient purity is being prepared.

78

5+5. INSTRUMENTATION AND CONTROLS

Data Acquisition System

The data acquisition system, which was sent to the fabricator for
repair and modification, was returned to the Laboratory in December.

A large number of original design errors were corrected and the system
was modified to conform more closely to the latest systems built by the
fabricator. The system was also completely tested before being returned.
The unit was placed in operation in January and has operated over 40O

hr in 8-hr cycles.

Some malfunctions remain, which, although not serious, cause some
trouble when checking the completed logging format, since the mal-~
functions result in characters being skipped when typed. This trouble
too has been traced to a design error, and a modification would be
required to correct it.

The system is now operating with eight recorder inputs that take
a total of 96 individual inputs. The eight recorders represent six
temperature recorders with a total of T2 thermocouples, cne level
recorder indicating 12 level points during a level cycle, and one
pressure recorder indicating 12 pressures. The unit is built to handle
120 points or ten 12-point recorders. It is now being checked out to
handle the maximum inputs.

Based upon the operating experience thus far, it appears that the
system can be used to record data from an operating system. It is
certain that the presentation of the data is far superior to that of
the regular recorder charts. The accuracy is almost (+1 digit) as good
as that of the Brown recorders, The system evaluation has resolved to
a problem of reliability, which has nct yet been established. With
further operation and changes a more complete evaluation of the useful-

ness of the system can be obtained.

Liquid-Metal-Level Transducers

The previously described; tests of level probes in NaK were con-
tinued until a leak occurred in the flange sealing one of the transducers,
When the test rig was shut down, more than 10,000 hr of operation had
been accumulated on two of the probes and over 5600 hr on the other.,

The flange was repaired and the system was checked in January. The
three probes were then reinstalled in the tanks, and the system was leak
checked and placed in operation to continue life tests on the level
probes at 1200 to 1400°F. The units have now operated an additional
period of 600 hr.

The similar system being tested with sodium was also stopped because
of the growing severity of a flange-seal leak., The flange was repaired
by permanently sealing it., Three new probes were then installed in this
test rig and operation was resumed. The new units have operated approxi-
mately 600 hr at low temperatures. The temperature will be raised to

1200 to 1400°F for life tests.

Thermocouple Development Studies

Thermocouples for Use at High Temperatures

An investigation of thermocouples for use at temperatures up to
4000°F in liquid metals, in fused salts, and in air and other gaseous
atmospheres was initiated.
Tests of the compatibility of thermocouple materials with insulating .
materials are under way, and initial results are presented in Table
3.5.1. The results for molybdenum and rhenium have not been corrected
for emissivity of the quartz sight glass, and in all cases the reaction

temperatures are accurate only to 15000.

Chromel-Alumel Thermocouple Life Tests

Life tests of Chromel-Alumel Heliarc-welded sheath-type thermocouples

1
p 94,

G. H. Burger, ANP Semiann. Prog. Rep. Sept. 30, 1958, ORNL-2599, -

80

e

Table 5.%.1. Results of Tests of Compatibility of Thermocouple and
Insulating Materials in a Helium Atmosphere

Insulating Materials

Thermocouple Metals Al,0. BeO Mgz0 ThO2 TiO UO2

Tungsten
Initial reaction
temperature, °C none 2140 2030 2645 1775 1785
Temperature at which fast
reaction begins, °C 2310 2L95 none 1780 1780

Tantalum
Initial reaction
temperature, °C 2045 2245 1790 2795 1785 2295
Temperature at which fast
reaction begins, °c none 2555 2000 none 1785 2420

Columbium
Initial reaction
temperature, °C 2000 1480 1760 2135 1185 2065
Temperature at which fast
reaction begins, °C none 2135 1855 none 1515 2065

Molybdenum
Initial reaction
temperature, °C 1805 2136 1465 2155 1675 2155
Temperature at whigh fast Yy 2%
reaction begins, C none none 2136 2155 1845 none

Rhenium
Initial reaction
temperature, °c 1415 920 1555 1550 1640 925
Temperature at which fast o
reaction begins, °C none 2175 2260 2280 1775 2295

2000

1575
2155

1185

2220

1955

2125

2035
2155

1555

2010

1430

2250

*
No reaction up to melting point.

18

¥
No fast reaction up to melting point.

in sodium were continued., The remaining 36 thermocouples, contained in
eighteen 1/L4-in.-dia Inconel sheaths, have now accumulated a total of
19,000 hr of operation. Most of the time the units are held at 1500°F,
but the temperature is reduced to 1300 and to llOOOF periodically for data
taking., Two of the original 38 thermocouples failed because of a leak
at the Heliarc weld on the Inconel sheath. The accuracy of the 36
thermocouples still in operation remains within the manufacturers specified
tolerance of +11°F at 1500°F,

Twenty-eight Chromel-Alumel thermocouples contained in 14 Inconel
sheaths with Coast Metals alloy brazed end closures are also being
tested in sodium. These thermocouples have accumulated approximately
7000 hr of operation. The last time the data were processed, it was
found that eight thermocouples in four sheaths had failed. The failures
were indicated by very low or zero output signals at lSOOOF. These
thermocouples have not yet been removed for examination.

The data plotted thus far show that all these thermocouples sharply
declined in output voltage during the first 1800 hr of operation. After
the initial 1800 hr, the output voltage became stable.

82

3.4, APPLIED MECHANICS

Basic Problems in Elasticity

Studies have been carried out on the elastic behavior of cylindrical
shellsl’2 and of tapered circular plates.5 In these studies, equations
for calculating stresses, displacements, and rotations were developed,
as well as tables of numerical values for all the functions,

Cylindrical shells can now be analyzed for any combination of
axisymmetrical edge forces and displacements in addition to axial loads
and uniform pressure. The calculational method was extended to include
the behavior of shells subjected to thermal loadings. The thermal
loadings can be in the form of temperature variations in the axial
direction and linear temperature variations through the thickness. The
various loads described can act singly or in combination. The equations
are applicable to shells where the behavior at one end is dependent upon
the conditions at the other end, as well as to shells where the two ends
behave independently.

The results of the tapered-plate studies can be used to examine a
plate with or without a central hole. Like the cylindrical shell, the
complete analysis of a plate with uniform pressure, a membrane force and

any combination of axisymmetrical edge loadings can be made,

Creep Buckling of Snells with Double Curvature

Experimental studies are being carried out at the Syracuse University

Research Institute to determine the creep~buckling behavior of copper

lF. J. Stanek, Stress Analysis of Cylindrical Shells, ORNL CF 58-9-2
(to be published).

2F. J. Witt, Thermal Stress Analysis of Cylindrical Shells,
ORNL CF 59-1-33 (Jan. 27, 1959).

3]F. J. Stanek, Analysis of Tapered Circular Plates, ORNL CF 58-7-9
(March 18, 1959).

hemispherical shells under external pressure loadings at temperatures
of 300 to 1+OOOF.LL Shells 10 to 12 5/8 in. in diameter have been tested;
the radius-to-thickness ratios ranged from 167 to 395.

The results of the test program have shown that shells with the
lower radius-to-thickness ratios tend to collapse much more symmetrically
than do those with the higher ratios. In cases where shells collapse
in highly unsymmetrical shapes, the buckling starts at reglions near the
supported edge because of localized bending stresses. There is some
indication that these bending stresses are reduced by creep effects
until they no longer influence the buckled shape. The evidence for this
lies in the fact that the mode of collapse accompanying long-time tests
tends to be symmetrical.

Some of the data obtained are plotted in Figs. 3.4.1, 3.4.2, and
3,4.3 where the critical pressure is shown as a function of time. The
numbers which appear beside the data points correspond to the numbers
of the collapsed shells, some of which are shown in Fig. 3.4.4. It
should be noted in Figs. 3.4.1, 3.4.2, and 3.4.3 that the initial portion
of the abscissa has a linear time scale, while the remainder has a
logarithmic one. The data of Fig. 3.4.1 show the effect of thickness
upon the critical pressure for 10~in.-diameter shells tested at 5OOOF.
Figure 3.4.2 gives similar results for 10-in.-diameter shells tested at
ACOOF, and Fig. %.4.3 indicates the small effect of temperature upon the
critical pressure for 12 5/8-in,-dia, 0.016-in.-thick shells.,

In the case of shell 13 (Fig. 3.4.1), for example, two points,
connected by an arrovw, are given. The lower point indicates the sustained
pressure loading and time, while the higher point indicates the instan-
taneous buckling pressure at the end of the loading period. By subjecting
the shell to a pressure which causes instantaneous collapse after it has
been loaded for a period to time, a measure of the reduction in instan-

taneous buckling pressure or loss of strength is obtained.

hR. V. Meghreblian, ANP Quar. Prog. Rep. Dec. 31, 1956, ORNL-2221,

p L.

84

UNCLASSIFIED
ORNL-LR-DWG 37263

o]
Q

)]
@]

CRITICAL PRESSURE (psi}

B
=}

Characteristics

[TTIm T TP T T 170 H'”
- . SHELL THICKNESS, 0015 in.
* SHELL THICKNESS. 0.019in.
- SHELL THICKNE SS,0.023in. _ Jm
g 150D
2511 L 3 I3
il s
el Rl [T
f 10 102 10> 10?
TIME (hr)

Results of Tests of Creep-Buckling
of 10-in.-dia Copper Shells at

UNCLASSIFIED
UNCLASSIFIED
200 | LR-DWG 37264 o ORNL-LR-DWG 37265
*SHELL THICKNESS, 0.019 in. B [T T T T TTIAm

=180 =SHELL THIGKNESS, C.030i i _ « TEST TEMPERATURE,300°F
& A ; i 50 « TEST TEMPE{RATURE,400°F
ul —

{160
2 ] i 40 il
) =) 17
w 140 34 o H % 21 20
4 _ L 30 {231+ 24 26
2 120 ;; & 128
S | = 20 hg—
=100 : A S 19
& e
&3 = 10

80 U4 5

60 : W © 2 3 4

GEG 103 0 050 { 10 {0 0 10
TIME (hr)
Fig. 3.4.2. Results of Tests of Creep-Buckling Fig. 3.4.3. Results of Tests of Creep-Buckling
Characteristics of 10-in.-dia Copper Shells at Characteristics of 125/8-In.-dia, 0.016-1n.-Thick
400°F. Copper Shells at 300 and 400°F.

85
UNCLASSIFIED
ORNL-LR-DWG 37408

(¢) SHELLS 10 in. IN DIAMETER, 0.023-in. WALL,

o
(g) SHELLS 10 in. IN DIAMETER, 0.015-in. WALL, (b) SHELLS 10 in. IN DIAMETER, 0.019-in. WALL, HEERR AU GRERD

TESTED AT 300°F TESTED AT 300°F

(d) SHELL 10 in. IN DIAMETER, 0.049-in. WALL,
TESTED AT 400°F

(e) SHELLS 42%g in. IN DIAMETER, 0.046-in. WALL, (f) SHELLS 12%g in. IN DIAMETER, 0.046-in. WALL,
TESTED AT 300%F TESTED AT 400°F

Fig. 3.4.4. Typical Collapsed Specimens. Numbers on data points in Figs. 3.4.1 — 3.4.3 refer to these specimens.

It is interesting to note that all three shells subjected to this

sort of test cycle showed relatively little reduction in the critical
pressure for instantaneous collapse. This is not surprising, for,
wherever asymmetries begin to develop as a result of creep, the local
stresses tend to increase rapidly and failure must proceed from the
incipient condition to complete collapse in a relatively short time.-
Thus it is quite uniikely that any shell that has been subjected to
extended endurance testing without failure will be on the verge of
failure at the moment of test interruption.

While at first glance the data appear to scatter widely, this sort
of scatter is characteristic of structures subjected to conditions under
which failure may occur through elastic instability. The data do show
that creep over an extended test period tends to cause shell collapse

at only 60 or 70% of the usual instantaneous buckling pressure.

87
3.5. ADVANCED POWER PLANT DESIGN

Vortex Reactor Experiments

Studies of the dynamics of vortical flow of gases in tubes for
both 100% radial throughput and fractional bleedoff (axial and radial)
have been continued. As discussed previously,l it was concluded that by
increasing the fraction of gas bled off axially at a position radially
removed from the tube centerline, a significant improvement in vortex
strength could be achieved with allowable exit mass flow rates.
Although considerable doubt existed as to whether this technique alone
would provide sufficient improvement to make possible a practical
vortex reactor, a series of experiments was undertaken to ascertain the
performance characteristics of a 2-in. tube with provisions for bleed-
off. Specifically, the separation of a mixture of helium and CBFl6
(perfluorodimethylcyclohexane) was determined for comparison with the
separation predicted by the theoretical analysis.2

Failure to achieve higher vortex strengths is believed to be due
to the existence of turbulence, particularly insofar as this results in
high shear stresses at the wall and between fluid layers. This situa-
tion is aggravated by turbulent injection of gas through the discrete
entry nozzles at high Mach numbers, which may, in part, explain the
failure to achieve boundary-layer stabilization by uniform wall suction.
In some experiments therefore a large number of discrete nozzles have
been utilized to provide laminar or near-laminar injection of helium as
a preliminary step toward an experiment involving laminar slit

injection with uniform wall suction.

1

p 101l.

2
J. L. Kerrebrock and J. J. Keyes, A Preliminary Experimental Study

of Vortex Tubes for Gas-Phase Fission Heating, ORNL-2660, p 35
(February 6, 1959),

J. J. Keyes, ANP Semiann. Prog. Rep. Sept. 30, 1958, ORNL=2599,

88

Separation Experiments

The theoretical expression relating the tangential Mach number at
the point of maximum mole fraction, Mm, to the exit mass flow rate of
light gas per unit of tube length,mnl, the mass ratio of heavy to light
gas,vmejnl, and the binary molecular diffusivity CDDlg)m at the point of
maximum mole fraction is, for laminar conditions,

- ! (1)

T oy (pnle)m(”j_e_ ) 1)
flfi

where y is the ratio cp/cv for the light gas.

It is to be noted that Mm for a given gas pair varies directly as
the square root of the mass flow and inversely as the square root of the
diffusivity. 1In view of the small magnitude of the term (PDlE)m at room
temperature, Eq. 1 can be satisfied for laboratory-feasible values of
Mm only if;nl also assumes low values. ,

The earlier separation experiments were qualitative and did not
yield very satisfactory results. Thus, in an experiment with helium and
bromine gas in which an annular separation zone was observed near the
center of a 2-in. plastic tube with 100% throughput, condensation effects
made interpretation difficult. It was felt at that time, although not
verified, that sufficiently high velocities were generated near the tube
center to satisfy Eq. 1. Later probe measurements of the local mole
fraction of C8F16 in helium also indicated an increase in the heavy
component near the tube center.

In order to increase the radius at which the CSFl6 mole fraction
peak will occur (so as to effect quantitative probe explorations), a
large bleedoff was employed. This made possible an increase in Mm
without a corresponding increase in the exit mass flow and thus
satisfied Eg. 1 at larger radii than were possible with 100% throughput.
The gas-sampling probe was a 0.008-in.=-0D tube introduced radially through

the tube wall; the gas was analyzed by thermal conductivity measurements.

89

The concentration of CBFl6 (molecular weight = 400) was kept below
100 ppm in order to minimize the possibility of condensstion in the
cold central region of the vortex.

Three typical separation profiles, which present only smoothed
data for different amounts and types of bleed, are shown in Fig. 3.5.1.
The lower dashed curve is for an axial bleed ratio of 17.5 (17.5 parts
by weight of gas bled off per part by weight of radial throughput) at a
radius, r'(= r/ro), of 0.4. There is a weak peak at a radius of
about 0.19. Moving the axial bleed closer to the tube center (solid
line) increases the peak strength but also moves it to a smaller radius
(0.15). In both cases the heavy trace component was introduced
uniformly through the porous tube wall. The third curve (upper dashed
line) was obtained with radial bleedoff through a 0.0l7-in. slit at
the tube wall. The bleed ratio was 9.L. The second broad peak, which
appears at r' ¥ 0.65, is questionable because the data were scattered
in this region, but the dropoff at the outer wall is distinctly present.
In this case the trace component was introduced through 3 of the 12
nozzles, and the possibility of droplet formation in the nozzle
effluent may explain the behavior of the data near the wall.

Tt is informative to compare the theoretical Mach number required
for a separation peak to occur (calculated from Eq. 1) with the experi-
mental Mach numbers existing in the tube at the radius of peak mole
fraction. The latter are obtained by calculation from the static pres-
sure distribution measured during operation under conditions identical
to those of the separation runs, including the effect of insertion of
the sampling probe. The data are summarized in Table 3.5.1 for five
typical runs, three being those depicted in Fig. 3.5.1l. In establishing
this comparison, an experimental measurement was made of the molecular

diffusivity, D n of C8F16 in helium. A preliminary value of D _ =

1 12
0.247 + 0.037 cm /sec at 80°F and 1 atmosphere was obtained, which was
1.75

then corrected for temperature by assuming a T variation and

isentropic expansion from the wall toward the tube center.

90

Table 3.5.1. Summary of Data from Some Helium-C,F Separation
Experiments in a 2-in.-dia Metal VorteX Tubes

Test Number

Test Parameter

Inlet mass Tlow, 7.

(1b/sec ft) L 0.028  0.030  0.023  0.013  0.030
Bleed ratio 11.3 17.5 15.6 4.8 9.l
Exit diameter (in.) 0.250 0.250 0.221 0.221 0.250
Bleed~off position 0.4 O.h 0.25, 0.k Wall

0.4 slit
Wall pressure (psia) 55 .6 53.0 56.3 40.8 40.8
Observed peak position, r' 0.15 0.19 0.15 0.16 0.10
Observed M _ 0.64 0.55 0.70 0.48 0.70
Calculated M_ (Eq. 1) 1.0 0.84 0.81 1.0 1.41

Ratio of observed Mm to
calculated M 0.64 0.68 0.86 0.48 0.50

Ratio of effective diffu-
sivity to molecular
diffusivity 2l 2.2 1.4 L3 4.0

The significant conclusion to be drawn from Table 3.5.1 is that the
observed tangential Mach number at the radius of peak formation is in
all cases less than that calculated from Eq. 1 for laminar flow. The
most plausible explanation for this discrepancy is that some degree of
turbulence must exist that results in an effective diffusivity, D*, which

is greater than the molecular diffusivity, D The last column in the

]_2.
table indicates that an effective diffusivity of from 1.4 to L.3 times
the molecular diffusivity will account for the observations; the degree
of turbulence required to cause these apparent diffusivities is quite

low as compared with fully developed turbulence in which the apparent

diffusivity might be of the order of 102 times the molecular diffusivity.

Laminar Injection

The effect of laminar, or near-laminar, injection is illustrated -
in Fig. 3.5.2. The ratio Mi':0.26/Mj may be considered to be the
effectiveness of :tilization of the Jjet velocity, Mj’ in generating the .
local velocity, M . The local velocity is determined by interpolation
from a two-point slope of the pressure profile and is at best an
approximation for comparison purposes only. It may be seen that for a
given mass flow, laminar injection (NRej = 2000) results in higher
Mach ratios than does turbulent injection; thus, more effective
utilization can be made of a laminar jet. The dashed curve for
nitrogen was obtained under conditions of inereasing jet turbulence and
is seen to fall away sharply from the laminar helium curve a‘(:“m.l
>0.001 1b/sec.ft.
The behavior of the ratio M*/Mj as Mj increases is an important
additional consideration which will determine the over-all effectiveness
of laminar injection. There is evidence that as Mj-—?l.o, M*/Mj
decreases for turbulent conditions; data for laminar entry conditions
are not available. A model is being fabricated, however, which will
make possible injection through a variable-width slit, and it is hoped
that data from this modified apparatus will be useful in further
evaluating laminar injection both with and without boundary-layer

suction.

Satellite Power Plant

Design studies of auxiliary power units for satellites have been
continued, with emphasis primarily on a comparison of over-all perfor-
mance and weight of generalized systems making use of nine different
working fluids ~ hydrogen, helium, air, aluminum chloride, water,
mercury, rubidium, potassium, and sodium. Although it is not feasible
to establish exact values for the weights of the various components, it
is possible to establish good relative values by making the analyses on -
the basis of a consistent set of design precepts. The work has not yet

been completed, but important results have emerged that are discussed v

o2

UNCLASSIFIED
ORNL-LR-DWG 37266

170 . ,
' : ‘ T T :
AXIAL BLEEDOFF AT /'=0.25 ANDO40D ‘
| m, (INLET) = 0023 Ib /sec-ft ‘ |
3 BLEED RATIO=15 ‘
3 1'50 M — 57 i o ‘ .. - —.‘:.—'-\ . o
/ N Pwa T 01 psia -~ -
i ; ! - \\
/ ~ - ~
Ll_i N ‘\h.._._ /' ~
O 430 | =" RADIAL BLEEDOFF, WALL SLIT \‘ }
2 i \ 74 (INLET)=0.030 Ib/sec-ft AN
~ — — hY
c : TN BLEED_RATIO'— 9.4 \
£ ot | yd AN Py = 31 psic N
T | |17 :‘ N 1 | N
."U_' ! i ] \\\ \‘\—JF I~ ]
o E’ ~ | e — ; —
E I / : T m——— ! |
= 090 [ i I - AXIAL BLEEDOFF AT /=040 .. [ L
w i m, {INLET}=0.030 Ib/sec-ft :
b i | BLEED RATIO=47.5
S R 4n PuaLL= 53 psia.
= ‘ |
= h i |
! ‘ : ;
050 | | E |
0 040 020 0.30 0.40 050 060 070 0.80 0.90 100

7" RADIAL POSITION

Fig. 3.5.1. Separation Profiles for Hellurn-CaF.“5 Mixtures in Vortical Flow
in a 2-in.-dia Tube with Bleedoff. (Gupge® with caption)

UNCLASSIFIED
ORNL-LR-DWG 37267
T T

0.9 ; T
| 1
OB 2000 ) ’
N !
0.7 **’J’"" L g 7t HELIUM 14,000
Mo, =2000/ / |
06 p /(©80C0 //__ SRR R
g / / L i
! Y/ . ]
X Y & TUBE PRESSURE = 27 psia
- Hey o TUBE PRESSURE = 54 psia
18.000 =g ® TUBE PRESSURE = 63 psia -- .- -- '
, _;,—Ar' | © N, 4 TUBE PRESSURE =54 psia | "
- . 1 : . [ ]
=40000.. ... | i . Ll . 4 95000
‘ NITROGEN 7 1 |
. | L
0 | |
0 0.006 0.042 0.018 0.024 0.030

M, MASS FLOW RATE (Ib/secft)

Fig. 3.5.2. Effect of Jet Reynolds Modulus on Variation of Local-to-Jet
Mach Number Ratio with Mass Floew Rate In a l-in.-dia Plastic Tube. T Y
with caption)

93

here. The effects of the cycle working fluid and cperating conditions
on beth the radiator specific weight and on the turbine specific weight
are described, and a preliminary layocut for a system employing rubidium

vapor as the working fluid is presented.

Radiator Design Considerations

The radiator and its design characteristics completely dominate the
choice of the thermodynamic cycle for a space-vehicle auxiliary power
plant designed to deliver 100 kw (electric), or more, because very sub-
stantial amounts of waste heat must be rejected, and this heat can be
re jected only by thermal radiation. The weight of such a radliator is
directly proportional to the thickness of the radiator tube walls.
Experience at ORNL with the construction of heat exchangers has led to
the conclusion that the design of* heat exchangers of this type must be
predicated on the use of material with wall thicknesses of at least
0.030 in. if a high degree of reliability is to be obtained. It is
important to note that design studies intended to establish an
"optimum" power plant are likely to lead to deceptive results if thinner
walls are presumed for the radiator. Some notion of the size of the
radiator required if a low-temperature system is used is given by
examining the radiator requirements for a conventional steam system
designed to produce 100 kw (electric). Such a system might have an
over-all efficiency of as much as 25% if high steam pressures and
temperatures were used together with a radiator temperature of lOOOF.
The resulting condenser would have a surface area of 7T0O0O ftz, vhich
would be equivalent to the surface of a cylinder 24 ft in diameter and
100 ft long. Such a condenser would clearly be difficult tc include
in a space vehicle, particularly if stringent vacuum tightness were
required.

The;e géems to be general agreement that rather high radiator tem-
peratures must be employed to obtaln acceptable system weights. This
in turn leads to the conclusion that the radiators must be made of an
iron-chrome-nickel alloy, such as a stainless steel, or some refractory

metal, such as molybdenum or columbium, which, while stronger at high

ok

temperatures, would have a higher density. Unfortunately, the techniques
of fabrication are not yet sufficiently advanced to provide large, thin-
walled, vacuum-tight structures from mavcerials other than the iron-
chrome~nickel alloys. Therefore, stainless steel was considered to be
representative of the material of which the radiators might be
fabricated. The radiator weight required as a function of radiator
surface temperature is shown in Fig. 3.5.3, in which each line is for a
representative thermodynamic cycle efficiency.

A number of radiator geometries can be used. The most obvious
arrangement incorporates passages in the skin of a cylindrical body, for
example, the shell of the vehicle. Unfortunately this arrangement poses
several problems, perhaps the most important being the difficulty in
obtaining a perfectly leaktight structure. This problem is aggravated
by the tendency of cracks to form as a result of thermal stresses from
differential thermal expansion between adjacent tubes operating at some-
what different temperatures. Temperature differences are likely to stem
from irregular variations in the flow distribution and hence in the
temperature drop in the fluid stream passing through the tubes. Another
objection to this arrangement is that the temperature within the space
enclosed by the radiating surface would inevitably be that of the fluid
from wvhich the heat was being rejected. Thus there is a strong incentive
to make the heat sink a separate unit placed some distance away from the
main body of the vehicle. It then could be built of tubing to give a
reasonably reliable, leaktight system.

A promising radiator configuration has been evolved and is shown
in Fige. 3.5.4. It makes use of a thin sheet of aluminum or titanium as
a reflector in back of each tube., The reflector could yield an effective
view factor for the entire tube surface area which would be close to
100%. Handbook data indicate that in the red and infrared regions the
reflectivity of polished aluminum surfaces is of the order of 80 to
98%, with the higher values being for the longer wave-length radiation.

In considering various shapes of curved surface for such a reflector,

it is not difficult to show that the optimum surface is an involute curve

(see Fig., 3.5.5). A line drawn tangent to the hot tube at any point such
as B will be normal to the involute curve at the point of intersection.
Any radiant ray from any portion of the tube surface between the tangent
BO and point A at the rear of the tube must impinge on the involute
reflector surface in such a way as to be reflected away from the tube.

In some instances, as is the case with the ray CO, the reflected radia-
tion strikes the involute again at a point farther out, but it is then
reflected on into space.

From the vehicle design standpoint the optimum configuration for a
radiator of this type is probably cylindrical (or conical), with the
tube and reflector arranged as in Fig. 3.5.4. This configuration makes
it possible to obtain the same effective radiating surface area with a
collection of small-diameter parallel tubes as with a large cylinder
having the same diameter as the envelope of the tube array; hence, no
more effective utilization of a cylindrical volume can be achieved than
with this array.

The radiator weight for a series of representative thermodynamic
cycles is given in Fig. 3.5.6; allowances were made for the effects of
radiator outlet temperature on the over-all thermodynamic efficiency of
the cycle. The peak temperatures in the cycles were kept consistent from
one cycle to another by considering the creep stress vs temperature curve
for type 316 stainless steel as typical of that for suitable materials
of construction, and by limiting the cycle peak pressure and temperature
to maintain a constant thickness-to-diameter ratic in the high-'
temperature piping. As might be expected, increasing the turbine outlet
temperature from a low temperature level would result in an initial
reduction in radiator specific weight per kilowatt of net electrical
output, but as the temperature increased further, a point would be
reached where reductions in thermodynamic cycle efficiency would more
than offset the effects on radiator weight of the increasing temperature.
The background lines running diagonally across Fig. 3.5.6 are identical
with the lines of Fig. 3.5.3. It is immediately evident from Fig. 3.5.6
that rubidium is the most promising working fluid for a thermodynamic

cycle.

96

UNCLASSIFIED
LR & ORNL-LR-DWG 37268
1000 T - T

|
N
500 N—

>

OVER-ALL THERMAL
N\ | EFFICENCY
200 Nl 5%

N\ KL o
100 N UNCLASSIFIED

NN (5% ORNL-LR-DWG 37269
AY

ol
STAINLESS STEEL
/ TUBES

P

RADIATOR WEIGHT (|b/kw OF ELECTRICAL QUTPUT)
- N
o (o]
| 7
Ut~
7
,/

] INVOLUTE PROFILE

REFLECTOR \\

//

20% =/\ N l\\‘\\ \\
5 " OO
30% 1 INAN MR\ N
2 40% RN ARNAN
HEO\\\\
1 50 /o \\ X‘\ ) \\ \\ \\
NS N
05 AAVARTAEAY
N \\ Y \\\\\\
N
0z AN \Q\
0.4 \\& \ A

500 700 4000 1500 2000
TEMPERATURE [°R)

Fig. 3.5.3. Radiator Dry Weight as a Function
of Radiator Surface Temperature for Various Fig. 3.5.4. Section Through a Radiator Fitted
Thermodynamic Cycle Efficiences Assuming a with a Reflector to Give a View Factor Approaching
0.030-in.-Thick Tube Wall. 100%.

UNCLASSIFIED
ORNL-LR-DWG 37270

INVOLUTE REFLECTOR

Fig. 3.5.5. Schematic Diagram of Involute Re-
flector.

97

ORNL-! !- DWG 372714

NN

= [ N
> . j=— WATER 7 1
a NN
= N\ T ]
=] 5 . \ N \ . __ S
3 \\ \ ]
) i \
T ! \ ALUMINUM CHLORIDE WITH
[FRREP R B ALLOWANCE FOR FILM DPOP | ___|
o | ALUMINUM CHLORIDE WITHOUT
. MERCURY ALLOWANCE FOR FILM DROP
‘lo 1 T
b4 i +
= ] b
P I R
TR B
5 o
o Af}tf RUB —
= i
a 2 t —
o
g
2
[ned 1 —
5
2 | \\
-1
10
400 600 800 1000 2000 4000 6000

MEAN EFFECTIVE RADIATOR TEMPERATURE (°R)

Fig. 3.5.6. Radiator Specific Weight os a Function of Meon Effective Radi-
ator Temperature for Typical Thermodynamic Cycles Corried Out in Systems

Designed to be Built of Type 316 Stainless Steel. NSNS . ith caption)

It may also be seen in Fig. 3.5.6 that aluminum chloride, which
had at first looked quite promising as a working fluid for the
thermodynamic cycle for a satellite power plant, has an important
disadvantage. A high film drop occurs in the radiator because of the
low heat transfer coefficient inherent in the low Reynolds number that
characterizes flow in the radiator for this thermodynamic cycle. A
second and equally serious consideration is the high melting point of
the aluminum chloride. While the high melting point (17300) would not
be a handicap under operating conditions, it would pose serious
startup problems because the radiator would inherently tend to be at
60 to lOOOF in the zero load condition. Solid deposits would form
which might block off the tubes or might form a film over the inner
surface of the radiator and absorb the available aluminum chloride
inventory. It is possible that sufficient aluminum chloride inventory
might be provided to build up a film of sufficient thickness so that
the temperature drop through the film would prevent the film thickness
from becoming excessive. Whether this would occur before complete

tube blockage took place could be determined only by experiment.

Turbine Design Considerations

The specific weight of a turbine in pounds per kilowatt of useful
power output depends on many factors. The weight of the turbine will
depend on the rotor diameter, the number of stages, the allowable
stress in the turbine casing, and the internal pressure which the
casing must withstand. The power obtainable from a turbine will
depend on its diameter, the number of stages, the working fluid
enthalpy drop per stage, the allowable tip speed, and the density of
the fluid flowing through the turbine. Each of these and other

pertinent factors have been examined carefully, and an expression for

turbine specific weight has been derived to indicate the relative turbine

specific weights for operation on different types of cycle with different
working fluids. The equation is
€ [ 1/2

Turbine specific weight = Sk <fi) ?
where the constant K5 is characteristic of the degree of detailed
design refinement of the turbine, S is the allowable stress in the
casing material at operating temperature, k is the ratio of the specific
heats for the working fluid at the outlet stage, m is the molecular
weight of the working fluid, and T is the absolute temperature at the
turbine outlet. This expression includes the effects of density, sonic
velocity limitations, the allowable stress in the casing, and the
effects of diameters up to diameters of the order of 4 ft. A
statistical analysis of data for aircraft gas turbines working on air
has yielded results that are consistent with the above expression. A
few scattered data for steam, mercury, and rubidium turbine designs
also seem to be consistent with the above formula. The derivation of
this formula is being included in detail in a report now being
prepared.3 Briefly, preliminary results indicate that the turbine
weight is a relatively small part of the total power plant weight,
except possibly for cycles employing aluminum chloride, helium, and

hydrogen.

Choice of Cycle

In choosing a cycle from those that meet the weight requirements,
perhaps the most important consideration is the simplicity and over-all
reliability of the system. In fact, it should be kept in mind that
this consideration is of even greater importance than weight in

choosing a thermodynamic cycle.

3A. P. Fraas, A Comparative Study of Auxiliary Power Sources for
Space Vehicles, ORNL-2718 (to be published).

100

It is essential in the design of any power plant that the materials
used be compatible. Much information is available on the compatibility
of sodium and NaK with stainless steels, and the limited data avai?able
indicate that rubidium should behave relative to the stainless stetls
in essentially the same way as sodium or Nak. Similarly, either
molybdenum or columbium should be compatible with rubidium if used as
structural materials. When fabrication techniques have been developed
to the point where a system can be built of molybdenum or columbium,
the rubidium cycle peak temperature could be increased to improve the
system performance. As the allowable temperature of operation of such
a system was increased, a point would be reached where the rubidium
vapor pressure would be so high that it would prove desirable to shift
from rubidium to potassium or sodium as the cycle working fluid.
Potassium would be particularly attractive from the activation stand-
point, since it should lead to induced activities in the fluid external
to the reactor that would be about 1/20 of the activity that would be

associated with sodium or rubidium.

Rubidium=Vapor-Cycle Power Plant

A preliminary rubidium-vapor-cycle power plant layout has been
prepared and is presented in Fig. 3.5.7 to give some idea of the
proportions of a system employing rubidium or potassium as a working
fluid. Pertinent system data are summarized in Table 3.5.2. The
rubldium serves as both the reactor coclant and as the thermodynamic-
cycle working fluid. The pressure envelope for the rubidium system
is relatively simple in mechanical design and should lend itself
readily to fabrication techniques that will assure good vacuum-
tightness.

The configuration was carefully worked out to facilitate filling
of the system prior to takeoff. While rubidium has a melting point of
lOEOF, the addition of around 1.5% sodium would reduce the melting
point to about lSOF. Thus the system could be readily filled when
mounted inéthe position it would logically occupy in the rocket with

the reacto% at the top and the radiator suspended vertically beneath it.

<01

ORNL-LR—DWG 37272

CONDENSER

5/ -in~00

N

TURBINE t

y
x —3/3-in=0D FEED LINE
B

VAPOR TO TURBINE —= _
[ CORE \
Yg-in.-OD RETURN LINE

NOTE: INSULATION OF VAPOR SEPARATOR, REACTOR L VAPOR SEPARATOR
SPHERE AND CORE TO SEPARATOR LINES. Rb FEED PUMP

FLOW DIAGRAM

VAPOR SEPARATOR

GENERATOR

CONDENSER
— a—f‘L JET PUMP

e ——— —_—

REACTOR CONTROL

i
- 12H 10in. OVER-ALL -

Fig. 3.5.7. Preliminary Design Showing a Promising Layout for a 20-kw Electrical Output Rubidium-Vapor Power Plant for Satellite Applications.

PSR v ith caption)
Table 3.5.2.

Summary of Estimated Size, Performance, and Weight Data

for Proposed Rubidium-~Vapor~Cycle Power Plant

Cycle Data

Generator power output

Reactor power output

Over-all thermal efficiency

Turbine efficiency

Generator efficiency

Rubidium temperature at turbine inlet
Rutidium pressure at turbine inlet
Rubidium pressure at turbine outlet
Rubidium tempersture at turbine outlet
Rubidium temperature into feed pump

Rubidium flow rate

Reactor Data

Reactor type

Core shape

Core diameter

Core height

Reflector shape

Fuel elements

Pressure drop in core circuit
Volume of core circuit

Volume of expansion tank—vapor serarator

Radiator

Surface area

Average surface heat {lux
Tube inlet inside diameter
Tube outlet inside diameter
Tube wall thickness

Total volume enclosed by tubes

20 kw (electrical)
100 kw (electrical)
189,

80%

90%

1600°F

66 psia

2.6 psia

1000°F

990°F

0.234 1b/sec

Boiling rubidium

Right circular cylinder
8 in.

10 in.

stherical

UO2 facing on BeO

> psi
0.25 ft
0.3 £t3

3

L5 £t°

6000 Btu/hr.ft°
0.94 in.
0.25 in.
0.32 in.
0.85 £t

103
Table 3.5.2. (Continued)

Total flow passage area at tube inlet end
Total flow passage area at tube outlet end
Mean vapor velocity at inlet

Mean liquid velocity at outlet

Sump volume

Turbine
Length
Diameter
Speed
Number of stages

Shield
Reactor-to=crew separation distance
Crew shield
Crew compartment dose
Reactor shield thickness

Reactor shield composition (by volume)

Component Weights

Reactor assembly

Shield assembly

Radiator assembly

Turbine and feed pump assembly

Generator

Expansion tank

Rubidium inventory

Connecting pipes, support structure, etec,
Total weight
Specific weight

0.1L45 ft°
>

0.0103 ft
100 ft/sec
0.3 ft/sec
0.03 £t3

50 ft
12 in. HéO

10 mr/hr (~100 rem/yr)
LO in,

20% Pb, 80% LiH

230 1b
680 1b

7L 1b

10 1b

43 1b

10 1b

38 1b

60 1b

1145 1b

57.3 1b/kw (electrical)

104
The expansion tank-—wvapor separator would be the highest point in the
system, and filling could be carried out with a minimum of difficulty.
There should be essentially no bubble formation in the lines during the
filling operation, and incomplete filling of loop segments or cavities
should not be a problem. After filling the system under helium, the
voids in the radiator, the expansion tank, and the turbine could be
evacuated and the system sealed. With a vacuum-tight system of this
sort, no component comparable to an air ejector in a feedwater heater
would be required, and aeration in the fluid passages leading to the
feed pump should not present a problem.

Provisions have been made for operation under conditions of
apparent weightlessness. In the design of Fig. 3.5.7 the fluid
leaving the reactor is directed tangentially into the expansion tank in
such a way as to induce a rather high swirl velocity. The resulting
centrifugal force at the liquid surface should be much greater than
the force imposed by gravity in a static power plant at the earth's
surface, This implies that the vapor-separating features of the power
plant can be tested in the laboratory under conditicns which would not
be seriously different from those that would prevail in orbit.

Provisions have been made for stabilization of the free liquid
surface in the condenser through the use of tapered tubes into which the
vapor is introduced at a moderately high velocity. As condensation took
place along the walls, the droplets would increase in size and
agglomerate until the drag forces of the vapor moving through the
tubes would sweep them toward the outlet. As they moved along they
would pick up other droplets and "snowball" to progressively larger
sizes as they moved toward the outlet. This would produce an irregular
"slugging" type of ligquid flow through the condenser tubes and give a
substantial liquid holdup in the condenser. Any vapor bubbles trapped
in the liquid toward the radiator outlet should be cooled and absorbed
by the liquid before they reach the outlet. This implies that the
condenser would have to be overdesigned so that the lower end would

always be filled with liquid. This would lead to a reduction in

105
over-all thermal efficiency at part load because the liquid in that
portion of the radiatocr would be subcooled substantially below the
condensation temperature., However, part-load efficiency would not be
an important consideration for this application,

Tests of this type of radiator could be carried out in the
laboratory with the tubes lying in a horizontal position or, possibly,
inclined in such a way that there would be a small rise between the
tube inlet and the tube outlet. If satisfactory fluid circulation were
obtained under these conditions in the laboratory, it seems likely
that operation would be satisfactory in orbit. A major objective of
laboratory tests would be the determination of operating performance as
a function of the vapor flow rate to the tube. It should be expected
that, at some high level of throughput, excegsive quantities of bubbles
would be found in the fluid at the outlet.

It would probably be desirable to start the system up before
takeoff, since it would be difficult to initiate rubidium circulation
in orbit unless a small sustained acceleration could be carried out to
stabilize the liquid surfaces.

Acceleration under takeoff conditions imposes a special set of
conditions. It may be necessary to accommodate accelerations of as much
as 5 g during launching conditions. The layout of Fig. 3.5.7 has been
designed to accommodate these accelerations insofar as operation of the
fluid circuit is concerned. The vibration commonly experienced during
launching and acceleration would cause severe structural stresses in the
radiator. These could be relieved if the entire radiator could be
immersed in a tank of kerosene which would not be burned until the
last phase of the powered portion of the climb into orbit.

A Jjet type of pump has been provided in the sump at the radiator
outlet to return the liquid rubidium to the feed pump inlet. This type
of pump was used, in part, to eliminate the moving parts of a separate
mechanical pump at the outboard end of the radiator. It was also felt
that this type of pump would be insensitive to damage from bubbles and

would serve to increase the pressure sufficiently so that any bubbles

106

in the fluid stream would collapse before they reached the main feed
punp. In view of the widely successful application of pumps of this
type in deep water wells, it is believed that it should give a relisble
arrangement. A similar Jet pump has been provided to circulate the
rubidium through the reactor core.

The generator cooling would be carried out by directing the
rubidium stream from the feed pump through the generator before sending
it to the aspirator type of pump employed for circulating the rubidium
through the reactor core. Similarly, a tube coiled in the interface
between the beryllium reflector and the beryllium oxide surrounding the
fuel would carry off the heat generated in the beryllium reflector.
Heat generated in the lead and lithium hydride of the shield would be

dissipated by thermal radiation to space.

Ma jor Development Problems

A number of major problems must be confronted before any extensive
design work can be carried out on a rubidium vapor power plant. First,
good data on the thermodynamic properties of rubidium should be obtained.
Calculations made to date have been carried out by interpolating and
extrapolating from a small amount of data obtained by an ORSORT group
during the summer of 1953. The data are meager and have never been
checked experimentally. A considerably more extensive range of vapor
pressures and temperatures should be covered, and data should also
be obtained in the superheated-vapor range.

Few data are available on heat transfer to boiling and condensing
liquid metals. Rough estimates of heat transfer coefficients have been
made for design purposes, but a much firmer basis for design is essen-
tial. It is believed that this information could be obtained from a
small heat transfer test rig that would include both a heating section
in which boiling heat transfer coefficients could be measured and a
condenser in which condensing coefficients could be measured. It might
be possible to include in this rig provisions for operating a condensing
leg in a horizontal position to investigate the feasibility of depending

on a relatively high vapor velocity to induce flow through the condenser.

PART L.

SHIELDING
4L.1. SHIELDING THEORY

Response Functions of Sodium Iodide Crystals

A calculation has been undertaken to determine functions character-
istic of the response of sodium iodide crystals to photons. This calcu-
lation will differ from previous calculations in that the effects of the
secondary bremsstrahlung and annihilation radiation will be treated in
detail. In addition, it will be possible to investigate the effects of
certain geometriceal variations from the typical cylinder, such as a
cylinder with one truncated conical end (see section of Chap. 4.3 on
"The Model IV Gamma-Ray Spectrometer") or a cylinder that has a hole
in it. The problem is presently being coded for the IBM-7O4 computing
machine.

A photon meking a collision in a sodium iodide crystal usually ejects
an electron or a pair of electrons, which, in turn, can make radiative
collisions that result in the emission of bremsstrahlung radiation. In
preparing the code the spectrum of this secondary radiation has been
computed by averaging the bremsstrahlung spectrum from an electron slowing
down in the sodium iodidel over the energy distribution of the electrons
ejected by a colliding photon. Typical spectra are shown in Fig. 4.1.1.
These spectra include the properly averaged contribution from electrons
produced by compton scattering, the photoelectric effect, and pair

production.

A Monte Carlo Code for Computing Fast-Neutron Dose Rates Inside

a Cylindrical Crew Com.partmen‘t2

A Monte Carlo code has been devised with which fast-neutron dose

rates inside a cylindrical crew compartment can be computed if the flux

lC. D. Zerby and H. S. Moran, ANP Quar. Prog. Rep. Dec. 31, 1957,
ORNL-24L0O, p 231.

2'I'his work was reported in greater detail in Monte Carlo Code for
Penetration of Crew Compartment, Report on ORNL Subcontract 931 by
Technical Research Group, New York.

111

AN

UNCIASSIFIED
2— 01— 059-404

10
10° \
FYANNN
W \\
>
© \\
; ! \*\Q\\-—:“\
G ] T~ QE\ £,= 200
g . \ "‘—-—._,_____. \: E' =15.0
x 10 \ N \Q- S
a S } e £,=100
-—.\

5 10"2 \\_‘_ E' = 5.0 \ \ %\\
o ,.—3 oy P —— ~ o
i;J 1o £,= ENERGY OF PHOTON MAKING iNITIAL COLLISION (mocz) 7:0 — \ \
< 2 1 ) =~
x k=ENERGY OF BREMSSTRAHLUNG PHOTON (moc ] £ =05 Fé%x\
8 1
207? | £,=0. T
=40 1 N \
w
@
o
R\
T

1078 \

w0’

Q0 [eX] 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
ENR/ED

Fig. 4.1.1. Avercge Spectra of Secondary Bremsstrahlung Photons Resulting from o Photon Making a Collision in Sedium lodide.
incident on the outside of the crew shield is specified. For these
calculations it can be assumed, to a very good approximation, that the
flux at the outside of the crew shield 1s the same as that which would
have resulted at the center of the crew compartment in the absence of the
crew shield. The distribution of alr-scattered fast-neutron fluxes from
point monodirectlonal, monoenergetic sources, with no crew shield present,
can be computed with a code recently developed at Convair.3 The present
code has been designed to accept the output of the Convair code as input
for the crew compartment calculation. Hence, by combining the two codes,
the air-scattered fast-neutron dose rates inside a cylindrical crew con-
partment may be obtained as a function of the initial energy and direction
of the source neutrons. The results can then be used to quickly deter-
mine total air-scattered fast-neutron dose rates inside a crew compartment
for an arbitrary neutron leakage distribution from a reactor shield.

The crew compartment penetration code, which has been designed for
the IBM-TO4 computer, is complete but has not been entirely '"debugged."
For this code it is assumed that the crew compartment cavity and the
outside of the crew shield are finitely cylindrical and coaxial. For
the time being the crew shield material must be considered to be either

water or CH,; however, the code could be adapted to other materials

whose crossesections are available.

The incident flux is assumed to be uniform over the surface of the
cylinder. Entry through the sides and ends can be considered either
separately or together by an initial option. The main output quantity
is the mean neutron dose rate in the cavity. The cavity can also be
subdivided into a number of coaxial cylindrical shells resting one in
another, all of the given cavity length, and the mean dose rate can be

found for each. Thus, a mean radial distribution of dose rate inside

M. B. Wells, "Monte Carlo Calculations of Fast Neutron Energy
Spectra,'" Fifth Semiannual ANP Shielding Information Meeting, May 1lL-15,
1958, C/25801.

the cavity can be found. Fractional variances are determined for the

over-all dose rate and for the dose rate in each region.

Comparison of LTSF Measurements of Thermal-Neutron Fluxes

in Water with Cornpone Reactor Code CalculationslL

One of the problems in shield design is the production of secondary
gamma rays by neutron interactions in the shield. In order to determine
the importance of these sources, it 1s necessary to obtain detailed
information concerning the spatial and energy distribution of the neutron
flux throughout the shield. Since an experimental determination of such
information would be very difficult and time consuming, a theoretical

p)

study making use of the Cornpone code,” a multigroup, multiregion reactor
code written for the Oracle, has been undertaken. In order to prove the
validity of the method, the results of the machine calculations for
representative configurations will first be compared with the results

of experimental measurements made at the ORNL Lid Tank Shielding
Facility.

Only the experimental and calculated thermal-neutron fluxes in a
plain-water shield have been compared thus far. The measured flux
reported is the "effective” thermal-neutron flux and is defined as the
integrated neutron number density in the energy region below about 0.300
ev times 2200 m/sec. This measured flux is determined by taking the
difference between measurements with bare and cadmium-covered gold foils.

The Cornpone program assumes that the reactor regions are symmetrical
about the origin, and fluxes along the centerline are calculated as a
function of distance from the origin. The neutron spectrum used with

the Cornpone program was divided into 31 energy groups ranging from

EThis study, which is being conducted by a staff member of the
Ordnance Tank-Automotive Command, Detroit Arsenal, Center Line,
Michigan, is largely supported by the U. S. Army.

5w. E. Kinney et al., Neutron Phys. Div. Ann. Prog. Rep. Sept. 1,
1958, ORNL-2609, p Sk.

114

thermal to 10 Mev, and cross sections from the Eyewash program were
used. As stated previously, the measured fluxes contain contributions
from all neutrons with energies less than approximately 0.3 ev. The
energy region from approximately 0.3 ev through thermal is covered by
groups 19 through 31 in the calculations; therefore, it is the sum of
the contributions from all these groups which should be compared with
the experimental results. It was observed, however, that group 31 (the
thermal group, E <0.02518 ev) alone gave approximately 97% of the total
contribution. Thus, since a very large amount of hand calculation was
involved in summing the groups, this was done for only one of the cases
vhich was investigated. In the remainder of the cases the contribution
from group 31 alone was compared with the experimental results.
Calculations were performed for the four configurations described
in Table 4.1.1. In all cases the slab source region was considered to

Table 4.1.1. Configurations Used in Calculations of
Thermal -Neutron Fluxes in Water

Thickness of Length and Width
Configuration Water Region of Configuration
(in.) (in.)
1 61.75 28
2 39.37 Infinite
5 39.37 23
L 39.37 2L .8

be made up of the same components as the LTSF source plate assembly and
to have the same thickness (2.75 in.). The four cases differed in the
thickness of the water region and in the width and length of the con-
figuration. Configuration 2, with infinite length and width, was
calculated to determine the upper limit of the flux curve. Case 3 was
calculated to determine the effect of reducing the water region

thickness from 61.75 in. to 39.37 in., and case 4 was chosen so as to

115
give a source area equal to that of the actual 28-in.-dia source in
the LTSF.

As may be seen from Fig. 4.1.2, the calculated curves for the
finite cases agree to within 10% with the experimental curve for the
first 32 cm from the source. The curve for case 1 is closer to the
experimental curve because of the larger area of the source. One reason
that the calculated curves differ from the experimental curve is that
the experimental data are accurate only to about *5%. The results for
case 3 are not included in the figure because they were the same as

for case 1.

Prediction of Thermel-Neutron Fluxes in the BSF from LTSF Data

In order to obtain the maximum usefulness from shielding data
collected at either the Lid Tank Shielding Facility (LTSF) or the Bulk
Shielding Facility (BSF), the power of the experimental source must be
accurately known. Furthermore, the data must be correctly converted
by geometrical transformations from the experimental source to the
reactor for which the shield is being designed. One method for checking
the powers quoted for the LTSF source (a disk-shaped uranium plate)
and the BSF reactor, as well as a method for checking on the validity
of the geometrical transformations, is to perform a calculation predicting
the neutron flux in the BSF on the basis of LTSF data transformed first
to a point-to-point kernel and then to the geometry of the BSF reactor.
A discrepancy between the predicted and the measured fluxes would indi-
cate either that one of the quoted powers was 1in error or that the
geometrical transformations were not properly derived.

A calculation6 that compared the fast-neutron doses was performed

in 1952. Since neither the BSF reactor power distribution nor the LTSF

GT. A. Welton and E. P. Blizard, Reactor Sci. Technol. 2, No. 2,

p 73 (1952).

116

UNCLASSIFIED

-LR-DWG R
103 ORNL-L DWl 35252R1
5
- L L
E S Bag
T 2 N
@ a
o
E 5 p
A —
f:’ 7 AN I
: e —,—— e —— . —_— —
E 5 B ANE I
> .2 _ \_{_ .
3 N2
o ~ Ng
= A
z \!
E 2
N\
2 ) \*
x 1 [ . S
5 CALCULATED POINTS N
o CONFIGURATION 1
& CONFIGURATION 2 N
© CONFIGURATICN 4 -
5 | | v EXPERIMENTAL
10°
0 5 10 15 20 25 30 35

2, DISTANCE FROM SOURCE (cm)

Fig. 4.1.2. Comparison of Calculoted and Experi-
mental Thermal-Neutron Fluxes in Water in the LTSF.

117
source plate power was accurately known at that time, it was not surprising
that the calculated and experimental results were not in agreement.

7 and the power of the

Re-estimates of both the power of the BSF reactor
LTSF source plate8 have since become available, however, and & second
calculation has been made in which the thermal-neutron fluxes at the two
facilities have been compared.

Before a conversion from LTSF data to BSF data could be made it was
necessary to obtain information about the self-attenuation of neutrons
inside the BSF reactor. This was done by placing an all-aluminum mockup
of a fuel element of the BSF reactor adjacent to the source plate in the
LTSF and taking thermal-neutron measurements in the water beyond the
mockup. Three configurations were used: no mockup; a nine-plate mockup
(one-half of an element); and an 18-plate mockup (a full element). The
resulting data for a plane-disk source were converted to a polnt-source
geometry by standard transformations for the three configurations
measured. The data were then extrapolated to a fuel element loading
thickness of up to six elements, which was the thickness of the BSF
reactor loading used, and the thermal-neutron flux in the BSF was calculated

from the following equation:

D(R) = S P(R,) G(Rl,R) av

Reactor
volume
where
D(R) = flux in water at a distance R from the surface of the
reactor,
Rl = distance from the face to the point inside the reactor,
P(R;) = power distribution in the reactor,

TE. B. Johnson, Pover Calibration for BSR Loading 33, ORNL CF 57-11-30
(Nov. 28, 1957).

8D. R. Otis, Lid Tank Shielding Facility at the Oak Ridge National
Laboratory, Part II, Determination of the Fission Rate ol the Source
Plate, ORIL-2350 (to be published).

118
G(R,,R) = attenuation kernel for a medium made up of an Ry thick-
ness of reactor material and an R thickness or water.

Since the power distribution through the BSF reactor was irregular,
a single mathematical expression could not be obtained for the entire
reactor. Therefore, the reactor was divided into 304 volume elements,
and the power in each volume =lement was computed. These calculations
were based on mappings of the thermel-neutron fluxes in the reactor
that were reported by Johnson.rI7

The predicted fluxes are consistently higher than the measured
fluxes for distances of 40 to 115 cm from the face of the reactor, as

may be seen in Table 4.1.2. As might be expected, the accuracy is some-

Table 4.1.2. Calculated and Measured Thermal-Neutron
Fluxes in the BSF

Distance Normalized Thermal-Neutron Ratio of
from Reactor Flux (nv/watt) Calculated Flux
(cm) Calculated Measured to Measured Flux
y ok
Lo 3.41 x 10 1.25 x 10 2.73
55 2.81 x 10° 1.22 x 107 .31
70 3.10 x 10° 1.40 x 10° .21
95 9.92 x 10° 6.3 x 10° 1.57
115 9.k x 107t 6.1 x 107F 1,54

what vetter as the distance from the reactor face increases. However,
on the whole, the results are no closer to the experimental measurements
than those previously calculated,6 although the discrepancies are in
the opposite direction. Those calculations resulted in doses lower
than the BSF data by factors varying from 2.4 at 40 cm to 1.6 at 110 cm.
As a first step in resolving this problem, plans are being made to
remeasure the BSF reactor fission distribution and to redetermine the

power by a calorimetric method.

119

A Monte Carlo Code for the Calculation of Deep Penetrations
of Gamma Rays

As reported previously,9 a code is being developed to calculate at
a point detector the angular and energy distributions of gamma rays
emitted from a monoenergetic, point isotropic or point monodirectional
source embedded in an infinite homogeneous medium of constant density.
Several Monte Carlo techniques have been investigated in this study in
order to determine, if possible, methods which can be used to calculate
the response of a detector at separation distances up to 20 mfp in a
three-dimensional geometry. The first method investigated involved a
process of importance sampling for the first-collision density where
the importance function was determined from a trial sample and set up
automatically during the course of computation. In conjunction with
this process, double systematic sampling from the source (quota sampling)
was also used, as well as splitting and Russian roulette and statistical
estimation. It was found that for a 20-mfp separation distance the con-
tribution to the detector from only the first two orders of scattering
was calculated with any reasonable accuracy.

A second method was then investigated which differed from the first
in that it incorporated an importance sampling scheme for every order
of scattering. The importance functions used were "guessed at" from an
intuitive knowledge of the problem but resulted in a rather small increase
in the accuracy. This meant, of course, that the "guessed at" functions
were not really close to the correct importance functions. The increase
in accuracy was noted by observing that the contribution to the detector
from five or six orders of scattering was very nearly correct.

A scheme of multiple systematic sampling was also devised and
investigated to some extent. A few test cases indicated that this

method would not offer any major improvement in the statistics. The

9S. K. Penny, ANP Semiann. Prog. Rep. Sept. 30, 1958, ORNL-2599,
p 155.

120
so-called "conditional" Monte Carlolo method is now being investigated
and has given some indication of being successful for large separation

distances.

Calculation of Bremsstrahlung Dose Rates which Result from
an Activated-Lithium Cooling System

In some of the Pratt & Whitney Aircraft fast-reactor systems which
are under consideration for aircraft applications, the natural lithium
which is used as the coolant passes directly from the core, through
columbium pipes, to the engine radiators without the use of an inter-
mediate heat exchanger. WNatural lithium contains approximately 92% Li7,
which may capture neutrons as it passes through the core to form Li .
The Li8 will decay with the emission of a beta ray with a half-life of
approximately 0.86 sec to form Be8, which, in turn, will usually split
into two alpha particles. The Li8(6-)Be8 reaction has a Q value of
15.987 Mev; hence, the electrons which are emitted by this process may
have rather high energies. The electrons themselves are easily stopped;
however, during the slowing-down process they may produce a considerable
amount of bremsstrahlung radiation which presents a much more difficult
shielding problem. Cycle times for the complete passage of coolant
through the Pratt & Whitney systems are rather short (~ 2 to 3 sec);
hence, the majority of the L18 decays will occur in thinly shielded
regions outside the primary shield, and the resulting bremsstrahlung
radiation may produce a serious biological dose-rate problem.

In view of this possibility, calculations have been performed to
estimate realistic upper limit dose-rate contributions from this source
for a typical supersonic aircraft configuration. The computed dose
rates were for points both inside and outside the shielded crew compart-

ment.

lOH. F. Trotter and J. W. Tukey, Symposium on Monte Carlo Methods,

p 64, University of Florida, 1954, H. A. Mayer (ed.), Wiley, New York,
1956.

121

Preliminary calculations of the dose rate produced from the
activated lithium indicated that the most important contribution at
the crew position would result from decays occurring in the hot pipes
that lead from the reactor core to the radiators. Therefore, the
contribution from this location was computed in somewhat more detail
than the contributions from other locations. The spectra of bremsstrahlung
radiation produced by initially monoenergetic electrons in lithium and
columbium were determined by means of an Oracle codell in which a
continuous slowing-down model for the electron 1s assumed and the
spectrum of radiation produced during the entire slowing-down process
is computed. These results were then applied to the spectrum of electrons
which result Irom the L18 decays to determine the total bremsstrahlung
spectrum produced by the passage of tnese electrons through the lithium
in the pipes. Next, the spectrum of electrons which reach the pipe walls
was determined, and this, in turn, was combined with the machine results
to give the spectrum of bremsstrahlung radiation produced in the pipe
walls. PFinally, the self-shielding of the pipes and the spatial variation
of the decays down the length of the pipes were included, and a calculation
of the dose rates was performed. Similar, though somewhat less exact,
calculations were performed to obtain the dose-rate contributions at the
crew compartment from decays in the radiators, radiator headers, and the
sump. It should be pointed out that all the computations were based on
a Pratt & Whitney calculation of 30,000 curies of total L18 activity
produced in the 350-Mw systems. The results of the calculation indicate
that the dose rate at the outside of the crew compartment from the
activated lithium should be approximately 2 to 9 r/hr, depending upon
the exact configuration assumed. The dose-rate contribution inside the
crew compartment from this source was found to be approximately 2 to

10 mr/hr. This is to be compared with a total design dose rate of 1.5

llC. D. Zerby and d. S. Moran, ANP Quar. Prog. Rep. Dec. 31, 1957,

ORNL-2440, p 231.

122

r/hr inside the crew compartment, which indicates that for this system

the dose rate Irom the activated lithium coolant does not create a really

serious problen.

123
4,2, LID TANK SHIELDING FACILITY

Experimental Flux Depression and Other Corrections
for Gold Foils Exposed in Water

One of the methods for determining the thermal-neutron flux in a
medium is to measure the activation induced in gcld foils. Both bare
foils and cadmium-covered foils are exposed in the medium, and the
thermal-neutron flux at any specified position is calculated from the
difference between the activations of a bare foil and of a cadmium-
covered foil at that position. This method is used frequently at the
LTSF. Since all thermal-neutron detectors at the LTSF are normalized
to the gold foils,l a knowledge of the corrections to be applied to foil
measurements is quite important.

The gold foils commonly used at the LTSF are 1 cm square and 0,002
in. thick. The cadmium covers on these foils are 0.020 in. thick. It
was felt that all applicable corrections could be contained in one
factor, namely, the ratio of the flux measured with infinitely thin
foils to the flux measured with 0.002-~in.~thick foils having the same
shape and area. In order to determine the values of this factor
experimentally, several l-cm-square foils of varying thickness were
exposed at the same position in the LTSF so that an extrapolation to
"zero'" thickness would be possible. The position chosen was such that
the neutron flux should have been isotropic. The foil thickness was
varied from 9.2 x 1077 in. (45 ug/cme) to 0.010 in. (L83 mg/cme). Foils
thicker than the usual 0.002 in. were used to make the interpolation
more reliable and because other facilities normally use thicker foils.
The thicker foils may even become desirable at the LTSF.

The experimental curves for both the bare foils and the cadmium-

covered foils, as well as the difference curve between the two, are

lD. W. Cady, Instrumentation at the Lid Tank Shielding Facility,
ORNL-2587 (to be published).

124

1

presented in Fig. L4.2.1., The value for "zero" thickness was obtained
by fitting the data for thicknesses less than 1.7 mg cm2 with straight
lines, using the method of least squares. From these data the experi=-
mental ratio of the saturated activity per unit mass of a foil of "zero"
thickness to that of a 0.002-in.-thick foil is 1.35 + 0.20.

Various calculational methods have been proposed from which the
flux depression factor can be estima.ted.2"5 Using the prescription of
Bothe2 results in a factor of 1.09; using that of Tittle,3 1.08; using
that of Skyrme,h 1.21; and, finally, using that of Bengston,5 1.30. 1In
all these calculations the self-shielding in the foil has been included.

The experimental value is obviously in best agreement with the
value calculated by the Bengston method. Since in that method it is
assumed that the foil area is infinite, an experimental check of this
assumption was made by exposing foils having different areal dimensions
(4.68 cm?, 7.92 cm2). Further, the validity of assigning an effective
radius to the square foil for calculational purposes was investigated
by exposing circular foils 1.1l cm in diameter (0.968 em in area)., If
the l-cm-square foil were indeed effectively infinite, then the
saturated activity per unit mass would remain constant for foils having
the same thickness but larger areas. The experiments confirmed this
proposition. Also, within the accuracy of the measurements (+2%), all
the circular foils exposed gave the same thermal-neutron flux as the
l-cm-square foils. Further work aimed at reducing the error in the

measurement is in progress.

“W. Bothe, Z. Phys. 120, 437 (1943).

3c. w. Tittle, Nucleonics 8(6), 5 (1951); Nucleonics 9(1), 60 (1951).

4?. H. R. Skyrme, Reduction in Neutron Density Caused by an

Absorbing Disc, MS=91 (no date).

5J. Bengston, Neutron Self-Shielding of a Plane Absorbing Foil,
ORNL CF 56-3-170 (March 1, 1958).

125

UNCLASSIFIED
2—MH—057—0— 461

J 1 [ 11 T I
S B
| IR | ]
} | 1 !
5 : e - : . - —
BARE FOILS ] —
o o /
v T o | ; s -
. { BARE FOILS
\ R
DIFFERENCE CURVE #flfit\
! —
; DIFFERENCE CURVE
10° |
!
w
@ 1
a B ; o
= = T77 —— 4 —)——]
= 1 \ | %
5 s CADMIUM-COVERED FOILS \
o
l ‘
. A
s ; \\ CADMIUM-COVERED FOILS
= i B
< 2 ! { . % o "N
a : i ‘ ! o
g i | | T
T \ ' i \'Q'-o__
2 ' o
8 ;
& 19 { |
! |
l
5 ‘ -
l
I
I
|
2 F
|
10’ *
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 0 100 200 300 400 500

GOLD FOIL THICKNESS {mg/cm?)

Fig. 4.2.1. Saturated Activities of Gold Foils Exposed in the LTSF.
4.3. BULK SHIELDING FACILITY

The GE~BSF Study of Radiation Heating in Shields

The experimental phase of the study of radiation heating in
laminated shields, which is being carried out at the Bulk Shielding
Facility (BSF) in cooperation with the General Electric Company, is
nearing completion, with the entire series of measurements teing made
for a second time. The shielding configurations, which are immersed in
a tank of oil, consist of slabs of beryllium and lithium hydride
separated by a gamma-ray shielding section of iron, lead, or Mallory
1000. Measurements of the temperatures of samples of the various
materials in the shields are made during and after the configuration is
exposed to the flux from the BSR. 1In addition, radiation measurements
are made within the configurations.

As was reported previously,l it was discovered after the initial
series of measurements had been made that leaks in the sample cases had
allowed slow infiltration of transformer oil into the air region sur-
rounding the samples. This not only introduced an additional foreign
mass which would absorb energy in the sample case but also changed the
heat transfer rate between the case and the sample, the effect of which
was difficult to determine. Modified sample cases were used in the
second series. A second uncertainty was introduced in the interpre-
tation of the original data by the unexpected bowing of the reinforced
wall of the containing tank. As a result, the oil layer between the
tank wall and the shield proper varied in thickness during the experi-
ments. Corrections based on parallel plumb bob readings taken at the
reactor face and at the shield face were made in the second gseries of
measurements to account for the bulging.

The previous reportl on this study included a discussion of the

experimental errors assoclated with the radiation measurements. These

lK. M. Henry, ANP Semiann. Prog. Rep. Sept. 30, 1958, ORNL-2599,
p 15h.

127

errors will be greater than reported previously because the residual
gamma-ray emission from the reactor and the now activated shield
produce a larger photoneutron effect,

As the experimental date are collected, they are forwarded to GE i
for analysis. The heating results obtained from this second series of
measurements have been reported by GE to be in reasonable agreement with

their predicted values of energy absorption.

The BSF Stainless Steel-~UQ, Reactor (BSR~IT)

The stainless steel-clad UOE-fueled reactore’3 designed for use at
the BSF and identified as the BSR-II was approved by the Advisory
Committee on Reactor Safeguards, and component fabrication is well under
way. Machining of the grid plate has begun and is expected to be
completed soon. The fuel elements should be available early in April.
Detailed design work on the control system is progressing, and control
system fabrication, which began in mid-January, will proceed as the
design permits. The control-plate drive is being fabricated and is
scheduled for completion the latter part of April. A prototype control-
rlate drive for life tests with a dummy element is nearly complete.
These tests will also provide information on plate insertion rates.

The initial low-power critical experiments with the reactor are
scheduled to be carried out at the BSF in May. Following these
experiments, the BSR-II will be tested at the SPERT-I Facility at the
National Reactor Testing Station to determine the capabilities of the
control system in overcoming short, positive reactor periods.

In order to select a suitable means for providing the required
positive periods, the following criteria for the reactivity insertion

device were established: (1) it should be capable of producing periods

2E. G. Silver and J. S. Lewin, Safeguard Report for a Stainless
Steel Research Reactor for BSF (BSR-II), ORNL-2470 (March 12, 1958).

3E. G. Silver, ANP Quar. Prog. Rep. March 31, 1958, ORNL-2517, p 1OL. .

128

of varying times to a maximum severity of 3 msec, (2) "ramp" as well as
"step-function" reactivity insertions should be possible, and (3) the
core should be as nearly homogenecus as possible after the reactivity
insertion. A regular BSR=II control-plate drive will be modified to
drive the reactivity insertion device. Space for the device will be
provided in a special core element which will resemble a control-plate
element with the control-plate channel cover plates removed. The widths
of the channels will be increased by removing the nearest fuel plate in
each of the two adjacent fuel elements (see Fig. L.3.1).

Several insertion devices were considered. The simplest method
proposed was to insert a number of standard fuel plates — the maximum
possible being six — into the two available slots; however, calculations
showed that this method would not afford a sufficient reactivity
increase. A decision vas made to use a "double-ended" device which
would have a lower poison section and an upper fuel section, since
inserting the fuel would e ject the poison and give a greater increase in
reactivity than would be realized by merely inserting fuel into a
water-filled channel, The present proposal is to use a device consisting
of six double-ended plates arranged so that three plates can be inserted
on each side of the special element described above (see Fig. 4.3.1).
Each plate will have a T-in.-long poison section, a l-in.~long transition
section, and a 15=in.-long fuel section. The two extreme positions of
this device in the core are shown in Fig. 4.3.2. Smaller initial
insertions of the poison will produce smaller changes in reactivity and
longer periods when the poison is ejected. Detail design work on the
device is under way.

The above "double~ended" design was chosen on the basis of several
calculations to predict the reactivity effects of different insertion
devices. The calculations were performed in two stages, the first of
which consisted in the determination of flux-weighted four-group
constants for the following energy ranges: (1) 10 Mev to 9.12 x 103 ev,
(2) 9.12 x 10° to 0.618 ev, (3) 0.618 to 0.0252 ev, and (4) thermal at
0.0252 ev., These constants were obtained with the IBM=-704 GNU-II code,

130

UNCLASSIFIED
2-01-058-0-448

7
SPECIAL L CONTROL
ELEMENT PLATES
c
) = £
< 2 3
o ] o
[ | [ o P il MODIFIED CONTROL
L ; Ll 7 PLATE ELEMENT
I ]
\I L] 2
+ [ N ' 1 STANDARD FUEL
’ < T ELEMENT (ONE FUEL
- PLATE MISSING )
Q
THREE -PLATE SECTION Q
OF INSERTION DEVICE S

Fig. 4.3.1. Configuration for BSR-Il SPERT-| Tests
Showing Position and Detail of Reactivity Insertion
Device (Top View).

UNCLASSIFIED
2-01-058-0-449

-
15in.{4 )
_____ | ___comerop_ __ ~1rm~2sin

FUEL
15in. FUEL
T I B

POISON IN FUEL iN

Fig. 4.3.2. Two Extreme Positions of Proposed BSR-Ii
Reactivity Insertion Device (Side View).

a one-dimension, spherical-geometry, diffusion=-theory code with which
reactor multiplication constants can be calculated using 32 lethargy
groups. In the second stage the constants were used as input data
for another IBM-70L4 calculation using the PDQ code, a few-group,
multiregion, two-dimension code which was used in X,y geometry.

Four different cases were considered, and the following

multiplication constants were obtained:

Multiplication
Configuration Constant
Case 1, Fuel in the two insertion
slots, no control plates 1.0207
Case 2. Iuel in the two insertion
slots, control plates in 0.8764

Case 3. DBoron poison in the two
insertion slots, no
control plates 0.9545

Case 4. Water only in the two
insertion slots, no
control plates 1.0115

The calculation with boron poison in the slots was performed by
assuming that rectangular boxes consisting of control-plate material
would be inserted which would have the same geometric cross section as
an insertion device.

The calculated multiplication constants indicate that the effect
of the fuel plates, as opposed to the effect of the water, is only
0.90% 8k/k. In order to achieve a 3-msec period, 3.2% ©®k/k would be
required, assuming a prompt-neutron generation time of 75 pusec. It is
seen, therefore, that merely inserting six fuel plates into the core
will not result in sufficient reactivity increases. Since the effect
of the poison sections as compared with the fuel plates is 6.5% &k/k,
a double-ended device will provide twice as much reactivity as will be

needed, and the required distance of travel of the drive mechanism and

131

thus the insertion time can be reduced for the SPERT~I experiments. A
comparison of Case 2 and Case 1 results shows that the worth of the .
control plates is 14% 8k/k.

About 75% of the electrical design work for the SPERT test is
completed. The results of the life tests of the control-plate drive

will be used to evaluate the drive and the insertion device designs.

The Model IV Gamma-Ray Spectrometer

A1l components for the model IV gamma-ray spectrometerh which will
be used for shielding measurements at the BSF have been delivered, with
the exception of the lead-lithium housing. The housing has been poured
and is being machined. The bridge crane and control mechanisms have
been completed and installed at the BSF.
It is still planned that the detector used in this spectrometer
will be the large NaI(Tl) crystal being investiganted at the BSF,5
which is approximately a 9-in.-dia right cylinder with a truncated
cone on one end. Recently, two additional crystals comparable in
size and geometry to the BSF crystal were obtained on a loan basis from
the vendor6 in order to compare their responses to gamma rays with the
response of the BSF crystal. One of these borrowed crystals, like the
BSF crystal, has a well in the cone along the axis of the crystal to a
depth of about 2 in. The other crystal has no well. In general, the -
pulse~-height distributions obtained with these crystals were
characteristically the same as the distribution obtained with the BSF .
crystal.
The responses of the BSF crystal and of the borrowed crystal

with no well to collimated gamma rays from a Nae,4 source are shown

hR. W. Peelle and T. A. Love, ANP Quar. Prog. Rep. June 30, 1957,
ORNL=-2340, p 303.

5G. T. Chapman and T. A. Love, ANP Quar. Prog. Rep. March 31, 1958,
ORNL-2517, p 108.

6The Harshaw Chemical Co., Cleveland, Ohio.

in Fig. 4.3.3. As may be seemn, double-peak distributions occurred
when the gamma rays were collimated into the conical end of either
crystal. Furthermore, the "low-energy" peak of the double-peak
distribution (that 1s, the peak on the low-energy side) coincides in
pulse height with the single peak obtained when the gamma rays were
collimated into the side of the crystal. The data of Fig. L.3.4
indicate that this characteristic holds true up to 6.1 Mev. These data,
along with considerations of other data, have led to the speculation
that the low-energy peak in the double-peak distribution is due to
gamma rays which interact in the cylindrical region of the crystal
and that the high-energy peak is due to those that interact in the
conical region., Other data have been obtained which indicate that

the optics of the crystal and reflector or of the reflector alone may
also contribute to the double-peak distribution. Since the reflectors
around the conical regions differ from the reflectors around the
cylindrical regions of the crystals, arrangements are being made to
remount the BSF crystal with a uniform reflector around both sections

in a further attempt to improve the response.

Cross Section for the Lis(n,o&)H3 Reaction for 1.2 <E 8.0 Mev

Ay

L

The Li6(n,cx)H3 reaction is well suited, in many respects, for
use in neutron detection and fast-neutron spectroscopy. In particular,
Li6-loaded nuclear emulsions or scintillation crystals can be used to
measure reactor neutron leakage spectra. A program has been under way
for some time to develop an Li6I(Eu) spectrometer for this purpose.7
Since the efficiency of such a crystal depends upon the probability of
interaction of an incident neutron with an Li~ nucleus to give an

Li6(n,a)H3 event, the interpretation of the data obtained with an

TR. B. Murray, ANP Quar. Prog. Rep. June 30, 1957, ORNL-2340,
p 303.

133

UNCLASSIFIED
2-01-058~0-4d46

6.0 ] T
! .85 Mev
ORNL CRYSTAL HARSHAW CRYSTAL -
(WITH WELL) {(NO WELL)
NORMALIZED IN J
1.85 Mev MAGNITUDE ONLV\
5.0 T fl
]
q
|
|
| ——— GAMMA RAYS COLLIMATED
| INTO CONICAL END OF CRYSTAL
4.0 :
; ——— GAMMA RAYS COLLIMATED :
INTO SIDE OF CRYSTAL |
|
7 ) |
z I
> |
2 A |
z Eopl )
2 ! t 276 M
- 1l " ‘R ev
E 3.0 I ,q i
Q I |
z I ’ J |
2 I | X
O II NORMALIZED IN I |
© MAGNITUDE ONLY ] 1
. “ Ny - : b
' 2.76 Mev
I B Ji
il | I
I j | 1
I | X
2.0 i I 1
I ! | H
|} i I
I I | I I -
i | | | h
! | I i
| & B -1
11
K | | I |
| [y || ) |
i | )
1.0 I | |
| R R 71 I
l l I o% | | |
(A I i i
o | | | ] I
o v (o] | | |
| ! I I |
] i I‘l L] l
o i VoL AV R T
O
= | _’/ ‘l o jl \\.I |
7
‘ / \ -~ !
| \ \
o)
0 20 40 60 80 0 20 40 60 80 100
PULSE HEIGHT PULSE HEIGHT

Fig. 4.3.3. Response of Two Different L.arge Nal(T!) Crystals to Na24 Gamma Rays.

UNCLASSIFIED
2-01-058—-0-447

8.0
O GAMMA RAYS COLLIMATED INTO SIDE OF CRYSTAL
® [OWENERGY PEAK OF THE DOUBLE-PEAK DISTRIBUTION; GAMMA RAYS o /p
7.0 b—o COLLIMATED INTO END OF CRYSTAL / #
O  HIGH ENERGY PEAK OF THE DOUBLE-PEAK DISTRIBUTION: GAMMA RAYS v
COLLIMATED INTO SIDE OF CRYSTAL ,/
| / / /
6.0
__ 50
>
@
=
>_
o
&
Z 4.0
[
>
I
o
t
<
:
g 3.0
o
2.0 |
|
|
1.0
0
o} 20 40 60 80 100 120 140 160 180

RELATIVE PULSE HEIGHT

Fig. 4.3.4. Gamma-Ray Energy as a Function of Pulse Height for Two Large Nal(T!) Crystals.

Li6I(Eu) detector depends upon a knowledge of the Li6(n,a)H3 cross

section. Previous measurements of this c¢ross section by Ribe give
data at a few points up to 6.5 Mev, with a single point at 14 Mev. In
view of the possible applications of Li~ counterg, it was felt that a
further study of the cross section in the Mev region would be of value.
Therefore, measurements of the Li6(n,a)H3 cross section were made in
the energy interval 1.2 éEneég.O Mev relative to the previously known

fission cross sections of U and Np237. Incident neutrons were
monitored by fission events in a 2% ionization chamber, and

Li6(n,a)H3 events were detected in an Li6I(Eu) scintillation crystal.
An essentially back-to-back geometry was used. The results of the
measurements, which are presented in Fig. L4.3.5, are in good agreement
with those of Ribe for the overlapping energy interval. A more
detailed description of the experiment can be found in a separate

9

report.

The Energy Spectrum of Prompt Gamma Rays Accompanying the Thermal
235 |

Fission of U

Since a knowledge of the spectrun of prompt gamma rays from the
thermal-neutron fission of U235 is important in the solution of
shielding and heat-removal problems for high-performance reactor
systems, a program to determine the spectrum between 0.4 and 10 Mev has
been under way for some time. The basic data have been obtained for
the entire spectrum, and some major parts of a final data analysis have
been finished. Some source-calibration experiments and the remainder

of the analysis are yet to be performed, however. Previous reports

%. L. Ribe, Phys. Rev. ;93, 4L (1956).

9R. B. Murray, H. W. Schmitt, and J. J. Manning, Neutron Phys. Div.

Ann. Prog. Rep. Sept. 1, 1958, ORNL-2609, p 1hL5.

136

UNCL ASSIFIED
ORNL -LR-DWG 37200

0.35
0.30 T 3
J & T(p,nHe®, EXPERIMENT NO. 1
. O T(p,n)He>, EXPERIMENT NO.2
T o T(p,n)He, EXPERIMENT NO.3
g e D(d,n)He>, EXPERIMENT NO.3
0.25 1 i } i %
0.20
‘; %
[
]
2
C)
)
b
0.5 )}% *
B
0.0 E’ > ‘[
—=&—— UNCERTAINTY IN ORDINATE = +7% *‘f\*\ }
0.05 f 1 T
o H
o) 1 2 3 a4 5 6 7 8

NEUTRON ENERGY (Mev)

LET

Fig. 4.3.5. Ll6(n,a)H3 Cross Section as a Function of Neutron Energy.

have described the procedures used to measure and eliminate back-

10,11 and have given results for the spectrum based on a

12,13

grounds
preliminary analysis of part of the data. This early analysis
indicated that gamma rays totaling (7.2 + 0.8) Mev/fission are given
off with a spectrum falling off roughly ;s 8 e-l'1E (Mev) (photons
fission-l-Mev_l). The spectrum exhibited a sharp cutoff for gamma-ray
energies above about 7.5 Mev,

The ma jor effort during recent months has been the combination of
results from the approximately 60 separate data runs and the various
experiments necessary for calibration of the spectrometer efficiency.
Combination of the data was possible only after careful evaluations of
internal consistency and of the slight differences between similar runs.
Experimental conditions such as pulse height vs energy and the count
rates changed by a few per cent from day to day. The most difficult
aspects of the efficiency calibration have been those involved in the
absolute coincidence counting experiments needed to check source
calibrations in the region below 3 Mev and to establish the strength of
a steady-state source of 7-sec N16 for calibration at 6.1 Mev.

Work is continuing on the remaining experimental and analytical
problems required to complete the spectrum determination for energies
greater than 0.4 Mev. An additional experiment is planned for the
region below 0.8 Mev, for which a single scintillation detector, and
g thin fission chamber will be used. These detectors will be separated
by a flight path sufficiently long to allow discrimination against

prompt neutron backgrounds in the gamma-ray scintillation.

lOT. A. Love, F. C. Maienschein, and R. W. Peelle, ANP Quar. Prog.

Rep. March 31, 1957, ORNL-227k4, Part 6, p 21.

11T. A. Love, F. C. Maienschein, and R. W. Peelle, Appl. Nuclear
Phys. Ann. Prog. Rep. Sept. 1, 1957, ORNL-2389, p 99.

12T. A. Love et al., ANP Quar. Prog. Rep. March 31, 1958,
ORNL-2517, p 1OL.

13T. A. Love, F. C. Maienschein, and R. W. Peelle, Neutron Phys.

138
L.4, TOWER SHIELDING FACILITY

TSF Studies in Support of the GE~ANP Program: Rechecks and
Extensions of the 2n Experiments

A series of experiments was carried out during the last quarter of
1956 at the Tower Shielding Facility by personnel of the General Electric
Company to obtain data to aid in the design of a shield for use at their
Idaho Shield Test Air Facility. The experiments, which were designed to
investigate the effect of ground-scattered radiation, consisted essen-
tially in measurements of radiation in a cylindrical GE crew compartment
both as a function of altitude and as a function of the shielding
configuration on the Tower Shielding Reactor 64 ft away. The reactor
was encased in the GE reactor shield No. 1, a cylindrically shaped shield
with a high neutron-to-gamma-~ray transmission ratio. For approximately
one-half of the measurements an additional borated water shield, called
a "2n" shield, was fitted over the lower half of the reactor shield.
Throughout the experiments the axes bf the crew compartment and the
reactor shield were held in the same horizontal plane; however, the
reactor assembly was rotatable and measurements were made for two reactor
shield positions. In the first position (the ©® = O deg position) the
reactor shadow shield was suspended between the reactor and the crew
compartment. In the second position (the 6 = 180 deg position) the
reactor shield was rotated a tull 180 deg so that the shadow shield did
not intercept the path of direct radiation. '

The results of the 27 measurements at the highest altitudes were
compared with predictions made by a GE analysis group.l Since the
measurements and the predictions were not always in agreement, GE
requested that ORNL perform a second series of measurements to check the
Tirst series and also to extend them. The extensions, included primarily

to aid in the calculations of secondary gamma-ray production, consisted

lR. H. Clark et al., Test Data from the 2x Solid-Angle Shield-

Cover Experiment, APEX-439 (Dec. 19, 1950).

139

in mapping the radiation leaving the reactor shield and a series of gamma-
ray spectral measurements inside several detector shields. (A few
spectra measurements were also made by GE in their original experiments.)
In the discussion below the results of the second series of measurements
are summarized and, where possible, compared with the results of the

first series.

Recheck of Fast-Neutron Measurements

Fast-neutron dose rates measured in either series without the 2=
shield on the reactor were 3 to 5 times higher than those measured with
the shield. The maximum dose rates with the 2x shield occurred at
ground level, whereas maximum values without the shield were observed
at altitudes of 50 ft for the ® = O deg case and at 30 £t for the
@ = 180 deg case.

The fast-neutron dose rates for corresponding cases in the two
series of measurements with the 2r shield in position were in agreement,
and the results from both series agreed fairly well with the calculated
dose rates. The corresponding measurements in the two series for the
cases without the 2n shield agreed within 15%; however, for these cases
the calculated value for © = 0 deg lies between the two measured values,
and the calculated value for © = 180 deg is about 50% lower than the
experimental results.

Recheck of Gamma-Ray Measurements

The maximum gamma-ray dose rates measured in the second series
occurred at ground level with the 2% shield on the reactor. Without
the shield the maximums were observed at 25 ft for the @ = 180 deg
case and at 4O ft for the © = 0 deg case. The gamma-ray measurements
in the second series agreed within 10% with those in the first series;
however, the experimental values from either series were as much as
a factor of 2 higher than the calculated dose rates for the € = 0 deg
case, both with and without the 2n shield, and as much as a factor of

2 lower than those calculated for the @ = 180 deg case.

140

Recheck of Thermal-Neutron Measurements

Thermal-neutron fluxes were measured in the second series with plastic-
covered BF5 counters to duplicate the test conditions of the first series,
and the fluxes observed in the two series were in agreement, both with
and without the 2n shield. When the plastic cover was removed, however,
the fluxes measured without the 2n shield increased 30%. This increase

was not observed while the 2n shield was on the reactor.

Mapping of Radiation Around the Reactor Shield

As mentioned above, the extensions of the 2xn experiments included
mapping the radiation around the reactor shield. In one group of
measurements unshielded detectors were rotated around the horizontal
axigs of the reactor shield in a vertical plane passing through the center
of the reactor. Cables attached to the detectors maintained a constant
separation distance of 50 ft between the detectors and the center of the
reactor. In another group of measurements the detectors were suspended
from a truss at fixed distances of 33, 59, and 100 ft along the truss,
and the reactor shield was rotated around its vertical centerline. The
truss was hinged where it joined the reactor support, and truss angles
of *45 deg from the horizontal could be achieved.

The fast-neutron and gamma-ray dose rates observed at the 100-ft
station increased by as rmuch as a factor of 7 as the reactor shield was
rotated and the lead shadow shield became ineffective. Under the same
conditions the thermal-neutron flux did not increase much more than a
factor of 2 since a large fraction of the flux was caused by air-scattered
rather than direct neutrons.

These mapping data will provide a check on the source terms used
in the calculations for the 2n experiments. In addition, the data will
contribute to a better understanding of the numerous secondary effects

produced by radiation emitted from the reactor shield.

Measurements of Gamma-Ray Spectra

During the first series of experiments performed by GE, measurements

of the spectra of gamma rays inside the crew compartment were obtained

141
with no water in the side shield compartments. As a result, neutrons
reached the sodium iodide detector and caused interactions which intro-
duced considerable error in the results. The measurements were repeated
in this second series with sufficient water side shielding to minimize
the neutron effects. The results show the usual 10.8-Mev nitrogen
capture peak plus a series of peaks near 10 Mev and a series of peaks
near 7.6 Mev. The sources for the peaks near 10 Mev are undetermined.
The peaks near 7.6 Mev are attributed primarily to gamma rays resulting
from neutron captures in aluminum in the crew shield and iron in the
vicinity of the shield in the form of cables, turnbuckles, beans, etc.
The hydrogen capture gamma ray at 2.23% Mev is easily distinguishable in
the spectra, and, despite a large buildup in the low-energy region,
there appears to be a peak just above 0.5 Mev. This peak probably
results from both boron capture gamma rays and annihilation radiation.
The spectra of gamma rays from the reactor snield were also
measured in this series, both with plain water and with borated water
in the reactor shield. The measurements were made at distances of 17
and 70 ft from the shield, and preliminary analysis indicates that the
high-energy portion of each spectrum is probably due to neutron captures
in the iron structure in the shield. Further analyses will be performed
by GE.

142

L.5. TOWER SHIELDING REACTOR II

Critical Experiments and Calculations

It was previously reportedl that the TSR-II fuel elements were
fabricated to contain 8.1 kg of U255, which, it was thought, would not
255

only include sufficient U to overcome the worth of the control grids
in their fully withdrawn position and the worth of the structural
materials in and around the core, but would also allow 1.6% 8k/k as

the amount to be controlled by the grid plates and 2.0% ak/k for
calculational error. Wnen the fuel elements were assembled for a
clean, water-reflected, room-temperature critical experiment, however,
it was discovered that the core was subcritical even in the absence of
the control grid plates. Subsequent critical experiments have shown
that the amount by which the core was subcritical was approximately
0.7% dk/k.

A series of investigations, both calculational and experimental,
was initiated in an attempt to discover why a core calculated to be
above critical by approximately T% was actually slightly subcritical,
and to determine what changes in the design of the reactor would be
required in order to use the elements. The case which had been used as
a basis for choosing the critical mass was one in which the temperature
of the internal water reflector was 5OOC, the temperature of the core
was 6OOC, and the temperature of the external water reflector was
20°¢ (room temperature). A new calculation was carried out with the
Cornpone code for the actual experimental configuration with a temperature
of 20°C throughout, and the multiplication constant was found to be
1.01l. From past experience and a knowledge of the behavior of this
type of reactor, it is felt that the large discrepancy in the multipli-

cation constants calculated for the two different temperatures is

lC. E. Clifford and L. B. Holland, ANP Semiann. Prog. Rep.
Sept. 30, 1953, ORNL-2599, p 10u.

unrealistic. PFurther investigations aimed at understanding these
variances are continuing.

In the meantime, other calculations and critical experiments have
been carried out to determine the increases in reactivity that could be
brought about by redesigning the internal reflector region. It has
been found that rather large changes in k (as much as 10%) can be
effected by introducing pieces of aluminum or air voids into this
region; in fact, the possibility of the sudden displacement of water
by air bubbles during a reactor operation must be considered I'rom
a safety viewpoint.

The size of the air pockets which might form in this region and,
thus, the severity of the reactivity change, can be limited by
placing a sphere in the internal reflector and letting the cooling
water flow between the fuel and the sphere. In order to investigate
this experimentally, spheres of varying diameters were placed in the
internal reflector region, and pieces of Styrofoam, simulating air,
were introduced into the remaining water. The only configuration in
which the Styrofoam did not produce a positive change in reactivity
was one in which the aluminum sphere was covered with boronated
plastic and the thickness of the water annulus between the boron and
the fuel was 1/4 in. It was also found that the control available
with the boron was slightly higher than that available with cadmium.

On the basis of the above experiments, it was decided to operate
the TSR-II control devices inside a 17-in.-dia spherical shell filled
with water and aluminum. The sphere will occupy most of the internal
water reflector region. In addition, the cadmium control grids
previously designed for the reactor are being replaced with BuC—loaded
plates. With these design modifications, any displacement of water by
air will be small enough to produce only small changes in reactivity,
and the increase in reactivity necessary for use of the fuel annulus
in the reactor will be effected. When the amount of excess reactivity
required for operation is determined in the final critical experiments,

adjustments in the reactivity will be made by insertling removable
BuC-loaded aluminum plugs in the aluminum.

14k

Several critical experiments have been performed to determine the
temperature coefficient of reactivity for the water-reflected TSR-IT
core without the sphere. It was found to be +L4.5 x lO_5 (SZk/k)/oC
from 25 to 40°C and to be =1.25 x 107> (8 k/k)/°C from 40 to 50°C. The
temperature coefficient remains negative through 6000, but an accurate
measurement was not made at that point. If the alterations in the
internal yeflector and the addition of the lead-boral shield do not
change the measured values appreciably, no excess reactivity will be
required to compensate for changes in the core temperature, since the

mean temperature should be in the range from 40 to 60°C.

Flow Distribution Studies

The results of the most recent test in the empirical studies for
determining the effect of screens on the distribution of water flow in
the annular fuel elements showed that at least the minimum required
flow distribution could be achieved in all channels in which
measurements were taken. There were indications, however, that the flow
distribution varied from fuel element to fuel element. In another
investigation of this problem, a clear plastic mockup of an annular
fuel element with a baffle plate is being used for visual studies of
flow distribution, and the results are promising. The gross flow
distribution appears to have the desired pattern, and there is positive
flow in all channels. Some tailoring of the flow will be necessary
when measurements are made in each channel. The pressure drop across
the element, including the baffle, was 6 psi for a flow equivalent to
900 gpm through the reactor,

The hydraulic flow test unit has been modified to study the flow
distribution through the fuel elements in the central cylinder. The
'lower fuel elements (dummy elements) and a 17-in.-dia sphere were
placed in the test stand, and the pressure drop was found to be 2 psi.
When the upper fuel elements are no longer required for critical
experiments, they will be placed along with the shield plug of the

ionization chamber guide assembly in the test unit to make a complete

mockup of the central cylinder region. The shield plug is the

section through which the 130 helical cooling tubes will pass. (In

a separate experiment the pressure drop through a three-tube gection
of the helical cooling tubes was found to be 4.5 psi.) Screens have
Peen placed between the ends of the helical tubes and the fuel elements
to prevent jet streaming through the upper fuel elements. The pressure
drop introduced by the screens is only about 0.5 psi. The over-all
pressure drop through the core 1is expected to be about 15 psi. The
pitot tubes that will be used to measure velocities in the central

fuel elements have been calibrated.

Heat Transfer Studies

An experiment performed to determine the friction factor versus
Reynolds number for a typical flow channel in an annular element showed
a smooth transition from laminar flow to turbulent flow; therefore, it
will be possible to use transitional flow to cool the annular elements.
Final calculations of the heat transfer from the annular fuel elements
under a 30-psi head and optimized flow distribution indicate that, even
with single-pass flow, less than 600 gpm is needed to keep the water
from exceeding the saturation temperature. Since the proposed flow
rate is 1000 gpm rather than 600 gpm, no allowance was made in the
calculation for hot-channel factors. A complete summary of these
calculations for laminar, transitional, and turbulent flow has been
reported by Lewin.2

An analytical study is under way to determine whether the TSR-II
fuel element temperature rise resulting from failure of the cooling
water system during reactor operation would be sufficient to melt the
aluminum fuel element cladding and allow the release of fission proddcts

to the atmosphere. The results of heat-transfer calculations based on

2J. Lewin, TSR-II Second Pass Heat Transfer, ORNL CF 59-1-111
(Jan. 28, 1959).

146

I .
ORI,

afterheat generation resulting from lO-hr operation at 5 Mw as
predicted by the Way-Wigner equation have indicated the following:

l. With the reactor in the vertical position, the total afterheat
would be sufficient to boil away only about one-«third of the volume of
water contained in the upper region of the reactor. The reactor core
would therefore remain immersed in 2120F water which would prevent an
appreciable fuel element temperature rise.

2. With the reactor in the horizontal position, a cooling-water
supply or return-line failure might cause partial loss of cooling water
in the core region. In order to establish the absolute minimum time
required for a section of fuel element located in the maximum power
density region of the reactor core to reach the melting point, a
calculation was performed in which zero heat loss and zero heat transfer
from a dry fuel plate were assumed. The results indicated that under
these conditions the fuel element cladding in the maximum power density
region would reach the melting point in approximately 12.5 min.
Therefore, if steps are taken to flood the reactor core with water in
less than 12.5 min, there will be no danger of fission-product release
due to melting of the fuel element cladding.

3. Since several heat transfer mechanisms act to retard the
temperature rise of a dry fuel element, a series of heat transfer
calculations that were based on certain simplifying assumptions were
performed to evaluate the effect of these mechanisms. The calculations
indicated that, within 15 min after shutdown, the heat losses will be
approximately equal to the heat generation and that the maximum cladding
temperature will be below the melting point. However, owing to the
nature of the simplifying assumptions, it cannot be definitely stated
that some melting of the fuel element cladding will not occur if the
reactor remains in the horizontal position and the fuel elements in the
maximum power density region remain dry. On the other hand, the
calculations indicated that extensive melting of the fuel element

cladding would not occur.

”fi'

In order to ensure that the reactor will be returned to the
vertical position if a power failure or other emergency should arige
while it is in the horizontal position, an air motor has been added to
the rotating unit. Bottled gas will be used to drive the reactor to
the vertical position when a solenocid is deactivated by a power failure

or other emergency signal.

e —

— X}

INTERNAL DISTRIBUTION

ORNL-2T711
C-85 — Reactors-Aircraft Nuclear
Propulsion Systems
M-3679 (22nd ed.)

l. J. W, Allen 37T« Re 8. Livingston
2. D. S, Billington 38. R. N. Lyon
3. F. F. Blankenship 9. H. G. MacPherson
4, E. P. Blizard 4. F. C. Maienschein
5. A. L. Boch 41, W. D. Manly
6. W. F. Boudreau Lo, A. J. Miller
7. G. E. Boyd 43, K. Z. Morgan
8. E. J. Breeding LL, E. J. Murphy
9, R. B. Briggs L5, J. P. Murray (Y-12)
10, A. D. Callihan L6, M. L. Nelson
1l1. C. E. Center (K-25) 4L7. G. J. Nessle
12. R. A. Charpie 48, P. Patriarca
1%. C. E. Clifford 49. 8. K. Penny
14, J. H. Coobs 50. P. M. Reyling
15. W. B. Cottrell 5. H. W, Savage
16. F. L. Culler 52. A. W. Savolainen
17. L. M. Doney 5%. E. D, Shipley
18. D. A. Douglas 54, M. J. Skinner
19. L. B. Emlet (K-25) 55. A. H. Snell
20. A. P. Fraas 56. E. Storto
21, J. H. Frye 57« C. D. Susano
22, R. J. Gray 58. J. A. Swartout
2%. B. L. Greenstreet 59. D. B. Trauger
24, W, R. Grimes 60. D. K. Trubey
25. E. Guth 6l. G. M. Watson
26. C. S. Harrill 62, A. M. Weinberg
27. T. Hikido 63, J. C. White
28. M. R. Hill 64, E. P. Wigner (consultant)
29. E. E. Hoffman 65. J. C. Wilson
30. H. W. Hoffman 66. C. E. Winters
3l. A, Hollaender 67. W. Zobel
32. W. H. Jordan 63-70., ORNL — Y-12 Technical Library,
33, G. W, Keilholtz Document Reference Section
34, F. L. Keller 7L-77. Laboratory Records Department
35. C. P. Keim 78. Laboratory Records, ORNL R.C.
%6. J. J. Keyes 79-81. Central Research Library

149

EXTERNAL DISTRIBUTION

82-85. Air Force Ballistic Missile Division
86. AFPR, Boeing, Seattle
87. AFPR, Boeing, Wichita ,
88, AFPR, Douglas, Long Beach
89-91. AFPR, Douglas, Santa Monica
92-9%, AFPR, Lockheed, Marietta
9k, AFPR, North American, Canoga Park
95. AFPR, North American, Downey
96. AFPR, North American, Los Angeles
97-98. Air Force Special Weapons Center
99-100. Air Research and Development Command (RDZN)
101. Air Technical Intelligence Center
102. Air University Library
105-105. ANP Project Office, Convair, Fort Worth
106, Albuguerque Operations Office
107. Argonne National Laboratory
108. Armed Forces Special Weapons Project, Sandia
109, Armed Forces Special Weapons Project, Washington
110~111, Army Ballistic Missile Agency
112, Army Rocket and Guided Missile Agency
11%. Assistant Secretary of Defense, R&D (WSEG)
114, Assistant Secretary of the Air Force, R&D
115-120. Atomic Energy Commission, Washington
121, Atomics International
122-124, Bettis Plant (WAPD)
125. Brookhaven National Laboratory
126. Bureau of Aeronautics
127. Bureau of Aeronautics General Representative
128. BAR, Aerojet-General, Azusa
123. BAR, Chance Vought, Dallas
130. BAR, Convair, San Diego
131. BAR, Goodyear Aircraft, Akron
132. BAR, Grumman Aircraft, Bethpage
1%%. BAR, Martin, Baltimore
134, Bureau of Ships
135. Bureau of Yards and Docks
136-137. Chicago Operations Office
138. Chicago Patent Group
159. Director of Naval Intelligence
140. duPont Company, Aiken
141-148. General Electric Company (ANPD)
149-150. General Electric Company, Richland
151, Hartford Aircraft Reactors Area Office
152, Idsho Test Division (LARQOO) .
153. dJet Propulsion Laboratory
154, Knolls Atomic Power Laboratory
155. Lockland Aircraft Reactors Operations Office N

150 ——

156-157. Los Alamos Scientific Laboratory
158. Marquardt Aircraft Company
159. National Aeronautics and Space Administration, Cleveland
160, National Aeronautics and Space Administration, Washington
161, Naval Air Development Center
162. Naval Air Material Center
163. Naval Air Turbine Test Station
164, Naval Research Laboratory
165. New York Operations Office
166. Oak Ridge Operations Office
167. Office of Naval Research
168, Office of the Chief of Naval Operations
169, Patent Branch, Washington
170-171. Phillips Petroleum Company (NRTS)
172-175., Pratt and Whitney Aircraft Division
176. Sandia Corporation
177. Sandia Corporation, Livermore
178-179. School of Aviation Medicine
180. Thompson Products, Inc.
181. USAF Headquarters
182-18%. USAF Project RAND
184, U, S. Naval Radiological Defense Laboratory
185-186. University of California Radiation Laboratory, Livermore
187-199. Wright Air Development Center
200-224, Technical Information Service Extension
225. Division of Research and Development, Atomic Energy Commission,
Oak Ridge Operations

152

ORNL-528
ORNL-629
ORNL-T68
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-17T7L
ORNL-1816
ORNL-186L4
ORNL-1896
ORNL-1947
ORNL-2012
ORNL-2061
ORNL-2106
ORNL-2157
ORNL=-2221
ORNL-2274
ORNL-23%40
ORNL-2387
ORNL-2440
ORNL~-2517
ORNL-2599

Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period

Inding
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

Reports previously issued in this series are as follows:

November 30, 1949
February 28, 1950
Mey 31, 1950
August 351, 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
Juns 30, 1957
September 70, 1957
December 31, 1957
March 31, 1958
September 30, 1958

R