ORNL-1515.txt

| CENTRAL RESEARCH LIBRAR
|

DOCUMENT COLLECTION
- BT ' o~ Tt

- - 4456 0353310 2 [0 IR

fi_}"é/fi A
== 4
Sy
AIRCRAFT NUCLEAR PMOJECT

QUARTERLY PROGRESS REPORT

CENTRAL RESEARCH LIBRARY
DOCUMENT COLLECTION

LIBRARY LOAN COPY

DO NOT TRANSFER TO ANOTHER PERSON

If you wish someone else to see this document,
send in name with document and the library will
arrange a loan.

OAK RIDGE NATIONAL LABORATORY

OPERATED BY

CARBIDE AND CARBON CHEMICALS COMPANY

A DIVISION OF UNION CARBIOE AND CARBON CORPORATION

ucc|

POST OFFICE BOX P
OAK RIDGE. TENNESSEE

SEceN
SECURITY ENFORMATION

ORNL-1513

This document consists of 200 pages

Copy é;;—of 247 ecopires Series A

Contract No W-7405-eng-26

ATRCRAFT NUCLEAR PROPULSION PROJECT
QUARTERLY PROGRESS REPORT

for Period Ending March 10, 1953

R. C. Briant, Director
J. H. Buck, Associate Director
A, J. Miller, Assistant Director

Edited by
W. B. Cottrell

DATE ISSUED

R 8 1853

0AK RIDGE NATIONAL LABORATORY
. Operated by
CARBIDE AND CARBON CHEMICALS COMPANY
A Division of Union Carbide and Carbon Corporation
Post O0ffice Box P

0ak Ridge, Tennessee

MARTIN MARIETTA ENEAGY SYSTEMS LIBRARIES

A - .. oo -

g“x‘ V.W%M"" « 1k "‘g
Fl + * + 117
4Tk«

L
] ]
%‘ R e i i e i T P Ag‘
s LY

~ - d - £ W
xi"- - : " Wy i a4t T ey 3

‘ PRE A TErt 2 ST 54

Yy 2 RRA
AN T L T
OO =1 O\ U P LI R

PR UFPFOIAEEMSE=QDIITNPUION-oDIOR

o=

llaender

PEOPAEOOEFFEEOIPELOEUDEOOCOSRUDFEAIIZIARIOEOOTO

ZVUEHEIRETVEDD

< \‘k-

Adamson
Affel
Baldock
Barton
Bettis
Billington
Blankenship
Blizard
Bredig
Briant
Briggs
Bruce

Buck
Callihan
Cardwell
Cathcart
Center
Cisar
Clewett
Clafford
Cottrell
Cowen
Eister
Emlet (Y-12)
Ergen
Fraas

Gall

Graham

Grigorieff (Cj

Grimes

Household-,;
Humes (K—;5)
Kellholr;'
Keim 4

. Kelleg
ertes f/
. Kigf
L:f%on
;7v1ngston

pLyon

”
o

INTERNAL DISTRIBUTION

42,
43.
44,
45,
46,
47.
48.
49.
50

>

Asultant)

0 0 M > O MmUY

ORNL-1515
Progress
W. I Manly
L. g Mann
W. . McDonald
JL. Meem
AN J. Miller
Z. Morgan
J. Murphy
F. Poppendiek
M. Reyling
W. Savage
D. Shipley
S1sman
P. Smith (consultant)
H. Snell
L. Steahly
W. Stoughton
D. Susano
. A, Swartout
E.N\{l. Taylor
F.'ég Uffelman
E. RVanArtsdalen
F. C. §onderLage
J. M. Wgde
A. M. Wepberg
J. C. Whi¥g
E. P. Wign& (consultant)
H. B. Willa\g
G. C. Wallaa X\
J. C. Wilson W
C. E. Winters §

ANP Library

Biology Library

Central Files

Health Physics Libr#y
Metallurgy Library

Reactor Experimental
Engineering Library
Technical Information
Department (Y-12)
Central Research Library

111
1lv

99-101,
102412,

114-12
122,
123-127.
128,
129,
130-131,
132,
133,
134-138,
139-160,
161-164.
165,
166.
167-174,
175.
176-183.
184-185,
186-188,
189,
190,
191-193,
194-197,

198.
199-200,.
201-202,

203,

204,
205-206.

207,

208,
209.210.

211,
212-215.

Y
SR Y

EXTERNAL DISTRIBUTION

Air Force Engineering Office, Oak Ridge
Argonne National Laboratory (l copy to Kermit A
Armed Forces Special Weapons Project (Sandia)
Atomic Energy Commission, Washington
Battelle Memorial Institute
ookhaven National Laboratory
Buyeau of Aeronautics (Grant)
Bugkgu of Ships

Cala ornia Research and Developme
Ch1cag Patent Group
Chief of Naval Research

duPont d%Tpany 4
General Electric Company, ANP’

Hanford Operations Office
USAF-Headquar% rs, Office
Idaho Operations, Office f
Iowa State College 4

dderson)

"Company

gf Assistant for Atomic Energy
copy to Phillips Petroleum Co.)

\ 4
Knolls Atomic Power Lgboratory

Massachusetts I gtltute of Technology (Kaufmann)

to A, Sllv/rst.eln)
National Adv1sory Committee for Aeronautics, Washington
New York Operations Office
North Amefican Aviation, Inc,
NuclearBDevelopment Associates (QDA)
Patent Branch, Washington
Rand Cdrporation (1 copy to V. G. nning)

Savangah River Operations Office, Adjgusta

Vifro Corporation of America
W“étlnghouse Electric Corporation

216-232.

233-247,

Wright Air Development Center

2

— = N =

1

copies to B, Beaman

copy of Col. P. L. Hill

copy to Lt. Col. M. J. Nielsen

copies to Consolidated Vultee Aircraft Corporation
copy to Pratt and Whitney Aarcraft Divasaon

copy to Boeing Airplane Company

copy to K. Campbell, Wright Aeronautical Corporation

Technical Information Service, Oak Ridge, Tennessee

vl

Reports previously 1ssued in this series are as follows

ORNL-528
ORNL- 629
ORNL-768
ORNL-858
ORNL-919
ANP-60
ANP-65
ORNL- 1154
ORNL-1170
ORNL-1227
ORNL-1294
ORNL-1375
ORNL-1439

Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period
Period

Period

3
.,
4 "{.:

Endaing
Endaing
Endang
Ending
Ending
Ending
Ending
Endaing
Ending
Endaing
Endaing
Endang
Endang

-~
e 1 3":'}3“‘ f“l’.«""

YR . PR ™

November 30, 1949
February 28, 1950
May 31, 1950
August 31, 1950
December 10, 1950
March 10, 1951
June 10, 1951
September 10, 1951
December 10, 1951
March 10, 1952
June 10, 1952
September 10, 1952
December 10, 1952

TABLE OF CONTENTS

FOREWOBD L] * L ] L ] L] L - L] L ) L] . . . . . L - . L ] . L L ] . L ] L]

PART I. REACTOR THEORY AND DESIGN

IN'TRODUCT I ON AND SlJ.MMA-RY L] L] L] L] L] L ] L] . L] L] L ° * L] L] ] L ]

].c

2-

CIRCULATING-FUEL AIRCRAFT REACTOR EXPERIMENT , . ., .

Fluid Circult « & ¢ o « ¢ o ¢ o o « ¢ o o s « o o o
Stress Analysis . ¢ o ¢ ¢ s ¢ ¢ 3 e e s e e 4 0 e
Reactor « « v v ¢ ¢ ¢ ¢ o o o o o ¢ o« s o o o « o o
Instrumentation « « o « o ¢ o s o o o o« o o « « « o
Off-Gas System . ¢ ¢« ¢ ¢ ¢« « o« o ¢ o « o o o o o« o
Reactor Control ¢« & & ¢« ¢« ¢ ¢ ¢ ¢ o ¢ ¢ « o o o o
Electrical System « « o ¢« ¢ ¢ ¢« s o o ¢« o « o o o o

EXPERIMENTAL REACTOR ENGINEERING . . . . . . « . . &

Pumps for High-Temperature Liquids . . « « « « « &

Centrifugal pump with combination packed and frozen

for fluoraides . ¢« « « ¢« ¢« o ¢« ¢ o & . .
Allis-Chalmers centrifugal pump for l1qu1d metals
Laboratory-size pump with gas seal . . . « « . &
ARE-s1ze sump pump « « « « o« « ¢ o « o o o o o o

ARE centrifugal pump .« « ¢« ¢« « ¢ « ¢ ¢ o« o « o« o
Electromagnetic pump .« « « « ¢ « o« o o o o s o =
Rotating-Shaft and Valve-Stem-Seal Development . .
Mechanical face seals . ¢« v & ¢« ¢« ¢ ¢ ¢« & ¢ « & &
Combination packed and frozen seal with MoS, .
Combination packed and frozen seal with graphite
Graphite packed seal ., . . .« ¢« & ¢ ¢ ¢ « ¢« ¢ o« &
Rotating-shaft seal test . ¢« « « ¢« ¢ ¢ ¢ ¢ & o« &
Packing penetration tests « « o « o « o o o o o

Frozen-sodium and frozen-lead seal for NaK . . . .

ARE valve test .« ¢ o ¢ o« o « ¢ o « o o o o o o
Heat Exchanger Systems .+ . ¢« ¢« 4 o ¢ o o ¢« ¢ « o &
Sodium-to-air radiator tests « o+« o« « o o ¢ ¢ o
Bifluid heat transfer loop .« « « ¢ ¢ ¢ ¢ ¢ o o &
ARE Fuel Circuit Mockup « « ¢« o ¢ s ¢ ¢ ¢ o ¢ o« « &
Vortexes and bubbles .« ¢« ¢« ¢« ¢« ¢ o ¢ o ¢ o o o &
Operational instabilaities « + ¢« &+ ¢ ¢ ¢ o ¢ o o &
Instrumentation « « « « ¢ « o o ¢ s ¢ s o ¢ s & o
Pressure measurement . . . . « o v s v s 6 e »
ARE leak detection i1ndicator tests . « « « o o« o

seal
e o o
« o s
.
« s o n
« o e
o« o
. o e
e e
.« .

Vil
[
L ]

Slip rings for temperature measurement . o o o o o« o o o o
Rotameter type of flowmeter « 4« o « o « « « o o o « o o o &
Handling of Fluorides and Liquid Metals . . . & ¢« « &« o o o .
Distazllatzon of NaK & o & & v v ¢ v v 4 v v v v 4 v o o o
Flame tests for NaK vapor 1n helium « + « & v & ¢ & « o o« &

3 L ] BEACTOR PHYS ICS . Ll L] L] [ ] Ll [ ] . L] L] L] L] L] L] L] L] L] L] . . L] L]
4. REFLECTOR-MODERATED CIRCULATING-FUEL REACTOR . . . « & +« « «

Stat1c Physics & v 4 v 4 v 4 4 b e e e e e e e e e e e e e
Reactors with various reflector-moderators . . . + « o« o &
Reactors with beryllium reflector-moderators . . « « « o &
Summary tables and graphs . « v ¢ ¢« ¢ ¢« ¢ ¢ 4 e 0 4 e .

Critical Experiments .« « ¢« o 4 ¢ ¢ o o « o o o o « o o o s
First critical assembly . & ¢« ¢ ¢« 4 ¢« ¢ ¢ o o « o o o o s o
Second critical assembly . . ¢« ¢« ¢« ¢ ¢« ¢ ¢ 4 0 0 e 0 o .

Design Characteristics « « o+ o « o « o« o o s « o o o s o o o
Factors affecting core diameter « . « « « o« « « o« « « « o« 4
Temperature, pressure, and StTeSS « o+ o« « o« o o « o o « o &«
Reflector heating « + ¢« 4 & ¢ & 4 ¢ ¢ ¢ ¢ ¢ o o o o o o o o

Shi1elding o o 4 4 4 ¢ ¢ ¢ ¢ ¢ o o o o ¢ o o o o ¢ o o o s o

Power Plant Des1gn « ¢« ¢« &« ¢« 4 4 ¢ ¢ ¢ o ¢ « o o o o o o o «

PART II. SHIELDING RESEARCH
INTRODUCTION AND SUMMARY . . . & & & ¢« v ¢ ¢ o o o o ¢ « ¢ o « « o 4
5. LID TANK FACILITY [ ] [ ] [ ] . . . Ll L] - L] - ] L] * L ] L} * L} L] L] * . L ]

Effective Fast-Neutron Removal Cross Sections « « « « « o o &
Mockup of the Unit Shield of the Reflector-Moderated Reactor

6. BULK SHIELDING FACILITY . v 4 ¢ ¢ ¢ ¢ o o« o « « o o o ¢ o o o« s
Air-Scattering Experiments . . ¢ ¢ ¢« ¢ ¢ 0 0 4 0 e e e e
Irradiation of Animals &« ¢ ¢ o 4 & o ¢ o o o o o o « o s o o
Neutron Spectroscopy for the Divided Shield . . . . . . « . .
Gamma Spectroscopy for the Divided Shield . . . . . « « « . .
Fission Energy and Power in the Bulk Shielding Reactor . . .

7 L] TOWER SHIE]-JDING FACILITY L] L] L ] - L] * - . e L] . - L] e L] L] L] - L]

Tower Des 1gn L ] [ ] L ] L L ] [ ] L ] L ) L ] L ] L ] L ] L ] L ] L ] L ] [ ] L L ] * L ] L ] - a
Construction Schedule « « ¢ ¢ ¢ % o « « o« o « « o s s o s o

8 L ] NUCLEAB thASI.IREMENTS L] -* L ] -* L ] L ] L ] L] - . * L] L ] L ] [ ] L ] L] L] 9 L] k]
Fast-Neutron Scintillation Spectrometer . .+ « « « « « « « o o
Measurements with 6-Mev Van de Graaff . . ¢« ¢« ¢« ¢« ¢+ ¢« ¢« « o &

vill

36
36
36
36
37

39

41
41
44
46
51
53
53
36
61
64
66
69
74
79

87

89
89
89

91
91
91
92
92
92

97
97
97

101
101
101
PART II1., MATERIALS RESEARCH

INTRODUCTION AND SUMMARY . . ¢ v & ¢ 4o ¢ v o ¢ « « « o o &

9.

10,

CHEMISTRY OF HIGH-TEMPERATURE LIQUIDS ., . . . « . . .
Fuel Mixtures Containing UF, v ¢ v v v v v « v o
LiF-ZrF, -UF, . . ¢ v v v v 6 v v e e e e 0 e
KF-LiF- BeF -UF ¢« s v e e 4 e e e e e e e e
Fuel M1xtures Contalnlng ThF « o e s 0 s s e e 4 e
Fuel Mixtures Containing UCl s o s s = % s & s a
NaCl-UCl, . & ¢ & 4 v v v v o v o o o o o s o o
KC1-UCl, . . .. . .. s e e e & s e s e s e o
Fuel Mlxtures Containing UF e s s 4 s e e e e e s
Addition of Reducing Agents to ARE Fuel . . . . . .
Coolant Development « ¢« « « « « « & o « « o « o o
LiF-ZrF, ¢ ¢ v v v v v i e e e e e e e e e
CsF-ZrF, . ¢« 4 ¢« v v v 0 v 0 v 0 e e vt e e e
KF-LlF-ZrF4 s 5 @« * e & 8 8 & 4 e ® s = & 9+ o+ o4 .
RbF-LiF-ZrF, . . ¢ ¢ ¢ v ¢ v ¢ v v 4 ¢ ¢ o 0 o o
KF-LiF-BeF, . . . . . s 4 4 e 4 s e e e e s
Production and Purlflcatlon of Fluoride Mixtures .
Laboratory-scale fuel preparation « « « « « « « &
Pilot-scale fuel purification « « « ¢ « « ¢ o + &
Fluoride production facility ., . « ¢ ¢« « « « « .
Hydrofluorination of Zr0Q,-NaF mixtures . . . . .

CORROSION RESEARCH . . & ¢ ¢ ¢ ¢ ¢ ¢ « ¢ o o « o o «
Fluoride Corrosion of Metals i1n Static, Seesaw, and
Rotating Tests « ¢ ¢ ¢« ¢ ¢ o o ¢ o o o o o o«
Crevice COrroS10n o« o « o « ¢ o o o o« o o o o « o
High-temperature pretreatment of Inconel . . , .
Effect of time of exposure of Inconel to fluoraide
mixture with and without ZrH, added . . . . .
Structural metal fluoride additives « « « 4+ « o
Hydrogen fluoride additive . « « ¢ « o« o« « o &
Fluoride corrosion i1n a rotating rig . . « « o+ o
Static tests on Incoloy and Inconel in fluorides

Fluoride Corrosion of Metals i1n Thermal Convection Loops

Effect of fluorade batch s1ze « « ¢« « « & « « o &
Fluoride pretreatment . « o« o o« o o o o o o o o o
Hydride additives o+ + 4o ¢ ¢« ¢« ¢ ¢ o ¢ o & « o o «
Metal addatives o« o o & ¢ ¢ ¢ ¢ ¢ ¢ o « o o o o o
Temperature dependence . ¢ ¢« ¢« ¢ ¢« ¢ « ¢ o o o
Crevice COTTOS10N « ¢« « « ¢ « « o o « o ¢ o « o o
Inserted corrosion samples .« o « o « o+ & o o o &

23

105

107
107
107
107
107
108
109
109
109
110
112
112
112
112
112
113
113
113
114
115
115

117

118
118
118

118
119
120
121
121
121
121
122
123
123
123
124
124

1X
11.

.

—w- —

Effect of exposure time . . . . . . .
Nonuranium bearing mixtures . . . . .

Liquid Metal Corrosion of Structural Metals .

Seesaw tests of sodium-lead alloys .
Static tests with lathium . . , . . .
Incoloy 1n sodium, lithium, and lead

Corrosion by lead i1n thermal convection loops « « « « « . &
Corrosion of Ceramics by Fluorides and Liquid Metals ., ., . .

Cermets in fluorides and liquid metals

Static tests of the ARE-BeO blocks in Na and NaK . . . . .

Seesaw tests of BeO in NaK . . . . .

Convection loop tests of BeO in NaK . . . . . . . . . . . .

SOIUblllty Of Beo in NaK « * e & e »

Effect of Atmosphere on the Mass Transfer of

N].Ckel in HydrOXIdeS . . . . " e .
Fundamental Corrosion Research . . . .

Identification of corrosion products from
Preparation of special complex fluoraides

Air oxidation of fuel mixtures . . .

METALLURGY AND CERAMICS . . . . . « « . .
Welding and Brazing Research . . . . .
Cone-arc welding . +« « « ¢« « « o« .« .
Fabrication of heat exchanger units ,
Brazing of copper to Inconel . . . .
Evaluation Tests of Brazing Alloys . .

Corrosion of brazing alloys by fluoraides
Effect of brazing time on joint strength

Effect of spacing on brazed joint strength

L] [ ] [ ] L] * L] L] L ) .

convection loops

Strength of brazed joints with various base metals ., . . .

Creep-Rupture Tests of Structural Metals
Preoxidized Inconel in argon ., . . .
Peened Inconel 1in argon « « « « . . .

Effect of Inconel grain size on time to rupture . . . . .+ .
Effect of environment on strain vs. time curves for Inconel
Type 316 stainless steel tube burst tests

Fabrication of Control and Safety Rods
Control rods for the G-E R-1 Reactor
Tower shielding facility safety rods

Columbium Research .« + « v & v o o « &
Gaseous reactions ¢+ s 4 & e e 4 s
Ox1dation & « « ¢ o o o « o o « o o 4

Fabrication of Metals . . . . « ¢« « « .
High-conductivity metals for radiator

fins

in argon s e e

SOlld‘phase bondlng * ¥ ¢ 8 e ® & ¢ 8 e &6 4 4 e+ & s e e e =

124
124
125
125
126
126
128
129
129
131
132
132
133

136
136
136
137
137

139
140
140
140
142
143
143
143
144
145
145
145
146
147
147
148
148
148
150
150
150
151
152
152
152

Extrusion of high-puraity tubing . . « « ¢« ¢« ¢ v ¢« ¢« ¢« ¢« ¢ « o & 152
Hot-Pressed Pump Seals . ¢ ¢« & ¢ ¢« ¢ ¢ ¢ ¢ ¢ ¢ ¢ o ¢ o ¢ « o o s 153

Ceramlcs € 8 @ e & © 4 e @& 4 8 & %8 & 4 8 & 8 8 6 8 e e s e e = s 154
Ceramlc Coatlngs fOI‘ Shleldlng s 8 & 8 & 8 & & & 8 8 4 & & =8 154
Development of cermet fuel elements « « ¢« ¢« « « o o ¢ o« o o o« o 154

Reduction of porosity of ARE beryllium oxide . « ¢ ¢ o ¢ & & & 155

12, HEAT TRANSFER AND PHYSICAL PROPERTIES . . . ¢ ¢« ¢ ¢ ¢ ¢ ¢ o« o« s o 157
Thermal Conductivity of Liquids + ¢ ¢ ¢ ¢« ¢ ¢« o « ¢ ¢ o o o « o « 157

Heat Capacity of Liquids . . . . ¢ s s s e e s e e e 157
Viscosity and Density of Alkala Hydrox1des c t s e a4 e et e 158
Vapor Pressures of Fluorides .« o « ¢ ¢ « o « o o ¢ o o o « « « & 159
Prandtl Moduli of Various Materials . « ¢« ¢« ¢ ¢« ¢« ¢ ¢ ¢ « ¢ ¢ o« « 160
Specific Reactor Heat Transfer Problems . . . . « ¢« ¢« ¢« ¢« « « ¢« & 160
Turbulent Convection 1n Annuli & & ¢ 4+ ¢ ¢« ¢« v 4 ¢ ¢ o ¢ o o o o 160

Clrculatlng“Fuel Heat TranSfer *+ 8 # 4 8 8 8 e 8 s 8 5 8 8 s s » 162
High Temperature Reactor Coolant Studies + « + « o ¢ ¢ o ¢ o o & 163

13 * BADIATION DAMGE L] * * L] L ] * - . * L ] L ] * L ] . L ] L ] a L ] L ] . L ] L L] . L ] 1 65

Irradiation of Fused Materials . & « « ¢ o o o o o s o o o s o & 165
In-Reactor Circulating Loops « « « & « ¢ o o o o o o s s o o o & 167
Creep Under Irradiation . « « ¢ o o v ¢ o o o o o o s o o o o s 167
14. ANALYTICAL STUDIES OF REACTOR MATERIALS . e e e e e e e e e e 171
Chemical Analysis of Reactor Fuels and Contamlnants e s 8 s e e e 171
ZITCONIUM o o s o o o « o o o o o o s s s s s s s o o s o s o s 172
Reduction products « « o « o « o o o ¢ o s o o o o s 2 o o o o 172
Ox1des & v ¢ & o 4 ¢« o 4 e s e s e s e e s e e s e e e e e e 173
Hydrogen fluoride . . ¢« & ¢ ¢ ¢ ¢ 4 ¢« ¢ ¢ o o o o o s o s s o = 173
Corrosion products « « « o ¢ s o s o o o s o o o o o o 4 o o 175
Preparation of Uranium Tetrachloride . « &« « ¢ ¢« o« ¢ ¢ ¢ o « o & 176
Petrographic Examination of Fluorides . « « ¢« ¢« ¢« ¢« ¢ ¢« ¢« ¢« « « « 176

UCl, -NaCl system .+ & ¢« ¢« o o o o o o o 2 o o o o s o s o s o s 176
UC1,-KCl System o« « o « o o o o o o o o o s o « o o o o o o o & 176
X-Ray Daffraction Studies . « ¢« « ¢ ¢ o ¢ o ¢ ¢ s o s o s o o s 177
Service Chemical Analyses 4 4 ¢ ¢ o« ¢ o o o s « o s s o o s s o« & 177

PART IV, APPENDIXES
15. LIST OF REPORTS ISSUED DURING THE QUARTER . . . . . ¢« ¢« « ¢« « &« &« & 183
TECHNICAL ORGANIZATION CHART OF THE ANP PROJECT . . ¢ ¢ ¢ ¢ « « ¢ « & & 187

ANP PROJECT QUARTERLY PROGRESS REPORT

FOREWORD

This quarterly progress report of the Aircraft Nuclear Propulsion Project
at OBRNL records the technical progress of the research on the circulating-fuel
reactor and all other ANP research at the Laboratory under 1ts Contract
W-7405-eng-26. The report 1s divided into three major parts I. Reactor
Theory and Design, II. Shielding Research, and III. Materials Research. Each
part has a separate introduction and summary.

The ANP Project 1s comprised of about three hundred technical and scientific
personnel engaged i1n many phases of research directed toward the nuclear
propulsion of aircraft. (The Project organization chart 1s included as an
appendix.) A considerable portion of this research 1s performed in support of
the work of other organizations participating in the national ANP effort. How-
ever, the bulk of the ANP research at OBNL 1s directed toward the development
of a circulating-fuel type of reactor.

The nucleus of the effort on circulating-fuel reactors 1s now centered upon
the Aircraft Reactor Experiment - a 3-megawatt, high-temperature prototype of
a circulating-fuel reactor for the propulsion of aircraft. This reactor
experiment 1s now being assembled, 1ts current status 1s summarized in sec. 1.
However, much supporting research on materials and problems peculiar to the
ARE will be found in other sections of Part I and Part III of this report,
along with the general design and materials research contained therein.

A survey report on a reflector-moderated type of circulating-fuel reactor
1s contained 1n sec. 2. Since the feasibility of such a reactor was indicated
by critical experiments completed during the previous quarter, considerable
research, as well as design effort, has been devoted to problems associated
therewith.

The shielding research report, Part II, 1s devoted almost entirely to
problems of aircraft shielding.

e
CABERE Rt Bidind g

ay _—
;:4 ek L ‘,fi
gi X y BT g e SRR AL W
4 geaent W% A '%II“
¥ k3 - y - » Lfli T oy N +
"3" A Pty ™ - s r
Part |

REACTOR THEORY AND DESIGN
INTRODUCTION AND SUMMARY

The Aircraft Reactor Experiment
(sec. 1) 1s rapidly taking form as
equipment 1s received and installed
in the ARE Building. There were no
significant changes 1n design or
concept during the quarter, and the
installation 1s proceeding according
to schedule. The fuel system heat
transfer loop and the fill and flush
tanks are completely installed, and
the sodium heat transfer loop 1s
nearing completion. Installation of
fuel and sodium piping 15 now 1n
progress. The control and graphac
panels have been completely instru-
mented and checked. The formal
presentation on the hazards involved
in the ARE was made to the Reactor
Safeguards Committee earlier in the
quarter, and in view of the subsequent
discussion of the Committee members,
1t 1s not expected that any modifi-
cations of the experiment will be
required.

Valves, pumps, i1nstrumentation,
and other components of the fluoraide-
fuel and sodium-coolant systems are
being developed and tested for the
ARE (sec 2). Both centrifugal and
electromagnetic pumps of ARE capacity
have been operated, the latter will
be employed i1n the by-pass loop for
preliminary tests, with NaK, of the
external fuel system. Impeller designs
for the centrifugal pumps are con-
ventional, but the pump seal, particu-
larly -for fluoride fuels, remains a
problem. The packed-frozen seals on
fuel pumps have widely varying leakage
rates, and adequate temperature
control of the frozen zone has not been
obtained. Both packing and face-seal
materials are being examined 1n an
effort to find materials with the
desired service and life charac-
teristics. Gas face seals for vertical-
shaft pumps are being investigated.
Considerable success has been experi-
enced with frozen-sodium seals 1n
sodium pumps, and frozen-lead seals

are being tested for use with fluoraides.
Frozen-sodium and frozen-lead seals
have proved to be unsatisfactory for
use on NaK pumps. The ARE bellows
type of valve was tested at 1100 and
1300°F and found to be satisfactory,
although excessive leakage occurred
after operation at 1500°F. The 1initial
mockup of the ARE fluid circuait
revealed instabilities and gas entrain-
ment that were elaiminated by mainor
system alterations. The rotameters,
level indicators, and pressure trans-
mitters have operated satisfactorily
for long periods of time under the
conditions expected in the Aircraft
Reactor Experiment.

The reactor physics studies (sec.
3) were concerned with the statics of
the reflector-moderated reactor and
damping of power oscillations. The
applicability of the multigroup method
to the calculation of the reflector-
moderated reactor was demonstrated by
the good correlation obtained waith
critical experiments. The damping of
power oscillations 1n a reactor 1s
demonstrated for a nonlinear case,
with delayed neutrons included, by
assuming (1) constant power extraction
or (2) a special case of cooling with
circulating fuel.

The reflector-moderated reactor
and an aircraft power plant assembly
employing this reactor are described
(sec. 4). In general, the reactor
consists of lumped regions of fuel
and moderator or reflector. An
essent1al element i1n this reactor 1is
the use of a thick beryllium reflector
that not only contributes to neutron
economy but also constitutes the farst
layer of the aircraft shield. This
reactor and the compact, spherical
arrangement permit realization of
what 1s practically a unit shield for
a 200-megawatt reactor and heat
exchanger combination that weighs
about 80,000 pounds. Craitical experai-
ments on a mockup of thais reactor
ANP PROJECT QUARTFRLY PROGRESS REPORT

assembly indicate a craitical mass of A preliminary weight estimate of the
about 20 1lb and a total uranium in- complete power plant assembly, in-
vestment of 100 lb when the compact cluding the turbojets, 1s about 115,000
external fuel system 1s 1included. pounds.

PERIOD ENDING MARCH 10, 1953

1. CIRCULATING-FUEL AIRCRAFT REACTOR EXPERIMENT
J. H Buck, Research Director’'s Division, E. S Bettis, ANP Division e

The major effort in the ARE project
during the quarter has been 1n 1instal-
lation of equipment in Building 7503.
The arrival of the heat exchangers
permitted the i1initiation of a large
amount of construction work that had
been postponed for several weeks
Valves began to arrive toward the end
of the quarter, and work could proceed
on the i1nstallation of the maain
plumbing system. The installation of
equipment will no longer be dependent
upon the arrival of components. The
installation work 1s not ahead of
schedule, but thus far 1t appears
that the completion goal of June 1953
will be realized

No new technical crises have arisen
1n the project. Tests of valves,
seals, pumps, etc. continue (sec. 2)
and, to date, reveal no insurmountable
difficulties. Seals have operated
satisfactorily, but their performance
has not been so good as might be
wished. The same condition exists
with respect to the valve tests, but
the valves have been shown to be
usable.

Although detailing of installation
drawings continues to be a problem,
1t has not yet actually retarded the
job. There have been the usual
modifications dictated by problems
arising i1n the field, but they have
been no more numerous or serious than
would be expected 1n a system of such
complexity as the ARE.

FLUID CIRCUIT

G. A. Craisty
Engineering and Maintenance Division

No significant design changes were
made in the fluid circuit. However,
since 1t appeared 1nadvisable for the
long drain lines 1n the fuel circuat
to stand full of stagnant fuel, two
valves were added to permit draining

An

these long lines so that they w¥Il
be empty during operation of the
experiment, ot

The heat disposal loops for the
fuel circuit, that 1s, the fuel-to-
helium heat exchangers, the helium-to-
water heat exchangers, the thermal
barriers, the helium ducts, and the
helium blowers, are now completely
installed. Figure 1.1 shows these
loops 1nstalled 1n the pat. Three
sections of the fuel pipe, together
with annuli, thermocouples, and
heaters (where possible), have been
prefabricated and can be welded 1nto
the system when the valves for these
lines are ready. Figure 1.2 shows one
of these sections of fuel pipe.

All the fill and flush tanks have
been put 1n the pit, and work has
been started on installing heaters on
these units. These tanks are shown
in Fig. 1.3.

The by-pass, which enables the
system to be run without the reactor
in the circuit, has been designed.
The electromagnetic pump, which wall
be i1nstalled 1n this by-pass for
circulating liquid metal, has been
tested by the Experimental Engineering
Group (sec. 2).

The surge tanks have been completed,
and tests of the mockup of the hydraulic
system are continuing (sec 2).

The helium-to-water heat exchangers
for the sodium coolant caircuit have
been received and are being installed.
The li1quid metal-to-helium heat
exchangers are to be delivered the
first week of March.

STRESS ANALYSIS

R. L Maxwell J. W. Walker
Consultants, ANP Division

The reactor by-pass circuit was
checked to see that previous pre-
stressing calculations would not be
ROD - COOLING]
DUCTING

IHEUUM DU

BARRIER MOTOR|
¢ (RAISE -LOWER)! HELIUM- TO-WATER

g4 HEAT EXCHANGERS

HELIUM BLOWER

THERMAL BAF?RVERS

£
FUEL TO-HELIUM L/
HEAT EXCHANGERS

140434 SSIYO0Hd ATHALYVNO LOAf0¥d dNV

Fig. 1.1.

Fuel System Heat Disposal Loops.
Fig.

1.2,

Fuel-Pipe Section.

UNCLASSIFIED
PHOTO 10852

d

INSULATION

0N
CLAM SHELL
ELECTRIC HEATERS

‘0T HOYVW ONIGNI doI¥dd

€561
P ‘UNCLASSIFIED
H g PHOTO 10907
TAL

Fig. 1.3. Fill and Flush Tanks.

LYOdA SSAYI0Ud ATHALYYND LDAf0¥d ANV
invalidated by 1ts 1ncorporation in
the circuit A report on the complete
analysis of the stresses 1s beaing
prepared

A design has been developed for
1ncorporating strain gages 1n the
system at the anchor points to check
the computed stresses These gages
w1ll be 1nstalled prior to the pre-
stressing, and the magnitude of the
prestressing forces will be measured
The stress relief as the temperature
1s raised will be measured until the
temperature reaches about 200°F,
above 200°F the gages will no longer
be usable

REACTOR

The reactor 1s ready for operation,
except for welding of the serpentine
coils and final assembly. Figures 1 4,
1.5, and 1.6 show the "status of parts
of the reactor All preliminary work
on the pressure shell has been com-
pleted The coi1ls have been received
and sample coi1ls are now being test
welded 1n ORNL shops It 1s planned
to assemble the reactor i1nto a test
circuit at the site and to circulate
li1guid metal through the fuel coils
at desi1gn temperature so that the
reactor can be i1ndependently checked
prior to 1ts i1nstallation Thais
pretesting of the reactor core will
also serve to clean the fuel passages
with hot liguid metal prior to 1ts
1ncorporation 1nto the system,

INSTRUMENTATION
R. G. Affel, ANP Division

The only real design modification
in the i1nstrumentation 1nvolved a
change 1n the leak detection circuit.
The change was a simplification to
minimize trouble without changing the
principle of operation of the system.
All the control room instruments have
been mounted. The graphic panel has
been 1nstalled and checked with a
dummy setup of pneumatic switches
that simulated valve operation.

PERIOD ENDING MARCH 10, 1953

Pneumatic lines have been run to
valves and sensing points 1n th'%
pits, but a small amount (2%) of work
remains to be done. Thermocouple
designs have been completed and orders
have been placed Several of the
thermocouples needed to allow instal-
lation work to proceed have been made
by the Instrument Department All
instruments and actuators have been
labeled The enunciator panels have
been 1nstalled and checked, except
for connection to the sensing points.

OFF-GAS SYSTEM

There has been no change 1n the
off-gas system, and installation in
the building 1s practically completed.
The final design for the shielding of
the radiation monitors has not been
completed, but this shielding will be
quite simple and will be assembled 1n
the field by using a stacking of lead
bricks around the monitors

REACTOR CONTROL

The model fission chamber for use
at high temperature has operated
satisfactorily for several hundred
hours at a temperature of 1400°F
This test has proved the acceptability
of the i1nsulator for this chamber.
The live chamber, with a U?3?% sleeve,
has not yet been tested.

The control room relay panel has
been completed. The wirang of the
console and the control board instru-
ments console has progressed as far as
the i1nterconnection terminal blocks
No interconnecting cables have been
run between the amplifier cabinet,
the relay cabinet control board, and
the control actuator assembly. This
work had just started at the end of
the quarter.

The mechanism for control rod
actuation 1s being assembled 1n the
crane bay area for testing. All
parts for this assembly are complete
and on hand i1n Building 7503, but no
signi1ficant progress had been made 1n
the assembly.

11
14

b | 235

in. 0D

PHOTO |o!95|

LY0dAd SSAYI0Yd ATHALYVNO IDAL0¥d dNV
PERIOD ENDING MARCH 10, 1953

REFLECTOR COOLANT
INLET

SAFETY ROD HOLES)

BOTTOM OF|
PRESSURE SHELL:

ITEMS NUMBERED
1 THROUGH 6 ARE 3
FUEL OUTLET LINES

Fig. 1.5. Reactor Pressure Shell.

13

R

ANP PROJECT QUARTERLY PROGRESS REPORT

P&o 10899)

WELD JOINT TO
PRESSURE SHELL

o I r—
.
'

FUEL INLET! ‘
HEADER /

L. 1
REFLECTOR COOLANT|

14
ELECTRICAL SYSTEM

”

The electrical work 1s proceeding
satisfactorily. The large job of
completing the heater circuit panels
has been accomplished. Heaters are
being installed on the fill and flush

PERIOD ENDING MARCH 10, 1953

tanks, and leads have been run from
the control panels to two of the faill
and flush tanks. Both motor generator
sets have been received and checked.
It was found that these sets raised
the noise level 1n the basement tp an
untolerable level, consequently they
w1ll be moved out of the building.

15
2. EXPERIMENTAL REACTOR ENGINEERING

H. W. Savage, ANP Division

The work of the past three months
on the liquid circuits of the ARE has
been devoted to minimizing corrosion
to values acceptable for 1000 hr of
continuous or 1ntermittent operation
at elevated temperature, determining
and applying a process of blending
fluorides for the ARE that will yield
a product with a quality equal to that
of the mixture used i1n the most satis-
factory corrosion tests, providing
pumps for the circulation of fluorides
and ei1ther sodium or sodium-potassium
alloy, developing seals for the pumps
that are reliable for 1000 hr of
operation, providing other circuit
components, and studying circuit
modifications that will assure accepta-
ble and operable fluid dynamics.
This work was carried out cooperatively
with the Metallurgy Division, the
Materials Chemistry Division, and the
ARE Section of the ANP Division,

Both centrifugal and electro-
magnetic pumps of ARE capacity ~ 50 to
100 gpm - have been tested. The
conventional pump i1mpellers are satis-
factory, but seals for mechanical
pumps are still a problem, A 100-gpm,
a-c, double-cell, electromagnetic pump
has proved to be entirely satisfactory
for pumping sodium-potassium alloy at

1500°F, and 1t 1s being tested for
pumping sodium,
Packings for horizontal-shaft

centrifugal pumps have not yet been
proved reliable. The most promising
packing was graphite and/or MoS,
retained with braid or graphite rings,
or possibly by magnetic means. Leakage
rates have varied widely, and friction
demands, 1n some cases, have been
excessive. Thermal control of the
frozen zone, needed for low-friction
operation and for remotely stopping
and starting the pump,
been achieved.

has not yet

O1l-lubricated, low-temperature,
face-type, gas seals have been reliably
operated for thousands of hours in
high-temperature, centrifugal, sump-
type pumps. An ARE-size sump pump,
similarly sealed, 1s undergoing a
water test, and there have been no
major difficulties, to date.

Frozen-sodium or frozen-lead seals
for NaK have proved to be impractical,
Frozen-sodium seals for ARE-s1ze sodium
pumps (2 1/2-1n.-d1a shaft) are now
being tested, and the difficulties
experienced to date appear to have
been due to thermal distortions,
Frozen-lead seals have been used
satisfactorily with fluorides, and in
several tests with molten fluoride 1in
contact with molten lead, there has
been no mixing of the fluoride and the
lead. High-temperature face seals for
use with molten fluorides are being
lnvestigated,
available.

Other components that have been
operated successfully for long periods
are high-temperature rotameters and
high-temperature liquid-level 1nda-
cators (both the rotameters and the
liquid-level 1ndicators use variable-
inductance pickups for signal gener-
thin-walled (0.005 1in.)
bellows 1n pressure stransmitters 1in
contact with fluorides, a pneumatically-
actuated, bellows-sealed, Stellite-
seated valve, and the slip-ring and
brush arrangements for removal of the
thermocouple signals used for measuring
temperatures 1n moving parts.

A study of the parallel circuits
and the pumping and expansion tank
arrangements of the ARE has revealed
operational steps that will have to be
taken 1n the event of failure of one
of the pumps so that loss of circu-
lation from the remaining pump and
expansion tank can be prevented. The

but data are not yet

atlon),

17
ANP PROJECT QUARTERLY PROGRESS REPORT

study also revealed the need for
piping modifications to eliminate gas
entrainment at the free surface in

the surge tanks.
Sodium-potassium alloy has been

distilled from closed systems to
simulate circuit cleaning of the ARE,
negligible quantities of alloy remained
in 1solated pockets.

PUMPS FOR HIGH-TEMPERATURE LIQUIDS

W. B. McDonald A. G. Graindell
W. G. Cobb G. D. Whaitman
W. R. Huntley A, L. Southern
J. M. Trummel P. W. Taylor

ANP Divisaion

Centrifugal Pump with Combination
Packed and Frozen Seal for Fluorides
A 50-gpm pump with a Stellite No. 6
coated, 2 1/2-1n.-di1a shaft was
operated for 85 hr with the ARE fuel,
NaF-ZrF,-UF, (50-46-4 mole %). The
seal consisted of a packing of 1/4-by
1/4-1n. monel braid wrapped in nickel
foi1l., Operation was unsatisfactory
and the test was terminated because of
shaft seizure i1n the packing area.
Examination of the shaft after the run
showed that very severe scoring had
occurred. Some grooves were 0,049 1in.
deep.

It may be concluded from this test
that small packings alone are
adequate. Compressive loading to
minimize leakage will be high and waill
result 1n excessive wear because the
poor penetration of liquid fluoraides
will prevent their acting as a sort of
lubricant. The packing i1n subsequent
tests 1ncluded braid as a component
and either graphite or MoS,, each of
which 1s a lubricant,

The pump has been repacked waith
Dixon’s Microfyne flake-graphite
powder, which 1s retained by close-
fitting, solid; APC-graphite rings at

in-

each end of the packing gland., Initaial
dry runs, without fluorides present,
showed this packing to be quite
sensitive to compressive loadaing.
Approximately 35 psi was found to be a

18

maximum packing loading for smooth,
dry operation without the seal heating
excessively. The repacked pump has
now operated at a shaft speed of 700
rpm for 300 hr at 1200°F, The pressure
against the seal has been varied from
5 to 18 psi. Leakage of solid fuel
from the seal has varied from 60 to 75
g/day during operation with pressures
in the 5~ to 10-ps1 range. Power
1input to the driving motor has been
quite smooth, with a variation from
2.6 to 3.0 kw, and under these con-
ditions only 1.5 to 1.8 kw 1s being
lost as friction in the pump seal,
Alli1s-Chalmers Centrifugal Pump for
Liquid Metals. The Allis-Chalmers
hydrostatic-bearing pump previously
reported(!) was shut down during the
quarter for alterations. A new, one-
piece shaft has been fabricated, the
bearing surface on the i1mpeller has
been hard coated with Stellite, and a
frozen-sodium gas seal has been 1n-
corporated i1n the pump. All fabri-

cation 1s complete and the pump 1is
ready for reassembly and further
testing.

Laboratory-Size Pump with Gas Seal
The laboratory-size gas-sealed cen-
trifugal pump operated 996 hr waith
fluoride fuel, NaF-ZrF,-UF, (46-50-4
mole %), at a temperature of 1500°F.
The pump has thus far operated a total
of 2300 hours. The pump was run at
3600 rpm and produced a 50-ps1 head
at 8 to 10 gpm.

Figure 2.1 shows the upper assembly
of the pump after 996 hr of operation,
The ZrF, condensation does not appear
to be excessive, however, 1in the
original assembly, probe shorting was
serious and the two probes in the pump
were rendered 1noperable after 50 hr
of operation, The shorting may have
been a result of fuel freezing around
and bridging the probe and radiation
shield assembly, shown 1in Fig. 2.1,
when an i1nadvertent liquid-level surge

(l)ANP Quar
1439, p 19

Prog Rep Dec 10, 1952, ORNL-
PERIOD ENDING MARCH 10, 1953

UNCLASSIFIED
PHOTO 6-6074

Fig. 2.1. Gas-Seal Pump Assembly After Operation.

19

ANP PROJECT QUARTERLY PROGRESS REPORT

occurred during operation or fillaing.
This condition has been alleviated by
1increasing the clearance between the
probes and radiation baffles and in-
serting a half cylinder in the cutaway
section 1n the baffle assembly,

The pump operated gquite satis-
factorily without probes, and the
liquid level 1n the pump bowl was
established by noting the pump per-
formance and temperature of the
face plate. The reason for termi-
nation was again a broken quide vane
1n the discharge bowl. The vanes have
been rewelded tothe bowl several times,
but the latest failure occurred in the
parent metal of the vane rather than
in the weld. The reliability of this
assembly can be greatly increased by
employing thicker vanes and adding a
reinforcing ring around the top edge
of the six vanes,

An 1dentical pump has been in-
stalled and has operated approximately
700 hr 1n a fluoride system that in-
cludes a fluoride-to-NaK heat exchanger.
Several design changes were made to
reduce or nearly eliminate the o1l
leakage past the rotary gas seal,

It was found that when the gas
volume above the pump was vented 1nto
a si1ght glass, the discharged helium
was heavily laden with o1l vapor. It
has not been ascertained whether this
vapor 1s harmful to fluorides, but, at
least, the contamination was un-
desirable.

In the original pump, o1l was used
to cool the drilled shaft and to flood
the rotary gas seal assembly both for
cooling and for lubrication. In the
present pump, the shaft and the
stationary seal ring are water cooled
by using a closed system. The rotary
face seal can be lubricated, 1f
necessary, by adding small amounts of
o1l, as liquid or mist, through an
opening 1n the bearing housing.

The rotary, Graphitar No. 30, gas-
sealing ring was replaced with a
si1lver-impregnated graphite (Morganite
MYIF) ring, and an attempt was made to

20

run the dry seal against hardened tool
steel. However, at a bearing pressure
of approximately 60 psi, this combi-
nation would not run unlubricated. It
has been necessary to add small amounts
of 011 to reduce gas leakage and to
prevent small power excursions 1in the
recorded demand of the pump motor. The
seal has required an average of 6 to 8
drops of o1l per day, and the gas
leakage has been negligible.

It 1s planned to use a Keller Aar
Line lubricator to atomize the lubrai-
cation i1nto the seal region through
the bearing housing. This unit may be
operated remotely as frequently as
experience dictates.

ARE-S1ze Sump Pump A modification
has been made to the existing model DA
pump(z) to convert 1t to a sump pump
with a shaft labyrinth seal from which
leakage wi1ll return by gravity to the
pump’s sump tank. The modified pump
1s now being tested with water and
other room-temperature ligquids, and 1t
wi1ll be tested with liquids that
simulate the density and viscosity of
the ARE fuel. Studies are being made
of the labyrinth seal during pump
operation, the effects of the pump
inlet suction bell, and the volume of
fluid and the bafflaing required in the
sump tank for satisfactory pump
operation,

The operation of the labyrinth seal
(0 009-1n. radial clearance) was 1in-
vestigated by measuring the leakage
of the seal at various pump speeds and
heads. As expected, the fountain
1ncreases with increase in pump speed
and in pump head. However, the amount
of the bypass fountain flow varied
from 2 to 10% of pump flow, which 1s a
reasonable fraction. A minor problem
was encountered in that the labyrinth
leakage splashed the surface of the
lIiquid 1in the sump tank and some of
the gas bubbles were carried deep
enough to be entrained in the fluid
entering the impeller.

(2)y 6. Cobb, A G

Whitman, ANP Quar
ORNL-1375, p 17.

Grindell, and G D
Prog Rep Sept 10, 1952
The pump has been tested 1n two
types of sump tank. The first tank
was a rectangular, plexiglas box, 14
by 16 by 30 in., in which the fluaid
1inlet to the tank was considerably
offset from the eye of the impeller,
The second tank was a plexiglas cylin-
der, about 12 in. 1n diameter and 14
in. high, in which the fluid inlet 1s
below and directly 1nto eye of the
1mpeller. When continuously submerged,
even 1f only to a small fraction of an
inch, the pump i1nlet suction bell
appeared to function satisfactorily in
elther sump tank under all conditions
of pump speed and flow.

Pump priming 1s not difficult to
accomplish 1n either tank. In the
1nitial test setup, the volume of gas
1n the system piping at startup was
approximately equal to the volume of
Ii1quid, and 1n the cylindrical tank,
the prime was broken by a sudden surge
of gas i1nto the impeller eye. Placing
an obstruction over the fluid inlet to
divert most of the gas bubbles away
from the eye to the free surface was
sufficient to maintain the prime. 1In
a system i1n which there 1s a small gas
volume compared with the liquid volume
at startup, priming 1s accomplished by
bleeding the trapped gas into the sump
tank or by filling under vacuum. In
the ARE system, the trapped gas 1s
forced out.

Tests were made 1in which the fluid
was gassed by admitting air, helium,
or argon 1nto the eye of the suction
bell until the fluid became milky.
The time required for the fluid to
become relatively clear was then
noted, In all tests, 1including tests
with flow rates of up to 100 gpm and
tests 1n which the cylindrical tank
without baffles was used, the fluaid
degassed 1nless than 5 min of operation.
When the 1inlet of the cylaindrical tank
was offset about 1/2 1n. with respect
to eye of the impeller, the entire
holdup volume of liquid in the tank
was given a rotary motion that 1in-
creased 1n velocity at high impeller

PERIOD ENDING MARCH 16, 1953
speeds and low flow rates. When the
baffles were 1n place, this circular
motion occasionally caused the for-

mation of vortexes and gas entrainment
in the liquid. Without baffles i1n the
cylindrical tank, the rotary motion
was present but no vortexes were
observed. The gas seal for this pump
will be either a frozen-sodium or
frozen-lead seal or a mechanical face

seal.
ARE Centrifugal Pump An ARE model

FP pump(3®’ has been constructed and
1incorporated 1n a test loop. The most
critical feature of this pump was the
frozen-sodium seal for which sodium
from an external source was fed to a
sealing annulus and frozen in rings on
each si1de of the annulus.,

Three tests have been run in which
NaK was pumped for about 15 min, 4 hr,
and 1 hour., FEach test was terminated
because of excessive seal leakage.
Chemical analyses of samples of leakage
material 1ndicated the presence of
from 1 to 7.5% potassium. Since thas
pump does not operate successfully
with NaK, 1t was modified for testing
with sodium. For operation withsodium,
the external supply of sodium to the
seal wi1ll not be needed and a larger
surface wi1ll be presented for cooling
of the seal.

Prior to installation of the new
sealing gland, a series of tests was
run on the externally supplied seal
with helium i1n the pump loop. During
these tests, pump speeds and seal
temperatures were systematically
varied, and considerable data were
taken to provide more adequate under-
standing of the frozen-seal mechanism.
Successful operation, as a gas seal,
was obtained over the available range
of speeds, 400 to 1920 rpm, and the
temperature of the sodium in the gland
groove varied from 300 to 400°F. Speed
and pressure conditions were determined
under which the seal would fail and
under which 1t would operate reliably.

(B)ANP Quar
1439, p 21

Prog Rep Dec 10, 1952, ORNL-

21
ANP PROJECT QUARTERLY PROGRESS REPORT

The sealing gland that has been
installed i1n the pump for operation
with sodium has a length-to-diameter
ratio of 2.3, 1n contrast to the
ratios of 0.6 and 0.2 for the 1insade
and outside glands of the previously
used externally supplied seal. Also,
the radial shaft-to-gland clearance
has been increased from 0.015 an. for
the externally supplied gland to
0.030 1n. for the longer gland. Initaial
tests of this gland and seal indicate
a satisfactory seal, but they also
1indicate the presence of excessive
thermal distortions that will result
in displacement of parts and eventual
bindaing.

Electromagnetic Pump The two-
stage electromagnetic pump that waill
be used to circulate NaK in testing
and cleaning operations of the fuel
circuit of the ARE has been tested
with NaK at 1500°F and 1s undergoing
similar tests with sodium. With Nak,
the pump delivery 1s approximately
90 gpm at approximately 10 psi, per-
formance curves are shown in Fig. 2.2,
The pump has givenentarely satisfactory
performance, except for a hard-soldered
Joint failure in the current loop,
which was easily repaired,
cooling, which was removed.

and excess

ROTATING-SHAFT AND VALVE-STEM-SEAL

DEVELOPMENT
W. B. McDonald R. N. Mason
W. C. Tunnell P. G. Smith

W. R. Huntley
ANP Division

R. E. Engberg
Long Range Reactor Planning Group

Mechanical Face Seals High-
temperature mechanical face seals for
li1guids such as fluorides are being
1investigated, Successful operation of
such seals may be dependent upon dry-
film lubrication, boundary lubrication,
or adsorbed-film lubrication, but
research has revealed only empairical

22

knowledge about the principles of
similar seals successfully used 1n
applications much less severe than
those i1mposed by ANP requirements.
The problem appears to be one of
selecting appropriate materials that
wi1ll retain optically flat surfaces at
high temperatures and will have the
proper degree of hardness and softness
relative to each other, Graphite 1s
being considered for the soft material
and materials such as tungsten carbade,
titanium carbide, and cermets are
being considered for the hard materaial,
The methods of fabrication and the
maintenance of high ambient tempera-
tures and low thermal gradients to
avoid distortions are also pertinent
problems.

The work of others has shown that
a principal factor in determining wear
and surface damage 1n sliding 1s the
tendency of the sliding or rubbing
materials to alloy. Although other
factors, such as the hardnesses of the
materi1als, the melting points, the
rubbing velocity, and the temperature
of operation, are considered as being
of secondary i1mportance, 1n some
instances they may become of controllang
1mportance. The use of MoS, as a high-
temperature lubricant 1s intriguing,
and considerable work has been done to
obtain surfaces coated or impregnated
with this material. Table 2 1 lasts
the attempts to establash a MoS, layer
by adding MoS, to the vehicle to make
a paste. It should be noted that MoS,
will begin to oxidize at about 700°F
1n air, but 1n the absence of oxygen
1t 1s stable to 2000°F. Tests of face
seal materials, made by rotating them
1n contact with the other specimen are
described in Table 2.2, The chatteraing
noted may have been due to forces
resulting from the geometry used 1n
the test apparatus. These experiments
indicate that of the materials tested,
graphaite 1s best for one of the
materials, under these test conditions,
when air 1s present,
PERIOD ENDING MARCH 10, 1953
\
UNGLASSIFIED
DWG 18713
* I T lC RRENT l
INPUT VOLTAGE TO CU NT LOOP
'\-Cl\ \ A 100
\r\ B 125
i6 Lt ) \ 0 150
\ ® 175
EUTECTIC NaK LOOP AT 750°F
12 ~
g \A\\ \ \D\ O]
> \ \ \
4
0
0 10 20 30 40 50 60 70 80 a0
FLOW (gpm)
UNGLASSIFIED
DWG 18712
20
INPUT VOLTAGE TO GURRENT LOOP
® 25
O 50
16 A 75
A 100
W 125
\‘.X 8 ‘50
175
g2 -~._E}__-- I fi‘"”“--.‘~h\~\ EUTECTIC NaK LOOP AT 1500°F
E "T}~§§\‘
[15]
h‘h“fi- *75\‘\“‘\55\\\\\\;::\\&\\\\\
ox \
0 10 20 30 40 50 60 70 80 90
FLOW (gpm)
Fig. 2.2. Performance Curves for Double-Stage Electromagnetic Pump with Nak.

Input voltage to magnet loop, 125 volts.

23
ANP PROJECT QUARTERLY PROGRESS REPORT

TABLE 2.1, DESCRIPTION OF ATTEMPTS TO OBTAIN A MoS, SURFACE

BASE METAL VEHICLE TREATMENT AND APPLICATION REMARKS

1 Type 316 G-E enamel Brushed on, air drled,o Good coating obtained,
atainless steel furnace heated to 700 F apparently

2 Type 316 Colloidal Brushed on, air draed, , Did not stick to base metal
stainless steel s1lver furnace heated to 700 F

3 Type 316 High- Brushed on, air dried, . Varnish burnt off, dad not stick
stainless steel temperature furnace heated to 700 F

varnish

4 Type 316 No 1202 varnish | Brushed on, air dr1ed,0 Varnish burnt off, did net stick
stainless steel furnace heated to 700°F

5 Type 316 Keystone grease Base metal heated to 700°F, Did not stick to base metal
stainless steel paste brushed on

6 Type 316 Permatex Ne 2 Base metal heated to 700°F, Showed promiae, fair coating
atainless steel paste brushed on obtained

7 Type 316 Ruby flux Base metal heated te TOOOF, Showed promise, fair coating
stainless steel psste brushed on ebtained

8 Type 316 Aquadag No 200 Base metal heated to 700°F, Did not stick to base metal
stainless steel paste brushed on

9 Type 316 Permatex No 2 Base metal heated te TOOOF, Remained soft, possibly net
stainless stesl thinned waith paste brushed on enough heat, indicates promise

10 Lapped tungsten
carbide

11 Etched tungsten
carbide

12 Type 316
stainless steel
13 Tungsten carbide

14 Tungsten carbaide

15 Type 316

Sebacate

None

None

None

Handi-flux

Ruby flux

Permatex No 2

Heated to 700°F, dipped in

dry M082

Heated to TOOOF, dry Hn82

sprinkled on

Heated to 700°F, dipped 1n

dry H082

Paste applied, heated with

torch

Paste applied, heated waith

torch

Base metal heated, brushed

Thin layer obtained, apparently

Thin layer obtained, apparently

obtained

Good layer

Adhered to

uneven

base metal, but

Soft and flexible, apparently
Good layer, apparently

Brass burnt, but apparently

stainless steel with Sebacate on 1n furnace
16 Brass Handi-flux Brushed on base metal,
heated 1n furnace
17 Type 316 Corn syrup Brushed on base metal,
stainless steel heated 1n furnace
18 Type 316 Celluflux Brushed on base metal,
stainless steel heated 1n furnace
19 Type 316 Pliobond Brushed on base metal,
stainless steel heated 1n furnace

some layer remained

Fair coating, shows premiase

Good coating, but uneven,
shows promise

Fair coating, hard and
brattle

An attempt has been made to establash
an upper temperature limit for silver-
impregnated graphite rubbing against
type 316 stainless steel. The assembly
ceased to seal helium at about 450°F,
and although the temperature was 1in-
creased slowly to 910°F, sealing daid
not occur again, When the temperature
was decreased to 450°F, sealing was

24

again accomplished.
w1ll be continued, and varicus sealing
pressures will be tried.

Another experiment 1s 1n progress
1n which an attempt 1s being made to
seal against water with silver-
impregnated graphaite vs, type 316
stainless steel atvarious differential
pressures and bearing pressures. To

This experiment
TABLE

2.2,

PERIOD ENDING MARCH 10,

SUMMARY OF TESTS OF FACE SEAL MATERIALS

BASE METAL NO.
{cf , Table 2.1)

MATERIALS

RESULTS

Operated for 2 hr before chattering
occurred

Operated for 2 hr before chattering
occurred

Operated for only a short period

Operated for 5 hr before chattering
occurred

Operated for 3 hr before chattering

1953

11 Etched tungsten carbide vs
type 316 stainless steel
10 Lapped tungsten carbide vs
type 316 stainless steel
13 Tungsten carbide vs tungsten
carbide
16 Brass vs tungsten carbide
12 Type 316 stainless steel vs
tungsten carbide
14 Tungsten carbide vs
type 316 stainless steel
15 Type 316 stainless steel vs

tungsten carbide

C-18 graphite vs
carbide

Graphite (No

carbide

tungsten

40) 1mpregnated

with MoS, vs tungsten
carbide
APC graphite vs tungsten

C-18 graphite vs tungsten
carbide
C-18 graphite vs tungsten

occurred, addition of dry MoS, stopped
chattering for about 15 min

Operated for 7 min before chattering

occurred

Operated for 50 min before chattering

occurred

Chattered immediately, surface not flat

Operated for 12 hr at self-generated
temperature of 225°F

Attempted to coat graphite with fuel
27, operated for 4 hours

Heated with torch to 475°F, chatter
developed

Dry MoS, added, heated to 913°F wath
torch, operation smooth, except
between 400 and 450°F. when chattering

occurred

date, the seal has not been consistent,
1t has leaked at times and in varying
amounts.

Other materials have been ordered,
such as boron nitride compacts, MoS,-
1impregnated copper, silver and stain-
less steel compacts, and a sintered
molybdenum compact, that will be
treated to form a MoS, surface.

Combination Packed and Frozen Seal
with MoS,. Because MoS, has high-
temperature lubricating properties,
an attempt was made to use 1t as a

high-temperature packing material for
sealing fluorides, fluoride leakage
was slight, operation was smooth, and
there were no power surges. However,
as 1n previous experiments, the MoS,
was too fluid and could not be con-
tained. The retainers used were
Graphitar No. 14 rings. The material
slowly leaked out of the stuffing box
and around the gland until the gland
was completely 1inserted,

For the next attempt to seal waith
this material, a packing of Inconel

25
ANP PROJECT QUARTERLY PROGRESS REPORT

braid impregnated with MoS, was used.
This experiment was terminated after
approximately 925 hr of operation with
fuel No. 30 at 10 psi and 1750 rpm.
Excessive leakage occurred because the
seal reached too high temperatures
when an attempt was made to restart
operation after binding had occurred.
The operation of the seal corresponded
to operation of seals 1n other tests
1in that the leakage of the seal was
apparently temperature sensitive. When
the seal was too hot or too cold,
leakage and power surges resulted, but
under certain conditions operation was
smooth and there was no detectable
leakage, There have been nointentional
start-stop tests, although the unit
has been accidentally stopped on
several occasions, and power surges
have overloaded the motor relays
several times, The unit has been
restarted with little difficulty on
each occasion by heating the packing
area and using a wrench to ‘““break’” the

shaft loose.
Combination Packed and Frozen Seal

with Graphite. Thecontinuous operation
of the graphite-packed seal, mentioned
1n the previous report, was terminated
after 725 hours. The seal had shown
no signs of failure, and after another
150 hr of periodic and stop-start
operation, the test was again termz-
nated. During the latter period of
operation, attempts were made to
control the seal temperatures and,
thereby, the zone in which freezing
was occurring, since 1t was believed
that most satisfactory operation would
occur 1f freezing took place within
the graphite region. Two fairly
stable temperature ranges of operation
were found, one set of temperatures
corresponded to that maintained durang
the 725-hr period of continuous
operation and the other set was about
150°F lower.

at intermediate temperatures.

Operation was unstable
At the
lower temperatures, operation was
smooth, there was no detectable
leakage, and the unit could be stopped

26

for as long as 20 min and started
again by motor power only. The seal
used 1n this test was similar to the
seals used 1n other tests i1n that
leakage occurred when the seal was
too hot or too cold.

Graphite Packed Seal. In an attempt
to determine the benefits of holding
the transition from fluid to solad
within the graphite, a test was set up
that i1ncluded a long (4 1in.), graphite-
packed, stuffing box. In order to
obtain more complete temperature data,
four thermocouples were installed in
the rotating shaft, and signals were
taken off through slip rings. After a
dry run, fluorides were i1introduced and
the seal i1mmediately started leaking
graphite. The Graphitar rings used
as retainers did not contain the
graphite at the top of the seal.
Approximately 3/4 in. of Inconel braid
was then added, and the graphite was
successfully retained. The motor for
this test was a 5-hp unit. Operation
started off fairly normally, but the
temperatures adjacent to the seal
began to increase. Cooling of the
bearing housing and the gland was
started, but the temperatures continued
to rise. The temperatures never
became stable, they fluctuated over a
range of 940 to 2190°F throughout the
seal regiron. The hottest point was
apparently near the middle of the
packing, and the lowest temperatures
were 1n the regions of the bearings
and the fluoride. The molten fluoride
was acting as a coolant in this test.
The temperatures continued to increase
until the test had to be terminated
after approximately 50 hr of operation
because the shaft .froze. The shaft
could be restarted by motor power
afterit cooled, but would not continue
to operate when higher temperatures
were reached. At no time was there
any detectable leakage, however,
severe shaft scoring occurred.

A picture of the seal and shaft cut
through the center line 1s shown in
Fig. 2.3 The extreme scoring of the
PERIOD ENDING MARCH 10, 1953

g
2
we
Lo
]
8o
3o
3L
£2
5

Shaft and Graphite Seal.

Section View,

2.3.

Fig.

A

27

ANP PROJECT QUARTERLY PROGRESS REPORT

1 3/16-1n -dia shaft 1s quite evident,
other portions of the shaft (also at
very high temperatures) showed no
markings whatsoever. The condition
of the graphite 1tself 1s of interest,
since 1t apparently packed i1n more or
less “concentric” radial layers that
varied i1n thickness along the length
of the seal. The layers adhered at
times to the shaft and rotated, and at
other times they adhered to each other
and remained stationary. The powder
was exceptionally fine and very
“greasy”, the “concentric” layers were
very slimy, and there was no visible
evidence of fluoride penetration.
Rotating-Shaft Seal Test. The
apparatus for the tests of the rotating-
shaft seal consists of a container
for the packing material being tested,
an upper compression gland that exerts
pressure on the packing, a lantern
gland located midway in the container,
and a shaft extending through the
assembly. The shaft rotates in a
bearing beneath the assembly and 1s
driven by a motor mounted above The
packing container 1s mounted on four
rollers and 1s restrained from turning
with the shaft by two tension springs
The torque necessary to turn the shaft
1s 1ndicated by a pointer and a
calibrated scale on the assembly
Fluorides are forced under pressure
from a fill tube into the container
through the lantern gland During
the tests, the temperature of the
midpoint of the packing container was
1500°F, the temperature of the lower
shaft opening was 1400°F, the pressure
was 20 psi. The packing materials
tested were MoS, and powdered graphite
The MoS, packing began to leak after
1 hr of operation. The torque required
to turn the shaft at 350 rpm was
2 ft-1b. The graphite test ran for
8 hr before leakage occurred, and the
same torque was required for rotation.
There was considerable wear on the
upper packing gland. The shaft had
to be started and stopped a number of
times because of severe vibration.

28

Packing Penetration Tests. The
apparatus used 1n the packing pene-
tration tests and the results of
several tests were described prevai-
ously ¢*) Further tests have been
made and the results are presented in
the following

Stainless steel braid impregnated
with MoS, was compressed and heated
twice to the annealing temperature,
with further compression after each
heating This packing held the
fluorides for 1/2 hr at 2 psi, but
when the pressure was i1ncreased to
5 ps1, leakage occurred. Examination
showed that the leakage path was along
the walls of the containers, the
screw stem, and the strands of the
braid It could not be detected that
any fluoride mixed with the MoS, or
penetrated through 1t.

A second boron nitride test was
run 1n an assembly with much smaller
clearances than those used for the
previous boron nitride test. In this
test, 1t was possible to raise the
pressure to 30 psi, and operation was
continued for 234 hr before leakage
occurred,

A test with J. T. Baker Chemical
Co. powdered graphite previously
reported(?) has continued for 240 hr
with no leakage. Analysis showed
that the packing was approximately
30% amorphous carbon. Three more
graphites have been tested, and ain
all cases 1t was possible to raise
the pressure to 30 psi1i without leakage
occurring. The analyses and results
are given 1n Table 2.3. There should
be a minimum of amorphous carbon 1in
the graphite because 1t would be
expected to increase the friction
between a rotating shaft and the
packing, however, the graphite with
the highest amorphous carbon content
sealed for the longest periods. That
the packing i1n the first test did not
leak whereas the packing in the later

(4)4ANP Quar Prog Rep Dec 10,
1439, p 23

1952, ORNL-
PERIOD ENDING MARCH 10, 1953

TABLE 2.3. PACKING PENETRATION TESTS

ANALYSIS OF AMORPHOUS HOURS OF OPERATION
PACKING MATERIAL CARBON CONTENT BEFORE LEAKAGE
(%) OCCURRED
Dixon No. 2 flake graphite 10 22
Dixon Microfyne graphite 50 75
Amer Graphite Co. No. 620
powdered graphite 50 63

tests did, although the physical
characteristics of the graphites were
the same, might be the result of

fuel No. 14 being used 1n the fairst
test and fuel No. 30 in the later
test,

Frozen-Sodium and Frozen-Lead Seal
for NaK. It was reported previously(%?
that tests were under way to determine
the feasibility of sealing a NaK pump
with an externally supplied frozen-
sodium or frozen-lead seal. Although
the preliminary tests were encouraging,
subsequent tests revealed that seal
life (the time before failure because
of alloying) was, 1n general, 1inversely
proportional to NaK temperature and
that the seals were unreliable for
periods of operation greater than
100 hours.

In the frozen sodium seal, the
sodium alloys with the NaK and forms
a low-melting-point alloy in the seal
that cannot be frozen with room-
temperature cooling water, as a result
the seal 1s completely lost. Re-
frigeration, with temperatures below
the melting point of eutectic Nak
(12°F), would be required to assure
seal retention.

In the frozen-lead seal, the lead
alloys with the NaK and forms a high-
melting-point alloy, as a result there
1s shaft seizure and 1t 1s impossible
to operate with reasonable power
requirements.

(S)AHP Quar Prog Rep Dec 10,
1439, p. 26

1952, ORNL-

ARE Valve Test. Preliminary gas
and liquid leakage tests have been
run on the ARE pneumatically driven
valve supplied by Fulton-Sylphon Co.
The valve 1s constructed of Inconel
and has a Stellite-to-Stellite valve
and seat. The shaft-sealing member
1s a four-ply Inconel bellows The
valve 1s adaptable to being eather
normally closed or normally opened,
and the position 1s reversed by
application of 15 psi1 of air pressure
to a large bellows actuator.

For the 1nitial test, the valve
was mounted vertically and adapted
for normally closed operation. Waith
the valve at room temperature, gas
leakage was small with test pressures
from 30 to 60 psi1 that were first
applied up against the valve seat
and then down on 1t. Raising the
valve temperature to 1050°F caused a
slight i1ncrease 1n gas leakage,
however, the leakage remained small.

Testing for liquid leaks with the
ARE fuel, NaF-ZrF,-UF, (50-46-4
mole %), at 1100°F revealed that there
was no leakage with test pressures
of 30 to 60 psi applied both above and
below the seat. This test was repeated
several times.

In a test at 1300°F, slight leakage
(0.3 an.3/hr or less) occurred at
various 1ntervals However, 1t was
concluded that the leakage at thais
temperature was not serious. The
initi1al checks at 1500°F were very
successful. Subsequent checks,

29
ANP PROJECT QUARTERLY PROGRESS REPORT

however, showed a very serious leakage
rate, 120 in.3/hr, that could not be
corrected by reseating the valve,
It was concluded therefore that the
valve might be usable for several
cycles at 1500°F but that 1t could
not be expected to hold tight 1f
cycled indefinitely.

Difficulty at the elevated tempera-
ture was experlenced 1n that the
valve stuck when left closed for short
periods However, 1n all cases to
date, the valve has eventually become
operable by increasing the pressure
up to 30 psi1 on the valve actuator
In one instance at 1100°F and another
at 1500°F, 1t was necessary to vibrate
the valve body to get the valve open.
When the valve was cut for 1inspection,
1t was found that some binding of the
Stellite faces had occurred

HEAT EXCHANGER SYSTEMS

G. D. Whitman D. F. Salmon
ANP Division

Sodium-to-A1r Radiator Tests.
The sodium-to-air radiator tested
had a core element with 30 fins per
inch, (®) This test ran for 200 hr
and was then terminated because of
poor heat transfer performance. The
maximum, over-all, heat transfer
coefficients obtained were of the
order of 6 Btu/hr*ft?+°F, which 1s
far below the predicted performance
It was discovered, after i1nlet sodium
temperatures of 1500°F were reached,
that very few of the Nicrobrazed
fin-to-tube joints were good. The
radiator could be seen through the
air exhaust, and 1t was observed that
with minimum air flow less than 10%
of the nickel fins showed any heat
color. The small amounts of braze
material used to avoid closing the
0.025-1n. fin-to-fin gaps were ap-
parently i1nsufficient to completely
join the fins and tubes. No oxide
plugging occurred 1n the radiator

(6 )ANP Quar
1439, p 21

Prog Rep Dec 10, 1952, ORNL-

30

tubes during this relatively short
run, and the bypass filter caircuzit
was operated without diffaiculty.

A Roots Connersville gas pump
driven by a 2-hp Varidrive has been
installed 1n place of the Buffalo
centrifugal blower so that a wider
range of air flows and better flow
regulation can be obtained.

Bi1fluid Heat Transfer Loop. The
b1fluid heat transfer loop,‘’’ which
has been 1n operation for approximately
two months, transfers heat from fuel
No 30 to NaK. The heat transfer
takes place 1n a concentric tube
section, the center tube contains
the hot fluoride and the NaK 1s 1in
the annulus. The center tube has an
inside diameter of 0.269 in, and an
L/D ratio of 40, whereas the annulus
has an L/D ratio of 22.

Heat transfer data on the fluoraide
fuel have been taken over a Reynold’s
number range of from 5,000 to 20,000,
which corresponds to velocities of
from 8 to 30 fps. The maximum fluoride
temperature was 1400°F, and the
temperature drop across the heat
exchanger was varied from 15 to 40°F.
The NaK-side data were taken at
Reynold’'s numbers from 20,000 to
100, 000.

The over-all heat transfer coef-
ficient for heat fluxes from 300,000
to 500,000 Btu/hr*ft? has varied from
1000 to 2500 Btu/hr<ft?+:°F, Satis-
factory separation of the individual
coefficients has not been accomplished
because of the necessity of evaluating
thermal entrance effects and because
of what appears to be fouling in the
exchanger. It 1s possible that
fouling occurred, because there was
flow restriction when the temperature
of the NaK entering the heat exchanger
was lowered to the fluoride melting
point (approximately 950°F).

The pump for circulating fluorade
fuel has operated satisfactorily with

a minimum of gas leakage. This pump

(7)ANP Quar Prog Rep Dec 10,
1439, p 28

1952, ORBNL-
1s a sump pump 1n which the liquad
level 1s controlled with a spark-
plug-probe. The seal of the pump 1s
of the water-cooled, rotary-face type
that consists of silver-impregnated
graphite running against tool steel.

ARE FUEL CIRCUIT MOCKUP
G. D. Whitman, ANP Division

A mockup of the ARE fuel circuait
has been constructed for investigating
the hydrodynamic stability of the
design and the characteristics of the
liquid flow in the surge tanks. The
system, shown schematicallyin Fig. 2 4,
was constructed from black iron paipe,
and 1t has viewaing windows 1in the
surge tanks and glass sections in the
lines leaving the surge tanks. The
surge tanks are of the size to be
used 1n the ARE, and the two 1dentacal
pumps have characteristics in the flow
range to be i1nvestigated (0 to 50 gpm)

PERIOD ENDING MARCH 10, 1953

that are quite similar to those of the
proposed ARE pumps. Speed variation
of the pumps 1s obtained by using
5-hp Varidrive units. The heat
exchangers are simulated by gate
valves, and the reactor can be simulated
by a Globe valve or by the glass model
that was previously described.

Thus far, water and a water -
glycerine solution have been used ain
the system. The results are therefore
based on observations made with water
of approximately one-third the density
and one-tenth the viscosity of the
fuel and with the glycerine-water
solution of approximately one-thard
the density of the fuel and the same
viscosity. Tests have begun with
1-1,2-2,bromoethane, for which the
room- temperature density and viscosity
are nearly the same as those of
fuel No. 30 at 1400°F.

The ARE system comprises two,
similar, liquid heat exchanger and

DwWG 18714

Q AIR SUPPLY

i
N

ANTIVORTEXING AND
PROBE ASSEMBLY

I VIEWING
U: SURGE Efl l]: SURGE | | WINDOW
TANK - TANK 1]
¢ T fett-— 3 -0 |IPS 3-in {PS f ‘
i \\L_ []ROTAMETER ROTAMETER [] i
GLASS PIPE-% [ ] [ ] ["—GLASS PIPE
Q ¢ ¢
S gl —l— S—
HEAT EXCHANGER HEAT EXCHANGER
l REACTOR
= 2 -in |PS
PRESSURE AC, 200 gpm, 65-ft HEAD
GAGE
(P PUMP PUMP q> /)
k AW pd j
—_— ————
FLUID FLOW
Fig. 2.4. Sketch of ARE Fuel-Circult Mockup.

31
ANP PROJECT QUARTERLY PROGRESS REPORT

surge tank circuits - one for moderator
coolant and the other for fuel. Each
circuit includes two parallel circuits,
each of which includes, 1n series,
a pairofheat exchangers (inparallel),
a pump, and a surge tank. As designed,
each surge tank was to receive and
deliver the entire flow of 1ts par-
ticular pump. However, the surge
tanks of the moderator coolant circuits
have now been provided with bypasses
to reduce the flow through them to
low levels and thereby reduce gas-
entraining turbulence to acceptable
values. The moderator coolant flow
rate 1s, 1n general, several times
higher than the fuel flow rate.
However, in the fuel circuit there
are strong 1ncentives to retain
complete flow control of all fuel,
1inasmuch as the system will be filled
initially with nonuranium-bearaing
salts and wi1ll have U?35_bearing salts
added to 1t until criticality 1s
reached. Once addition starts, no
&alt can be removed or drained, and
complete and uniform mixing must occur.
Consequently, the expansion or surge
tanks must provide sufficient capacity
to receive the enriched fuel (about
15% of the system volume) and also to
provide for fuel thermal expansion
from 1000 to 1500°F (about 10% of
system volume). The thermal expansion
w1ll be about 200°F more than that
anticipated. There 1s also a strong
incentive for minimum volume holdup of
enriched fuel in each surge tank.

Vortexes and Bubbles. The system
was operated 1initially with water,
and the surge tanks were 1nvestigated
for vortexing above the discharge
lines. The surge tanks had no baffles,
and entrained gas was observed in the
discharges at all flow rates above 10
gpm with the tanks half full of liquid.
This gas entrainment was due to vortex-
1ing of the liquid leaving the tanks.
At lower liquid levels, vortexing
occurred at flow rates of less than
10 gpm.

Many geometries were tried for
breaking up this vortexing, and a

32

baffle design was finally developed
that allowed water flow rates of up to
40 gpm to be put through the tanks
without visible gas entrainment 1f a
minimum level of approximately 4 ain.
was malntained in the 10-1in.-di1a tanks.

This antivortexing baffle consists
of a flat plate, 6 by 8 1n., attached
to the bottom of the probe well that
1s centered over the discharge laine.
The probe well was cut off so that the
plate could be located 2 in. from the
bottom of the tank. There are three
vertical fins attached to the plate,
one 1n the center and one near each
edge, so that they are approximately
1/8 1n. from the bottom of the cy-
lindrical tank. This device serves to
guide the flow from the inlet to the
outlet with less velocity loss than in
the open tank, and the flow 1s re-
strained from going 1n on the sides
and edge near the end of the tank where
the more severe vortexing always
occurred.

When the glycerine solution was
tried, there appeared to be less
tendency to entrain gaé 1n the surge
tanks. At rated flow, 40 gpm, lower
fluid levels could be maintained 1in
the surge tanks without there being
visible gas entrainment 1in the dis-
charge.

It had been felt that Froude’s
modulus should be applicable, since
the fluid mockup and the ARE fuel
circuits were similar, and that
gravity forces should predominate on
the free surfaces in the tanks. Since
the mockup surge tanks are full ARE
si1ze, the ARE fuel flow rates were
maintalned with the water and the
glycerine-water solution for this in-
vestigation i1n order to maintain
hydraulic similarity,

The moderator circuit in the ARE 1s
quite similar 1n design to the fuel
circuit. The Na or NaK flow (about
100 gpm) cannot be put through a surge
tank of ARE dimensions without there
being very severe gas entrainment 1in
the discharge. This condition can be
remedied by bypassing most of the flow
around the tanks or by using the tanks
as stand pipes only. The first remedy
would be the more desirable, since gas
removal would be more effective,

It was found that the surge tanks
removed visible gas bubbles in the
water and glycerine-water solution.
Air bubbles of the order of 1/8 in. 1n
diameter and larger were removed 1in a
very few cycles, and those so small as
to be barely visible were removed after
an hour or so of circulation at low or
rated flows., When the glycerine-water
solution was first transferred to the
loop, 1t was nearly opaque because of
the finely divided gas bubbles, but
the solution was clear after circu-
lation through the surge tanks for
approximately 1 hour. The system
tested for gas removal had a capacity
of 15 gallons.

Operational Instabilities The
system has also been i1nvestigated with
respect topump failures and operational
instabilities. During operation with
each pump delivering 30 to 40 gpm at
approximately 45 psi head, 1f one pump
1s shut down the remaining pump 1im-
mediately becomes gas bound, starts
pumping 1intermittently, and severely
gasses the loop. The gassing of the
pump that remains operable results
from the transfer of liquid from the
surge tank 1n the operating loop to
the surge tank in the loop that was
shut down, that 1is, when one pump 1is
shut off, 1ts associated surge tank,
which 1s essentially at pump suction
pressure, 1s subjected to the dais-
charge pressure of the operating pump
and therefore gas volume 1s compressed
and causes transfer of liquid from the
operating loop to the loop that was
shut down.

When a pump operating at or near
rated flows stops, eirther the loop
must be immediately 1solated from the
system or the remaining pump stopped.
The loop containing the 1noperable
pump must be valved out of the system
before flow 1s restarted. Also, at
relatively high flow or pressure rise

PERIOD ENDING MARCH 10, 1953

across the pumps, impeller speeds must
be the same for both pumps, or excessive
amounts of liquid will be transferred
from one surge tank to the other,

In an attempt tomaintain equilabrium
between the two tanks 1n case of a
pump failure, they were connected by a
liquid line. It was never possible to
operate satisfactorily with this line
open because one pump would develop a
slightly greater head and take over
the entire load and thus 1induce back
flow through the other pump. At such
times, there was heavy flow from one
tank to the other through the liquad
line connecting the two surge tanks.
Flow through the heat exchanger cair-
cuits would not drop off apprecaiably.
The pump carrying the load could easily
supply both heat exchangers and a
small back flow through the other pump.

The head flow characteristics of
the pumps are shown in Fig. 2.5, It
will be noted that between shut-off
and 50 gpm there 1s little or no drop
in head, and at some speeds a very
slight pulsation point can be detected.
These flat curves explain the abilaty
of one pump to handle the total flow
of the reactor and heat exchanger, as
previously mentioned., One pump run-
ning at a slightly higher speed can
easily buck the other pumps and handle
the total loop load without losang
sufficient head to fall below the
shut-off head of the other pump.

During operation below 25 gpm per
pump, oscillations have been observed
1n the flow. These have never been
serious and could always be reduced to
negligible amplitude by increasing the
flow rate of the system. The osc1l-
lations are evidenced by an 1ncrease
1in flow rate and liquid loss 1n one
pump circuit and a simultaneous de-
crease 1n flow rate and liquid gain
1n the other circuit. The condition
then reverses and the period of the
oscillation 1s approximately 2 seconds.
The maximum change in flow rate has
been t2 5 gpm at flow rates of 20 gpm
per pump. Thas disturbance can be

33
ANP PROJECT QUARTERLY PROGRESS REPORT

HEAD (psi)

UNCLASSIFIED

60 | DWG 18715
® NO 1 PUMP
O NO 2 PUMP
35 A ARE PUMP
3880rpm
50
J 363Orpmd J
i S— — i — —_————)—————_ e SE—— —— —
a5 b) Q ) P O o
40
ZF_-——- —— —igierpm 3300rpm
—_H e —_—
'P——* —l—-._._._‘ !5- .
e —— . !
35
30
2780 rpm —_— Y o W W o W
25
2475rpm
20
—
1400 rpm
15 I20(3rpm —
\ - - —- --.“. -
ZS-‘--_ . |
1900 rpm ) S ———
——— — — — —— —— — C— S —— — c _““D_—c)— ———-—O—-—
10
1335 rpm P
0
o 5 10 15 20 25 30 35 40 45 50
FLOW (gpm)

Fig. 2.5. Head Flow Characteristics of Pumps.

34
explained by the pump characteristics
and the absence of check valves in the
discharge lines of the parallel pump
arrangement.

In this particular system, when one
pump takes the entire load, there 1s a
subsequent pressure rise i1n the surge
tank of the stalled pump. This 1s
then seen as an increase 1in discharge
pressure of the stalled pump, whach
then overpowers the pump handling the
load and momentarily reverses the
Ssi1tuation,

The actual reversal of flow occurs
only 1n the line connecting the surge
tanks through the pumps, that 1s, the
suction lines and the common dis-
charge line., The flow through the
heat exchangers 1s unsteady but does
not reverse or stop. The flow through
the reactor appears tobe quite steady,
even though 1t 1s being supplied by
alternate pumps.

An Esterline Angus recording watt-
meter was 1nstalled on one pump motor,
and 1t was established that duraing
these oscillations the pump alternately
carried full load and shut-off head
demand. This condition can be ef-
fectively corrected by 1nserting a
resistance 1n the common pump dis-
charge line sothat stable pump charac-
teristics are obtained. A 2-psi 1n-
crease 1n pressure 1s sufficient to
result i1n a discharge curve which
that the head decreases with
flow from shut-off to maximum flow.
This was demonstrated by partly closing
one of the throttle valves 1n the
pump discharge line during an oscail-
lation, the disturbance immediately
stopped.

shows

The system 1s now being tested waith
tetrabromoethane, which has physical
properties nearly the same as those of
the fluorade fuel (viscosity, 9.27 cps,
specific gravity, 2,95 at T7°F). The
mockup wi1ll again be used to check gas
removal and vortexing in the surge tanks,
filling and draining the reactor, pump
characteristics and flow i1nstabilities.

PERIOD ENDING MARCH 10, 1953

INSTRUMENTATION
W. B. McDonald D. R. Ward
P. W. Taylor A. L. Southern
P. G. Smath

ANP Divaision

Pressure Measurement Two tests of
Moore Nullmatic pressure transmitters,
operating with the 0.005-2n. wall
bellows completely submerged in the
ARE fuel, NaF-ZrF, -UF, (50-46-4 mole %),
at 1100°F, have logged enough operating
time to make 1t seem likely that thas
method of pressure measurement 1s the
most advantageous ofany tried to date.
The transmitters are ‘“upside down’' in
the sense that the cavity containang
the bellows and the liquid under
pressure 1s open at the top. A nickel
transmitter has been cycled every 1/2
hr between 10 and 30 psi1 for 1900 hours.
A type 316 stainless steel transmitter
has operated similarly for 1500 hours.
A third transmitter of type 316 stain-
less steel filled with lead and topped
with a layer of the molten fuel has
1500 hr of operating time, to date,
and 1s running slightly cooler than
the others. The initial zero shift ain
each case was approximately 1% of full
scale at the operating temperature,
and dri1ft over the 1500-hr period was
approximately 2%. However, the zero
position can be adjusted easaily, af
necessary.

Several other Moore transmitters
have been installed in dynamic loops
for circulating fluorides or liquad
metals. These transmitters operate
“right side up,’’ and there 1s trapped
helium around the bellows to protect
them from the liquid. This method of
protecting the bellows, although
operable 1n most cases, does not seem
as satisfactory as complete submersion.
The ARE fuel 1in the nickel transmitter
mentioned above has been rapidly
frozen and then rapidly remelted three
times, with no apparent damage to the
1nstrument. This 1s not possible with
the trapped-gas instrument because of
fouling of the bellows. It was also

35
ANP PROJECT QUARTERLY PROGRESS REPORT

found that when the pressure transmitter
1s filled with static fuel, the bellows
temperature can be maintained at 1100°F
or below. Operation of the bellows at
the reduced temperature reduces the
corrosion rate.

Failure of the bellows i1n one of
the type 316 stainless steel trans-
mitters after 500 hr of operation at
1200°F was apparently caused by oxi-
dation, Helaium instead of air 1s now
used as the continuous-flow-balancing
gas 1n the pressure transmitter tests,

ARE Leak Detection Indicator Tests
There 1s i1nterest in developing a
secondary method for detecting leaks
in NaK or fuel systems in the ARE to
supplement the halogen leak detector.
Experimentation revealed that a so-
lution of phenol red can be made very
sensitive to changes in pH by adjust-
1ng the solution to a reddish brown.
When helium 1s passed over a surface
of hot fuel and then over the surface
of the i1ndicator solution, the solution
changes color. Likewise, 1t was found
that a cold lump of the fuel dropped
into the indicator solution changed
the i1ndicator toward the acid direction,

Slip Rings for Temperature Measure-
ment Three tests of slip raings for
measuring the temperature i1nside a
rotating shaft were started. Because
si1lver-graphite brushes were notavail-
able, carbon brushes were used. One
test was completed i1n which copper
rings and carbon brushes were used.
When the brushes were well run-ain, an
accuracy 1n the temperature measure-
ments of +1°F at 1200°F was obtained.
The combination of carbon brushes and
carbon rings 1s satisfactory 1f a
wiping brush 1s mounted on the raing.
If the ring 1s not kept clean, the
resistance across the contact increases
with time, and the error may increase
to t40°F or more.

A second combination of materaials,
coin silver and silver-impregnated
graphite, was placed 1n operation
before being tested, and performance
has been satisfactory for several

36

hundred hours. In a third test that
1s being conducted with type 316
stainless steel rings and carbon
brushes, an error of +20°F has de-
veloped, but 1t 1s a constant error.
The carbon brushes will be replaced
with si1lver-i1mpregnated graphite
brushes, which should reduce the error
slightly. Commerical slip ring as-
semblies have been ordered that waill
work on either a 7/8- or a 2 1/4-1n.

Shafto
Rotameter Type of Flowmeter. Testing

has continued on the rotameter type of
flowmeter reported previously, (%) and
approximately 1000 hr of operation has
been logged 1n the temperature range
of 1100 to 1300°F. This 1instrument
will measure flows of fuel No, 30 with
10% accuracy over the range of 9 to
60 gpm. Two i1nstruments of this type
have been tested, and the results
1indicate that this instrument will be
reliable for ARE operation.

HANDLING OF FLUORIDES AND LIQUID METALS

L. A. Mann J. M. Cisar
F. M. Grizzell
ANP Division

Distillation of NaK. NaK has been
chosen as the precleaning fluid for
the ARE before introduction of fluoride
salts. Since only a small fraction of
NaK (less than 0.5 wt %, cf., sec. 9)
can be tolerated in the fuel without
reduction of UF, to UF3, studies have
been made to determine how much NakK
remains after the system 1s drained
and how effective distillation 1s 1n
removing 1t. Since the vapor pressure
of potassium at any temperature 1s
higher than that of sodium, upon
distillation the alloy undergoes rapad
depletion of potassium, and there 1s a
corresponding enrichment of sodium,
Therefore each distillation 1s pro-
gressively more difficult, The boiling
point of sodium 1s 1621°F at 1 atm,
hence, vacuum distillation 1s used to
avolid excessive temperatures. An

() 4NP Quar
1439, p 29

Prog Rep Dec 10, 1952, ORNL-
absolute pressure of 100 mm Hg, which
corresponds to asodium boiling temper-
ature of 1285°F, was selected for the
distillation. After some 1nitial
difficulty with gas leaks, several 5-1b
batches of NaK were distilled, and the
following observations were made

1. There was no difficulty an
completely distilling the Nak.

2. Most of the distillation occurred
in one surge 1n each of the four runs,
and a part of the condenser was heated
to almost the distilling pot tempera-
ture.

3. The adjustable probe (a probe
wired through a stuffing box of
s1licone-rubber “washers’'’) worked
excellently, and gave reproducible
level readings that checked to within
1/32 an. or closer.

NaK has been distilled from ARE
pump loop, but because of excessive
oxidation of the type 316 stainless
steel 1n the loop, a sufficiently low
pressure could not be maintained on

PERIOD ENDING MARCH 10, 1953

the system for a good test of the
efficiency of the distillation process
to be obtained. Pockets of NaK re-
mained 1n some of the traps, par-
ticularly the pump. In the Inconel
ARE system, such corrosion should not
be encountered, hence, essentially
complete removal of NaK by distillation
should be effected.

Flame Tests for NaK Vapor 1n Helium
(D. R. Ward, ANP Division). Tests
are being conducted to determine the
reliabality of flame tests for detect-
1ng the presence of NaK 1n helium. One
method used was to introduce helium,
from the system from which NaK was
being removed by the distallation
process, 1nto an otherwise colorless
hydrogen flame, the brightness of the
resulting flame was then measured by a
phototube circuit. The results of six
tests indicate that this method may be
used to determine the degree of
cleanliness of a system from which Nak
has been removed by the distillation
method.

37
-

3. REACTOR PHYSICS

W. K. Ergen,

From the standpoint of the physics
of nuclear reactors, the main event
during the past quarter was the
correlation between the IBM multaigroup
calculations and the critical experi-
ment on a fairly realistic, reflector-
moderated, reactor mockup. Because
of the approximations i1nvolved in the
calculations, 1t previously had not
been clear how close this correlation
would be. That the experimental
critical mass turned out to be only
slightly higher than the computed
value and was well within the limits
acceptable with respect to uranium
investment and fuel chemistry, consti-
tutes a milestone 1n the development
of this type of reactor. Furthermore,
there was, except at the boundaries,
good correlation between the computed
and the measured flux Thais applies
to the flux distribution in space and
in energy, but the only experimental
indications of the energy distribution
are the thermal value and the cadmium
ratio, which are effectively only two
points on the energy distribution

ANP Division

curve. Some significant discrepancies
exi1st at the boundaries, but the
simple diffusion theory employed in
the calculations obviously cannot be
expected to hold at the i1nterfaces
between widely different materials.
The nuclear studies pertinent to the
reflector-moderated reactor are
included 1n sec 4, ‘“Reflector-
Moderated Circulating-Fuel Reactor.”

As to reactor kinetics, 1t was
shown that the delayed neutrons
introduce damping of reactor power
oscillations, even 1n the case of
large initial amplitudes, and, further-
more, one typical example indicates
that this damping does not interfere
destructively with the damping caused
by fuel circulation. Also, the damping
can be demonstrated i1n a typical case
in which the power and flux vary along
the path of a circulating fuel. The
techniques employed are very similar
to those reported i1n the previous
ANP quarterly reports and are therefore
not repeated here.

39
-
¢+ 3W# 4. REFLECTOR-MODERATED CIRCULATING-FUEL REACTORS

A. P, Fraas, ANP Division
C. B. Mills, ANP Division
A, D, Callihan, Physics Division

Homogeneous reactors are the most
simple nuclear reactors with respect
to 1nternal structure because there
are no structural elements 1n the
reactor core. Although the con-
s1derable experience achieved with
liquid fluoride fuels i1ndicates that
they do not lend themselves to the
design of small, homogeneous reactors
because of their poor moderataing
properties, 1t has been shown that, by
using thick and efficient reflectors,
1t 1s possible to design a small,
fluoride- fuel reactor that will be
relatively free of structural com-
plexities. Further, the combination
of such a reactor with a spherical-
shell heat exchanger would provide a
reactor, heat exchanger, and shield
package that would be much smaller and
lighter than any reactor now being
considered for the nuclear propulsion
of aircraft,

This section summarizes the work
recently carried out on this type of
power plant. The first part covers
the work done on the static physics of
the reactor, including the effect on
the reactor of core size and the use
of various materials 1n the core and
reflector, The second part presents
the results of the critical experi-
ments and correlates them waith the
multigroup calculations. The thaird
part covers the mechanical design
envisioned and the developmental work
1nitiated to provide a basis for the
detalled design of a full-scale reactor.
The fourth part covers the shielding
work todate, and the last part presents
some possible full-scale aircraft
power plant arrangements, including
engines and radiators,

(I)D K Holmes, The Multigroup Method as Used
by the ANP Physics Group, ANP-58 (Feb 15, 1951)

STATIC PHYSICS
C. B. Mills, ANP Division

The most simple reflected reactor
1s a small sphere of fissionable ma-
terial surrounded by a concentrac
sphere of moderating material. The
fast neutrons leave the fuel-bearing
region easily and diffuse ainto the
reflector. If the reflector 1s
relatively thick and does not capture
the essentially thermalized neutrons,
the chain reaction can be supported
because of the sufficiently great
probability that the low-energy
neutrons will diffuse back i1nto the
central fuel-bearing region. The
fission cross section,as well as the
other cross sections 1n the fuel
region, 1s so haigh that the slow
neutrons re-entering the reactor cannot
easily re-escape.

This reactor has been briefly
evaluated with respect to moderator,
fuel, and geometry effects to obtain a
qualitative understanding of neutron
diffusion and loss. The analysais
involved the use of the ORNL-ANP
multigroup method,(!? supported by
critical experiments.

Neutron diffusion 1n a reflector-
moderator can be most easily described
by referring to the results of the
first multigroup calculation for a
32-1n.-dia fuel region surrounded by a
12-1n,-thick beryllium oxide reflector.
Figure 4.1 shows the leakage spectrum
as a function of lethargy for neutron
leakage from the central fuel-bearing
region i1nto the reflector. The net
current 1s very high across the boundary
for high-energy (low-lethargy) neutrons.
Most of the fission neutrons move
immediately into the reflector, where
they are moderated to thermal energy.

41
ANP PROJECT QUARTERLY PROGRESS REPORT

DWG 18T7i6
025
020
0 REACTOR 97 (SEE TABLE 4 1)
3 N
4 e”r i
|._
2 y 235
u 251ib OF U
g 015
<L
E
Z
Au
T
a 010
> ey
(L)
vl
<
I
= -0 2509
- THERMAL
'% 0 05 b———| ESGCAPE AT r
z 17 =18 6\
)
o
u'l B
(4]
<
=
Y 0 LH ——
-005
20 18 16 14 2 10 8 6 4 2 0
LETHARGY, v

Firg. 4.1.

They then either diffuse back i1nto the
fuel region, since the fuel region at
the center 1s a sink for thermal
neutrons, or out of the reactor across
the outer reflector boundary.

Figure 4.2 shows the spatial distra-
bution of nevtrons at three lethargies.
The thermal flux 1s very high in the
nonabsorbing berylliumoxide reflector.
The lethargy distribution of fissoning
absorptions and neutron escape for an
assumed neutron source distribution

42

Leakage Spectrum to Reflector vs.

Lethargy.

!

(Fig. 4.3) 1s given on Figs. 4.4 and
4,5. Intermediate lethargy fissioning
processes are 1mportant, but 50% of
the fissions results from thermal
neutrons streaming into the central
region from the reflector. The escape
spectrum, which 1s i1mportant for the
reactor shield, shows that avery small
fraction of fast neutrons escapes from
the reflector. The relative values of
fast-neutron escape through a l-in.-
thick layer of boron carbide for a

¢
L% )

PERIOD ENDING MARCH 10, 1953

DWG 18717

REACTOR 97 (SEE TABLE 41)

¢"vs n

25 |b OF U235

/ \ v=18 6, THERMAL

10 v=7 TO 10

v=17015

= \

NEUTRON FLUX FOR AN AVERAGE OF ONE FISSION /cm® sec

5
//

‘\\\\‘--_

0 2 4 6 8 io 12 14 {6

18 20 22 24 26 28 30 32 34 36

SPACE POINTS,n(r=214 n,cm)

Fig. 4.2.

beryllaum reflector-moderated reactor
as compared with a reactor with a 6-
in.-thick beryllium oxide reflector
and a moderating core are shown ain
Fig. 4.6. The fast escape 1s even
smaller for some of the subsequent
reflector-moderated reactor designs
described below.

In summary, the neutrons are born
1n the central region, spend the major
part of their life in the reflector,
and, after moderation, diffuse into
and out of the fuel-bearing region.
The prompt-neutron lifetime 1s about
2 x 10°% which 15 a relatavely
high value.

There 1s a positive component 1n
the temperature coefficient of re-
activity of this reactor because there
1s self-shielding of the fuel and a

sec,

Neutron Flux Distribution for Three Lethargy Values.

part of the fissions 1s caused by fast
neutrons 1n the “Doppler-region,’ The
positive component arises because the
increased thermal motion of the U??°
atoms effectively broadens and flattens
out the resonance peaks 1in the fission
cross sections, particularly those
peaks 1n the high-neutron-energy
region. This decreases the self-
shielding with respect to neutrons of
energles corresponding to the peaks
and i1ncreases the reactaivity.
Theoretical estimates and critical
experiments have thus far failed to
guarantee that this positive component
of the temperature coefficient of
reactivity 1s small compared with the
negative component expected from the
fuel expansion. However, this reactor
1s noworse with respect to the Doppler

43
ANP PROJECT QUARTERLY PROGRESS REPORT

DWG 18718

th3 7z 3
22

20

®
*

NORMALIZED TO ONE SOURCE
NEUTRON PER ¢m® OF CORE

REACTOR 97 (SEE TABLE 4 1)

16 25 |b OF U238

/

o8

I

ASSUMED

e T 77

NUMBER OF NEUTRONS PRODUCED PER cm3® OF CORE %
n

06 l
CALCULATED [
0a |
fotl— CORE ' REFLECTOR ~——t——1m
|
0 2 4 6 8 10 12 14 ie 18 20 22 29 26 28
SPACE POINT, n
Fig. 4.3. Spatial Power Distribution.

temperature coefficient of reactivity
than the other reactor being studied.
A small amount of U238 3dded to the
fuel or placed 1n the structure near
the fuel, would compensate for a
positive temperature coefficient, be-
cause with U238 3t 1s the absorption
that 1s self-shielded, and the decrease
in the self-shielding with increasing
temperature wouldreduce the reactivity,

Very low absorption and very good
moderation must be obtained in the
reflector. Structural material at the
core-reflector interface should also
have a low absorption cross section,
although the importance of the cross
section at the i1nterface 1s reduced by
the small probability of several
thermal-neutron transits because of

44

the high absorption and low albedo of
the fuel region. The structural ma-
terial in the fuel-bearing region will
compete with the uranium for thermal
neutrons,

Reactors with Various Reflector-
Moderators Computations have been
made for a number of possible reactor
core si1zes and reflector-moderator
compositions, and the results are
summarized i1n the following

1. The leakage of fast neutrons
from the reflector as a function of
the slowing-down power, 523, of the
reflector 1s shown in Fi1g. 4.7. A
factor of 1/5 1n gzs results 1n a
factor of 400 in the total escape of
fast neutrons, that 1s, neutrons with
lethargies smaller than 4. Figure 4.8
PERIOD ENDING MARCH 10, 1953

&N
DWG 18719
012
2 o1
Q REACTOR 97 (SEE TABLE 4 1)
5 010
w N
- n VGF
= 17
Z 235
= 008| 92% THERMAL 251b OF U
al
x 007
a
% 006
G
-
a—j 005 |-'1 = | i
|
~ 004
2
x 003
Ll
a
w 002
=
=}
@ 00t
* 0
19 18 17 16 15 14 13 12 1 1O 9 8 7 6 5 4 3 2 1 0
LETHARGY, v
Fig. 4.4. Fi1ssion Spectrum vs. Lethargy.
(.£ DWG 18720
(=]
el
=
-}
L
P-4
3 REACTOR 97 (SEE TABLE 4 1)
= v
= 0 2985 £
& 0007 — THERMAL v 035
x ESCAPE AT 251b OF U
o 0006— u=186 [
> -‘"qu
[©]
€ 0005 =
I
T 1|
ly 0004 - .
- et
Z 0003 —
&
W 0002
Ll
2 ooo
4
u 0
18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 {4 O
LETHARGY, v

Fig. 4.5. Leakage Spectrum to Shield vs. Lethargy.

shows the reflector thickness for a
given neutron current escaping from
the reflector as a function of the
moderating properties of the reflector
material, These properties are charac-

terized by the age-to-thermal in the
reflector material,

2. The absorption in the reactor
fuel 1s sensitive to the slowing-down
power, £, of the reflector only for

45
ANP PROJECT QUARTERLY PROGRESS

REPORT

s \
a vtz DWG 18721
4 8X10
—
4 4
CIRCULATING -FUEL ANP REACTOR
S 40 (REACTOR 69) —
L]
“;:; 36 e = =me == Be0 REFLEGTED=-MODERATED REAGCTOR
& (REACTOR 129, SEE TABLE 4 1) -
g 32
5
£
= 28
S o
1
x 24
L&)
Z 20
[
5
16
2 B -]
12
os
04x10'2
o —— eamam Spmass m— _1|_-_“--‘_fl
19 18 {7 16 15 14 13 M 10 9 8 7 6 5 4 3 2 { o}

Frg. 4 6.

LETHARGY, v

Comparison of the Neutron Current from Two Reactors Into the

Bprate%-Water Shield at the 400-Megawatt Power Level.
3 R

poor moderators, The absorption de-
creases rapidly for &2 < 0,08 (Fag.
4.9),

3. The escape spectrum vs. lethargy
1s shown in Fig, 4.10 for the two best
reflectors, beryllium and beryllium
oxide. The differences in the spectra
are characteristic of the differences
in the cross sections and the moderat-
ing power of the two materials, Beryl-
Iium 1s a much better reflector ma-
terial than beryllium oxide.

4. The effect of reflector material
and thickness on the escape of fast
neutrons 1s shown in Fig. 4.11. 1In
order of desirability, the materials
are beryllium, beryllium oxide, 6 1in,.
of beryllium plus 6 1n,
beryllium oxide aggregate, 4 in, of
beryllium plus 8 in. of carbon, and
carbon, NaOD 1s a good reflector only
with regard to neutron escape, The
critical mass with a NaOD reflector 1s
high (60 1b) because of the sodium ab-
sorption,

of carbon,

46

5. The effect of the moderator
indicates that there 1s little de-
pendence of the power distribution on
the reflector material in a saimple
system. The presence of structure and
the effect of operating temperature
are expected to favor beryllium as the
reflector-moderator, 1f a minimum ratao
of peak-to-average power density 1s to
be obtained.,

6. The effect of reflector thick-
ness on reactivity (Fig. 4.12) shows
that the beryllium or beryllium oxide
reflectors must be over 10 in. thick
but that little 1s to be gained by
exceeding 16 in, in thickness.

Reactors with Beryllium Reflector-
Moderators. The results given above,
as well as other considerations, indi-
cate that beryllium 1s the best re-
flector-moderator. Several beryllium
reflector-moderated reactors differing
1n si1ze and design were studied, and
the reactivity coefficients were

estimated, Although the numbers given
PERIOD ENDING MARCH 10, 1953
v R
DWG 18722
010
W st CARBON
oY b
"
=
>
o
oS
g 0 ol
3 &
3 8 l‘
z £ \
[ s
£ > ‘
oD o
2 ¥ \
LR e{-kBERYLLIUM OXIDE
v e
u g \
[« TS
2 - \
< ¥ ooo \
o ©° AY
w o AY
E F AN
2 AN
L N\
™
§ ‘\\3::,—BERYLLIUM
S ~~
\\
000 — -
O 02 03
2
Fig. 4.17. Total Neutron Leakage of F1ssion~Energy Neutrons vs, Slowing-
Down Power.
DWG 18723
30 ‘
CARBON"——mgy
25 ,/'
E /,‘-’ DWG 18724
20 ’,/’ g 10
[92] P -
: g
35 e—BERYLLIUM OXIDE g
F15 /’ w
- :
& F~—BERYLLIUM z o T e °
510 50° 3
o - B
w h ol
o Qo
5 2
)
=
S
0 = 0
0 100 200 300 400 0 01 0z 03 04
FERMI AGE {cm?) %
Fi1g. 4.8. Reflector Thickness for
Constant-Fast Neutron Leakage vs. Age- Frg. 4.9, Neutron Absorption 1n

to-Thermal of the Reflector.

the Fuel vs.

Slowing-Down Power.

47
ANP PROJECT QUARTERLY PROGRESS REPORT

OWG 187!5

e
(]
ut
N
3
Z &
S g oot
" S 0009 THERMAL
S e ESCAPE AT BERYLLIUM OXIDE
. ° 0008 v=186 (REACTOR 101, TABLE 4
o EJ Be0=033t3
g » 0007 Be = 02982 | /
5 8 0006 Y
a l //
=~ w
) _I A
% 3 -
o 0004 I
Ll
zZ 0003 Th,
8 “—u...._‘_
5 o002 1
w S
z N\BERYLLIUM T
0001 (REACTOR 104, TABLE 4 1) Ifi
o | | =
19 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0
LETHARGY, v
Fig. 4.10 Neutron Escape Spectrum to the Shield for Beryllium Oxide and

Beryllrum.

below refer to the specific reactors
considered, the general conclusions
are believed to be valid for a varaiety
of reflector-moderated reactors.

1. The ratio of the peak-to-average
power density can be decreased by re-
pl'acing the central volume of the fuel
with a moderator, The ratio of the
peak-to-average power can be reduced
to about 1.3, and the critical mass
can be reduced to about 1/2 the corres-
ponding value for a comparable reactor
without a center 1island.

2, The effect of changes 1n various
components on the ratio of peak-to-
average power density 1s of interest,
For a basic design, in which the d1-
ameters of central moderator, fuel
region, and reflector were 14, 22, and
46 i1n., respectively, and there was
2.8 vol % Inconel in the NaF-UF, fuel-

coolant (to simulate structure between

48

fuel and reflector), the relative
changes are indicated by the fractions
gf%en bel ow.

fractional change i1n the parameter
noted. The numerator refers to the

The denominator 1s the

ratio of the peak power density to the
average power density and represents
the fractional change of this ratio
caused by the change indicated in the
denominator.

PARAMETERS

NP/P

Amount of U233 in the fuel ———z—-= ~ 0,16
N/M
NP /P

Fuel layer thickness -——1—-= ~ 0.34
AT/T

Struct ght fract /P 0.067

ructure wei raction —— =~ (),

No/p
AP/P

Radius of central moderator —— = ~ 0,117
OR/R
OWG 18726

5 \\
%ARBON
2 R\ \
2 \ =—4m Be+8in C
0 N—A——
< ALY 1 I
v AN AN i |
3 g ST \-BeC AGGREGATE
v VY 3§
o BeO Py \
W \\\ \ h
g \Y‘& \F—Slln Be+6mn C
: ° A\ W AN,
a \V—A‘NUOD
Zz 102 HIGH-DENSITY N
o CARBON —— s
=) X
w A
£ .
% \
<
u.
2
\-—BERYLLIUM
o) 4 \‘
\
5
\
. \\
1073
(o} 4 B 12 16 20

REFLECTOR THICKNESS (in)

Fig. 4.11. Fast-Neutron Leakage

vs. Reflector Thickness.

10
REFLECTOR 120
=
X 0
< R\REFLECTORS
147 AND 101
REFLECTOR 110
-10
o} 10 20
REFLECTOR THICKNESS (in)
Fig. 4.12. Effect of Reflector

Thickness on Multiplication Constant.

PERIOD ENDING MARCH 10, 1953

The computed values of peak-to-average
power density at the two sides and at
the center of the fuel-coolant reg:on
are tabulated in Table 4.1.

3. The reactivity coefficients of
desi1gn interest are, where p1s density,
M 1s weight, and T 1s thickness

Ak /R

For core structure, ——— =~ 0,052 ,
Do/
Ak/k

For fuel layer thickness, —— =~ -0.17 |,
AT/T
Ak /R

For beryllium reflector, ——— =~ (.52 |,
bo/p
Ak /k

For U?35 content, ——JL- =~ 0,22 ,
/M
Ak [k

For NaF coolant, =~ 0,106 .
Do/p

4, The substitution of 6 or 8 in.
of carbon for beryllium 1n the re-
flector increases the critical mass by
15 and 25%, respectively,

5. Most of the structural materials
that might be used between the fuel
and moderator have a high absorption
cross section, and therefore some of
the thermal neutrons coming from the
reflector are absorbed. The loss of
all the thermal neutrons would increase
the critical mass by a factor of at
least 10, A 1/4-in.-thick layer of
Inconel might result in a reactivity
loss of as much as 0.24, which implaies
an 1ncrease 1n Crltlcal mass Of the
order of 100%.

6. A negative component of the
temperature coefficient of reactivity
1s obtained from the thermal expansion
of the fuel, for two reasons (1) the
loss of scattering centers for fast
neutrons and (2) the loss of fuel from
the active volume. If an expansion
coefficient for the fuel of 2x10°*/°F
1s assumed, the temperature coef-
ficients resulting from these two
effects are, respectively, 0.2 X
10°*/°F and 0.4 x 10°*/°F, a total of
0.6 X 10*/°F, The first of these
reactivity coefficients 1s smaller for
some other reactors (for example, the

49
08

TABLE 4.1. STATIC PHYSICS OF SEVERAL REFLECTOR-MODERATED REACTOR DESIGNS

REACTOR FUEL REGION REACTIVITY COEFFICIENTS POWER DENSITY RATIOS ESCAPE
CALCU- REACTOR REFLECTOR Outarde | Tnarde | Cmaach Freatal | Thermal Base U3 Mass Over all(®) | Inside | Outside | Mim TOTAL OF FAST
LATION TYPE (1fe thick) Diameter | Diameter (15 ) %) Ok/k Ak/k Ok/k Peak Peak- man NEUTRON NEUTRONS(® )
NUMBER o . te to ° ESCAPE o2 10
(2o} Gn ) &8 F) DM/M Ag(°F) Average | Average | Average ¢ Lo Mev)
103 Circulating fuel NaOD 22 14 a 60 40 96 x 109 0 20 126 116 0 90 0 397 0 00228
3 regions
121 Clrcul:(fl.}ing fuel Be(Na cooled) 22 25 45 06 x10° 0 20 -7 x 108 2 16 0 52 0 299 0 0002996
(str €¢)) 2
regions
124 Fuel plates (Ne Be{Na cooled) 19 9 17 k1] 75 %x 108 0 25 119 1 59 0 75 0 264 0 0002596
cooled) 3
regions
117 Circulating fuel Be(Na cooled) 22 14 14 8 59 65 x 10 ® 0 42 -8 x 10§ 118 135 083 0 299 0 000291
(etr )} 3
reglons
118 Carculating fuel De (Na cooled) 19 11 12 58 176 x10° 0 27 114 1 29 0 85 0 335 0 000328
(str } 3
regions
119 Circulating fuel Be(Na cooled) 22 11 17 52 69 x 108 Q26 1 14 129 0 85 0 303 0 000280
(atr )} 3
reglons
120 Carculating fuel Be{Na cooled) 16 ain 22 14 13 60 115 = 10 ¢ 0 25 115 1 29 0 86 0 222 0 0000149
(str ) 3 thick
regrons
105 Circulating fuel Be(Na cooled) 22 14 9 65 87 x10°¢ 0 30 -8 x 10 % 114 128 0 86 0 308 0 00029
3 regions
95 Circulating fuel BeQ 16 22 28 44 x10° Q 27 -7x10 % 1 95 0 50 0 447 0 0043
2 regions
97 Carculating fuel BeO 32 18 5 53 -21 %10 ¢ 0 47 -12 x 190 % 17 0 60 0 345 0 0020
2 regions
101 Carculating fuel BeO 22 14 9 4 60 -0 5 x 10 ¢ 0 30 -8 x 10-% 110 1 30 0 8BS ¢ 190 0 0025
3 regions
109 Carculating fyel BeO 48 24 160 40 14 x10° 2 34 211 0 55
(air cooled) 3
regions ~
108 Fuel plates {Na BeO 22 14 a5 40 19 x10 ¢ 0 20 -9x 105 121 1 69 071 0 352 0 0029
cooled) 3
regions
e
115 Circulating fuel BeO aggregate 22 14 15 30 7 x 10 § 0 20 107 121 0 90 ¢ 473 0 01037
3 regions
102 Circulating fuel c 32 20 33 40 -17 x 10 ¢ 0 20 -5x 10 % 13 12 0 90 QO 450 0 040
3 regions
£22 Carculating fuel C Be (high p} 22 14 13 57 65 »10°8 0 56 115 129 0 86 0 297 0 00167
(etr ) 3
regions
127 Circulating fuel Be{(4 1n ) C(8 1n ) 16 31 35 408x10% 0 20 1 86 0 52 0 488 0 010744
2 regions
128 Circulating fuel Be(6 1n ) C(6 1n ) 16 25 5 kL] 70l x10°% 0 20 19 0 49 0 440 0 00436
2 regions
129 Circulating fuel Be (No cooled) 19 8 23 47 8 02 x10°* 016 -4x10°* 0 92 1175 0 65 0 343 0 000501
(str )} 3
regions
137 Critical experi- Be 16 10 30 47 100 x 10 ¢ 127 1 00 o 87 0 367 0 001191
ment 20 mal
foil end sodium
cans 3 regions
140 R 118 with p{NaF) Be(Na cooled) 1% 11 15 36 105 x10 % 0 96 1 34 0 83 0 343 0 0003866
X 1/ 3 regions

(a)poes not include the Doppler effect
(b)Through 1 1n of boron carbide

c)In these calculations, structure was added to the fuel to simulate structure between fuel and moderator

LUYOdIY SSTYI0Ud ATHALYVNO I1DAf0Yd dNV
ABE¥$%§¢§ factor of 10, but the second
1s approximately the same,

7. The mean lifetime of the prompt
neutrons 1s about 4 X 10°* sec for the
beryllium-reflected reactor with a
center 1sland. About 1/2% of the
prompt neutrons has a mean lifetime of
the order of 10°? sec because of the
low absorption and leakage probability
1n the reflector.

8. A cooling system must be pro-
vided for the moderator because of
gamma and neutron heating, about 4% of
the total power appears in the moderator
volume. Insertion of such a system
means an i1ncreased amount of structure
1n the reflector, which, an turn,
causes a loss 1n reactivity and a gain
1n gamma-ray intensity., This problem
1s being studied.

9, One aspect of importance from
the shielding standpoint 1s the energy
spectrum of the neutrons escaping from
both the surface of the thick reflector
and through the fuel-circulation
passages at the reactor ends. Thas
spectrum 1s strongly thermal for the
reflectors of interest. The addition
of 1 1n. of boron carbide at the
reflector boundary reduces the total
leakage through the sides of the
reactor to the values listed in Table
4.1 as “escape of fast neutrons.” The
use of a nonpoisoning, heavy material
(lead or bismuth) as the reflector
coolant may be beneficial from the
shielding standpoint because 1t should
moderate fast neutrons by 1nelastic
scattering and also serve as gamma
shielding. The fast leakage through
the fuel passages and from fissioning
1n these passages must be minimized by
suppression of fission and by the
addition of extra shielding materaial,
This problem 1s not easily adaptable
to calculation, and hence craitical
experiments will be performed to
evaluate leakage control methods.

10. A feasible reflector-moderated
reactor can be built by replacing the
circulating-fuel coolant (for example,

PERIOD ENDING MARCH 190, 1953

NaF-UF,) withuranium-bearing stainless
steel fuel plates and sodium coolant.
A design 1ncorporating this feature
has been proposed by A. S. Thompson.
Shim control for such a reactor can be
accomplished by varying the concen-
tration of potassium 1n the sodium
coolant, By this means, the relative
burnup could be made very large.
These results imply that the goal of a
simple, structureless core 1s 1im-
practical because the fuel solvent 1s
1nadequate as a moderator. An 1island
of solid moderator in the center of
the core seems to be desirable, and
the usefulness of such an 1sland 1s
not entirely nuclear. As will be shown

later, the hydrodynamics of a re-
flector-moderated reactor are also
aided by a central 1sland.

Summary Tables and Graphs. The

multiplication constants of two- and
three-region reflector-moderated
reactors as a function of uranium
weight in the fuel region for several
core and 1sland sizes and moderator
compositions are given 1in Fig. 4.13.
The various reactors referred taq 1n
this figure are described in Table
4,1. The figure summarizes a number
of point values for reactivity. To
faci1litate the extrapolation, the
curves drawn through point values
indicate the manner in which the
multiplication constant varies with
urantum weight for the partaicular
reactor to which the point value
refers. If only two-region reactors
are considered, the shapes of the
curves vary systematically with the
position of the point value 1n the
uranium weight vs. keff plane, Hence,
only a few of the curves were actually
computed, and the others were drawn by
analogy. Likewise, a few keff Vs,
uranium weight curves can be computed
for three-region reactors and then
simi1lar curves can be drawn through
the point values. The shapes of the
curves for three-region reactors are,
of course, different from those for

51
ANP PROJECT QUARTERLY PROGRESS REPORT

DWG (18728
12
l//pw4
123
1 97.
//}05
101
122
{
‘o / )
V&'
® /
«° /108
g 1104
Z // 14
%
g 09 //nemfl
Q
=
o
—
I
o
-J
a 08
|_
-
e )
= SEE TABLE 41 FOR DESCRIPTION
OF REACTORS BY NUMBER
15
07
103
o6
o 5 10 15 20 25 30 35
URANIUM WEIGHT (Ib)
Fig. 4.13. Uranium Weight vs. keff for Several Reflector-Moderated Reactors.

the two-region reactors. (Three-
region reactors have center 1slands.)

The curves described above are not
expected to be exact. It should be
noted that the IBM multigroup method
starts with an assumed power distrai-
bution, and, 1fthe calculation results
1n a power distribution different from
the assumed one, an 1teration must be
made by using the computed distrai-
bution as a start, Such 1terations
were carried out only where essential.
Whenever possible, fuel-coolant con-
stituents were held constant to
emphasize the effects of main interest.
It 1s to be noted that any fuel self-
shielding or poisoning effect, or any
change 1n reactor size, 1s reflected
by a rapid change 1n reactivity and a

52

rapid change i1in the slope of the curves
of keff vs. U235 pass,

Typical results of the multigroup
solution of the neutron diffusion
processes are given in Figs. 4,14 to
4,17, Figures 4.14 and 4.15 give
spatial power distribution and flux
spectra, respectively, for one extreme
case — a reactor with a small (8-in.-
dia.), central, moderator region, a
thick (5 1/2-1n.), fuel-bearingregion,
and sodium-cooled fuel plates. The
relatively high peak-to-average ratio
1in the power distribution curve,
Fig. 4.14, emphasizes the value of a
thin fuel layer. Figures 4.16 and
4.17 give the computed power and flux
distribution for the first craitical
PERIOD ENDING MARCH 10, 1953

DWG 18729
20 | i
¥ let— ASSUMED
w 18 | |
x
S let—CALCULATED
w16
" REACTOR 129 (SEE TABLE 41)
©14
o
(V8]
N \
a2
(&
o] \ // ¥NORMALIZED TO ONE SOURCE
% 10 \ / NEUTRON PER c¢cm3 OF CORE
2 \ \ //
% g Ar=2032
06
@]
o
@04 BeO } | |
3 e CENTRAL | FUEL -l -l
2 MODERATOR | REGION i BeO REFLECTOR |
o) 2 4 6 8 10 12 14 16 18 20 22 24 26 28
SPACE POINT, n{r=nlAr.cm)
Fig. 4.14. Spatial Power Distribution for a Three-Region Reactor with a

Thick Fuel Annulus.

experiment calculation, for which 1-
by 3- by 3-in. sodium-filled cans and
10-m1l-thick, 3-1in.-dia, U235 fuel
disks were used. A comparison with
the experimental results 1s given 1in
the next section,

CRITICAL EXPERIMENTS

D. V. P. Wailliams R. C. Keen
J. J. Lynn

Physics Division
Dunlap Scott, ANP Division Experiments
C. B. Mills, ANP Division Computations

First Critical Assembly. A pre-
liminary craitical assembly of the
reflector-moderated circulating-fuel
reactor was described previously. (%)
Some i1nformation already reported 1is

repeated here to givea unified picture
of the first reflector-moderated
reactor critical experiment., The
fuel consisted of 0.020-1n.-thaick
pieces of U?3% petal lumped between
l-in. layers of sodium canned 1n
stainless steel. This arrangement
was shown to be 1nefficient 1in neutron
utilization because of the self-
shielding in the thick uranium layers.
The self-shielding of the fuel was
measured experimentally by replacing
one of the 20-m1l fuel disks with
10 disks that were each 2 mils thaick
with aluminum catcher foils between
them. The activaity of the catcher
foi1ls showed that the 20-mi1l layers

2)p ¥y P Willaems, R C Keen, J J Lyan,
D Scott, and C B Mills, ANP Quar Prog Rep
Dec 10, 1952, ORNL-1439, p 48

53
ANP PROJECT QUARTERLY PROGRESS REPORT

30

£
DWG 18730

22
u 20
& N\
O REACTOR 129 (SEE TABLE 41) / \ LETHARGY GROUP ./ = 92
. [ 1\
2]
€ 16
14 /' \
g, Ar=2032cm \
2
: / \
D2
o / \
W \
Z {0
O
z / \
o 8 N\
% N N=3 !
3 6 \\ T T—
w N=12 Iy
55 S
E 4
a N=20 !

0 l — ‘ ::_= \
0 2 4 6 8 10 12 {4 16 18 20 22 24 26 28
SPACE POINT, An{r=nar,cm)
Fig. 4.15. Spatial Flux Distribution for a Three-Region Reactor with a

Thick Fuel Annulus.

were 66% effective, the corresponding
theoretical estimate was 63%.

The critical mass was found experi-
mentally to be 15 kg of U?35, With
this loading, amultiplication constant
of 1.03 was calculated by the multai-
group method. The usual self-shielding
correction for the lumped fuel was
used. The calculation was carraied
out for a spherical shape, whereas
the actual geometry was rectangular.
This difference was accounted for by
reducing the fuel volume for calcu-
lation purposes by 10%. The computation
did not include any poisoning effect
of i1mpurities i1n the materials.

54

The neutron-flux traverses measured
radially through the mid-plane of the
reactor were given previously.(3)
The activations of bare- and cadmium-
covered-i1ndium foils and their differ-
ences were shown. These results are
typical of those obtained along other
traverses, and they confirm the
prediction of high neutron flux in
the moderator 1sland and reflector.
This effect 1s, as expected, particu-
larly pronounced for thermal neutrons.

The cadmium fraction, derived from
the 1ndium-fo1] activation data and
defined as the ratio of the activation

(3)

Ibid , Fag 5 3, p 51
PERIOD ENDING MARCH 10, 1953

DWG 18731
16
t 4
»*
W REACTOR 137 (SEE TABLE 4 1)
© /
S 12
Q
u Aslsumsol /
m_ {0 | |
g CALCULATED \ 47
"]
& osg
o
S * NORMALIZED TO ONE SOURCE NEUTRON
§ose PER c¢cm® OF CORE
[+
a
204
o
= ar=1757
D
Wwo?2
[T
o
& 0
m
=
3 |
=2
CENTRAL s | |
o MOBERATOR —J-_ COR? et REFLECTOR _7
1 }
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32
SPACE POINT, n{r=nar,cm)
Fig. 4.16. Calculated Spatial Power Distribution for the First Critical
Assembly.

by neutrons of energy below the cadmium
cut-off to the activation by all
neutrons, was given previously.(*)
The values of this cadmium fraction,
as calculated from the IBM multigroup
computations, were also plotted.
Theoretical and experimental values
practically coincide i1n the center of
the fuel and in the bulk of the
reflector. The discrepancy at the
fuel surface 1s not surprising, since
all practical calculation methods,
including the "age” multigroup method,
are not exact near boundaries,

The fissioningdistribution through-
out the fuel region 1s shown 1in
Fig. 4.18. The experimental values
are catcher-foi1l measurements. Since
the measurements were not normalized,

(4)

Ibud., Fig 55, p 53

there 1s no significance in the close
agreement of the absolute values of
the theoretical and experimental
curves. However, the agreement 1in
the peak-to-average ratios of power
density does seem significant, The
discrepancy in the shape of the curves
inside the fuel layer can be attributed
to the fact that there 1s more moderator
in the rectangular center island than
was assumed i1in the computation based
on spherical geometry.

To determine the reactivity loss
expected from structural material at
the core-reflector 1nterface, an
experiment was performed (Fig. 4.19)
with one-half of one of the four
outside surfaces of the core covered
with 1/4-1n. stainless steel plates.
The loss 1n reactivity was 220 cents,
which 1s consistent with the theoretical

55
ANP PROJECT QUARTERLY PROGRESS REPORT

AxT.
s

OWG 18732
20
m
> 18
t REACTOR 137 (SEE TABLE 41)
O
o '8 oV
G \ LETHARGY GROUP, ¥ =92
x4 \\
a
2
L \ TN
2/
W10 \ / \
L)
g / \\
(NS
o
5 6 N
o \\ \\\
fi
5 g yd SN
x N=12 N
a N=20 —
R <N S \
N=3 — -
ol ——] [~ — N

0 2 4 6 8 10 12 14

SPACE POINT, # (r=1 757 n, cm)

Fig. 4.17.

value of 24%, for an Inconel layer
over both the 1nner and the outer
fuel -reflector 1nterfaces.

A l-in.-thick layer of lead or
bismuthat the core-reflector interface
(where 1t seems to be most useful) on
one of the four outer surfaces of the
core cost 0.24% in k, for bismuth
and 0.91% for lead, as compared with a
voaid.

Second Critical Assembly. A second
assembly with fuel properties that
more closely resemble those of a
possible full-scale reactor was
constructed. In this second assembly,
the fuel was a mixture of 66 wt% ZrO,,
24 wt % NaF and 10 wt % C (the Zr0,-C
combination simulates ZrF,), with
sufficient enriched UF, added to
make the U?35 density 0.2 g/cm3. The
well-homogenized powder mixture was
packed 1n 1 1/4-1n. square aluminum

56

Calculated Flux Spectrum for the First Critical

Assembly.

tubes that were arranged around the
central beryllium core and were, 1n
turn, surrounded by the beryllium and
graphite reflector These regions
may be seen in Fig. 4.20, which shows
the loading 1n a plane through the
ax1s of the reactor. The maximum
cross section of the beryllium 1sland
was 9 by 9 in., and the maximum
thickness of the fuel region was
4 1/2 inches. (With materials availa-
ble, 1t was not possible to completely
fill the fuel region at all axial
locations.) The thickness in the
fuel region was decreased i1n both
directions from the center to simulate
the i1nlet and outlet ducts of the
full-scale reactor. Figure 4.21 1s a
photograph of the assembly at section
AA of Fi1g. 4.20, This assembly was

critical with a loading of 7.7 kg of
y23s,
&1; B DWG 18733
\\"f—_

W
Z 10
}_
<
-
i
x
> EXPERIMENTAL
5 (NOT NORMALIZED)
=
l
]
(U]
2
&05
n
W
w

0

9 10 " 12 13 14
SPACE POINT, 2 (r=1 757 n,cm)
Fig. 4.18. Measured Fissioning Dis-

tribution 1n the Fuel Region of the
First Critical Assembly. :
Neutron flux distributions measured
with bare- and cadmium-covered-indium
foi1ls along a horizontal traverse
lyang in the mid-plane of the assembly
are shown 1n Fig. 4.22. Note the
spatial distribution of the cadmium
fraction, that 1s, the fraction of
neutrons, detected by indium, that
have energies below the cadmium
cut-off. Both the experimental values
for the cadmium fraction and those
computed by the multigroup method are
presented. The difference 1s withan
experimental error, except at the
core-reflector interface where 1t 1s
about 40%. The experiments indicate a
neutron energy lower than that computed
for this point. (Recent data aindicate
that this diafference may be caused by

PERIOD ENDING MARCH 10, 1953

DWG 18734
225 /)
200 /
175 /
150

1g

LOSS OF REACTIVITY (cents)
ro
(]

T
o L
/

o] n 6 '/B 3 16 ‘/4

THICKNESS OF STAINLESS STEEL (i)

Fig. 4.19. Reactivity Loss vs,
Thickness of Stainless Steel at the
Fuel-Reflector Interface of the First
Critical Assembly.

the void at the interface, which 1s
peculiar to the structure of the
critical experiment.)

The experimental cadmium fractaion
curve shown in Fig. 4.21 1s repeated
as curve A 1n Fig. 4.23, which gaves
the distribution of the fraction of
neutrons, detected by indium, waith
energies below the cadmium cut-off.
Curve B of Fig. 4.23 was obtained
from the first critical experiment.
Comparison of the two curves shows
the similarity of the spectra an the
centerof the fuel i1n the two assemblies.
It 1s to be remembered that the fuel
layer 1n the first mockup was only
3 in. thaick.

A preliminary measurement of the
neutron leakage spectrum gave values
of the cadmium fraction for indium-
detected neutrons at the outside edge
of the reflector, both on the axais
and at the side at the mad-plane.
The respective values were 0.4 and

57
ANP PROJECT QUARTERLY PROGRESS REPORT

L
B

&
s

DWG 18735

-

)

BERYLLIUM

----------------------------------------------------------------------------------

i

Fig. 4.20.

0 9 A measure of the fission rate
or power distribution i1n a direction
parallel to the reactor axis has been
obtained from the i1mprovisation

1llustrated in Fig. 4.24. Uranium
metal disks, 1.4 i1n. 1n diameter
and 0.002 in. thick, 1in contact with

aluminum disks, 0.005 in. thick, were
placed adjacent to one of the fuel

containers, and the activity on the

o8

Vertical Cross-Section of the Second Critical Assembly.

aluminum, caused by recoiling fission
fragments, was counted. The low value
at the 17-i1n abscissa point probably
resulted from the shielding of ex-
ternally reflected neutrons by the
layer of fuel that was thicker there
than at the 20-i1n abscissa poaint,

In a similar manner, an attempt
was made to determine the fission rate
pattern i1n a direction perpendicular
Fig. 4.21.

PERIOD ENDING MARCH 10, 1953

Photograph of Second Critical Assembly at Section AA of Fig. 4.20.

59
ANP PROJECT QUARTERLY PROGRESS REPORT

4 DWG 18736
- = T .
< BERYLLIUM I FUEL - BERYLLIUM - GRAPHITE ™
iISLAND VoD REFLECTOR To 27
RADIAL INDIUM TRAVERSE
30
© BARE INDIUM, EXPERIMENTAL
A CADMIUM-COVERED INDIUM, EXPERIMENTAL
* EXPERIMENTAL CADMIUM FRACTION
25 A CALCULATED CADMIUM FRACTION
"
o |
oS W T, ‘o
s =
E o}
2 — 5
<
i v
c [P
g 15 \ L =
© = |
s
(o]
<
O
10 o 05
s A /
[ ]
0 — - o
0 5 10 15 20 25
DISTANCE FROM REACTOR AXIS (tn )
Fi1g. 4.22. Radial Indium Activation and Cadmium Fraction i1n the Second

Critical Assembly.

| 0 DWG 18737
BERYLLIUM FUEL BERYLLIUM GRAPHITE
ISLAND REGION REFLECTOR /o— REFLECTOR
08
=
o
o 06
& B -
w v (o]
S 04 ® o
2 T rheo T S
51
O "One 0&
~o 4
\‘“0 /7
02 ‘\%
6 4 2 0 2 8 10 12 14 16 18
DISTANCE FROM CENTER OF FUEL REGION (in)
Fi1g. 4.23. Comparison of the Cadmium Fractions i1n the Two Critical Assemblies,

60
PERIOD ENDING MARCH 10, 1953

DWG 18738

BERYLLIUM REFLECTOR

x4

L i

~

BERYLLIUM ISLAND

-~

7727777777727/ o
722222 2227277777702

T~

0 002-in URANIUM DISK
AND ALUMINUM CATCHER
FOIL

;
/

AN

T

H

DN

FISSION FRAGMENT ACTIVITY {counts per sec x10°3)
(9] o

n

0 5 10 15

20 25 30 35

DISTANCE FROM REACTCR INTERFACE (in)

Fig. 4. 24.
Assembly.

to the axis along a horizontal line
lying 1n the mid-plane. Uranium
disks and small (5/16 in. in diameter)
aluminum catcher foils were placed
between the fuel containers The
results show a center-to-edge activity
ratioof 0.45, that 1s, a 55% depression
in the power across a fuel layer
3.75 in. wide. By direct measurement
with 20-mi1l cadmium 1n the first
critical assembly, 1t was found that
70% of the fissions was caused by
neutrons with energy below the cadmium
cut-off. From the similaraity of the
spectra, 1t 1s suspected that about
the same distribution occurs in the
second case. No direct measurement
was made.

DESIGN CHARACTERISTICS
A, P. Fraas, ANP Divaision

Although the many complex and
interrelated considerations underlyaing

Measured Power Distribution Parallel to Ax1s of Second Critical

the design of the full-scale reactor
shown 1n Fig. 4.25 should logaically
precede a detailed description, an
exposition of these considerations
will be greatly samplified by early
reference to a fairly specific system.
Therefore, this section begins with a
descraiption of the reflector-moderated
reactor design and subsequently takes
up the more significant of the many
factors that have entered 1nto that
design.

Much work remains to be done on
reflector-moderated reactors to give
better bases for detailed designs,
but the construction i1ndicated by
Fig. 4.25seems to be the most promising
one considered, to date, for an
aircraft power plant. This cross
section through the reactor core,
moderator, and heat exchanger shows a
series of four concentric shells,
each of which 1sa surface of revolutzion,

61
¢9

INCONEL PRESSURE SHELL HEAT
EXCHANGER

SODIUM
7/ EXPANSICN TANK
7 T

SODIUM PUMP

FUEL EXPANSION TANK

— R
CONTROL ROD

BaC /

CONTROL ROD

BaC

INCONEL CORE SHELLS . I3
Yg-1n THICK \\ ¢

Bfin FUEL PUMP
L4
SODCIUM TG Nak
HEAT EXCHANGER
NaK OQUTLET
NoK INLET
10 6 12

INCHES

F1g. 4.25. Three Region Reflector-Moderated Reactor.

DWG 4830

LHOddY SSAYI0Ud ATHALYVNAO 12dAf0dd dNV
Thé Twd® inner shells surround the fuel
region at the center, that 1s, the
core of the reactor, and separate 1t
from the beryllium 1sland at the
center and from the outer beryllium
reflector. The fuel cairculates
downward through this region in which
the fissioning takes place and then
downward and outward to the entrance
of the spherical shell heat exchanger
that lies between the moderator outer
shell and the main pressure shell.
The fuel flows upward between the
tubes 1n the heat exchanger into two
mixed flow pumps at the top. From
the pumps, 1t 1s discharged inward to
the top of the annular passage leading
back to the reactor core. The total
fuel volume 1n the system for thais
design would be approximately 7 ft?3,
of which approximately 1 1/3 ft3 would
be 1n the reactor core. Most of the
remaining fuel would be i1n the 1n-
terstices between the tubes 1n the
heat exchanger. The moderator was
designed to be cooled by sodium
flowing downward through the annular
space between the beryllium and the
enclosing shells and back upward
through passages i1n the beryllium.
Two centrifugal pumps at the top
circulate the sodium first through
the moderator and then through the
small toroidal sodium-to-NaK heat
exchangers around the outer peraiphery
of the pump and expansion tank region.
A horizontal section through the
pump and expansion tank region 1s
shown 1n Fig. 4.26. Note that sump
pumps with gas seals are used. A
pump of this type recently completed
1000 hr of very successful operation
in a fluoride system, with pump inlet
temperatures of about 1400°F,

The praimary construction material
1s Inconel because 1t seems to be the
material that 1s most resistant to
fluoride corrosion, Beryllium was
chosen as the moderator material,
partly because 1t seemed to be the
best material obtainable, from both
shielding and craitical mass stand-
points, and partly because 1ts physical

PERIOD ENDING MARCH 10, 1953

properties seem to be superior to
those of any other material that maght
be used in that location, other than
graphite. It was decided that the
vase-shaped 1sland in the center
should be used, partly because 1t
reduced the critical mass and improved
the power distribution 1n the fuel
regiron, and partly because hydro-
dynamically 1t promised to give the
simplest and most desirable fuel
passage The 12-in.-thick beryllium
reflector followed by 1 in. of boron
carbide was chosen originally to
keep the neutrons escaping to the
heat exchanger region to a level
approximately equal to that of the
delayed-neutron flux from the circu-
lating fuel in that region. It has
since proved to be nearly an optimum
configuration from both cratical mass
and shielding standpoints.

The spherical shell heat exchanger,
which makes possible the compact
layout of the reactor-heat exchanger
assembly, 1s based on the use of tube
bundles curved i1n such a way that the
tube spacing 1s uniform, irrespective
of latitude.®3) The i1ndividual tube
bundles terminate in headers that
resemble shower heads before the
tubes are welded i1n place. Thas
arrangement facilitates assembly
because a large number of small
tube-to-header assemblies 1s made
leaktight much more easily than one
large unit. Furthermore, these tube
bundles give a rugged flexible con-
struction that resembles steel cable
and 1s admirably adapted to service
in which large amounts of differential
thermal expansion must be expected.
This basic tube bundle and spacer
construction was used 1n a small
NaK-to-NaKheat exchanger that operated
satisfactorily for 3000 hr with a Nak
inlet temperature of 1500°F. (%) Two

{(5)A P Fraas and M. E LaVYerne, Heat Ex-
changer Design Charts, ORNL-1330, Dec 7, 1952

(6)G H Cohen, A P Fraas, and M E LaVerne,
Heat Transfer and Pressure Loss trn Tube Bundles

for High Performance Heat Exchangers and Fuel
Elements, ORNL-1215, Aug 12, 1952

63
ANP PROJECT

- -

QUARTERLY PROGRESS REPORT

D!G 18739

Na TO ISLAND

FUEL TO CORE

——

N

S
&

>
<$

NN

Na K QUTLET

Fig. 4.26.

sodium pumps and two sodium-to-NakK
heat exchangers are provided so that
failure of one pump or one heat

exchanger will not completely disable
the reactor.

Factors Affecting Core Diameter,
The first factor considered i1n es-
tablishing the reactor core diameter
(the fuel region) was the effect of
diameter on critical mass,
much work remains

Although
to be done to
establish the core diameter, 1t
appears, at this time, that the critical
mass 1s essentially independent of
the core diameter. A more important
factor affecting critical mass 1s the

64

FUEL PUMP
IMPELLER
—_/

<%
q%mmmflfl/

B,

FUEL EXPANSION TANK

»
_~
<>
L
s““‘
<
<o
S

)
e

Na FROM PUMP

Na PUMP IMPELLER

Na PUMP VOLUTE

ORE INLET
PASSAGE

)

Na TO REFLECTOR

N AN R NN N

|

\\\,\\\\\\

NN
N

Na-TO-NaK HEAT EXCHANGER

FUEL PUMP VOLUTE AND CORE INLET PLENUM CHAMBER

Cross Section Through Pump and Expansion Tank.

presence and size of the i1sland. It
appears that the use of a beryllium
1sland with a diameter one-half that

of the fuel region cuts the critical
mass approximately in half. At the
same time, the 1sland reduces the
ratio of the peak-to-average power
density by a factor of about 1.5,
Further work on the effects of 1sland
si1ze and fuel

region thickness
under way.

1S

A careful scrutiny of the data

that have been obtained on the melting
point, viscosity, and thermal conduc-

tivity of various fluoride melts
indicates that 1t will probably not
be possible to obtain a fluoride fuel
with more than 4 mole % uranium that
has close to optimumphysical properties
from the heat transfer standpoint,
in fact, from the standpoaint of
physical properties, there appears to
be an i1mportant incentive for keeping
the uranium concentration less than
2 mole %. On the basis of critical
mass data from the calculations and
the critical experiments, 1t appears
that a fuel-region diameter of at
least 18 1n. wi1ll be requaired to
satisfy this condition.

A major consideration 1n es-
tablishing the core diameter 1s the
power density. Work 1s currently
under way to determine the effect of
power density on radiation damage.
Although the experimental results are
difficult to interpret and therefore
no clearly defined limit to the
allowable power density has been
established, 1t 1s entirely conceivable
that radiation damage considerations
w1ll not be the limiting factor.

The kinetics of reactor control
are very complex. Work carried out
thus far on control of the reactor
indicates that transient power surges
may cause unacceptably large tempera-
ture excursions 1f the temperature of
the circulating fluoride fuel exceeds
1000 to 2000°F per second. A tempera-
ture rise of 2000°F/sec in the fuel
would i1mply a power density of approxi-
mately 3 kw/cm®. This, in turn, would
mean that a core diameter of about
21 i1n. would be required for a 200, 000-
kw reactor.

The problemof cooling the moderator
becomes more severe as power density
1s 1ncreased. As the power density
1s 1ncreased for a given reactor
power output, the percentage of the
fission energy that appears as heat
1n the moderator 1s increased and the
intensity of heat generation in that
portion of the moderator closest to
the fuel regionis essentially directly
proportional to the power density in
the fuel. Althoughit seems practicable

PERIOD ENDING MARCH 10, 1953

to design and build the moderator
cooling system for a reactor having
power densities of up to 8 or 10 kw/cm?,
moderator cooling considerations
seem to 1ndicate that power densities
higher than 10 kw/cm® would be 1inad-
visable

Temperature variations in the fluid
fuel 1n the core will be a function
of both the velocity distribution and
the power density With a gaiven
velocity distribution, 1t would be
expected that certain localities such
as the boundary layer or the centers
of eddies would run substantially
hotter than the bulk mean temperature
of the fuel The magnitude of thas
temperature differential will depend,
to some extent, on the thermal conduc-
tivity of the fluid, but the rate at
which heat 1s dispersed by turbulent
interchange of particles between
hot and cold regions will probably
be much more important The Reynold’s
number for a high-power reactor will
be somewhere between 10° and 10°,
depending on fuel viscosity and design
values for the reactor core diameter
and the fuel temperature rise. At
present, 1t 1s not possible to make
more than the roughest sortof estimate
of the magnitude of the temperature
differenti1al between these hot spots
and the free stream. The problem 1s
being attacked both theoretaically
and experimentally, and 1t 1s hoped
that results from at least the latter
wi1ll place an upper limit on the
magnitude of the effectsto be expected.
One possible 111 effect with respect
to control could arise from these
hot spots If boiling were to take
place, 1rregular fluctuations ain
power level that would probably be
similar to those experienced 1n
boiling experiments made with the
LITR(?) would be expected to result.
There 1s also a possibilaty that

(Dw M Breazeale, T E Cole, and J A. Cox,
Journal of Reactor Science and Technology, Vol 1I,
l?o 2), TID-2002 (1952), Vol II, No 4, TID-2004
1952

65
ANP PROJECT QUARTERLY PROGRESS REPORT

coupling of the oscillations 1n power
with the periodic fluctuations 1n
fluid flow that caused the eddies or
other hot spots might take place.
After the various considerations
outlined above were examined, 1t
appeared that only i1n a full-scale
experiment can the maximum, allowable,
power density in a fluid-fuel reactor
core be definitely established. Thas,
in turn, i1mplies that the first
experiment of this sort should be run
with the smallest practical core size
because the cost of the experiment
would be largely a function of the
total amount of heat to be dissipated.
Hence, the cheapest experiment to
yield a definitive answer will be the
one carried out with the smallest
reactor core that might reasonably be
expected to yield a power plant of
adequate size Since fuel chemistry
and critical mass considerations
indicate that an 18-1n -dia core 1is
the smallest that can be used with
fluoride fuels that have desairable
physical properties, most design work
has been based on reactor core sizes
ranging from 18 to 21 i1in in diameter
Temperature, Pressure, and Stress.
Two major tenets of the design phi-
losophy have been that the pressures
throughout the fluid systems should be

zones, and that all structure should
be cooled to a temperature approxi-
mately equal to or below that of the
secondary coolant leaving the heat
exchanger. The temperature, pressure,
and stress values calculated for
various stations in the reactor are
indicated in Fig 4 27. The stresses
1n the structural parts have been
kept to a minimum Thermal stresses
were not included in this diagram,
since 1t 1s felt that they will anneal
out at these temperatures and, at
worst, will cause a little distortion
that should not be serious. Table 4 2
gives strength data on Inconel and
columbium Recent tests indicate
that 1n the absence of oxygen the
creep rate of Inconel may be two or
three times the rate given ain the
table for a given stress level.
Examination of Fig. 4.27 will show
that the stresses have been kept so
low that the creep rate should not
prove to be a serious problem. Note
that the highest stresses are in the
heat exchanger tubes.

A small, experimental, heat ex-
changer has been designed for in-
vestigating the problem of fabricating
tube bundles and header connections of
the type required. Although small,
the heat exchanger should be capable

kept low, particularly i1n the hot of transmitting 1 5 megawatts of heat
TABLE 4.2. STRENGTH OF INCONEL AND COLUMBIUM AT ELEVATED TEMPERATURES
STRENGTH OF INCONEL STRENGTH OF COLUMBIUM
(ps1) (ps1)
AT 1200°F | AT 1400°F | AT 1600°F | AT 1800°F | AT 1600°F | AT 1800°F
Ultimate tensile strength 65,000 27, 500 15, 000 7,500 25, 000 20, 000
Yield strength (0 2% offset) 26, 500 17, 000 9,000 4,000
Stress-rupture at 100 hr 21,000 8,500 4,500 2,800
Stress-rupture at 1,000 hr 12, 000 5, 500 2,700 1,700 6,000+
1% creep in 10,000 hr 9,300 3,500 2,000 560
0 1% creep 1n 10,000 hr 5, 500 1,700 1,700 340
Endurance limait (108 cycles) 26, 500 15, 500 10,000 7,000

66
PERIOD ENDING MARCH 10, 1953

-l
OWG t83094A

U = x
[y Z 3
[a] g% o Q
"3 8o |8
: " g - |
w " o * o
238 zea |E
~uwe g U 4
g = o > =0 8
;52 w
owv |
n < ud
& -
> J u-
w "
|
™
g
"
Na PUMP OUTLET LAk gt Nak INLET
MP OU TEMP :=1100°F
TEMP = 1150 °F | / PRESS =100ps!
PRESS = 60psi E [ i v -
IN " N
7 f
’ Al il“)
PRESS SHELL
TUBE WALL
gggs;uggfl-' i § sk Tsap - 1130°F
*IVpst No FROM REFLECTOR FUEL TO PUMPS STRESS = 450 pst
TEMP = 1150°F TEMP = {190°F
PRESS = 20ps PRESS =10 psi
FUEL TO CORE
TEMP = t190°F
[=] PRESS = 45ps!
[=]
[+
)
(o]
[
.—
=
[e]
[ ]
BERYLLIUM

FUEL

PEAK TEMP =1200°F

INCONEL CORE SHELLS
Yg-n THICK

INNER CORE SHELL
TEMP - 1200°F -
STRESS = 240 pst

[
ECY
1‘1

OUTER CORE SHELL
TEMP = {200°F
STRESS = 360pst

PRESS SHELL
TEMP = {500°F TUBE WALL
TEMP ={530°F

STRESS = 370 pst
\ STRESS =100 pst

NaK OUTLET
TEMP =1500°F
PRESS =60psi
FUEL TO HEAT EXCHANGER
t 0 6 12 TEMP =1590°F
INCHES PRESS = 40 ps

Fig. 4.27. Temperatures, Pressures, and Stresses Throughout the Reflector-
Moderated Reactor Operating at 200 Megawatts.

67
ANP PROJECT QUARTERLY PROGRESS REPORT

from the fluoride to NaK. It is
expected that, ultimately, six, small,
heat exchangers will be built for
investigating the effects on endurance
life of operation at various tempera-
ture and pressure levels. Figure 4.28
shows one tube bundle for this heat
exchanger. Six of these bundles will
be used in a 5-in.-dia cylindrical
annulus instead of in the larger and
more complex spherical shell array
envisaged for the full-scale reactor.

The design temperature level in
the sodium circuit for cooling the
moderator region is quite tentative.

-

Tube Bundle for Spherical
(a) Tube-to-
(b) One

Fig. 4.28.
Reactor Heat Exchanger.
header welded connections.
end of a completed tube bundle and
“shower head’’ assembly.

68

The temperatures specified in Fig. 4.27
were based on the highest temperatures
that appear reasonable from the stand-
pointof compatibility of the beryllium-
sodium-Inconel system. Higher tempera-
tures would be likely to give trouble
because of mass transfer and because
of attack of the Inconel canning
material by the beryllium. The
temperature levels required should
not be difficult to attain, since the
moderator heat would be removed at a
high temperature level. Two engines
could be fitted with small auxiliary
radiators placed between the compressor
and the main radiators. These could
be employed to subcool a portion of
the secondary circuit NaK to, perhaps,
900°F. This cooler NaK could then be
fed to the Na-to-NaK heat exchanger
at the top of the reactor. There, by
cooling the sodium from the moderator
cooling circuit, it would be heated
to about 1100°F so that it would then
flow to the main heat exchanger at
the same temperature as that of the
NaK from the other radiator circuits.
Thus the weight penalty attached to
cooling the moderator region would be
only about 400 lb of radiator core
plus possibly 200 1b of extra reactor
shield and 100 1b of pump and line

weight.
Lead, bismuth, a nonuranium bearing
fluoride, sodium, and NaK were all

given serious considerationas coolants
for the beryllium moderator. The
metallurgists felt that lead or
bismuth would be likely to pose a
serious mass transfer problem. The
relatively high neutron absorption
cross section of the potassium in the
NaK made it quite undesirable from
the critical mass standpoint. Rubidium
might be used in place of potassium
but, because of little demand, it is
currently very expensive. Thus sodium
seemed to be the best choice for a
moderator coolant. Since corrosion
and mass transfer are likely to occur
in a Be-NaK-Na system, it seemed
essential that the beryllium be clad

1n some fashion. Work at Battelle(®’
indicates that beryllium can be
chrome-plated to give satisfactory
resistance to sodium attack at 932°F,
However, this work was carried out on
small specimens, and there 1s a
question as to whether adequate
protection to the very large surface
areas required would be practical.
An alternate possibility would be to
can the beryllium in thain-walled
Inconel cans and to fill the small
1nterstices between the beryllium and
the can with stagnant sodaum. This
arrangement appears to be the more
promising of the two, but both possi-
bilities are being investigated. The
reflector could be constructed of two
large hemispheres of beryllium 1f the
canning technique were used Cooling
passages couldbe rifle-drilled through
the beryllium and lined with thain-
walled tubes, which could be welded
into headers at the ends The Brush
Beryllium Co. has indicated that the
fabricationof these large hemispheraical
shells would probably be no more
difficult than the fabrication of
large flat slabs The personnel of
the Y-12 beryllaum shop state that 1t
would not be difficult to rafle-
dri1ll holes 3/16to 1/4 i1n. 1in diameter
and as much as 40 i1n deep, with the
hole diameter held to within 0.001 in.
and the hole center location held to
within 0.010 inch. Since 1t would be
very difficult to plate the 1insides of
holes, an alternate construction that
appears attractive, 1f chrome-plated
blocks prove practicable, 1s the use
of a large number of wedge-shaped
segments, which would be shaped much
like the sections of an orange. These
sections could be made with shallow
grooves 1n their surfaces that would
form passages for cooling streams of
sodium 1n the assembly.

The choice of a secondary coolant
was di1fficult., Careful examination of
the components i1n the external system

(B)J G Beach and C L Faust, Electroplating
on Beryllium, BMI-732 (Apr 1, 1952)

PERIOD ENDING MARCH 10, 1953

indicates that about 40 ft® of fluad
would be required as a minimum. This
means that 1f lead or bismuth were
used, the weight required would be an
the range of 30,000 pounds. A number
of molten salt mixtures was considered,
but the melting points 1n all cases
were above 300°F, which 1s objectionably
high. Other media were rejected be-
cause of corrosion considerations.
Sodium or NaK seemed to be definitely
superior to any other coolant con-
si1dered except for the objectionable
neutron activation and resultant gamma
activity of sodium, and the fare
hazard associated with a possible
leak. Shielding characteristics
indicate that for the most promising
distribution of shielding material
between the reactor and the crew, gamma
activity from the sodium would be less
troublesome than the prompt-gamma
activity from the reactor core. Con-
si1derable experience with sodium has
indicated that 1f the plumbing 1s
designed to keep the stresses low so
that burst types of failure will not
result, leaks that do occur as the
result of fatigue cracks or corrosion
pits develop slowly and do not result
1in serious fires. In fact, saince
sodium does not explode (so long as no
water 1s present), 1n many ways 1t
seems no more hazardous than gasoline.
After all the factors were weighed,
the most promising secondary fluad
appeared to be NaK., A 56% sodium and
44% potassium alloy, which has a melt-
1ng point of 56°F, was chosen for the
analysis. Its low melting point,
excellent heat transfer characteristaics,
good corrosion characteristics 1n
1iron-chrome~-nickel alloys, and low
densi1ty made 1t appear to be the best
choace.

Reflector Heating (A. H. Fox, R. W.
Bussard, ANP Division). The heating
of the reflector of a reactor by the
absorption of gamma radiation and the
slowing down of neutrons originating
1n the core presents a heat removal
task that 1s only an order of magnitude

69
ANP PROJECT QUARTERLY PROGRESS REPORT

smaller than that of cooling the core
1tself.,

The distribution of the heat
generated 1n the moderator depends on
the strength and location of the
sources and on the attenuation of the
radiation between the sources and the
point 1n question. For the gamma-ray
energies of interest, the mechanism of
degradation i1nvolves principally the
Compton collisions. The mechanism of
degradation of energy of the core
gammas 1s complicated, even with the
most simple geometry of source and
reflector, for amore complex geometry,
such as that of the reflector-moderated
reactor, many simplifying assumptions
are necessary before even an approxi-
mate estimate of the rate of heat
generation 1in the beryllium may be
obtained,

The first approximation used 1in-
volved the straight-ahead theory of
gamma absorption for whichit 1s assumed
that Compton collisions merely degrade
1in energy and do not scatter the
photons. Only exponential attenuation
with a coefficient that 1s charac-
teristic of the material through which
the photons pass need be considered.
When a photon passes from one medium
to another, no refraction 1s con-
sidered. Use of a buildup factor to
allow for ordinary scattering 1s being
investigated.

Another saimplifying assumption
concerned the source distraibution.
Only the total amount of gamma flux
was considered, and variations from
point to point 1n the core were
neglected, Thus an average power
density for the core was determined
simply by dividing the total reactor
power output by the volume of the fuel
region of the core. (An 1investigation
of sources distributed as indicated by
the critical experiments showed only
a small variation in results.)

The particular case considered was
that of an 18-1n.-dia, spherical,
fluoride-fuel region surrounding a

9-1n.-d1a, central, beryllium 1sland

70

and enclosed by al2-in.-thick beryllium
reflector, A 3/16-in.-thick layer of
Inconel was considered as being the
separating material between the fuel
region and each of the two moderator
regions. The reactor power output was
taken as 200 megawatts, and the energy
evolved 1n the gammas was taken as
12 Mev/fission, which gave a total
power from gammas of 12 megawatts and
a gamma source density i1n the fuel
region of 275 watts/cm>. The average
gamma energy was assumed to be 1 Mev,
and the reciprocal attenuation lengths
for the fuel, Inconel, and beryllium-
moderator regions were taken as 0.09,
0.30, and 0.16 cm™!, respectively.

The power density was computed at
points spaced about 1 in, apart along
a radius from the center of the island
to the outside of the reflector. The
resulting values are given 1in Table
4,3 (case A) and Fig. 4.29., Table 4.4
(case A) shows the integrated values
of the power, in the various regions
of i1nterest, obtained by graphical
integration of the power density curve.
The fact that the total gamma heating
indicated by Table 4.3 1s only 10,6
megawatts 1nstead of the 12 megawatts
assumed 1s due, 1n part, to leakage
from the reflector and,
the approximations made for the
computations. However, 1t 1s believed
that the distribution of the gamma
heating 1s approximated quite well by
Fig. 4.29., One important result of
this work 1sthat for this configuration
more than 60% of the gamma energy goes
into heating the fuel. A second
computation was made with different
values for the reciprocal attenuation
lengths, that 1s, 0.06 and 0.13 cm™!
for the fuel and beryllium, re-
spectively, and the results are shown
1n Table 4.3 as case B, As can be
seen, this relatively large change had
no great effect on the distribution of
the gamma heating. The total power

in part, to

for both case A and case B 1s given 1n

Table 4.4.
PERIOD ENDING MARCH 10, 1953
DWG 18740
450
ISLAND FUEL REFLECTOR
400
ri
"2
350
[t—INCONEL SHELL—
fm——Na ANNULUS —=4
INCONEL CAN
~ 300
L]
E
(5]
~
@
3 250 %
TOTAL HEATING
> }/ TOTAL HEATING
tT') \<
= /
wi
& 200 N
x
w
& TOTAL HEATING
150 y HEATING” "\ % L
/ I
/ \
NEUTRON /|
HEATING y,
100
’//// \ NEUTRON HEATING
#:;fijéiiEAflNG &\ \ y HEATING
50 N
--____ I, \
NEUTRON S
HEATING
o |
0 2 4 6 8 10 t2 14 16 18 20

RADIAL DISTANCE FROM REACTOR CENTER LINE, » (in)

Fig. 4.29.

The power density resulting from
neutron moderation within the beryllium
of the 1sland and reflector can be
obtained directly from multigroup
results by using the flux distribution
¢(r,pn) or (W:/n) C. F. (correction
factor).¢!) The energy loss for each
lethargy group 1s the average energy
loss per collision times the number of
collisions i1n that group at a given
radius or space point, The spataial
distribution,

t h W; cE . .
£2.,

107

Radial Power Density from Neutron and Gamma Heating.

1s normalized to the total power lost
by moderation (2 1/2% of reactor power)
by the use of the integration operator
Q.

The power densities resulting from
neutron moderation and from gamma

heating, as obtained by the methods
described above, as well as the
resultant total power density, are

shown i1n Fig. 4.29. Severe peaks 1in

the Inconel shells and a rapid drop
with distance from the fuel region are
shown,

A number of factors must be con-
si1dered 1n the design of a system of

71
ANP PROJECT QUARTERLY PROGRESS REPORT

s . TABLE 4.3. POWER DENSITY IN VARIOUS REGIONS
RADIAL DISTANCE FROM RATE OF HEAT GENERATION
CENTER OF FROM GAMMAS
REGION REACTOR CORE (watts/cm3)
r (im } r (cm) Case A Case B
Island beryllium 0 0 47 5 61 9
2 5 08 60 65
3 7 62 80 74
4 10 16 101 93
4 31 10 95 120 142
Island Inconel 4 31 10 95 223 330
4 5 11 4 354 497
Fuel 45 11 4 106 99
5 12 7 167 137
55 14 186 148
6 0 15 2 182 152
6 75 17 1 184 151
8 20 3 157 132
8 5 21 6 141 116
9 22 9 73 65
Reflector Inconel 9 22 9 215 325
9 19 23 3 150 227
Reflector beryllium 9 19 23 3 80 98
9 5 24 1 53 7 57
10 25 4 37 7 44 6
10 5 26 7 25 4 31 9
11 27 9 19 0 24 4
12 30 5 10 3 14 0
13 33 0 5 4 8 3
14 35 6 29 49
15 38 1 1 67 30
21 53 3 0 066 0 19
TABLE 4.4. TOTAL INTEGRATED POWER moderator cooling passages. The volume

IN VARIOUS REGIONS

TOTAL POWER
REGION {megawatts)
Case A Case B

Island beryllium 0 46 0.51
Island Inconel 0 24 0 29
Fuel 6 71 5 58
Reflector Inconel 0 57 0 85
Reflector beryllium 3 14 3 44

12

of both the sodium and, especially, the
Inconel must be minimized to keep
parasitic neutron absorptions withain
reasonable limits. The neutron flux
curves show that from the neutron
economy standpolnt coolant passages 1n
the beryllium are much more i1mportant
several inches from the fuel region
than at the fuel-moderator interface.
Thermal stresses will be set up by
temperature variations in the beryl-
lium. Since an elongation of about
30%.ca&n be obtained in beryllium at
the operating temperatures envisioned,
thermal stresses should not lead to
cracking but might cause distortion
that could become a problem after a
number of thermal cycles of the system.
For this reason, the temperature
variation i1n the beryllium between
adjacent coolant passages was held to
50°F. The pressure drop through the
various coolant passages was limited
to 40 ps1 to keep pressure-induced
stresses low, The maximum beryllium-
sodium 1interface temperature was held

o o)
o
Q © ©
O 0 o O )
o)
?O0o0o
C 0 o )
()OOOO OOOO
) 0
= Q. 28
i B . . O O
"-\.\ 2 o O
N ::‘-\»-\.9 ©
‘\.
q00%800 o
ISLAND ooooooo
0500 %R
OOOOO%%O
O.Oe.(2 i
INCONEL SHELLS
AND Na ANNULUS
Fig. 4.30.

PERIOD ENDING MARCH 10, 1953

below 1200°F to reduce mass transfer
1in the Be-stagnant-Na-Inconel system.

Several detail designs were 1n-
vestigated that favored first one and
then another of the various require-
ments, that 1s, minimum poison, minimum
variation 1n beryllium temperature,
minimum beryllium-sodium interface
temperature, minimum sodium system
pressure drop, etc. Figure 4.30 shows
the hole pattern in the beryllium for
a promising arrangement, and the re-
sulting temperature distribution 1is
shown i1n Fig. 4.31.

DWG 18744

PASSAGE FOR INLET Na

TO REFLECTOR (2 REQ'D)

026~ 1n -DIA HOLES IN Be, CONTAINING

025-i1n ~OD TUBES FOR Na COOLANT FLOW

Cooling Hole Distribution in Reflector-Moderator.

73
ANP PROJECT QUARTERLY PROGRESS REPORT

S
DWG 18742

1600 '

ISLAND FUEL

1400

REFLECTOR

7

TEMPERATURE

1200 “nuj'/ /U U\" DISTRIBUTION Um

(\f\(\/‘\;/

__ 1000
n y & \\\
< Na COOLANT
& HOLES Nao COOLANT HOLES
|
'.-.
% 800 <~ INCONEL SHELL——
&fi e,
g i Na ANNULUS ——=it
M INCONEL CAN
600
400
200
0
0 2 4 6 8 12 14 16 18 20
RADIAL DISTANCE FROM REACTOR CENTER LINE (n)
Fig. 4.31. Temperature Distribution Across Midplane of 200-Megawatt

Reflector-Moderated Reactor.

SHIELDING

A. P. Fraas, ANP Division
J. B. Trice, Solid State Division

The shield design philosophy that
has been the basis of the work on
the reflector-moderated reactor shield
has been that an operational airplane,
1in the military sense of the word, can
be achieved only with great difficulty
with the use of a divided shield
because a divided shield would give an
unprotected man 50 ft from the reactor
a lethal dose i1n 15 seconds. Those
experienced in aircraft operation and
maintenance know that perhaps 50% of
the maintenance work 1nvolves non-

74

Sodium 1inlet,

1000°F.

scheduled operations, the character
and detailed nature of which are
1mpossible to predict without an
extensive background of experience.
Provision of an adequate set of equip-
ment and facilities to take care of
all contingencies promises to 1nvolve
expenditures of the same order as
those required for the nuclear power
plant 1tself, It was felt that 1f
the radiation dose at full power
could be cut to about 10 r/hr at 50 ft
from the reactor, the greater majority
of the nonscheduled maintenance jobs
that might delay a flight at the last
minute or require emergency attention
immediately following a landing could
be handled without special equipment.
For this reason, effort has been
directed toward the development of
something closely approaching a unit
shield. It 1s fully realized that a
welght penalty 1s inevitable.

It was found that the 1nitaial
startup and warmup of the reactor can
probably be carried out at power
levels that, i1n general, do not exceed
10% of the full power of the reactor,
thus the radiation dose during startup
and warmup would be 10% of that at
full power. If in the tuneup process
the operations must be carried out
close to the reactor, the reactor
might be 1dled down from 10% to
perhaps 1% of full power output so
that the radiation dose at any partai-
cular point would be only 1% of the
full-power dose., It 1s true, however,
that after a long flight the accumu-
lation of fission products will
constitute a substantial gamma dose
even after the reactor 1s shut down
completely. This dose w1ll still be
only of the order of 1 or 2% of the
total radiation dose at full power
output because the decay gammas are
fewer 1n number and softer than the
prompt gammas. A drain valve will be
provided at the bottom of the reactor
so that the fluoride fuel can be drained
1into underground tanks. It 1s hoped
that about 98% of the fuel can be
removed by a simple draining operation
and that most of the remaining fuel
can be flushed out by using a non-
uranium-bearing fluoride. Precisely
what can be done 1n this direction
1s difficult to predict, but one of
the major 1tems of information ex-
pected from ARE test work 1s the
extent to which the radiation from a
circulating-fluoride-fuel reactor can
be reduced by draining and flushing.
The degree to which the Inconel
structure wi1ll be activated and waill
represent an i1mportant source of
radiation after shutdown 1s difficult
to estimate, If the shield 1s not
disassembled, rough estimates indicate

PERIOD ENDING MARCH 10, 1953

that the Inconel activation should not
be a problem. Here again, the ARE
w1ll yield much valuable information.
The shield for the 200-megawatt
reflector-moderated reactor has some
characteristics that are peculiar to
this particular reactor configuration.
As 1ndicated i1n an earlier section,
the thick reflector was selected on
the basis of shielding considera-
tions.{%? The two major reasons for
using a thick reflector are that a
reflector about 12 in., thick followed
by a layer of boron-bearing material
w1ll attenuate the neutron flux to
the point where the secondary gamma
flux can be reduced to an unimportant
level for a quasi-unit shell., Thas
thickness also reduces the neutron
leakage flux from the reflector to
the heat exchanger to the level of
that from the delayed neutrons ap-
pearing 1n the i1ntermediate heat
exchanger as delayed neutrons from
the circulating fuel. An additional
advantage of the thick reflector 1is
that 99.98% of the energy developed
in the core will appear as heat 1in
the high-temperature zone inside the
pressure shell. This means that
very little of the energy produced
by the reactor must be disposed of
with a parasitic cooling system at a
low temperature level. The materaal
in the spherical-shell i1ntermediate
heat exchanger 1s about 70%as effective
as water for the removal of fast
neutrons, so 1t too 1s of value from
the shielding standpoint. The delayed
neutrons and the decay gammas from
the circulating fuel 1n the heat
exchanger region might appear to pose
a serious handicap. However, each
has an attenuation length that 1s much
shorter than the corresponding attenu-
ation length for radiation from the
core. Thus, from the outer surface
of the shield, the intermediate heat
exchanger appears as a much less
(g)A P Fraas, Three Reactor-Heat Exchanger-

Shield Arrangements for Use with Fused Fluoride
Circulating Fuel, ORNL Y-F15-10 (June 30, 1952)

75
ANP PROJECT QUARTERLY PROGRESS REPORT

1ntense source of radiation than the
more deeply buried reactor core.

It 1s true that this arrangement
violates one of the precepts of the
matched shield, namely, that each
layer of heavy material should be
placed as close to the reactor core
as possible, However, the loss in
weight associated with the disparaty
between this i1deal case and the shield
layout of Fig., 4.32 1s not too great,
whereas the engineering advantages
derived are of crucial importance.

A series of Lid Tank experiments
1S 1nh progress to determine the
optimum arrangement and thickness of
the various beryllium, boron carbide,
1ron, lead, and borated water layers
for a shield of the type shown in
Fig. 4.32, which represents the best
configuration tested to date, When
this series of tests 1s completed,
the resulting shield will be used as
a basis for comparison. Further tests
will be run with more unusual materaials,
Rough calculations i1ndicate that as
much as 10,000 1b of shield weaght
mi1ght be saved through the use of
some of the special materials. When
the test work 1s completed, 1t 1s
hoped that a fairly sound basis for
a decision on the use of these special
materials will be provided by a com-
parison of their cost and attendant
welght savings relative to the simpler
shield configuration of Figure 4,32,

One of the major pieces of 1in-
formation obtained from the Lid Tank
experiments has been the confirmation
of the original estimates of the
degree of activation of sodium 1n the
heat exchanger region. It had been
suspected that in a relatively than
slab of material having the poor
moderating properties of the heat
exchanger matrix, neutrons of 0.1 Mev
or higher would tend to escape from
the slab long before they could be
slowed down. Although 1t 1s not possible
to simulate the effects of the cir-
culating fuel i1n the Lid Tank experi-

76

ments, the spectrum of the delayed
neutrons 1s not too different from
that of the bulk of the neutrons that
escape from the fuel region through
the thick beryllium reflector and the
heavy boron carbide curtain between
the reflector and the heat exchanger.
For this reason, 1t 1s felt that the
sodium activation data from the Lid
Tank experiments give a good basis for
estimating the activation of the sodium
1n a full-scale reactor of the type
shown 1n Fig. 4.25, On this basais,
1t appears that the sodium activity
after a long period of full-power
output will give a dose of about 1
r/hr at 50 ft from the reactor. The
activaity of the potassium in the Nak
would be much less than that of the
sodium.(!%) The NaK could be drained
after operation and the activated
sodium allowed to decay. Several
weeks should suffice to bring the
activaity of the sodium down to a
negligible level,

The design described 1s believed
to give enough shielding all around
the reactor to give a dose of 5 r/hr
at 50 ft from the center of the reactor
at full power. An additional lead
layer would be employed at the front
of the reactor to act as a light
shadow shield for the crew. Further
shadow shielding would be employed
at the back of the crew compartment
to attenuate the radiation from both
the reactor and the NaK radiators to
give a dose of approximately 1 r/hr
1nside the crew compartment., Tables
4.5 and 4.6 give detailed data for

such a configuration for several
reactor core diameters, power out-
puts, and power densities. Further

work covering the effects on shield
weight of various dose levels for
both ground and flight crews 1s under
way.

(1°)w. K Ergen, Potasstium as Coolant tn Con-
nectton witth Ground-Safe Shields, ORNL Y-F20-.4
(Dec 6, 1950)
LEAD SHIELDING

REACTOR PRESSURE SHELL

THERMAL INSULATION

MAKEUP FUEL TANK

VALVE

THERMAL INSULATION

NaK DUCT

1 ra
l”’fll/lll///”l/fll//l”/I///I/////
ra

THERMAL

N

_——

0'6?
o7 >
N
[V
o~
L
™
t"‘\a
~ O
\X 3

°o

=k H—:-}ILL_— -

NaK DUCT

REACTOR PRESSURE SHELL

\\ THERMAL INSULATION
LEAD SHIELDING

PERIOD ENDING MARCH 10,

PUMP QUILL SHAFT

CONTROL ROD ACTUATOR
,/PUMP QUILL SHAFT

llllllll

A A A A A4 Z 7

w7 27 z

TUBE BUNDLE

SELF—SEALING RUBBER TANK

HORIZONTAL SECTION

BORATED WATER

\\WEB OF CANTILEVER BEAM
SUPPORTING REACTOR AND SHIELD

0 6 112 24
|

l | -

1

SCALE IN INCHES

COOLING PASSAGE

BORATED WATER

DRAIN VALVE

FUEL DRAIN LINE

THERMAL
INSULATION

TUBE BUNDLE

SELF-SEALING RUBBER TANK

VERTICAL SECTION

F1g. 4.32. Shield Configuration for 200-Megawatt Reflector-Moderated Reactor

1953

77
TABLE 4.5. DOSAGES AT VARIOUS LOCATIONS FOR THE SHIELD OF TABLE 4.6
DOSE (r/hr)
® &
LOCATION At At Full At Shutdown
Startup* Power NaK not NaK
Drained Drained
50 ft from reactor (except
1n front cone) 0.5 6 1 0.1
Crew compartment 0.1 0.2 0.02
16 ft from reactor (except
1n front cone) 6 60 10 1
Shield surface (except 1in
front cone) 60 600 30 10

*Ten per cent of full power

**Fifteen minutes after shutdown from long period at full power

TABLE 4.6. PRELIMINARY SHIELD WEIGHT ESTIMATES SHOWING EFFECTS OF VARIATIONS
IN REACTOR POWER, CORE DIAMETER, AND POWER DENSITY IN THE FUEL REGION®*

Power, megawatts 200 200 400 200 400 800
Reactor core diameter, in, 18 22.7 22,7 28.5 28,5 28,5
Power density, kw/cm? 5.5 2.78 5.5 1,37 2.75 5.5
Pressure shell outside

diameter, 1in. 59 62.3 68.8 66,6 72.4 8l.6
Lead layer outside

diameter, 1n. 72 75.3 82 79.6 85.4 94.6
Reactor shield outsaide

diameter, 1in. 120 119 128 121 129.5 140
Total weaght, reactor,

heat exchanger, and

shield, 1b 74,000 (84,000 | 108,000 {93,000 (115,000 | 147,000
Shadow disk at crew

(7 ft daia), 1b 3,600 3,600 3,600 3,600 3,600 3,600

*Computed from Lid Tank data for the beat Be-B4C-NaF-B4C-Fe-Pb-H20 layer configuration tested up to

March 13, 1953

POWER PLANT DESIGN

A. P. Fraas, ANP Division

Some early work on nuclear power
plants 1nvolved detailed designs of
power plants in which the engines were
widely separated from the reactor. The
acute problems of differential thermal

expansion, thermal lag and system
control, and the plumbing weight
associ1ated with these arrangements
1ndicated that i1t would be advantageous
to use compact units, Thus, the
current trend toward compact power
packages, followed by both the General
Electric Co. and the Pratt and Whitney
Aircraft Division, seems 1in order.

79
ANP PROJECT QUARTERLY PROGRESS REPORT

In considering the types of appli-
cation of greatest 1nterest, 1t
appeared that there are three repre-
sentative cases namely, aB-36 flight-
test-bed installation, a Mach 0.9 sea-
level bomber, and a Mach 1.5 high-
altitude bomber. In sketching pre-
liminary layouts, 1t appeared that
there was little difference 1n the
requirements for the first two, and
the third differed 1n not too serious
a fashion., As a result, the layout
of Fig. 4.25 was prepared for the
Mach 0.9 sea-level bomber on the
basis that 1t would require a reactor
heat output of 200 megawatts.
presumed that the same power plant
might be used for preliminary test
work at a reduced power, say 130
megawatts, in a B-36 flight-test-bed
installation.

Table 4.7 gives the weight of
various components of the installation
shown in Fig. 4.33. If an 1increase
1n power were made, the weight of the
reactor-heat -exchanger-shield assembly

It was

would i1ncrease as indicated in Table
4.6, and the weight of engines,
radiators, pumps, and lines would be
roughly proportional to the power

A

to be required at an altitude at whach
the air-swallowing capacity of the
engines was cut in half, the turbojet
engine weight would be doubled. The
weight of the radiators would be
increased by, perhaps, 50%, but the
weight of the pumps and lines would
remain substantially the same. The
power plant of Fig. 4.33 represents
only a preliminary layout for showing
the proportions that might be expected.
The thrust that can be obtained from
this power plant 1s heavily dependent
on the allowable reactor operating
temperature. By careful design of the
moderator-cooling system, 1t will be
possible to keep the temperature of
all structural parts down to the
temperature of the secondary fluid
leaving the intermediate heat ex-
changer. This should make 1t possaible
to operate such a power plant, con-
structed of Inconel, at NaK outlet
temperature of about 1500°F for 500
hours. A take-off or war-emergency
rating of 1600°F might be permitted
for perhaps 20 to 50 hr of the 500
hours.

A brief description of the essential
features of the design shown 1in Fag,

output. If the same power output were 4.33 follows. The reactor 1s the same
TABLE 4.7. MAJOR WEIGHT COMPONENTS OF THE 200-MEGAWATT
AIRCRAFT POWER PLANT OF FIG. 4.33
NO WEIGHT
OF UNITS COMPONENTS (1b)
4 Turbojet engines 12,400
16 Radiators (filled with NaK) 6,000
16 NaK pumps, header tanks, and turbine drive assemblaes 1,600
16 Piping for NaK caircuits 2,000
Thermal insulation for NaK caircuats 1,300
NaK 1n NaK circuits (except in radiators) 2,400
4 Fuel and sodium pump drives 300
Turbojet i1nlet and outlet ducts and support structure 8,000
Total power plant weight excluding reactor and shield 34,000

80
18

e
OWo. (8744

FUEL PUMP DRIVE TURBINE e e A

Na PUMP DRIVE TURBINE

AIR TURBINE

NoK EXPANSION TANK
WEB OF CANTILEVER BEAM

FROM REAR WING SPAR ~\(

NoK TO INTERMEDIATE HEAT EXCHANGER (100°F)

f— EXTERNAL SHIELD (RUBBER CONTAINER
FILLED WITH BORATED WATER)

LEAD SHIELD

INSULATION

REACTOR

NaK PUMP

NaK TO ENGINES (1500°F)

TAIL PIPE

NaK-TO-AIR RADIATOR
HELICAL BAFFLE

ENGINE DATA
MODIFIED WRIGHT TURBOJET
COMPRESSION RATIO 4:4 (CORRECTED FOR SEA LEVEL) ",
AIR FLOW 220 Ib/sec (CORRECTED FOR SEA LEVEL)
DIAMETER = 44 Y in.
LENGTH =140 in.
ENGINE WEIGHT = 3100 Ib (WITHOUT RADIATOR)
RADIATOR WEIGHT = 1500 Ib (WITH NaK) REEE 6

COMPRESSOR

Fig. 4.33. Aircraft Power Plant (200 Megawatt).

‘0T HOMVW ONIGNYI aorydd

€S6T
ANP PROJECT QUARTERLY PROGRESS REPORT

as that shown i1n Fig. 4.25. The
shield was designed so that 1t might
be i1nserted through an opening 6 ft
8 1n., wide in the bottom of the fuse-
lage. Thus 1t would lend 1tself to
installation in the bomb bay of the
B-36 just aft of the main spar. Two
cantilever beams placed on 6-ft centers
were employed as the primary structure,
The front ends of these beams could
be bolted or pinned to fittings on the
rear spar, and the reactor and the
shield structure could be attached to
the rear ends. The shield structure
would be divaided
sections, each of which would consist
of an i1nner layer of lead and an
outer tank of rubber that would be
filled with borated water. The top
and bottom sections would be separated
from the four sections in the central
region by surfaces of revolution for
enclosing the NaK ducts through the
shield. A large, bowl-shaped, rubber
tank would be placed at each side.
These tanks could be deflated while
the power plant was being inserted 1in
the airplane and could be filled with
water after the power plant was 1in
place. The top section of the shield
would be penetrated by five tubes.
Four of these tubes would carry quill
shafts to transmit power from bleed-
off air turbines to the pumps at the
top of the reactor, and the fifth
tube would carry the control rod
actuating mechanism.

into s1x major

Each of the engine radiators would
be coupled to a separate tube bundle
1in the i1ntermediate heat exchanger in
such a way that there would be 16
separate secondary fluid circuits.
The expansion tank, filter, and pump
for each of these circuits would be
located at the point of greatest
elevation and lowest temperature ain
the system, that 1s, at the top of
the shield and just outside 1t. Thus
the NaK would leave the pumps and
pass i1nto the shield, through the
1ntermediate heat exchanger, back out

82

through the shield down to the ra%fator,
and back up to the pump. Thejpipes
passing to and from the radiator
would be approximately 3 1/2 in, in
diameter, The pressure drop through
the radiators would be about 35 psa,
and that through pipes fromthe reactor
would be about 7 psi. In all cases,
the pressure has been kept low to
keep the stresses low and, hence, to
minimize the likelihood of a burst
type of failure., Thus, even though
small leaks might occur as a result
of fatigue cracks or small corrosion
pits 1n welded or brazed joints, 1t
seems likely that a major leak (except
for gunfire) could be avoided.

The turbojet engines indicated in
Fig. 4.33 are modifications of a new
engine being developed by the Wraight
Aeronautical Corp. The original
engine 1s a high-compression-ratio
two-spool machine. Since fuel economy
1s less important than engine weight
for a power plant of the type shown, (11)
an estimate of the weight of a modified
engine was prepared by the Wraght
Aeronautical Corp., and estimated
performance curves for these engines
are given i1n Fig, 4,34, The radiators
used 1n these turbojets are of the
same type as those that have been
tested at ORNL in the form of core
elements during the past year. One of
these radiator cores operated for over
1000 hr at 1500°F, or above, including
over 40 hr at 1700°F.¢(11) T¢ ;5
expected that developmental work now
in process will produce lighter units
with better heat transfer performance
than the radiators of this preliminary
design,

The procedure that might be followed
to assemble the complete peower plant
would be to mount the reactor on the
cantilever beams that support 1t and
the shield assembly, The turbojet
engines would then be mounted to

(ll)w S Farmer, A P Fraas, H J Stumpf,
and G D Whitman, Preliminary Design and Per-

formance Studies of Sodiua-to-Air Radiators,
ORNL- 1509 (to be published)
AIR FLOW {ib/sec)

HEAT CONSUMPTION {kw)

TOTAL THRUST (Ib)

320
280
240
200
160
120

80

40

64,000
56,000
48,000
40,000
32,000
24,000
16,000

8000

Fig.

. PERIOD ENDING MARCH 10,

SRR
OWG 18745

I f | { 1 | ! ! {
CORRECTED COMPRESSION RATIO= 40 1
ENGINE WEIGHT=3100 {b

ENGINE LENGTH =140 n

15,000 ft ——

35,000 ft e |t

|
45,000 ft —T

55,000 1

T i
RADIATOR WEIGHT =4500 Ib
RADIATOR DEPTH =1{2.n

No TEMPERATURE THROUGH RADIATOR = 400 °F

|

-
SEA LEVEL "
_—-—-"'/

——’
15,000 11 | odue—t

35,000 ft |t "

|
45000 ft

55,000 ft

ENGINE DIAMETER = 44 5 n
SEA LEVEL AIR FLOW =220 Ib/sec

RADIATOR INLET FACE AREA =16 7 ft2

T
[ SEA LEVEL

Pl

15,000 f1

I l

35,000 ft
45000 11
|

55,000 f1

o] 03 06 o9 12 15
MACH NUMBER

4.34. Estimated Turboj)et Engine Performance Curves.

1953

83
structure 'attached to these same
cantilever beams. The pipes connecting
the engine radiator to the reactor
and heat exchanger assembly could then
be welded i1nto position, along with
the NaK pump and header tank assemblies,
The entire system could then be
pressure checked carefully and examined
closely for leaks. It would probably
prove desirable to install NaK in the
secondary circuit, sodium 1n the
moderator circuit, operate the pumps,
and check all components carefully for
leaks before the thermal insulation
was 1nstalled. An auxiliary supply
of compressed air would be required
to run the sodium and NaK pumps at
perhaps half speed for this check.
The system temperatures 1in the Nak
circuits would not have to exceed room
temperature, but an auxiliary burner
arrangement should be provided to heat
the two radiators coupled to the two
sodium-to-NaK heat exchangers for
the moderator cooling system. These
heaters could be used to bring the
sodium system temperature up to 300
or 400°F., Upon satisfactory completion
of this test work, thermal insulation
could be installed around the reactor
pressure shell and the connecting
lines and pumps of the sodium circuit,
By using the same heat source, the
sodium system could be brought up to
perhaps 1000°F and a nonuranium-
bearing fluoride salt could be pumped
into the reactor. The fluoride-fuel
system could then be checked and
drained, and the lead shielding and
the rubber tanks for the borated water
could be i1nstalled to complete the
assembly. After the power plant unit
had been installed in the airplane,
the rubber tanks could be filled with
borated water, and the necessary
connections could be made for the
instruments and controls.

84

In starting up the reactor, 1t 1is
expected that the NaK circuits would
be used to heat the sodium to bring
the reactor temperature up to about
1000°F. A fluoride fuel melt con-
taining less than the desired amount
of uranium would then be poured ainto
the fuel system and circulated.
Gradual additions of a fluoride melt
containing a rather high percentage of
U%3% could be made until the reactor
became critical. The temperature of
the reactor could then be allowed
to 1ncrease to the desired value -
probably an average temperature of
1300°F - and the load could be gradually
increased by starting up first one
and then another of the turbojet
engines. It will probably be possible
to depend upon the self-stabilizing
characteristics i1nherent 1in the
fluoride fuel reactor to keep the
mean fluoride temperature essentially
constant so that, i1in effect, the
reactor would become a slave to the
turbojet engines and would act as a
constant-temperature heat source,
Control of the individual turbojet
engines could be accomplished by
varying the amount of air allowed to
bypass the radiators through adjustable
louvers i1n the helical baffles between
In effect, this would
give a variable, mean, turbine-air-
inlet temperature. Coupled with this
control, 1t would probably be necessary
to have an adjustable jet exit nozzle
to give good performance and to avoid
compressor stall over a wide range
of operating conditions. Fine control
of reactor temperature could be
accomplished through the use of one
or two control rods in the central
1sland or i1n the reflector region.
Coarse shim control could be obtained
by varying the uranium concentration
in the fluoride fuel,

radiator banks.

Part 1l

SHIELDING RESEARCH
INTRODUCTION AND SUMMARY

E. P. Blazard

J. L. Meem, Associate

Physics Divasion

With the development of a reactor
design 1n which the fissions are
confined to a small volume, and an
which 1t 1s hoped the radiators can
be kept relatively nonradioactive,
the hopes for a laght unat shield
have been revived. Accordingly, a
unit-shield mockup was recently
installed i1n the Lid Tank Facilaty
and 1s now being optimized with regard
to the location of the shield com-
ponents. In addition, an intensive
search 1s beaing carried out for
promising materials that could be
used 1n the unit shield. Measurements
of relative shielding effectiveness
1indicate lathaium to be a promisaing
component for neutron shielding, and
inquiries have been 1nitiated to
determine whether uranium could be
made available and fabricated as a
gamma shield (sec. 5).

The air-scattering experiments at
the Bulk Shielding Facility have
been completed. Earlier difficulties
have been ascribable primarily to
unsuspected background effects, and
1ndications now are that the weights
described 1n the report of the Shieldaing
Board (1950) were approximately
correct. The neutrons were somewhat
low and the gamma rays somewhat high,
with the weight differences approxai-
mately cancelling. The i1rradiataon
of animals in the Bulk Shielding
Facility 1s now complete, and the

animals involved have been returned
for long-range observation of the
biological effects of these exposures.
Neutron spectroscopy on the divaded
shield i1n the Bulk Shielding Facilaity
has now commenced, and the first few
spectra have been measured with the
proton-recoi1l spectrometer. The
results are only prelaminary, hence,
no data are yet available. However,
the gamma spectral measurements from
the divided-shield mockup are beang
tabulated for machine calculations
(sec. 6).

The Tower Shielding Facilaity design
1s being developed rapidly. The
tower structure design 1s final and a
building 1s fairly completely laad
out. The Reactor design features and
some of the instrumentation are yet
to be developed. Indications are that
the facility will begin operataion
by the end of the calendar year, and
tests with the first shields will
commence shortly thereafter (sec. 7).

A neutron spectrometer 1s being
developed that utilizes scintillation
in a lithium fluoride crystal. This
apparatus gives promise of appreciably
greater sensitivity than recoil-proton
spectrometers 1f developmental diff1-
culties can be overcome. No cross-
secti1on measurements were completed
on the 6-Mev Van de Graaff during the
quarter because the machine was being
moved to 1ts permanent location an the

X-10 Area (sec. 8).

87
5. LID TANK FACILITY

J. D. Flynn
G. T. Chapman

J. N. Miller
F. N. Watson

Physics Division

The Lid Tank Facility has been
used during the past quarter praimarily
for measuring the fast-neutron removal
cross sections of a number of con-
ventional and potential shieldaing
materials. In addition, a mockup of
the unit shield of the reflector-
moderated reactor has been assembled.

EFFECTIVE FAST-NEUTRON REMOVAL
CROSS SECTIONS

The following effective fast-
neutron removal cross sections have
been measured

Al 1.19 barns/atom
Be 1.12 barns/atom
Cu 2.08 barns/atom
Fe 1.93 barns/atom
LiF 2.80 barns/molecule

W 3.08 barns/atom

The data on lithium fluoride are
particularly interesting. If fluorine
1s assigned a cross section of 1
barn, as would be indicated from a
comparison with previously measured
values for oxygen and carbon, 1t
would appear that lithium exhaibats
a cross section of 1.8 barns. Thas
is exceptionally high for such a light
nucleus, and, 1f verified, reveals a
real i1ncentive for using this element
in aircraft shields. Accordingly, 1t
1s planned to measure the cross section
of fluorine as soon as possible by
usingslabs ofa saturated fluorocarbon.

Measurements have also been made
on the effect of replacing water near
the source with a slab of transformer
o1l (CH,). By comparison of these
measurements with those for graphite,
1t 1s possible to determine the effect
of the oxygen that the o1l replaced
1n the water close to the source.

This represents the first measurement
of the true effective removal cross
section for this element. Previous
estimates have been dependent upon a
somewhat different definition, because
the oxygen that was measured was
spread throughout the shield as in
water It 1s 1nteresting to note
that, as might be expected, the oxygen
exhibits a somewhat greater effectave
removal cross section 1in the locataion
near the source.

It had been proposed by Sleeper of
Brookhaven National Laboratory that
deuterium might prove a more effective
shi1eld component than hydrogen because
of the more nearly isotropic scattering
(laboratory system) 1t exhibits
because of 1ts i1ncreased mass. To
explore this point, a tank of D,0 was
inserted next to the source 1n the
Li1d Tank Facility, and an effectave
removal cross section was measured
Results 1ndicate a difference 1n
effective removal cross section
between normal and heavy hydrogen of
0.1 barn. This corresponds very
closely to the difference 1n thear
total cross sections 1n the range of
about 2 to 5 Mev and indicates that
no added premium accrues from the
increased angle of scattering or, at
least, that this 1s counterbalanced
by the greater energy degradation of
the light hydrogen.

MOCKUP OF THE UNIT SHIELD OF THE
REFLECTOR-MODERATED REACTOR

Recently a mockup has been installed
in the Lid Tank Facility to simulate
the reflector heat exchanger and the
unit shield of the Fireball reactor.
The beryllium reflector 1s simulated
by a large slab of material supplied
by KAPL plus some additional material

89
ANP PROJECT QUARTERLY PROGRESS REPORT

that was available at the Y-12 sate,
The heat exchanger 1s simulated by
sodium fluoride and aron, the sodium
fluoride 1s loaded into large, than,
1ron boxes. The prototype for the
shield will be of lead and water, and
tests currently under way are designed
to determine the optimum location of
the lead within the water.

90

An investigation will also be made
of the desirability of borating part
of the water. These experiments have
just begun, and therefore no firm data
are yet available, however, there are
indications that the over-all reactor-
shield weight will be very low. This
1s a direct result of the small volume
to which the fissions are confined.
PERIOD ENDING MARCH 10, 1953

6. BULK SHIELDING FACILITY

J. L. Meem

R. G. Cochran

M. P, Haydon

K. M. Henry

H. E. Hungerford

E. B. Johnson

J. K. Leslaie

T. A. Love

F. C. Maienschean
G. M. McCammon

Physics Division

The air-scattering experaiments and
the program of i1rradiating monkeys
have been concluded. In addition,
some preliminary neutron spectral
measurements have been made, but
further work will be postponed untal
measurements have been completed on
the mockup of the top plug of the
SIR shield. The gamma spectral
measurements on the reactor part of
the divided shield are being applied
to calculations of a divided shield.

AIR-SCATTERING EXPERIMENTS

The air-scattering experiment at
the Bulk ShieldingFacility, originally
carried out last summer,(!*2?) has
been extended and i1mproved 1in an
effort to understand the serious
discrepancy that appeared to be extant
between this experiment and the calcu-
lations of the Shielding Board.(?)
With the 1increased power now available,
100 kw having recently been approved
for the BSR, 1t 1s now possible to
eliminate much of the extrapolation
previously required. In addition, 1t
was discovered that the radioactaivaity
1in the pool water was responsible for
a large part of the observed dose 1n
the mocked-up crew position. Measure-
ments were taken to determine the
effect of spurious radiations scatter-
1ng from the pool walls, as well as
from the reactor support structure.

(L) ANP Quar Prog. Rep June 10, 1952, ORNL- 1294,
p 46.

(Z)J L, Meem and H, E, Hungerford, Air-
Scattering Experiments at the Bulk Shielding
Facility, ORNL CF-52+7-37 (July 8, 1952)

(3)Report of the Shielding Board for the
Aivrcraft Nuclear Propulsion Program, ANP-353,
p 64 ff {Oct 16, 1950)

These were found to contribute only
negligibly to the observed neutron and
gamma doses.

Although the experiment has been
completed so recently that 1t as
1mpossible at this time to give a
complete report, the following con-
clusions can be drawn

1. The neutron dose 1n the crew
compartment appears to be lower by a
factor of 5 than the ANP-53 calcu-
lations i1ndicate. The advantage
accruing from this amounts to about
5000 1b for the standard crew-shield
design.

2. The gamma-ray dose in the crew
compartment appears to be higher than
indicated by the ANP-53 design by a
factor of 3.5. The weight penalty
associated with this 1s about 6000
pounds.

The conclusions differ from the
earlier ones both i1n the magnitude
of the discrepancies and 1n the amount
of material required to make up the
added attenuation. Since this experi-
ment 1s basically so very crude,
further exploitation of the Bulk
Shielding Facility in this type of
work 1s not considered worthwhile,
the present comparison with the ANP-53
calculations 1s considered adequate.

IRRADIATION OF ANIMALS

In addition to the two groups of
monkeys mentioned 1n a previous
quarterly report,¢*) a third group has
been i1rradiated by the Bulk Shielding
Reactor. The complete series of
experiments 1s summarized 1n Table 6.1.

(4)ANP Quar
ORNL- 1375, p 66.

Prog Rep Sept 10, 1952,

91
ANP PROJECT QUARTERLY PROGRESS REPORT

\
TABLE 6.1. IRRADIATION OF MONKEYS IN THE BULK SHIELDING FACILITY

IRRADIATION CONDITIONS SERIES 1 SERIES II SERIES III
Exposure rate,* rem/hr 1 0.25 4
Number of exposures 8 16 8
Time per exposure, hr 16 8 16
Time between exposures, days 7 7 7
Total exposure, rem 128 32 512
Number of animals used 12 12 12

*One-half dose in neutrons and ene-half dose 1n gamma rays

All exposures have been completed
and the animals have been returned to
the USAF School of Aviation Medicine
at Austin, Texas, where they will be
held for observation., It 1s expected
that the third group, which received
512 rem, will develop eye cataracts,
but that the first and second groups
will not If the experiment turns
out as anticipated, the threshold
for cataracts should be definitely
bracketed

A report 1s being prepared in
cooperation with the Health Physiecs
Division that gives details of the
dosimetry during the experiment.
Figure 6.1 shows one of the animals
ready to be placed 1n a watertight
cage for submersion 1n the pool.

NEUTRON SPECTROSCOPY FOR THE
DIVIDED SHIELD

Some preliminary neutron spectra
have been run with the proton-reco1l
spectrometer developed by Cochran
and Henry.{(5) Fast-neutron data have
also been taken with nuclear plates
and threshold detectors. All results
are 1n preliminary form and are not
vet suitable for reporting. The
experiments will be continued as
reactor time permits.

(S)R G Cochran and K M Henry, A Proton
Recoil Type Fast-Neutron Spectrometer, ORNL-1479
{in press)

92

GAMMA SPECTROSCOPY FOR THE DIVIDED
SHIELD

The data on the energy and angular
distribution of gamma rays from the
divided-shield mockup are being
tabulated for machine calculations.
The energy, angular distribution, and
total intensity of the direct, as well
as scattered, photons arriving at the
crew compartment will be computed.

FISSION ENERGY AND POWER IN THE
BULK SHIELDING REACTOR

A report on the power distribution
1n the reactor with a beryllium oxide
reflector has been completed, (%) and
a report on the determination of the
energy released per fission 1s beaing
prepared. Figures 6.2 and 6.3 1llus-
trate the time decay of neutrons and
gamma rays from the reactor after
shutdown. The data are normalized to
unity for operations at 1 and 100 kw
and plotted against time after scramming
the reactor. It 1s interesting to
note how the (¥,n) reactionon beryllium
keeps the neutron level fairly high,.
Without the beryllium oxide reflector,
the neutrons decay with periods that
are characteristic of the delayed-
neutron emitters.

(G)J L Meem and E B Johnson, Determination
of the Power of the Bulk Shrelding Reactor,
Part I, OBNL-1438 (an press)
PERIOD ENDING MARCH 10, 1953

PHOTO 10829

Fig. 6.1. Monkey and Irradiation Cage for the Bulk Shielding Facility Experiment.

93
14

FRACTION OF FULL POWER

DWG 17500

\ | _~GAMMA DECAY AFTER 1-kw OPERATION FOR 2 hr
|

L

5 [T —— _—GAMMA DECAY AFTER 100-kw OPERATION FOR 16 hr
D s

5 —-NEUTRON DECAY AFTER 1w OPERATION FOR 2 hr

10-3 NEUTRON DECAY AFTER 100-kw OPERATION FOR {6 hr

5 o ——

106
0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360
{hr 2 hr 3 hr 4 hr 5 hr & hr

TIME AFTER SHUTDOWN {min)

Fi1g 6.2. Decay of Neutrons and Gamma Rays from 0 to 6 hr After Shutdown

LU0ddY SSIY90dd ATHILYVNO LIIf0¥d dNV
S6

FRACTION OF FULL POWER

107!

107

-«
DWG 17504

TIME AFTER SHUTDOWN (sec)

Fi1g 6 3 Decay of Neutron and Gamma Rays from 0 to 400 sec After Shutdown

- GAMMA DECAY AFTER 1kw OPERATION FOR 2 hr
\ D — e
P —— e
[ —
v
P~ GAMMA DECAY AFTER 100 kw OPERATION FOR 16 hr
\\
\ /NEUTRON DECAY AFTER 1-kw OPERATION FOR 2 hr
\
T —— =
AY
‘\
N\
\\
AN
~.
N
J
™~
\
\ NEUTRON DECAY AFTER 100 kw OPERATION FOR 46 hr
\\</
\\‘-_-&
20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400
t min 2 mm 3 min 4 min 5 min 6 min

‘0T HOMVW ONIONA aorvad

€61
W
7. TOWER SHIELDING FACILITY

C. E. Clifford

T. V. Blosser

Physiecs Division

J. Y. Estabrook, ANP Division

The preliminary design of the
Tower Shielding Facility proposed
by the firm of Knappen-Tippetts-
Abbett-McCarthy (KTAM), architectural
engineers, was accepted by ORNL early
1n January. Some delay was necessitated
by the discovery that the tower design
first considered by KTAM proved
exceptionally heavy 1n structural
steel, particularly in the long
overhead members from which the loads
were supported by a vertical hoist,.
Since the heavier members raised the
background from structural scattering
beyond reasonable limits, a design was
proposed that eliminated the overhead
structure and reduced the weight of
the legs bya factorof approximately 2.
The designs and specifications of the
tower and associated buildings are
now scheduled for completion by KTAM
and the OBNL Engineering Department
on March 1.

Design, procurement, and con-
struction of the reactor and instru-
ments, as well as engineering of the
site, road, and utilities, are being
done by ORNL. This work 1s proceeding
ahead of schedule, at present.

TOWER DESIGN

The tower design 1s shown 1n
Figs., 7.1 and 7.2. The structure
1s 315 ft hagh and has a 200-ft insaide
clearance 1n the north-south cdirection
and a 101-ft i1nside clearance i1n the
east-west direction. The four tower
columns are 9 ft square and weagh
approximately 400 1lb per running foot.

The loads will be raised by saix
individual hoists - two for the

reactor shield and four for the crew
shield. The reactor shield can be
raised 200 ft off the ground, and 1t
may be moved laterally to approach:
the tower legs. This lateral movement
will permit measurement of the radi-
ations scattered by the tower structure,
The four hoists supporting the crew
compartment will allow considerable
flexibility 1n the crew compartment
position and orientation, 1including
the possibility of a variation of
from 35 to 100 ft in the reactor-to-
crew separation distance.

CONSTRUCTION SCHEDULE

The tentative schedule for outsaide
engineers and contractors for completing
their various phases of design and
construction of the Tower Shieldaing
Facility have been somewhat optimisti-
cally estimated as follows

February 24 Design completed and submitted

to bidders

March 17 Bids received
March 20 Contracts awarded
June 15 Construction of the roads, tower

foundation, and guy anchors

comp leted

November 15

All outside constructioncom-
pleted

December 31 Reactor critical

It appears that ORNL will eather
meet or be ahead of the time schedule
1n 1ts share of the Tower Shielding
Facility work. The following 1s a
list of ORNL’s functions, together
with the present status of accomplaish-
ment.

97
ANP PROJECT QUARTERLY PROGRESS REPORT

FUNCTION TARGET DATE  SER CENT
Engineering of site, building, and pool 3/1/53 100
Engineering of reactor, tank, and mechanical controls 3/1/53 100
Engineering of reactor controls (electrical portion) 7/1/53 20
Engineering of instrumentation (mechanical) 5/1/53 50
Engineering of instrumentation (electrical) 3/1/53 100
Construction of reactor, tank, and mechanical controls 9/1/53 5
Construction of reactor controls (electrical portion) 12/1/53 20
Construction of instrumentation (mechanical) 8/1/53 5
Construction of instrumentation (electrical) 2/1/53 20

98
Sy

Lt SN TNLTINL~TNLT

PERIOD ENDING MARCH 10,

Fig. 7 1

Tower Shielding Fac1lity (Perspective)

1953

DWG G-121178

99
ANP PROJECT QUARTERLY PROGRESS REPORT

~v g R

=— ¢ TOWER AND
GONTROL HOUSE

I —

| —

T ™\

~=— GONTROL HOUSE

e

b \06‘3// L/\\/,
1 5
\ )
l\\_
T DR
\ \ ; il ) B i |
\ ( T L. ) \ \ - _4\ ‘
! \ [\ TOWER STRUGTURE
® // > \ Ti///
o]
RAMP =1\ \
HOIST HOUSE JX \ ©,
\ ° g
bl \ \ Q
8 . ) e A \1
& o / ¢} / X %3
/ POOL — } 7 ‘Tl \Q POOL, TOWER,
o PAD |0 AND PAD\
" |
HOIST HOUSE & 0
/ / o
o // /
- /5’ ; & é@/ //
a /// / TOWER STRUGTURE
/ ‘ E_4 _1/ / // o / ]/ ’/ / [/ /
| &= -1-J / |-n; Tj u)
6////// T -—gff///,//>pz_,/i:i 9t // g
\96 /t 7101 ft ©
/S °
/S s
S/
/S /
;r_/M J/
//// 30 0 30 60 90
== — __——____—

SCALE IN FEET

Fig. 7 2.

100

Tower Shielding Facility (Plan View)
PERIOD ENDING MARCH 10, 1953

8. NUCLEAR MEASUREMENTS

A fast-neutron scaintillation
spectrometer employing UF crystals,
which have low gamma-ray response,
1s being developed, The 6-Mev Van de
Graaff has been set up in the High-
Voltage Laboratory, but no research
has yet been undertaken in the new
location.

FAST-NEUTRON SCINTILLATION
SPECTROMETER

J. Schenck F. J. Muckenthaler
Physics Division

The work on the development of a
scintillation spectrometer for fast
neutrons has appeared so promising
for shielding work that two research
personnel have been assigned to the
project to accelerate the work.
Although europium-activated lithium

10odide crystals have been used to
date, lathium fluoride would be much
more desirable because of 1ts lower
gamma-ray response. Jhe work 1s now
concentrated i1n an effort to grow a
lithium fluoride crystal with suitable
scintillation properties.

MEASUREMENTS WITH THE 6-MEV
VAN de GRAAFF

H.B. Willard, Physics Divisaion

The 6-Mev Van de Graaff was moved
during the quarter from 1ts temporary
location in Building 9201-2, Y-12
site, to 1ts permanent quarters in the
High-Voltage Laboratory, Building
4500, X-10 site. A beam was fairst
obtained late in the quarter, but no
cross—section measurements were
undertaken.

101
Part 111

MATERIALS RESEARCH
INTRODUCTION

The research on fused fluoraide
systems (sec 9) has been largely
concerned with the development of
techniques for purifying and preparing
enriched ARE fuel and fuel carrier
The fluoride treatment procedure
1involves successive purging of the
liquid with HF, H,, HF, and finally
He. The treated fuel then contains
only traces of HF, NiF,, and FeF,,
which are primary factors i1n fluorade
corrosion The beneficial effect on
the corrosion of the reducing agents
has been substantiated, and the
quantity of the various reductants
that may be tolerated 1n the fuel
has been defined. The consequences
of the attendant reduction of some UF,
to UF; are being investigated. For
the longer-range fuel development
program, several ternary and quaternary
fluoride systems containing LiF are
being 1nvestigated, as well as some
binary chloride mixtures containing
ucCl,

The corrosion rksearch (sec. 10)
has been primarilyon the determination
of the corrosion characteristics of
fluoride mixtures. The 1ncreased
attack experlenced 1n recent tests
with the ARE fuel mixture 1s believed
to be due to faulty purification
rather than to handling techniques
Zirconium hydride as an additive to
the fuel mixture reduces the fluoraide
attack on Inconel by a factor of 2
Apparently the mechanism 1s the
reduction of the metallic fluorides
Ni1F, and FeF;, which are present 1in
the fuel, because addition of these
fluorides to the fuel 1ncreases the
corrosion. Measurements of the concen-
tration of NiF,, FeF,, and CrF,; as a
function of exposure time of Inconel
in the fuel have been made as an
1index of corrosaion The nickel
content remains approximately constant
and the 1ron content drops sharply to

a constant value, however, the chromium

AND SUMMARY

content rises during the first 100 hr
and then levels off After the
addition of ZrH,, the only significant
change 1n these concentrations was 1n
the chromium content, which then
changed 1n the same manner as the 1ron
content. The addition of chromium
metal was not effective i1n inhibiting
fluoride corrosion, apparently because
of the low solubilaty of chromium 1in
the fluoride Although Inconel and
stainless steel are severely attacked
by lead 1n tests 1in convection loops,
molybdenum and columbium are relatively
1mmune However, static tests of
lead-sodium alloys show that sodium
additions decrease the otherwise
severely corrosive action of the lead
on both Inconel and stainless steel

The stability of BeO 1n NaK 1s stall
being investigated. It appears that
there 1s some solubility of BeO 1in NaK

The metallurgy and ceramics research
(sec 11) includes the fabrication of
various reactor components for use at
high temperatures, creep rupture tests
of structural metals, the development
of special alloys, cermets, and ceramic
coatings, and tests of welding and
brazing alloys Control or safety rod
inserts are being prepared for the
G-E reactor and for the Tower Shielding
Facilaity Chrome-plated, high-
conductivity fins have not proved
adequate for high-temperature radiator
use, but Inconel or stainless-steel-clad
copper fins may now be satisfactorily
brazed. Recent creep tests of stain-
less steel and Inconel have shown that
surface oxidationis not the controllang
factor i1n the longer rupture times
observed 1n air as compared with the
rupture times 1n hydrogen and argon
The cone-arc welding technique 1s
being adapted to the welding of
dished headers for heat exchangers
because of the unequal arc distances
encountered. Because of dilution and
embrittlement of the Nicrobraz used 1in

105
ANP PROJECT QUARTERLY PROGRESS REPORT

brazing a heat exchanger element
other alloys have been considered for
this application. The evaluation of
high-temperature alloys with respect
to temperature, strength, and corrosion
resistance 1s continuing

The physical property measurements
were made primarily for the determi-
nation of various properties of
several fluoride mixtures, and the
heat transfer studies were concerned
with experimental systems, as well as
aircraft components (sec. 12). The
heat capacity of the ARE fuel was
found to be 0 31 cal/g*°C from 550 to
850°C. Vapor pressure measurements of
the enriched fuel mixture were lower
than expected and i1ndicated the
formation of complex 10ons at high
temperatures. Preliminary measurements
were made of both the thermal conduc-
tivity and the viscosity of sodium
hydroxide at elevated temperatures.
A minimum-weilght analysiswas completed
for an aircraft radiator with the
optimum fin spacing and thickness and
the optimum tube spacing and diameter
Experimental apparatus to determine
the fluid velocity profiles and the
fluid temperature structure 1s being
assembled One study 1s under way to
determine temperature distrabution 1in
entrance regions, whereas another
study will compare the various high-
temperature heat transfer fluids.

Creepunder 1rradiation and corrosion
by sodium 1n an in-reactor loop are

106

being investigated, however, 1irradiation
of fluoride mixtures comprises the
greater part of the radiation damage
studies (sec 13) In recent 1r-
radiations of the ARE fuel mixture 1in
the MTR, i1n which the power generation
1s 20 times that expected in the ARE,
corrosion and changes 1n the fuel
composition occurred that were 1n
excess of those observed 1n control
tests. However, the Inconel and the
fuel were at temperatures 1n €XcCess
of 1500°F during the tests The
recent i1n-reactor creep measurements
confirmed previous conclusions that
irradiation had little effect on the
creep strength

studies of reactor
materials (sec 14) included chemical,
petrographic, and x-ray diffraction
1dentification of impurities, corrosion
products, reduction products, and
constituents of reactor fuels Volu-
metric methods for the determination
of zirconium 1n the presence of
uranium and for the simultaneous
determination of uranium trifluoraide
and zirconiummetal have been developed.
An apparatus has been built to aid in
the determination of traces of oxides
1n fluorades With petrographic and
x-ray-diffraction studies, 1t has been
possible to determine and define
compositions of the compounds and
eutectics present ain UCl,-NaCl,
UC1,-KCl, and NaF-ZrF,-UF, systems.

Analytical
PERIOD ENDING MARCH 10, 1953

9. CHEMISTRY OF HIGH-TEMPERATURE LIQUIDS

W. R. Grimes, Materials Chemistry Division

The major effort of the ANP Chemistry
group continues to be devoted to the
study of fluoride mixtures for use
as aircraft reactor fuels and coolants.,
A large fraction of this effort has
been devoted to the specific materials
that comprise the fuel solvent, the
concentrated fuel, and the final fuel
mixture for the ARE.

The final processing plans for
preparation of the fluoride mixtures
1n pure form are nearly finished
Progggsires for removing all except
traces of HF, N1F, and FeF, from these
materials are in the final develop-
mental stages. Contrary to previous
beliefs, hydrogen does cause partial
reduction of UF, to UF; 1n these melts
at 800°C. The beneficial effect of
reducing agents on the corrosion of
fluoride mixtures has been well sub-
stantiated, and the quantities of
various reductants thatmay be tolerated
1n the fuels has been defined.

Phase equilibrium and other studies
of systems 1n which all or part of the
UF, has been replaced by UF; are still
being carried out as a part of the
study of the effect of added reducing
agents and as an aid 1n the 1dentifi-
cation of corrosion products Adda-
tional materials have been synthesized
that contain trivalent uranium and
seem to be 1dentical with species
produced during corrosion. In the
hope of developing other and superior
fuel mixtures for future applacations,
an extensive study of phase equilibria
1n systems containing lithium fluoride
1s being carried out. Some study of
UCl, systems 1s also 1n progress

FUEL MIXTURES CONTAINING UF,

L. M Bratcher C. J. Barton
Materials Chemistry Division

L1F-ZrF,-UF,. The L1F-ZrF, -UF,
system was investigated with UF, con-

centrations of up to 45 mole % and
ZrF, concentrations of up to 50 mole
%. The lowest melting point observed
was 440°C for the mixture containing
68.5 mole % L1F, 26.5 mole % UF,, and
2.0 mole % ZrF,. Thermal halts were
noted with this composition, as well
as with other compositions nearly the
same, at about 400 to 415°C, however,
1t cannot be ascertained definitely
from the data that have been obtained
whether this effect 1s due to a
eutectic of unknown composition or to
a solid transition

KF-LiF-BeF,-UF,. The effect on the
melting point of varying the UF, con-
centration of one composition 1n the
KF-Li1F-BeF, -UF, system has been
determined. The data show that when
UF, 1s added to the 40% BeF,, 20% KF,
and 20% LiF mixture the melting point
goes through a minimum at a low con-
centrationof UF, and then rises rather
rapidly wath increasing UF, concen-
tration. The thermal effect noted at
330 to 340°C may indicate the existence
of a low-melting-point eutectic 1n
the four-component system, a great
deal of work will be required to

determine the composition of the
eutectic. The lowest melting tempera-

ture observed to date 1s 380°C at
2.5 mole % of UF,.

FUEL MIXTURES CONTAINING ThF4

L. M Bratcher C. J. Barton
Materials Chemistry Divisaion

Other 1nvestigators{(!) have shown
that the LiF-ThF, equilibrium diagram
1s quite similar to that of the LiF-UF,
system, and unpublished data obtained
1in this laboratory indicated that the
BeF, -ThF, system 1s similar to the
BeF,-UF, system Since the LiF-BeF,
binary 1s common to both systems, 1t

(I)J O Blomeke, An Investigation of ThF4-

Fused Salt Solutions for Homogeneous Breeder
Reactors, ORNL-1030 (June 19, 1951)

107
ANP PROJECT QUARTERLY PROGRESS REPORT

seemed reasonable to assume that the
Li1F-BeF, -ThF, equilaibrium diagram
would be comparable to the diagram for
the LiF-BeF -UF, system described an

the previous report.{(2?) The data
shown 1n Fig. 9.1 confirm this ex-

pectation. The higher melting point
of the L1F-ThF, eutectic (550°C com-
pared with 490°C for LiF-UF,) 1s
reflected 1n the contours. A com-
position containing 5 mole % ThF,, 67
mole % LiF, and 28 mole % BeF, melts

(2)L M Bratcher and C. J Barton, ANP Quar
Prog Rep Dec 10, 1952, ORNL-1439, p 105
-~

at approximately 430°C., From the
limited data obtained with the lower
ThF, concentrations, 1t appears that
compositions i1n this system that melt
below 400°C will contain a low con-
centration of ThF, and a high concen-
tration of BeF,.

FUEL MIXTURES CONTAINING UCI4
R. J. Sheal C. J. Barton

Materials Chemistry Division

The problem of obtaining pure UCl,
has not yet been completely solved.
Efforts to prepare the pure material

. W \
s Lo \ 5
‘0 ., \G DWG 18746
}&\ Y \& 1114 °C
\"x g
\
1}
30 X
(¥}
U\T- ‘" .
A M
X 3\‘-1;’- .~ u‘/ - 1000°¢
r-". »,\.lb '% i
?y- s \D ’ ‘\‘{ o ' \
\Va J/a/ 950
900
850°C
g00™°

550°C

50T

LiF

543°C
Fig. 9.1,

108

LipBeF, 845

475°C

The Systenm LlF-Ber-ThF4.
by sublimation from the available
crude UCl, have not been successful.
Synthesis of UCl, by liquid-phase
chlorination with hexachloropropyl-
ene{3) has been used to provide
material for recent studies. Although
analytical data from which to evaluate
this material are not yet available,
1t appears that the material 1s not
of good quality The data reported
briefly below may therefore be revised
when purer UCl, becomes available.
NaCl-UCl,. Thermal data on 18 com-
positions i1n the NaCl-UCl, system have
been reported by Kraus.?4) He con-
cluded, on the basis of the data
obtained from cooling curves, that
the compound 2NaCl-UCl, was the only
compound formed 1n this system and
that 1t melted 1ncongruently at
430 £+ 5°C. It also appeared from
hi1s data that there was only one eu-
tectic, which melted at 370 * 5°C. The
present,lncomplete investigation seems
to show the existence of two eutectics
and two compounds in this system. The
first eutecticis believed to beat about
30 mole % UCl,, and 1t appears to
melt at 420 %+ 10°C. One of the
compounds, Na,UCl,, 1s believed to
melt congruently at 440 * 10°C. The
second eutectic 1s believed to be
at approximately 45 mole % UCl,, and
1t appears to melt at 370°C, 1n good
agreement with Kraus’' results. In
the 65 to 90 mole % UCl, range, a
series of breaks at 420 + 10°C probably
represents the i1ncongruent melting
point of a compound such as NaU,Cl,
or NaU,Cl,, The pure compound has
not yet been prepared, and 1t cannot
be definitely 1dentified on the basais
of information available at this time.
Kcl-Ucl,. Thermal data on 14 com-
positions i1n the KCl-UCl, system, 1n
the 20 to 75 mole % UCl, range, have
also been reported by Kraus.(*) These

(3)B M Pitt et al , The Preparation of TCl4
vith Hexachloropropylene, C-2 350 3 (July 27,
1945)

(4)C A Kraus, Phase Diagraa of Some Complex
Salts of Uranium with Halides of the Alkal:i and
Alkaline Earth Metals, M-251 (July 1, 1943)

PERIOD ENDING MARCH 10, 1953

data showed that there was one com-
pound, K,UCl ., which melted congruently
at 650 + 5°C, and two eutectics 1in the
system. The KC1-K,UCl, eutectic was
stated to be at 25 mole % UCl,, and
1tsmelting point was given as 5601 5°C,
the K,UC1 ,-UCl, eutectic was 1ndicated
to be at 53 mole % UCl,, and 1ts
melting point was 350 + 5°C. Thermal
data and petrographic examination of
the soli1d phases show strong evidence
for the existence of three compounds
and three eutectics 1n this system.
The present results are in fair agree-
ment with those of Kraus on the
location and melting point of the
KC1-K,UCl, eutectic, the melting point
of K,UCl,, and the melting point of
the higher eutectic. However, 1t
appears that a K,UCl,-KUCl, eutectic
with 42.5 £ 2 5 mole % UCl, occurs 1n
this system and that 1t melts at
320 £ 10°C The compound KUCI,
probably melts congruently, but a
reliable value for 1ts melting point
has not yet been obtained. The thermal
evidence for the existence of a com-
pound with more than 1 mole % UCIL,
per mole of KCl 1s a series of breaks
1in the 70 to 85 mole % UCl, region at
400 + 5°C. The thermal effect at thas
temperaturemay indicate the 1incongruent
mel ting point of this compound of
unknown composition. As mentioned 1n
the section on NaCl-UCl,, and 1in line
with experience with fluoride systems,
1t has been found rather difficult to
prepare incongruently mel ting compounds
in sufficiently pure state to permit
their positive 1dentification. The
composition of the KUCl.-KU Cl, .,
eutectic has not been accurately
determined.

FUEL MIXTURES CONTAINING UF,

V. S. Coleman C. J. Barton
Materials Chemistry Division

Work on the NaF-UF3 and KF-UF3

systems has not yet revealed com-
positions that melt below 700 and
680°C, respectively. Complications

109
ANP PROJECT QUARTERLY PROGRESS REPORT

from UO, and UF, 1n the melts have
been minimized byusingalkali fluorades
that have been treated with HF at
high temperatures and by minimizing
the top temperature to which the melt
1s heated to decrease disproportiona-
tion of the UF,.

These systems are
present,

studied, at
in nickel capsules welded
under an argon atmosphere. The
examination 1s hampered because the
KF-UF, compounds decompose readily in
the laboratory air. Petrographic
examination 1s accomplished by grinding
the material i1na dry, inert atmosphere
and transferring 1t to microscope
slides under o1l. Although the thermal
data are as yet 1nconclusive, 1t
appears that the compound formed 1n
NaF-bearing melts 1s 3NaF-2UF,.

ADDITION OF REDUCING AGENTS TO ARE FUEL

J D. Redman V. S. Coleman
L. G. Overholser D. C. Hoffman
K. J. Kelly F. F. Blankenship

Materials Chemistry Division

Decreased corrosion of Inconel 1in
thermal convection loops has been
demonstrated to result from small
additions of NaK, ZrH,, or Zr° to
fluoride mixtures of i1nterest as ARE
fuels., When these reducing agents
are added 1n large quantities to
materials containing UF,, they produce
UF,, which 1s only sparingly soluble
in the melt at 600°C. Therefore the
reactions of possible ARE fuels with
various reducing agents have been
studied i1in an effort to discover a
safe upper limit to the amount of such
additives.

Previous experiments had demon-
strated that addition of sufficient
reductant resulted i1n separation of
UF, as a discrete phase, which could
be removed by filtration at 600°C.
The resulting filtrates revealed no
trace of the easily detectable UF,
after solidification, but they showed
considerable quantities of various

110

complex compounds of UF,. Identical
mixtures (with the same reducing
agents) that were not filtered showed
considerable quantities of UF,. If
a sufficiently small quantity of
reducing agent was added so that no
UF, could be filtered at 600°C, there
was no UF, detected in the cool,
solidified material

Various ZrkF,-bearing mixtures, with
and without added UF,, have been
equilibrated with predetermined small
quantities of NaK (78 wt % K), ZrH,,
and Zr° 1n sealed capsules of Inconel
by rocking for 16 hr at 4 cpm with the
ends of the capsules at 800 and 650°C,
respectively. The capsules were then
heated at 800°C 1n a vertical position
for 2 to 6 hr to permit settlaing of
insoluble material and, especially,
unreacted reducing agent., Cooled
samples were sectioned longitudinally
for petrographic examination.

The upper tolerance laimits an
number of equivalents (one equivalent
1s the quantity of reductant needed
to reduce all UF, to UF,) are listed
in Table 9.1.

Reducing agents added to NaZrF; 1in
amounts 1n excess of 0.2 wt % showed
opaque nonmagnetic masses, and chemical
and x-ray tests indicated the materaial
to be, at least in part, Zr°. Appar-
ently, trivalent zirconium does not
exi1st 1n fluoride melts under these
circumstances.

Verificationof the tolerance values
for ZrH, 1n the ARE fuel mixture has
been obtained by equilibration of ZrH,
with the fuel mixture at 800°C 1n
nickel equipment, followed by equila-
bration at 600°C and filtration of the
resulting melt at 600°C through
sintered nickel. The data obtained
by chemical analysis of the filtrate
and by petrographic examination of the
filtered residue are shown in Table 9.2,

The data indicate that no solad
materi1al exists i1n the liquid at
600°C 1f less than 0.9 wt % ZrH, as
added to the mixture. Agreement
between the data shown 1s probably
within the error of such experiments.
The failure to observe UF,, as such,
in any of the filtrates affords con-
firmation of the assumptions used to
interpret the previous data.

When small quantities (up to about
0.4 eq) of reducing agent are added,
an olive-drab phase (average refractave
index of 1.558) occursin the NaF-ZrF, -
UF, mixtures.

Reducing agents added to compo-
sitions 1n the NaF-ZrF,-UF, system
produced an olive-drab phase (average

PERIOD ENDING MARCH 10, 1953

refractive 1ndex of 1.558) at the
0 2-eq concentration. The quantaty of
olive-drab phase generally increased
upon addition of 0.4 eq of reducang
agent, With larger additions of
reducing agent, a decrease 1n the
olive-drab phase was observed, along
with an i1ncrease inthe red-orange phase
(UF,-2ZrF,). Synthesis and petro-
graphic examination have i1ndicated
that the olive-drab phase may be
UF3-2ZrF4, in which tetravalent uranium
has been partially substituted for

TABLE 9.1. AMOUNTS OF REDUCING AGENTS TOLERABLE IN FLUORIDE MIXTURES
AMOUNT TOLERABLE
FLUORIDE USED NaK ZrH, 2c®
Equivalent Weight % Equivalent®| Weight % Equivalent Weight %

NaZrF5 02 02 02
NaF-ZrF4-UF4

(46-50-4 mole %) 06 07 12 10 12 09
NaF-ZrF4-UF4

(50-46-4 mole %) 0.4 035 08 07 12 10

*Hydrogen considered neutral in calculation of equivalent weight,

TABLE 9. 2. DATA OBTAINED BY CHEMICAL ANALYSIS AND PETROGRAPHIC
EXAMINATION OF ARE FUEL®* AFTER TREATMENT WITH ZrH2
LRANIUM CONTENT
er2 ADDED OF FILTRATE PETROGRAPHIC OBSERVATIONS
(wt %)
(wt %)
0 8.62
0.2 8.72 Uranium reduction occurred,
0.5 8.68 no discrete UF; observed
0.7 8.65
0.9 6.58 UF; found on filter and in
1.0 7.12 reaction chamber
1.5 5.95
1.5 8.64 Presence of Zr0O, indicated
an oxygen leak

‘NaF-ZrF4-UF4 (50-46-4 mole %).

111
ANP PROJECT QUARTERLY PROGRESS REPORT

trivalent uranium. Petrographlc
examinations revealed that UF,-UF,-
4ZrF, was an olive-drab, pleochroic
phase, with an average refractive
1index of 1.564, a birefringence of
about 0.008, and an optic angle of
80 degrees. Another olive-drab
material (refractive index of 1.574)
was obtained when these materials
were mixed to correspond to UF,-3UF,-
8ZrF,. A mixture corresponding to
3UF,-UF,-8ZrF, produced striated
crystals of olive-drab and red-orange
materials.

The 50 mole % NaF composition
developed free UF, upon addition of
1.2 eq of Zr®° or ZrH,. The 46 mole %
NaF composition formed free UF; upon
the addition of 1.6 eq Zr° or ZrH,.
The usual olive-drab and orange-red
phases were observed 1n these com-
positions with less than 1.2 eq of
reductants.

It was noted throughout this study
that the 46 mole % NaF, 50 mole %
ZrF,, 4 mole % UF, composition toler-
ated slightly more reduction without
precipitating UF, than did the 50
mole % NaF, 46 mole % ZrF,, 4 mole %
UF, composition. Further, 1t was
noted that both compositions are more
tolerant to Zr° and ZrH, than to NaK.
These observations support the mecha-
nism suggested previously, (%) which
postulates that NaF and UF, compete
for the ZrF, present. The Na(K)F
formed by the NaK reduction reaction
1s capable of breaking the UF,:2ZrF,
complex to form Na(K)ZrF, and free
UF,. On the other hand, ZrF4 formed
by the Zr° or ZrH, reduction reaction
1s capable of complexing more UF,.
This suggests that a fluoride com-
position with a ZrF,-to-NaF ratuio
larger than unity would be desirable
1f only because of 1ts capacity to
complex UF,.

(S)F F Blankenship, D C Hoffman, K J
Kelly, and T N McVay, ANP Quar Prog Rep
Dec 10, 1952, ORNL-1439, p 119

112

COOLANT DEVELOPMENT

.. M. Bratcher C. J. Barton
Materials Chemistry Division

LiF-ZrF,. Study of the LiF-ZrF,
system, mentioned 1n the previous
progress report,(s) has been continued
by thermal analysis andby petrographic
and x-ray-diffraction examination of
solidified melts. Although Li,ZrF  1s
the only compound that has been
positively 1dentified, there 1s some
evidence for one additional compound
with less than 33 mole % ZrF, and one
with more than 50 mole % ZrF,.

CsF-7ZrF,. Both Cs,ZrF, and CsZrF,
have been 1dentified by petrographic
examination of melts in the CsF-ZrF,
system, and three eutectics have been
shown to exist. The lowest-melting-
point (412 * 10°C) eutectic contains
about 40 mole % ZrF,. Further work 1s
1n progress to complete the equi-
librium diagram for the system.

KF-L1F-ZrF,. The KF-LiF-ZrF, system
has been examined over a ZrF, concen-
tration range of 5 to 60 mole %. A
ternary eutectic melting at about
385°C appears close to the KF-ZrF,
eutectic. The exact composition of
this eutectic has not yet been estab-
lished.

RbF-L1F-ZrF, Melting-point deter-
minations have been made for com-
positions in the RbF-LiF-ZrF, system
that have 5 to 60 mole % ZrF,. The
results were similar to those obtained
with KF-LiF-ZrF, mixtures. The lowest
melting point shown to date was 370°C
for amixture containing 45 mole % RbF,
45 mole % ZrF4, and 10 mole % La1F,
this composition 1s probably close to
a ternary eutectic. Thermal effects
at about 550°C were observed with a
number of compositions i1n this region.
These effects are believed to be due
to the presence in the fused mixture
of complex oxyfluorides that may
result from the hygroscopic nature of

the RbF.

(6) M Bratcher and C J Barton, ANP Quar
Prog Rep Dec 10, 1952, ORNL-1439, p 113

KF-LiF-BeF,. The data obtained
thus far with compositions 1n the 4 to
57 mole % BeF, range of the KF-LiF-BeF,
system seem to indicate that only
slight lowering of the melting point
results from the addition of KF to
LiF-BeF, mixtures. The lowest melting
point that has been confirmed by both
heating and cooling curves 1s 350°C
for the mixture with 54 mole % BeF,,
10 mole % KF, and 36 mole % LiF, and
1t 1s only slightly lower than the
melting point (365°C) of the 50 mole %
L1F, 50 mole % BeF, composition.

PRODUCTION AND PURIFICATION
OF FLUORIDE MIXTURES

F. F. Blankenship G. J. Nessle
Materials Chemistry Division
H. W. Savage, ANP Division

Study of the hydrogenation-hydro-
fluorination procedure previously
described¢’) for fuel preparation has
been continued on laboratory and pilot
scales. Operating conditions to
provide optimum purity of the molten
mixtures are now reasonably well
defined.

Construction of the production-
scale equipment for producing 3000 ib
of ARE fuel solvent (NaZrF.) has been
delayed to some extent by lack of
manpower. This equipment should be
avallable for testing early in Aprail,
and the material should be prepared
during May and June. All raw materials
for this operation have been received,
and chemical and spectrographic
analyses of each batch indicate that
the materials are of satisfactory
purity.

Laboratory-Scale Fuel Preparation.
(C. M, Blood, F. P. Boody, R. E. Thoma,
Jr., Materials Chemistry Division).
Corrosion by selective removal of
chromium from Inconel 1n fluoride
melts may occur by any combination
of several mechanisms. Three of the

(TYF F Blankenmship and G J Nessle, ANP
Quar Prog Rep Dec 10, 1952, ORNL-1439, p 122

PERIOD ENDING MARCH 10, 1953

most likely of these processes are

Cr® + 3UF, < CrF, + 3UF; , (1)

2C:° + 3HF < 2CrF, + 3H, (2)
and

2Cr° + 3N1F, < 3N1° + 2CrF,

2Cr® + 3FeF, < 3Fe® + 2CrF, ,  (3)

Cr® + FeF, < Fe° + CrF,

Addition of the strong reducing agent
ZrH, would be expected to minimize
the corrosion, regardless of whaich
mechanism was effectave.

Corrosion by the first mechanism
can be minimized by reduction of UF
to UF, by some reductant such as ZrH,
to the extent tolerable without
excessive deterioration of the physical
properties of the fuel. Corrosion by
the other mechanisms can, 1nprancaiple,
be eliminated by the proper techniques
for removing the reagent materials.

The hydrogenation-hydrofluorination
process for fuel purification was
originally chosen as the one most
likely to produce very pure fluoraide
liquids. The schedule for this process
was prescribed to provide what was
considered to be large excesses of
time for reaction and stripping at
each stage. In recent experiments,
attempts were made to quantitatively
evaluate the chemical changes during
each process stage. The need for such
data 1s obvious. Incomplete treatment
at any stage 1s reflected by poor
corrosion characteristics, however,
long treatment times are uneconomical
1n terms of equipment laife, production
rate, material consumed, and manpower
required and, in addition, result in
slight alteration in fuel composition
because of volatilization of the ZrF,.

Stripping with 1inert gas should
serve as a rapid and reasonably com-
plete method for removing HF, because
the acid fluorides are unstable at
temperatures above 500°C. Since any
residual HF should lead to enhanced
corrosion, the progress of HF removal

113
ANP PROJECT QUARTERLY PROGRESS REPORT

has been studied i1ntensively by
analyzing the exit gas stream for HF.

Initial experiments revealed that
plots of the logarithm of HF concen-
tration in the helium vs. volume of
helium passed were essentially linear,
with half-life volumes between 40 and
70 liters when 0.5 to 1.2 liters/main
of helium was passed through about 1
liter (3 kg) of the melt. From tests
of the stripping of the HF by He, 1t
may be concluded that typical batches
of fuel prepared by the standard
procedure described previously should
have been stripped to about 10°% mole
of HF per liter of He and should have
contained about 10 ppm of HF in the
melt, There 1s no detectable difference
in stripping rate with He between the
NaZrF, and the NaF-ZrF,-UF, mixture
containing 4 mole % UF,. The effect
of flow rate on attainment of equi-
librium vapor concentration has been
only partly explored.

When hydrogen 1s used to sweep the
melt after passage of HF, the acid
content of the effluent gas 1s con-
siderably higher than when helium 1s
used. The following reactions are
almost certainly responsible

NaF, + H, = N1° + 2HF, AF® 000k = —65 keal ,
2FeF,<+ H, — 2FeF, * 2HF, AF° ;0% = =31 5 keal ,
FeF, + H, — Fe® + 2HF, AF° 000% = 1 5 kecal ,

2UF, + H, = 2UF; + 2HF, AF®;,00°g = 22 O keal .

In treatment of NaZrF, with hydrogen
at 800°C half-li1fe volumes of about
70 liters are found. Since most of
these experiments are performed 1in
nickel containers, appreciable amounts
the HF result from reduction of NiF,.

In stripping of the ARE fuel mixture
with H,, the contrast between He and
H, 1s even more noticeable. When H,
1s used, the HF content of the gas
drops sharply, as expected, but 1t
levels off at about 10°* mole HF per
liter of H,. The reduction of UF, to
UF; 1n the solid state 1s considered
negligible below 900°C. However, UF,

114

has been shown by petrographic tech-
niques to be present in ARE fuel
batches after prolonged H2 treatment
at 700 to 800°C. There seems to be
no doubt that the high values for
HF i1n the exit gas are a consequence
of this reaction. From the rate of
HF evolution, 1t appears that the rate
of reduction 1s about 3% of the UF,
present per hour.

Filtration of the melt after
hydrogen treatment and HF stripping
serves to reduce the nickel content
of the fuel batch to less than 150
ppm. It has not been possible, to
date, to bring the i1ron content of
the melt to such a low level. It
appears, however, that low values for
N1F2, FeF,, and HF can be achieved
without excessive exposure times and
without undue reduction of the UF,.

Pilot-Scale Fuel Purification (G.
J. Nessle, J. E. Eorgan, Materials
Chemistry Division). During the past
quarter, a total of 326.4 kg of mixed
fluorides, comprising 24 small batches
of approximately2 kg each and 11 large
batches of approximately 25 kg each,
has been processed by the group in
Building 9928 and dispensed to re-
questing personnel in the ANP Division.
Heating the control panels to 125°F
has resulted 1n marked improvement 1in
the smoothness and efficiency of
operation. The addition of two
pressure gages, one directly on the
reactor can and one directly on the
receiver can, has eliminated the
difficulty i1n discovering possible
danger points, and minor failures
during a run are now practically
nonexistent,

The process for treatment of mixed
fluorides has been modified to permit
1.5 hr of stripping with H, and 1.5 hr
of stripping with He after the HF
treatment and before transfer of the
molten fluorides. A gas throughput of
about 15 liters/min 1s accomplished
by pressurizing to 10 lb (gage) and
releasing to 0 1lb (gage) 30 times per
hour. The HF content of the straip

gas 1s usually about 10”% mole/later
at the time of fuel transfer. Thais
probably corresponds to about 10 ppm
of HF 1n the melt. The combaination
of H, treatment and filtration through
sintered nickel has reduced the soluble
nickel 1n the melt from 900 to less
than 150 ppm.

Fluoride Production Facility (G. J.
Nessle, Materials Chemistry Division,
H. W. Savage, ANP Division). The
production facility for preparing in
excess of 3000 1lb of fuel solvent 1is
scheduled for completion during the
month of March. Shakedown runs and
other preliminary tests of the appara-~
tus should require less than four
weeks, the production operation could,
1f necessary, be finished by mid-June.

The low-hafnium ZrCl, has been con-
verted to ZrF,, and the equipment for
this process has been cleaned and put
into standby condition for future
use, 1f needed. At present, the
stockpile contains 3000 1b of ZrF,.

All batches of the ZrF, have been
sampled and analyzed by chemical
met hods for zirconium and fluoraine.
Complete spectrographic analyses,
with special shots of each batch,
for hafnium and boron are available.
The boron content of the batches
ranges from 1.8 to 0.3 ppm, with an
average of 0.8 ppm, and seven of the
nine batches show less than 85 ppm
hafnium. The other two batches,
comprising about 25% of the total,
show 150 and 550 ppm hafnium.

PERIOD ENDING MARCH 10, 1953

About 1000 1b of NaF from one
production lot has been stockpiled
for use 1n fuel production. Chemical
and spectrographic examinations of
samples from three containers indicate
that the material 1s satisfactory for

use
Hydrofluorination of Zroz-NaF

Mixtures (R. E. Thoma, Jr., C. M.
Blood, Materials Chemistry Division).
The feasibility of producing NaZrF,
by a short treatment of NaF-ZrQ,
mixture with HF at a high temperature
has been demonstrated. Best results
are obtained when some previously
prepared NaZrF; 1s added to the charge
so that a liquid phase 1s present
during the 1nitial stages. An amount
of NaZrF¢ that 1s equivalent to 10% of
the total charge appears to be suffai-
cient to bring about complete con-
version of the mixture in 4 hr at
800°C. Utilization of about 30% of the
HF delivered has been attained.

There are still some problems in
connection with the equipment for thas
operation. Frequent rupture of the
HF input line occurred when nickel
tubing was used. Metallographic
examination revealed that the attack
was due to sulfur, which was probably
present in the Zr0O, feed. Graphaite
liners 1n the nickel reactors have
been used satisfactorily, however,
there has been difficulty in the use
of graphite dip lines for HF because
of breakage during runs.

Further study of this procedure,
which promises to reduce materially
the cost of ZrF, as NaZrF,, 1s planned.

115
10.

CORROSION RESEARCH

W. D. Manly, Metallurgy Division
W. R. Grimes, Materials Chemaistry Division
H. W. Savage, ANP Division

Corrosion phenomena are usually
examined first 1n static or seesaw
tests, and then selected systems or
phenomena are examined 1n the convection
loops. Crevice corrosion, effect of
exposure time, and the effects of
additives have been studied in both
static and dynamic tests during the
past quarter No significant increase
1in corrosion was noted ain static
tests However, the crevice corrosion
found 1n thermal convection loops
was about three times that found 1n
adjacent tubing, the mechanism by
which crevices 1ncrease corrosion
1s not yet understood. Corrosion, as
evidenced by the chromium metal content
of the fluoride, has been shown to be
fairly constant after the first 100 hr
of exposure Convection loops that
were run for 1000hr at 1500°F 1ndicated
that the amount of corrosion decreased
with time, although there was a small
increase 1n the depth and intensaty
of attack.

The work on the dynamic testing of
the fluoride fuels 1n Inconel thermal
convection loops has continued, with a
major effort being made to determine
why the corrosion 1n the
fluoride systems has been 1increasing.
It has been shown that neither the
increase i1n fluoride batch size nor
poor handling techniques during
filling were responsible, retreating
the fuel batches has produced contra-
dictory results. Evidence indicates
that the i1ncrease 1n corrosion 1s
being caused by faulty purification,
since the corrosion was deeper 1n
loops 1n which hydrofluorinated
fluoride mixtures were circulated
than 1t was 1n the loops that used
as-melted fluoride as the fuel. One
Inconel thermal convection loop was

Inconel -

operated with a hot-leg temperature of
1650°F, and no large increase 1n the
depth of attack was noted 1n comparison
with that found 1n loops run at
1500°F A whirlagig rig, patterned
after one developed by NACA, with
which fluoride flow velocities of
10 fps can be obtained, has been
placed 1n operation. Preliminary
testswith this apparatus with fluoride
fuels 1n Inconel show only a slight
1ncrease 1n corrosive attack compared
with that found 1n static capsule
tests. It was confirmed that ZrH, 1s
an effective corrosion inhibitor when
used with NaF-ZrF, -UF, (46-50-4 mole %),
TiH, will also inhibit corrosion, but
not quite so effectively as ZrH,.
It now appears that the solubilaty of
chromium in NaF-KF-LiF-UF, (10.9-43.5-
44.5-1 1 mole %) 1s about 3000 ppm,
which 1s not so high as originally
expected Considerable attack was
found 1n an Inconel loop 1n which the
fluoride had been saturated with
chromium metal prior to circulation.

The compatibility of BeO in Nak
continues to be examined, and a
correlation between the density of
the BeO and the corrosion behavior 1s
being studied Various surface
treatments for the BeO are being
evaluated. There 1s some evidence
that BeO 1s soluble 1n NaK to an
extent that 1s dependent on the
temperature of the system The
corrosion of two cermets was examlned
1n both fluorides and liquid metals.
Al though the tests 1n liquid metals
showed little or no attack, at least
one fluoride test resulted 1n signifi-
cant corrosion

The mass-transfer characteristics
of various metals in high-purity
lead are being studied by usaing

117
ANP PROJECT QUARTERLY PROGRESS REPORT

quartz thermal convection loops

The results show that both columbium
and molybdenum have very good re-
si1stance to mass transfer and corrosion
in liquid lead On the other hand,
severe corrosion and mass transfer
were found i1n lead-Inconel systems,
even thoughthe lead had been carefully
deoxi1d1zed Materials to be tested
in liquid lead include type 304, high
and low carbon, stainless steels,
type 410 stainless steel, and Armco
1iron. Tests of stainless steel and
Inconel 1n sodium-lead alloys continue
to show that additions of sodium to
lead decrease the severe corrosive
action.

The effect of a hydrogen atmosphere
in minimizing mass transfer of nickel
1n sodium hydroxide was demonstrated
by using a 50-50 mixture of helium and

hydrogen However, the use of atmos-
pheres of CO,, forming gas, wet
hydrogen, or dry air under a hydrogen

pressure of one-half atmospheric
pressure resulted 1n considerable mass
transfer.

A number of simple and complex
fluorides of the structural metal has
been prepared for studies of thear
effect on corrosion by fluorade
mixtures The complex fluorides
formed 1n corrosion tests are 1dentified
by either petrographic or x-ray-
diffraction techniques,or both The
oxidation of molten NaUF, yielded no
uo,r,, UO,, or UF,. However, 1t 1s
evident from the resulting products
that the container plays an important
part in the mechanism of air oxidation
of uranium 1n the molten fluorides.

FLUORIDE CORROSION OF METALS IN STATIC,
SEESAW, AND ROTATING TESTS

D C Vreeland J E. Pope
E. E Hoffman L R. Trotter
Metal lurgy Division

F Kertesz C R. Croft
H. J. Buttram R. E. Meadows
Materials Chemistry Division

Crevice Corrosion. In connection
with the crevice corrosion tests

118

being run, 1t was suggested that a
more realistic method of testing
would employ a tapered or V-shaped
crevice 1n the bottom of the testing
tube Accordingly, a ji1g was designed
for putting tapered crimps 1n test
tubing With this jig, crevices can
be made that are approximately 1 1/4
in long and vary 1in width up to
approximately 0 3 inch. Several
static tests with various materials
for containing the NaF-ZrF,-UF,
(46-50-4 mole %) have been rumn to
test the crevices Corrosion 1n the
crevices seemedto be somewhat errataic,
with some surfaces being apparently
unattacked and others havaing the
usual subsurface voids However,
accelerated corrosionain these crevices
was not noted, the attack that was
found 1n the crevices was not beyond
what might be expected for the materials
tested.

High-Temperature Pretreatment of
Inconel. On the basis of results
from some previous tests with NaF-KF-
LiF-UF, (10 9-43 5-44 5-1 1 mole %)
at high temperatures (1200 to 1300°C),
1t was reported that there was no
surface layer attack on Inconel.
It was thought that perhaps a pre-
treatment at high temperature would
render Inconel i1mmune from corrosion
at the usual temperature of 816°C.
Accordingly, two Inconel tubes were
loaded with the fluoride, heated for
4 hr at 1250°C, and then heated for
250 hr at 816°C. Light attack was
noted 1n each tube 1n the form of
subsurface voids that extended to a
depth of 1 mil in one tube and 0.5 m1l
in the other tube.

Effect of Time of Exposure of
Inconel to Fluoride Mixture with and
Without ZrH, Added. When the ARE
fuel mixture NaF-ZrF,-UF, (50-46-4
mole %), without added material, 1s
tested 1n Inconel under an atmosphere
of helium, the severity of attack and
the depth of penetration seem to
1ncrease very slightly with exposure
times of between 100 and 250 hours,
Differences between corrosion observed
after 25 and 100 hr seem to be real.
A plot of the structural metal content
of the fluoride mixture vs. exposure
time 1s shown 1in Fig. 10.1. It 1s
obvious that the chromium content
shows a rapid i1nitial rise and then
tends to level off at a high value,
whereas the 1ron content falls rapidly
at first and then very slowly. It
seems certain that reactions such as

FeF; + Cr —— CrF; + Fe,

with consequent deposition of the
metallic 1ron on the walls, are
responsible fora part of the corrosion
observed.

When this fuel mixture, with
0 5 wt % ZrH, added, was tested 1n
Inconel, the corrosion obtained was
considerably less severe, but the
corrosion changes with exposure time

PERIOD ENDING MARCH 10, 1953

were similar., Figure 10 2 shows the
results of analysis of the fuel
mixture for the structural metals.
In this case, FeF; and N1F, contents
dropped to very low values because of
the strong reducing agent, and,
apparently, most of the CrF, was
also reduced. However, with increased
exposure time, the CrF; concentration
increased slightly.

These results are encouraging 1n
that they indicate that the relatively
large corrosion rates 1nitially
observed are drastically reduced at
longer exposure times

Structural Metal Fluoride Additives.
Chemical analyses of fluoride materials
before and after corrosion testing 1in
Inconel have repeatedly demonstrated
that the 1ron and nickel contents of
the mixture are reduced and the

D
1600 WG 18656
CHROMIUM —
/#
1400
//

T 1200
Q
b
.—
&
W 1000
=z
(o]
o
|
2 800
LJ
=
-
<
£ 600
...,.
Q
0
x
7

400

200 |

IRON
NICKEL
0
0 100 200 300 400 500
EXPOSURE TIME (hr)
Fig 10.1 Structural Metal Content of NaF-ZrF4-UF4 (50-46-4 mole %) As a

Function of Exposure Time.

119
ANP PROJECT QUARTERLY PROGRESS REPORT

1000

-
DWG 18657

STRUCTURAL METAL CONTENT {ppm)

M~ - ——--IRON _ |
CHROMIUM

0 NICKEL

0 100 200

300 400 500

EXPOSURE TIME ({hr)

Fig. 10 2.

The

- e
chromium content 1s 1ncreased.
reactlons

3FeF, + 2r —— 2CrF, + 3Fe
and
3N1F2 + 2Cr ——> 2CrF,; + 3Na

are obviously involvedin the corrosion
mechanism, Since both 1ron and
nickel fluorides are likely to be
present to some extent i1n the final
ARE fuel and since considerable
amounts of these compounds have been
present 1n some of the batches tested,
a serles of experiments has been
conducted to demonstrate the effects
of the fluorides of 1ron, chromium,
and nickel i1n considerable amounts on
the corrosion of Inconel by the ARE
fuel and the ARE fuel solvent.
Addition of either FeF, or NaF,
1increases the severity of attack and
depth of penetration. The addition
of CrF, apparently has no effect on
the corrosion rate. When 1ron or
chromium fluorides are added, metallo-
graphic examination reveals metallic
deposits on the capsule walls,
Chemical analyses of the solidified
melts, after exposure, reveal that the

120

Structural Metal Content NaF-ZrF,-UF, (50-46-4 mole %), with 0 35
wt % Zrflz,As a Function of Exposure Time.

addition of CrF; does not appreciably
alter the amounts of 1ron and nickel
normally found i1n the mixture. The
addition of NiF,, however, greatly
increases the amount of CrF; found an
the melt, the very large quantities
of soluble nickel added are nearly
completely reduced and are apparently
deposited on the walls of the capsule
so that they are not removed with the
fluoride mass. The addition of FeF,
also serves to raise the chromium
content of the fluoride mixture, but
only a part of the FeF, seems to be
reduced to metal.

The handlaing, samplaing, and ana
lytical techniques are not sufficiently
preclise to permit obtaining accurate
material balances from the equations
shown above, i1n general, however, the
balance 1s within 120%, from which it
appears that a considerable fraction
of the corrosion 1s, as was expected,
due to other actions.

Hydrogen Fluoride Additive. Hydro-
fluoric acid 1s one of the reagents
used 1n the process for purifying
the fluoride melts. This reagent
may be present in the melts at low
concentrations, even though con-
si1derable careis used i1n the stripping
operation.

Study of corrosion of Inconel by
a batch of ARE fuel, NaF-ZrF4-UF4
(50-46-4 mole %), to which HF had
been added as NaHF,, showed that
heavy subsurface void formation
resulted when as much as 0.3 wt % HF
was present. This attack was shown
to be as much as 3 mils deep, whach
1s at least twice the depth expected
without addition. However, the use
of 0.1 wt % HF resulted i1n less severe
attack, with penetration to only
15 mls, and the addition of 0 04 wt %
HF yielded a light metallic deposit
and relatively light corrosive attack.

The fuel mixture used 1n these
experiments had been treated waith H,
and, presumably, contained small
amounts of UF;. It 1s possible that
the reaction

2UF, + 2HF =— 2UF, t H,

may have masked the deleterious
effect of small HF additions.

Fluoride Corrosion i1n a Rotating
Rig. The whairligig ri1g¢!’ 1s now
running and operational ‘‘bugs” are
gradually being worked out In thais
device 1t will be possible to run
relatively high-velocity corrosion
tests with the fluoride mixtures
contained 1n Inconel and other suitable
tubing. At the present time, a fluaid
velocity of approximately 10 fps 1s
being employed. Upon metallographac
examination after a 100-hr test waith
NaF-ZrF,-UF, (46-50-4 mole %) 1n
Inconel at a temperature of 816°C
(no appreciable thermal gradient
employed), 1t was noted that the
attack was no more severe than that
often encountered 1n ordinary static
corrosion tests, The attack was
quite uniform and was in the form of
subsurface voids to a depth of 2 mils.

(I)L G. Desmon and D R Mosher, Preliminary
Study of Circulation i1n an Apparatus Suitable for
Determining Corrosive Effects of Hot Flowing
Liquids, NACA-BM-E-51D12 (June 29, 1951).

PERIOD ENDING MARCH 10, 1953

Chemical analysis of the fluoride
after the test revealed the presence
of 0.025% Fe, 0.003% N1, and 0.141% Cr.
Other tests are being run with various
additions to the fluoride mixtures,
and attempts are being made to vary
the heating arrangements on the tubing
so that sizeable temperature gradients
can be attained.

Static Tests on Incoloy and Inconel
1n Fluorides. Incoloy (32 % N1-21%
Cr-47% Fe) was tested in NaF-ZrF,-UF,
(46-50-4 mole %) for 100 hr at 816°C.
In a few places, subsurface voids to
a depth of 2 mils could be noted. A
photomicrograph of the Incoloy,
taken after testing, 1s shown 1in
Fig. 10.3.

The descaling properties of the
fluoride on oxidized Inconel were
tested at 816°C, and 1t was found that
the oxidized Inconel was cleaned very
well at the 1-, 4-, and 8-hr test
times that were tried.

FLUORIDE CORROSION OF METALS IN
THERMAL CONVECTION LOOPS

G. M. Adamson, Metallurgy Division

Effect of Fluoride Batch Size.
As discussed 1n the previous report, ( ?)
the corrosion has been i1ncreasing 1in
the Inconel loops circulating NaF-
ZrF,-UF, (46-50-4 mole %). The
increases 1n themselves were not
enough to cause trouble, but they are
dangerous when regarded as a trend
that 1s not understood. Consequently,
a series of loops was run to determine
whether the 1ncrease in batch size 1n
the production of the fluoride mixture
was responsible for the corrosion
increase. Loops were run with fluoride
mixtures prepared in 5- and 50-1b
batches for 100- and 500-hr periods.
In both the 100- and 500-hr tests,
more corrosion was caused by the small
batches than by the large batches.
Obviously, the i1ncrease 1in batch

s1ze was not the reason for the

(Z)G M Adamson, ANP Quar Prog Rep Dec 10,
1952, OBNL-1439, p 133 ff

121
ANP PROJECT QUARTERLY PROGRESS REPORT

Fig. 10.3.
hr at 816°C.

increase in attack. Furthermore,
even though extreme care was taken
during filling of the loop, the
attack was still extensive; therefore
poor handling during filling was
also eliminated as the cause.

Fluoride Pretreatment. Two small
batches of NaF-ZrF,-UF, (46-50-4
mole %) were taken from the 50-1b
batch and given a complete retreat-
ment.¢?) When these batches were
circulated in Inconel loops, the
attack in one instance was considerably
greater than average, whereas in the
other it was less; therefore it may be
concluded that some major variable
still has not been controlled.

G)c, M, Blood, A. J. Weinberger, F. P. Boody,

and G. J. Nessle, ANP Quar. g Rep. June 10,
1952, ORNL-1294, p. 97.

122

v

Static Test of Incoloy in NaF-ZrF,-UF, (46-50-4 mole %) for 100

The attack found in Inconel loops
in which NaF-ZrF,-UF, (46-50-4 mole %)
circulated was much less than had
been found when NaF-KF-LiF-UF, (10.9-
43.5-44.5-1.1mole %) was the circulated
material.(*) Since the NaF-ZrF,-UF,
had received more extensive pre-
treatment with H, and HF, its higher
purity is a logical explanation for
its lower corrosiveness. Accordingly,
two special batches of the NaF-KF-LiF-
UF, mixture were given the same
special pretreatment. These batches
produced attack on Inconel that was
almost twice that heretofore observed.
Two additional batches of the same
fluoride were then specially prepared
and circulated in Inconel loops, and

(4)G, W, Adumson, ANP Quar. Prog. Rep. Sept. 10,
1952, ORNL-1375, p. 103 ff.

the attack was even worse than 1in
the first test,

In all the tests with pretreated
fuel, the operator reported a strong
odor of hydrogen fluoride during the
filling of the loops. Since hydrogen
fluoride, as an additive 1in static
tests, has been shown to 1ncrease
corrosion, 1t 1s most probable that
the HF used at one step 1n the fuel
pretreatment, as well as the NaiF
produced during this treatment, 1is
not being completely removed and 1s
causing the increased corrosion.
These tests show that poor purifi-
cation may do more harm than good.

Hydride Additives It has been
confirmed that zirconium hydride wall
reduce the attack of NaF-ZrF, -UF,
(46-50-4 mole %) on Inconel. An
addition of 1/2% zirconium hydrade
1n one thermal convection loop reduced
the depth of attack to between 1 and
4 mils. This attack 1s lower than
normal but 1s deeper than that found
when zirconium hydride 1s added to
NaF-KF-LiF-UF, (10.9-43.5-44.5-1.1
mole %). The coolant underwent
considerable transformation, as
evidenced by the presence of a brown
material.

Titanium hydride (1/2%) was added
to the NaF-KF-LiF-UF, mixture and
circulated 1n Inconel. The maximum
penetration of 2 5 mils, although
much less than that found when no
additive was used, was not quite so
low as that obtained when zirconium
hydride was added. Surface layers
were found 1n both the hot and cold
legs These layers must receive
additional study 1f any hydride 1s
to be used as an inhibitor.

Metal Additives. The three most
common 1mpurities found in the fuel
are nickel, 1ron, and chromium. One
loop 1n which chromium metal powder
had been added to NaF-KF-LiF-UF,
(10.9-43.5-44.5-1.1 mole %) was
discussed 1n the previous report.(?)
This test was repeated, and again no
change 1n the average depth of attack

PERIOD ENDING MARCH 10, 1953

was found, no unusual layers were
found, and the chromium content of the
fluorides from the loop was normal,
so again the chromium metal did not go
into solution. Chromium metal powder,
in sufficient quantity to make 0.5%
in the solution, was then added to a
transfer pot before the fluorides
were placed anto 1t. After the
fluorides had been added, the pot was
kept hot and helium was blown through
1t for 30 minutes. Again, average
attack was found after this materaial
had been circulated in an Inconel
loop. The chemical analysis of thas
fluoride showed a chromium concen-
tration of 2500 ppm, which 1s about
twice the normal concentration,
Further work 1s scheduled to determine
whether the solubility of chromium 1is
much lower than-was expected. This 1s
likely, since the chromium content
did not 1ncrease greatly during
circulation and yet attack had taken
place

As another approachto this problem,
a section of Inconel pipe was chromium
plated and welded 1nto the hot leg
of a loop. After the NaF-KF-LaF-UF,
mixture had been circulated, the depth
of attack was the same 1n adjacent
plated and unplated sections. No
traces of the chromium plate remained
on the walls. A metallic appearing
layer was found in the cold leg. The
chromium content of the fluorides
varied from 2500 to 3000 ppm, whaich,
although higher than normal, 1s about
the same as that found in the loop
discussed above.

Temperature Dependence. A loop
filled with the fluoride fuel, NaF-
ZrF,-UF, (50-46-4 mole %), was operated
for 500 hr with a minimum cold-leg
temperature of 1500°F and a hot-leg
temperature of 1650°F., The hot-leg
attack was moderate to light, with a
maximum depth of 11 mils and an
average depth of 8 mils. A concen-
trationof voids in the grain boundaries
had taken place. The results may be
compared with those of a similar

123
ANP PROJECT QUARTERLY PROGRESS REPORT

loop test in which the hot-leg temper-
ature was 1500°F The hot-leg attack
1n this loop was from 3 to 10 mails
deep. Comparison of these two loops
indicates that the corrosion mechanism
1snot extremely temperature sensitive.
One disturbing fact 1s that with the
hotter loop there was some evidence of
uranium segregation in the cold leg.
This 1s now being checked in another
loop.

Crevice Corrosion. Another loop
operated 1n the study of crevice
corrosion¢?) was an Inconel loop,
with two crevices built i1nto the hot
leg, 1n which NaF-ZrF,-UF, (46-50-4
mole %) was circulated. The upper
hot-leg section showed a maximum
attack of 16 mils and an average
attack of 12 mi1ls, whereas the attack
in the lower hot-leg section had a
maximum depth of 9 mils. The maximum
attack i1n the crevices was 19 mails
and the average was 12 mils. The
intensity of attack 1n the crevice
was several times that in the straight
section. This confirms previously
obtained results i1n that the attack
1n the crevice 1s only slaightly
deeper but the i1ntensity of attack
1s greatly increased.

Inserted Corrosion Samples. Me-
tallographic data received for the
previously discussed(?) loops 1n which
thermocouple tubes were i1nserted into
the center of the hot and the cold
legs confirm the results obtained
with flat samples. With both NaF-
ZrF,-UF, and NaF-KF-LiF-.UF,, the
attack on the loop walls i1n the hot
leg was normal, but the hot-leg
inserts were attacked to a depth of
onlyl mil, no satisfactory explanation
has been offered for this phenomenon.

Effect of Exposure Time. An
Inconel loop was used to circulate
NaF-ZrF,-UF, (50-46-4 mole %) 1000 hr,
and the resulting hot-leg attack was
moderate to heavy in 1ntensity and
extended to depths of 5 to 11 mils.
The attack was present both as general
attack and as a concentration of the

124

voids 1n the grain boundaries For
comparative purposes, these data are
summarized i1n Table 10 1, along with
simi1lar data for loops previously
operated for 100 and 500 hours
Although the depth and intensity of
the attack i1ncreases with the exposure
time, the rate of attack definitely
decreases

TABLE 10.1. EFFECT OF EXPOSURE TIME
ON DYNAMIC CORROSION OF INCONEL BY

NaF-ZrF4-UF4
TIME OF
OPERATION | MAXIMUM ATTACK | cpposyoN INTENSITY

(mils)

(hr)

100 4 Light to moderate

500 10 Light to moderate

1000 11 Moderate to heavy

Nonuranium Bearing Mixtures. Since
NaF-ZrF, (50-50 mole %) will be used
during the 1nitial testing of the
ARE, several loops have been operated
to study 1ts effects These are
standard loops, that 1s, Inconel
loops operated for 500 hr with a
hot-leg temperature of 1500°F. The
upper section of the hot leg of one
loop showed light attack that extended
to depths of 4 to 8 mils,
the lower section was practically
unattacked The second loop 1s still
operating. A third loop, 1n which
NaF-ZrF, (52-48 mole %) was circulated,
ylielded scattered i1ntergranular attack
of from 3 to 8 mils 1n the top section
of the hot leg. The lower section
showed general pitting 1 mil deep,
with an occasional patch of pits up
to 6 mils deep No deposit was found
in the cold leg

Zirconium hydride (1/2%) was added
to another batch of this coolant,
which was then subject to the standard
loop test The hot-leg surface of
this loop was pitted, but the attack,
as shown by the subsurface voids,
extended to a maximum depth of only

whereas
1.5 mils. A two-phase layer 1 mil
thick was present on the hot-leg
surface, and a tightly adhering
metallic-appearing layer 0.1 mil
thick was found in the cold leg.
Again the zirconium hydride acted as
an inhibitor, but hot-leg deposits
were formed.

One Inconel loop circulated 500 hr
with NaF-BeF, (57-43 mole %). The
maximum hot-leg attack was 9 mils and
the average was 4 mils. More voids
were concentrated in the grain bounda-
ries than are normally found, and the
voids were larger. A nonmetallic
deposit was found in the cold leg.
With both NaF-ZrF, and NaF-BeF,, the
attack mechanism seems to be the same
as for the other fluorides.

10.4.
100 hr at 850°C.

Fig.
100x.

PERIOD ENDING MARCH 10, 1953

LIQUID METAL CORROSION OF STRUCTURAL
METALS

E. E. Hoffman

J. V. Cathcart L. R. Trotter

D. C. Vreeland J. E. Pope
Metallurgy Division

G. P. Smith

Seesaw Tests of Sodium-Lead Alloys.
A series of seesaw tests of sodium-
lead mixtures has been run in Inconel
and types 430 and 304 stainless steel
tubing for 100 hours. The lead for
these tests was purified by bubbling
hydrogen through molten lead at 750°C
for 1 1/2 hours. All Inconel tubes
were attackedin the hot zone. Severity
of the attack decreased with increasing
sodium additions, but the attack was
still appreciable with a 30 wt %
sodium addition, as shown in Fig. 10.4,

UNCLASSIFIED
Y-7804
[ 5% =y N,

<
S

Inconel After a Seesaw Test ina30 wt % Na-70 wt % Pb Mixture for

125
ANP PROJECT QUARTERLY PROGRESS REPORT

which illustrates the type and extent
ofattack of these sodium-lead mixtures.
In no case was attack or deposition of
crystals noted in the cold zone upon
metallographic examination; however,
in several of the tubes, crystals
could be detected in the bath material
In the test with the type 430 stain-
less steel tubing, there was very
little corrosive attack by the sodium-
lead mixtures. Type 304 stainless
steel showed exceptionally good
resistance to these corrodants.
A section of the hot zone of a tube
after exposure to 5% Na-95% Pb is
shown in Fig. 10.5. There was es-
sentially no attack in any of the
tests with type 304 stainless steel.
A summary of these tests is given in

Table 10.2.

Fig.
Pb Mixture for 100 hr at 830°C

100X.

126

Static Tests with Lithium. Two-
component static tests of several
materials have been run with lithium
at 1000°C to furnish a basis of
comparison with three-component tests
run previously. Severe attack had
been noted in the three-component
tests, 10 to 15 mils penetration
being quite common. It should be
emphasized, however, that the more
recent tests were static tests and
the attack might be less severe than
in the dynamic systems. Results of
these tests are presentedin Table 10.3.

Incoloy in Sodium, Lithium, and
Lead. Incoloy (32% Ni-21% Cr-47% Fe)
was tested for 100 hr at 816°C in
sodium, lithium, and lead. No attack
was noted in either sodium or lithium,
although some crystal deposition was

UNCLASSIFIED
Y-8144

10.5. Type 304 Stainless Steel After a Seesaw Test in a 5 wt % Na-95 wt %
TABLE 10, 2.

PERIOD ENDING MARCH 10, 1953

SEESAW TESTS OF VARIOUS MATERIALS RUN IN SODIUM-LEAD
MIXTURES FOR 100 HOURS

BATH COMPOSITION AVERAGE ¥FHPERAJUBE
MATERIAL (%) (¢ METALLOGRAPHIC NOTES
Na Pb Cold Zone Het Zone
Inconel S 95 530 §20 Terminated after 18 hr, no cairculation,
12 mils of intergranular attack in het
zone
10 90 330 870 Intergranular attack entirely through
35-m1l wall 2n hot zone
15 85 315 860 15 m1ls of heavy intergranular attack
in hot zone
20 80 395 855 12 mi1ls of intergranulsr attack in hot
tone
25 75 470 850 S mils of intergranular attack in hot
zone
30 70 485 850 6 mils of intergranular attack 1n het’
zone
Type 304
stainless steel 5 95 560 830 Essentially no attack 1n either hot or
cold zonme i1n any of the type 304 stain-
lesas steel tests
10 90 575 840
15 85 545 830
20 80 585 825
25 75 600 835
30 70 600 835
Type 430
stainless ateel 5 95 565 850 Essentially ne attack in either hot or
cold zone 1n any of the type 430 gtaine
less steel tests
10 90 580 845
15 85 575 820
20 80 575 810
25 75 550 805
30 70 610 835

127
ANP PROJECT QUARTERLY PROGRESS REPORT

TABLE 160.3.

RESULTS OF TWO-COMPONENT STATIC TESTS WITH LITHIUM AT 1000°C

MATERIAL TIME(EE TEST METALLOGRAPHIC NOTES

Type 316 stainless steel 100 No si1gn of attack on sections exposed to liquid
phase, 2 5 mils of intergranular attack on
sections exposed to vapor phase

Inconel 100 Some mass-transfer crystals could be noted on
the surface, no attack in vapor zone, 3 to 4
mi1ls of subsurface voids on tube i1n bath zone

Type 304 stainless steel 400 l to 2 mils of intergranular penetration and
transformation from austenite to ferrite

apparent 1n the test with lithium.
In the test with lead, 1 to 2 mils of
intergranular penetration could be
noted.

Corrosion by Lead 1n Thermal
Convection Loops. A series of tests
has been made to determine the extent
of mass transfer and corrosion 1in
thermal convection loops containing
liquid lead. The systems lead-Inconel,
lead-niobium, and lead-molybdenum
have been studied. Results are also
avalilable for the lead-type 304
stainless steel system, but are
incomplete.

Since the results of previous
investigations of mass transfer and
corrosion 1n lead thermal convection
loops have been subject to criticism
because of oxide i1n the lead or on
the metal tubing, 1n the present
series of experiments, the liquid
lead was deoxidized with hydrogen
prior to contact with the test metal.
Care was also taken to avoid heavy
oxidation of the metal sections of
the loops. The apparatus and experi-
mental procedure were described in
the previous report. (5)

The quartz tubing used in the
loop 1s virtually i1nsoluble 1n lead
at the test temperatures, the concen-
tration of 3510, 1n the lead after one

(5)ANP Quar Prog Rep Dec 10, 1952, ORNL-
1439, p 148,

128

test was only 17 ppm, and the S10,
concentration 1n a lead sample that
had not been 1n contact with glass or
quartz was 15 ppm

Loop tests were run with Inconel,
columbium, molybdenum and type 304
stainless steel samples suspended 1n
the circulating lead The results for
Inconel have been reported previ-
ously, (%) the results for the other
metals are summarized i1n Table 10 4.

The columbium specimens exhibited
no mass transfer and virtually no
corrosion even after almost 600 hr
of loop operation. The small decrease
in wall thickness was probably due to
the solution of the small amount of
columbium required to saturate the
lead. No mass transfer occurred with
molybdenum, but the corrosion, although
slight, was a little greater than
that found for columbium. Little or
no intergranular penetration was
observed. The reason for the dis-
crepancy 1n the wall-thickness measure-
ments for the second lead-molybdenum
test 1s not known.

The molybdenum specimens used were
not actually sections of tubing but
were made of molybdenum foil bent into
a cylinder. The foil had a specified
thicknessof 10 mils, which was checked
by micrometer readlngs at two points
on the foi1l. It 1s possible that there
was a varlation 1n thickness 1n the
PERIOD ENDING MARCH 10,

1953

TABLE 10.4. CORROSION OF STRUCTURAL METALS BY LEAD IN QUARTZ THERMAL
CONVECTION LOOPS
TEST TEMPERATURE (°C) SPECIMEN WALL THICKNESS (in.) TIME OF
METAL NO OPERATION
* | Hot Zone | Cold Zomne.| Original | Hot Zone | Cold Zone (hr)
Columbium 1 800 575 0.035 | 0,032 0.034 270( )
2 800 565 0.035 0.033 0.033 572(%)
Molybdenum 1 800 450 0.010 | 0.006¢¢) | o0.006(¢) 473(®)
2 800 595 0.010 | 0.011 0 010 305(8)
Type 304, high carbon,
stainless steel(?) 1 800 500 111
(a)Stoppage caused by failure of hot leg heater, no plug formed
(b)Scheduled termination
(C)D1screpancy probabl y caused by variation 1n tube thickness.
(d)Examxnat1on incomplete
fo1l and that the original foi1l used CORROSION OF CERAMICS BY FLUORIDES

1n test No. 2 was slightly greater
than 10 mils thaick.

The results of the test with
type 304 stainless steel are i1ncomplete.
On the basis of the thermal record of
the experiment, however, 1t appeared
that circulation 1n this loop was also
stopped by plug formation. The
type 304 stainless steel specimen
used had a carbon concentration of
0.087%.

In summary, both columbium and
molybdenum showed very good resistance
to mass transfer and corrosion in
liquaid lead. On the other hand,
severe corrosion and mass transfer
occurred 1n the lead-Inconel systems
despite the careful deoxidation of
the lead with hydrogen. The tests
have i1ncluded a nickel-base alloy
(Inconel), columbium, molybdenum, and
an iron-base alloy (type 304 stainless
steel). It 1s planned to extend the
study to other iron-base alloys, such
as type 410 stainless steel, to a
low-carbon type 304 stainless steel,
and to Armco 1iron.

AND LIQUID METALS

D. C. Vreeland L. R. Trotter
E E Hoffman J E. Pope
Metallurgy Division

E. E. Ketchen L. G. Overholser

Materials Chemistry Division

L. A. Mann J. M. Cisar
D. R. Ward
ANP Division

Cermets 1n Fluorides and Liquid
Metals. Two cermets were prepared by
the Ceramics lLaboratory, one consisted
of TiC plus 20% N1 and 15% CbC plus
TaC, and the other consisted of ZrC
plus 20% 1ron. These cermets were
tested 1n various corrosive mediums
by sealing them,together with the
corroding agent, under vacuum 1n
Inconel tubes. After the tubes had
been heated for 100 hr at 816°C,
specimens were removed, checked for
weight and dimensional changes, and
examined metallographically. 1In the
tests with sodium and lead, there was
little or no attack (Fig. 10 6). The

129
ANP PROJECT QUARTERLY PROGRESS REPORT

UNCLASSIFIED
PE Y8209

ol

Vit §

Fig. 10.7. Titanium Carbide Cermet After Exposure to NaF-ZrF,-UF, (46-30-4
mole %) for 100 hr at 816°C.

TABLE 10.5. RESULTS OF STATIC TESTS OF TiC AND ZrC CERMETS IN SODIUM,
LEAD, AND SOME FLUORIDE MIXTURES

CERMET CORRODANT b i METALLOGRAPHIC NOTES
TiC | Na -0.0065 No attack apparent
Pb Pb adhering
to specimen No attack apparent
NaF-KF-Li F-UFy Fluoride No attack apparent
(10.9743.5-44.5-1.1 mole %) | adhering to
specimen
NaF-ZcFy-UF, -0.0119 Bonding material appeared to be
(46-5024 mdle %) leached from specimen to a depth
of 5 mils
zC | Na -0.0037 Specimen showed that some uniform
solution had taken place, but no
thickness change was involved
NaF-KF-Li F-UFy 0.0158 Surface appeared to be slightly
(10.9°43.5-44.5-1.1 mole %) roughened

130

Fig.

mixture NaF-KF-LiF-UF, (10.9-43.5-
44.5-1.1 mole %) appeared to have
little effect on these materials, but
in the one test with NaF-ZrF,-UF,
(46-50-4 mole %), the specimen was
attacked to adepth of 5 mils (Fig. 10.7).
No significant dimensional changes
could be noted in any of the tests.
The weight loss data and the results
of metallographic examination are
given in Table 10.S5.

Static Tests of the ARE-Be0 Blocks
in Na and NaK. Two tests have been
performed with ARE beryllium oxide
moderator blocks heated for several
hundred hours at 810°C in a type 347
stainless steel pot containing Na or
NaK. In the first test, the beryllium
oxide was in 255

immersed in.? of

PERIOD ENDING MARCH 10, 1953
L3 S UNCLASSIFIED
.‘ . Y-8210
8
x
2
005
%
ol
3|

10.6. Zirconium Carbide Cermet After Exposure to Sodium for 100 hr at 816°C.

sodium for 213 hr at 816°C. Although
no crackingor spalling of the beryllium
oxide was noticeable immediately
after the test, the surface im-
perfections apparent in part a of
Fig. 10.8 occurred after the test
specimen had been soaked in alcohol
and water.

In the second test, the NaK was
transferred to the pot with the
beryllium oxide when both the NaK and
the beryllium oxide were at room
temperature. The test was run for
500 hr at 816°C with the beryllium
oxide immersed in 255 in.® of NaK.
After the test, the beryllium oxide
was badly cracked, as can be seen
in part b of Fig. 10.8. The weight
and dimensional changes noted for
both tests are given in Table 10.6.

131
ANP PROJECT QUARTERLY PROGRESS REPORT

Fig.
816°cC.

10.8.
(a) In Na for 213 hr;

Seesaw Tests of Be0 in NaK. Be-
ryllium oxide specimens 1/4by 1/4 by 1
in, were tested in NaK in the rocker
furnace for 100 hr, with a hot-zone
temperature of approximately 830°C
and a cold-zone temperature of ap-
proximately 450°C. Specimens were
crimped into the hot end of the tube
to prevent their sliding from the
hot end to the cold end. Specimens
indicated in Table 10.7 as‘‘low density”
had a density of approximately 2.27;
those indicated as “high density” had
a density of approximately 2.83.

132

UNCLASSIFIED
Y-8297

ARE Beryllium Oxide Blocks After Static Tests in Na and NakK at
(b) in NaK for 500 hours.

Specimens indicated as ‘“dehydrated”
were dried for 24 hr at 125°C before

test. The information derived from
these experiments is tabulated in
Table 10.7. It is evident from these

data that after the specimens were
stripped in alcohol there was some
slight evidence of cracking in the
high-density specimens and some slight
evidenceof spalling in the low-density
specimens.

Convection Loop testsof Be0 in NakK.
Eight specimens of BeO 1/4 by 1/4 by
1/2 in. were placed in a wire basket

1n the surge tank of a thermal con-
vection loop, and NaK was permitted
to flow over the specimens at a
velocity of about 8 fpm for 212 hours.
The temperature of the NaK was 1150°F
in the cold leg and 1000°F in the
hot leg, and the samples were 1n
contact with NaK at 1500°F. The NaK
specimens 1n the second test were
pretreated before exposure to the
hot NaK i1n the following manner-
specimens 1 and 2 received no special
pretreatment, specimens 3 and 4 were
treated with MgNO,, specimens 5 and 6
were 1mpregnated with AlF,; and CaF,,
specimens { and 8 were treated with
BeNO,. The test results are given 1in
Table 10. 8.

TABLE 10.6.

PERIOD ENDING MARCH 10, 1953

It 1s noteworthy that specimens 1,
3, 5, and 7 were located above speci-
mens 2, 4, 6, and 8, respectaively, 1in
the basket during exposure to the Nak,
and 1n each case the upper specimen
suf fered the greater loss 1n weaght.
Special treatment of BeO to increase
1ts resistance to hot NaK appears to
offer possibilaities, and further
tests are to be conducted. The effect
of the density of the specimen could
not be correlated to the weight losses
1n these tests,

Solubility of Be0 1n NaK. Various
workers have 1nitiated experiments
to determine, 1f possible, the effect
of NaK on BeO at temperatures of
approximately 800°C 1n both static and

WEIGHT AND DIMENSIONAL CHANGES OF ARE-Be0 BLOCKS AFTER

STATIC TESTS IN Na AND NaK AT 816°C

BEFORE AFTER CHANGE AFTER CLEANING IN CHANGE
TEST TEST (%) WATER AND DRYING (%)
First Test, 213 hr ain Na
Length, in 5914 5917 +0 05
Width, ain 3 706 3 708 +0 05
Weight, g 3 154 3 221 +2 1 3 205 +1 6
Second Test, 500 hr in NaK
Length, 1in 5 909 5 899 -0 17
Width, ain 3714 3712 -0 05
Weight, g 3 160 3 205 +1 4 3 187 +0 85

TABLE 10.7. RESULTS OF SEESAW TESTS OF Be0 IN NaK AFTER 100 hr AT 816 °C

WEIGHT WEIGHT | TOTAL WEIGHT TOTAL BeO FOUND BY ANALYSIS

SPECIMEN LOSS LOSS LOSS OF NaK FILTER AND CONTAINERS
(g/an.?) (%) (ng) (mg)
Low densaty, as received* 0 0454 2,38 30.7 14.26
High density, as received 0.0286 1 25 18 9 4,97
Low densaty, dehydrated 0.0893 5.15 47.0 9.91
High densaty, dehydrated 0 0104 0.52 6.1 6.02

*This test run for only 16 hours.

133
ANP PROJECT QUARTERLY PROGRESS REPORT

dynamic test systems. The weight
losses of the BeO blocks observed in
a number of 1nstances prompted a
search for the beryllium that had been
removed from the block. 1In those
tests for which the NaK was filtered
at room temperature, the beryllium
found 1n the filtrate corresponded to
approximately 3 ppm of Be(O, which was
a very small part of the total beryllium
removed from the blocks. Further
examination revealed that the be-
ryllium or the BeO, or both, was
located pramarily on the walls of the
containers to which 1t adhered tightly,
The question then arose as to‘'whether
or not the beryllium was removed from
the block and deposited on the wall
by some process involving solubilaty
of Be or BeO i1n the NaK at 800°C.

Two tests were performed to measure
the solubility i1n NaK at 800°C. 1In
the first test, BeO was contacted with
NaK at 800°C for 6 hr in the reaction-
filtrationrig previously described. ()
The NaK was filtered at approximately
800°C, and the filtrate was collected
in a nickel receiving tube. Based on
this single experiment, the solubility
of berylliumin NaKat 800°C corresponds
to roughly 100 ppm, but 1t 1s not
known whether the material dissolved

was Be or BeO.

(6)ANP Quar Prog BRep Dec 10,
1439, Fag 10 7, p 121

1952, ORNL-

TABLE 10.8.

In the second test, a stainless
steel thermal convection loop was
used to circulate NaK over BeO blocks
suspended in the surge tank After
212 hr of circulation at a flow rate
of 6 to 10 fpm, a weight loss of about
25% was measured for the BeO blocks

(original weight of 4 g). After the
NaK had been drained, the loop was
sectioned as shown in Fig. 10.9. The

operating temperatures of the loop
are given on the figure, along waith
the analytical results for the BeO
determinations on the respective
3 1/2-1n sections.

The analytical results show that
the three coldest (and lowest) sections
of the loop contained the greatest
concentrations of beryllium. Since
the analytical data showno accumulation
of BeO fragments i1n the coldest part
of the tank, 1t 1s suggested that a
dissolution process, governed by the
temperature gradient, 1s responsible
for the transfer of the material
from the BeO block to the walls of
the cold region.

Such a mechanism could be represented
by the following reaction

2NaK + 2Be0 == Na,0 + K,0 + 2Be ,

which, 1f the reaction proceeds
slightly to the right at 800°C and
the beryllium formed 1s soluble 1n
NaK, would account for removal of

RESULTS OF TESTS OF Be0 IN Nak

SPECIMEN NUMBER
1 2 3¢e) | 4le) 5 6 7 8
Average loss 1in dimensions
per surface, 1in 0 006 0 004 [ 0002 (0002 | O 000 | 0 001 | 0 005 |0 004
Loss of weight, g 0 152 0 120 | 0 091 |0 076 | O 069 | O 062 | 0 162 |0 119
Loss of weight (%) 11 § 9 6 8 6 0 49 47 11.1 9.6
Density of specimen 2 60 2 50 54 2 45 2 82 2 65 2.66 2 45

(a)
alcohol bath

134

Much or all of the weight loss may have been due to the shedding of a thin surface film 1n the
PERIOD ENDING MARCH 10, 1953

UNCLASSIFIED

DWG 48658
/\/\/\,PM
0] 20
22 WEIGHT OF BeQ FOUND IN VARIOUS SECTIONS
> AVERAGE
19 SECTION NO BeO (mq) TEMPERATURE (°F)
_1 - 1 113 1500
3 137 1360
3 5 552
18
- 6 1250
7 217
9 2208
4 1 10 248 4 1140
17 1" 70 8
13 5 76
] — 14 % 1860
5 15 364 1370
\ 17 414
— 6 18% 1440
19 474 1800
20 1480
24 672
22 27 48
¥ 14 and 18 LOCATED UNDER HEATER

Fi1g. 10 9. Schematic biragram of Convection Loop for NaK-Be0 Test Showing
Temperature and Be0 Recovered 1n various Sections.

135
ANP PROJECT QUARTERLY PROGRESS REPORT

beryllium from the BeO block. If the
solubi1lity of beryllium 1s less at
600°C than at 800°C, deposition or
alloyingin the cold area would occur.

EFFECT OF ATMOSPHERE ON THE MASS
TRANSFER OF NICKEL IN HYDROXIDES

H. J. Buttram C. R Croft
F Kertesz
Materials Chemistry Division

The 1inclined-tube technique previ-
ously described¢’’ was used for further
study of the effect of various expera-
mental factors on the mass transfer of
nickel by hydroxides. With a tempera-
ture gradient of 150°C (800°C at the
bottom and 650°C at the top) under
helium atmosphere, sodium hydroxide
transported enough nickel to the
liquid level of the inclined nickel
tube to plugit completely i1n 48 hours.
As was previously mentioned, hydrogen
gas sweeping over the surface of the
molten hydroxide reduced the mass
transfer considerably, only polishing
of the hot end and roughening of the
cold endof the tube could be observed.
The temperature gradient was then
increased to 300°C (500°C at the top
and 800°C at the bottom), and a
hydrogen atmosphere was used. Similar
satisfactory®results were obtained,
the hydrogen largely suppressed any
crystal deposit at the liquid level.

Attempts were made to determaine
the minimum pressure at which mass
transfer 1s effectaively i1nhaibited by
hydrogen. A 50 vol % helium-50 vol %
hydrogen mixture over the hydroxide
gave satisfactory results. On the
other hand, with an absolute hydrogen
pressure of 38 cm Hg - established as
a manually adjusted vacuum - the
hydrogen system affordedno protection,
possibly because of the technique
employed.

A number of other atmospheres was
also applied over sodium hydroxide 1in
the 1nclined tube test. The use of

(T)ANP Quer Prog Rep Dec 10,
1439, p 142 {f

1952, ORNL-

136

carbon monoxide and forming gas (10%
H, + 90% N,) did not have a beneficial
effect. Water in the hydrogen, even
1in small quantities, resulted in
considerable mass transfer, Bubbling
the hydrogen through water at various
temperatures resulted in nickel
crystal deposits 1n amounts similar
to those obtained under a helium
atmosphere. A continuous vacuum over
the system (which would remove any
hydrogen formed during the test)
resulted 1n extremely heavy corrosion
and mass transfer, and left the
hydroxide with a nickel concentration
of up to 3% However, when the tube
was evacuated and sealed, the mass
transfer was much less severe. Dry
air was definitely less harmful than
wet hydrogen. The use of commercial
sodium hydroxide, which contains up

to 2% water, resulted in greatly
reduced mass transfer under dry
hydrogen. One effect of the dry

hydrogen may be to strip the water
from the hydroxide.

In addition to the tests with
sodium hydroxide, experiments were
made with sodium hydroxide-potassium
hydroxide eutectic, with pure potassium
hydroxide, and with lithium hydroxaide.
With none of these materials could
the i1inhibiting effect of hydrogen on
the mass transfer of nickel be ob-
tained, the reason for this 1s not
yet clear. More closely controlled
experiments are planned.

FUNDAMENTAL CORROSION RESEARCH

Identificationof Corrosion Products
from Convection Loops (D. C. Hoffman,
Materials Chemistry Division). Cooling
jets that impinged on the cold leg
of an Inconel thermal convection loop
in which the LiF-NaF-KF eutectic
containing 1.8 mole % of UF, was
circulating, produced partial plugs
which persisted after the cooling jets
were removed and could not be liquefied
at temperatures of up to 650°C. X-ray
analysis of the plugs showed the
presence of KF and some material that
gave a slightly shifted spectrum of
K,NaCrF, Petrographic investigation
substantiated the presence of this
compound and also revealed some bright
metal that was shown to be nonmagnetic,
Inspecti1on under binoculars (45X)
showed a network of crystals, presumed
to be the K,NaCrF,, growing inward
from the tube walls.

The compound K,NaCrF, has a solu-
bility 1n the NaF-KF-LiF eutectic of
450 ppm Cr(III) and 2000 ppm Cr(III)
at 500 and 650°C, respectively. It
therefore appears probable that
K,NaCrF, has been responsible for
plugging in this loop, as well as
1n others, 1n which the amount of
metal transferred was insufficient to
account for the phenomenon. ©Of the
ten stainless steel harps that were
plugged while circulating this mixture,
s1x showed higher chromium content 1in
the cold leg, one showed higher
chromium content in the hot leg, and
the other three showed no systematic
variation,

The K,NaCrF, crystals apparently
form an open network that gradually
stops the flow without developing,
under usual operating conditions,
easlly detectable local concentrations.
The observation that increased operating
temperatures postpone the onset of
plugging 1n stainless steel loops 1s
consistent with this explanation of
the plugging phenomenon.

Preparation of Special Complex
Fluorides (B. J. Sturm, L. G, Overholser,
Materials Chemistry Division). Prepa-
rations of K,LiCrF,, KNaLiCrFy
Li,NaCrF,, Na,LiCrF,, and NaCrF; have
been made to furnish specimens for
x-ray and optical crystallographac
data. Continued study of the complex
fluorides of nickel has not yvet
established whether definite compounds
or solid solutions exist in the
composition range between K,NiF, and
Na,N1F,. Also, several simple fluorides
of the structural metals, including
chromofluoride, ferrofluoride, nickel
fluoride, and ferrous oxide, have been

PERIOD ENDING MARCH 10, 1953

prepared for studies of their effect
on corrosion by fuel mixtures.

Chromic fluoride can be prepared
by thermal decomposition of (NH,),CrF,
at temperatures not exceeding 900°C
some CrF, 1s formed at higher tempera-
tures. Chromous fluoride 1s prepared
by heating (NH,),CrF, or a mixture of
hydrated CrF; and NH,KF, to 1200°C 1in
graphite crucibles. The product from
either of these procedures 1s i1dentical
with that obtained by hydrofluorination
of metallic chromium at 800°C, Ferric
fluoride 1s prepared by hydrofluora-
nation of “anhydrous” FeCl; at 200 to
300°C. Ferrous fluoride may be
prepared by heating (NH,) FeF, at
750°C or, preferably, by hydrofluorai-
nation of FeCl, at a maximum tempera-
ture of 500°C. Nickel fluoride 1s
best prepared by thermal decomposition
of (NH,),N1F, Ferrous oxide 1s
prepared by decomposition, under an
atmosphere of CO,, of ferrous oxalate
precipitated from an aqueous solution
of ferrous ammonium sul fate by ammonium
oxalate. The ferrous oxide produced
1s very finely divided and 1s easily
oxidized by room temperature air, all
handling of this compound requires
a dry, 1inert atmosphere.

Alr Oxi1dation of Fuel Mixtures
(R. P. Metcalf, Materials Chemistry
Division). Solad UF, 1s known to
yield UO,F, and UF; as predominant
reaction products when treated wath
oxygen at high temperatures.,(%:9)
However, when molten NaUFS in a nickel
boat was treated with 40 to 60 cc/min
of dry air for 2 to 6 hr at 750°C,
no gaseous uranium compounds were
observed

X-ray-diffraction techniques reveal
the presence of N10, U;0,, and Na,NiF,
among the solid products of the
reaction. In addition, several un-
1denti1fied products that may be

(8)

5. Fried and N R Davidson, The Reaction

of UF, with Dry 0 A Nev Synthesis of UFG' AECD-
2981 ?Nay 1945).
S S Kirslis, T. S8 McMillan and H. A.

Bernhardt, The Reaction of Urantum Tetrafluoride
with Dry Ozxygen, K-567 {March 15, 1950)

137
ANP PROJECT QUARTERLY PROGRESS REPORT

double fluoraidesof nickel are observed.
The absence of UO,F, and UO, and the
lack of evidence for volatilization
of UF, are noteworthy.

These results, as well as the
rapid corrosion rate of nickel an
contact with molten fluorides ain air,
are consistent with the followaing
reaction mechanism

138

2N, + O2 ——> 2N10

2N10 + UF4 —> U0, + 2N1F,
3UO2 + 0, ——> U,0,

2NaF + N1F} —_— NaleF4

It appears that the container plays
an 1mportant part in the mechanism
of air oxidation of uranium 1n the
molten fluorades.
11.

Wo Dn Manly

. PERIOD ENDING MARCH 10, 19353

METALLURGY AND CERAMICS

J. M. Warde

Metallurgy Division

Cone-arc welding 1s now being
applied to the production of the many
tube-to-header joints needed 1n the
construction of the complicated heat
exchangers for aircraft reactors,
Since the heat exchangers are to be
operated at a moderately high pressure
for the temperature i1nvolved, dished
header sheets are desired. The dished
header sheets i1mpose a new problem of
unequal arc distances 1n cone-arc
welding., Studies are under way to
select the proper conditions tominimize
this effect so that consistently sound
welds can be produced. In the fabri-
cation of large heat exchanger test
units by Nicrobrazing, difficulties
due to dilution and embrittlement by
the brazing alloy were encountered.

It 1s possible to substitute other
alloys for this brazing cycle that
will not cause severe dilution and
embrittlement, and the evaluation of
high-temperature brazing alloys 1s
continuing. The melting point of the
palladium-nickel brazing alloy was
lowered by additions of germanium.
The effects of time at the brazing
temperature and of joint spacing on
the strength properties of Nicrobrazed
Joints have been determined, A summary
of the butt-braze tensile strength
data obtained on the various brazing
alloys 1s presented.

A search for a satisfactory brazing
alloy for the joining of stainless-
steel- or Inconel-clad copper fins to
Inconel and stainless steel tubes has
proved fruitful, and the oxidation
characteristics of the alloys that
show good flowability on Inconel are
now being studied. However, 1t has
been demonstrated that a chrome-plated,
high-conductivity, copper fin does not
have sufficient oxidation resistance
for this application.

The effects of environment and of
surface treatment on the creep and
stress-rupture properties of Inconel
are still being studied. Preoxidized
specimens were tested i1n argon, and 1t
was found that the creep properties
were quite similar to those determined
in argon, therefore surface oxidation
1s not the controlling factor in the
longer rupture times observed i1n the
air tests.

Control rod inserts of a mixture of
boron carbide and i1ron are being
prepared by powder metallurgy tech-
niques for the GE-ANP program. The
control rod inserts must be metal-
lurgically bonded to the outside
stainless steel tube, and various
brazing alloys are being studied for
this application. Boron carbide
safety rods for the Tower Shielding
Facility were also produced.

Special heats of a high-puraty
Inconel are being prepared for cor-
rosion tests, In the subsequent
extrusion of Inconel tubes, one of the
biggest difficulties to overcome 1s
the selection of a proper extrusion
lubricant. A satisfactory tube has
been produced by the use of glass as a
lubricant. In other studies, the
oxidation characteristics of columbium
are being i1nvestigated, and some
exploratory runs have been made on the
production of materials for use as
pump seals of the face-seal type being
studied by the Experimental Engineering
Group.

An enamel with high boron content
was applied on a shield test plate,
but a part of the surface was not well
covered and a second attempt was
necessary. Two cermet fuel elements
are described. Impregnation of
hexagonal beryllium oxide blocks with
beryllium and magnesium nitrates and

139
ANP PROJECT QUARTERLY PROGRESS REPORT

calcium and aluminum fluorides was
attempted. The fluorides filled the
pores, and the magnesium impregnation
shows promise.

WELDING AND BRAZING RESEARCH

P. Patriarca V. G. Lane
G. M. Slaughter C. E. Schubert
Metallurgy Division

Cone-Arc Welding. The design of
recent fuel-to-NaK heat exchangers for
reactor research has necessitated the
fabrication of small, Inconel, tube-
to-header subassemblies that can be
built individually and then be welded
into one, larger, test assembly. It
is obvious from the large number of
tube-to-header joints to be heliarc
welded that a semiautomatic method
would be extremely desirable. As a
result, the cone-arc welding technique
was applied to preliminary experiments
in this subassembly fabrication. Since
the heat exchanger is to be operated
at a moderately high pressure for the
temperature involved, dished header
sheets are desirable. If the plane of
the bottom of the header is kept level
during welding, it is obvious that
unequal arc distances will prevail
around the periphery of many of the
tubes. Thus the ability to make con-
sistently sound welds on the dished
header depends, to a large degree,
upon the selection of conditions that
will minimize these variations in arc
distance. Offsetting the tungsten
electrode from the center of the tube
prior to welding and using a ball-and-
socket joint arrangement from which
the header can be pivoted are obvious
means of equalizing the arc distance
around the tube periphery.

Preliminary experiments on this
problem have consisted of investigating
the variables of arc current, arc
time, and electrode distance required
to produce consistently leak-tight
welds. An experimental, cone-arc
welded, test specimen is shown in
Fig. 11.1, which illustrates the

140

An Experimental Heat

Fig. 11.1.
Exchanger Subassembly After Cone-Arc
Welding.

tendency for uneven melting around
some of the tube peripheries when the
tungsten electrode was centered over
the tube with the header plate axis
perpendicular to the electrode axis.
The diameter of the header plate is
2 1/4 in. and the plate thickness is
0.125 inch. The tubing has an outside
diameter of 0.148 in. with a 0.025-in.
wall, and the hole center-to-center
distance is approximately 0.378 inch.

Some difficulties have been en-
countered with joints that are not
pressure-tight, but it is expected
that further experiments will enable
the production of subassemblies that
will be completely sound. In each
test assembly that has been fabricated
thus far, only one or two joints were
unsatisfactory; the other joints were
pressure-tight to air at approximately
60 psig. This indicates that an
improvement in technique or a slight
variation in welding conditions should
help to improve the quality of these
cone-arc welded joints.

Fabrication of Heat Exchanger Units
A second large sodium-to-air heat
exchanger assembly was fabricated by

Nicrobrazing and the over-all appear-
ance, as shown in Fig. 11.2, was much
better than that of the first as-
sembly.(') Changes incorporated in
the brazing procedure for this second
unit were the use of lesser quantities
of Nicrobraz, the use of several
aspirators to promote more even hy-
drogen flow between the fins, the
buildup of the whole assembly off the
the drastic
initial heating rate, and the use of a
slightly lower brazing temperature.
Pressure testing with helium, however,
revealed the presence of a leak in
the tube-to-fin matrix. A technique
was developed to seal off this tube on
each side of the leak by rebrazing,

can bottom to overcome

(L)ANP Quar. Prog. Rep. Dec. 10, 1952, ORNL.
1439, p. 164,

UNCLASSIFIED
Y-8580

Fig. 11.2. Sodium-to-Air Radiator
After Nicrobrazing

PERIOD ENDING MARCH 10, 1953

but upon performing this operation,
several other leaks appeared in other
tubes.

It is hoped that the damaging
effects caused by subsequent Nicro-
brazing operations can be eliminated
by adopting a modified design that
would permit the heliarc welding of
the tube-to-header joints and the
joints in the heavier manifold sections.
By the use of the single-braze method,
any leaks in the tube-to-fin matrix
after brazing can be eliminated from
the coolant circuit by plug welding
the proper tubes. The tube-to-header
welds could then be made by manual
heliarc welding, as could the other
joints in the manifold circuit. With
some minor changes in radiator design,
it seems probable that completed units
could be fabricated with less chance
of obtaining leaky joints. By elimi-
nating several high-temperature brazing
operations, grain growth in the stain-
less steel tubes could be minimized,
as could embrittling of the base metal
by brazing alloy diffusion.

It is well known that certain
Nicrobrazed stainless steel joints
exhibit brittleness to a high degree.
This factor, coupled with the alloying
away of the tube wall by brazing alloy
dilution and diffusion, is thought to
be responsible for a major portion of
the leaks encountered in heat exchanger
fabrication. Since it seems probable
that joint brittleness may also be
associated with dilution and diffusion
phenomena, a systematic study of this
problem has been initiated.

Nicrobrazed tube-to-fin joints are
being prepared by using 0.010-in.-
thick, type 302 stainless steel fin
material and 0,150-in.-OD, 0.016-in.-
wall, type 304 stainless steel tubing
for studying, by metallographic ex-
amination, the effects of (1) the
quantity of brazing alloy used, (2)
the brazing temperature for a given
time, and (3) the brazing time for a
given temperature. Small, medium,
and large amounts of brazing alloy and

141
ANP PROJECT QUARTERLY PROGRESS REPORT

the time at temperature, which may
vary from 10 min to 18 hr (for an
overnight brazing cycle), are being
investigated., The choice of brazing
temperature 1s being varied from 20°C
above the melting point to a maximum
of 120°C above the melting point.

Since 1t may be advisable to find a
substitute for Nicrobraz in the con-
struction of sodium-to-air heat ex-
changer, two other brazing alloys are
being subjected to similar diffusion
and dilution i1nvestigations. An 82%
Au-18% N1 alloy 1s typical of the
lower-melting, ductile, oxidation-
resistant brazing alloys that, un-
fortunately,are i1ncompatible with
sodium but may be used for tube-to-fin
construction 1f dilution and diffusion
can be controlled. A 60% Mn-40% N1
alloy has been shown to be compatible
with sodium, but i1t 1sslightly attacked
in high-temperature oxidation tests.
It 1s also somewhat brittle in the as-
brazed condition.

Brazing of Copper to Inconel The
need for a satisfactory brazing alloy
for joining copper fins to Inconel
tubing and for edge-sealing sheared,
Inconel-clad, copper fins has been

be resistant to oxidation at 1500°F,
should preferably serve as a diffusion
barrier against copper penetration
into the Inconel during service, and
should have a relatively high strength
at 1500°F. It 1s likely that some
alloy other than a copper-base alloy
would best fill these requirements.
An experimental, modified, Nicrobraz
alloy, which melts at approximately
1850°F, appears to be very promising,
as do alloys of 82% Au~18% N1 and 90%
Au-10% Co. Oxidation tests at 1500°F
on brazed Inconel joints indicate no
appreciable attack., A list of several
alloys, currently being 1investigated,
that melt i1n this medium temperature
range 1s given 1in Table 11.1. Flow-
abi1lity tests on Inconel have been
made, and the results of these tests
are also listed in Table 11.1.
Edge-sealing experiments on chromium-
plated copper disks are being con-
ducted, and techniques are being
studied for obtaining a satisfactory
edge seal on these disks. At the
present, the fin edges are being
suspended 1n a slurry of the modified,
low-melting-point, Nicrobraz alloy and
slowly rotated until an even coating

emphasized. The resultant braze should of brazing alloy 1s deposited. The
TABLE 11 1 PROPERTIES OF THE MEDIUM-TEMPERATURE BRAZING ALLOYS
CURRENTLY BEING INVESTIGATED
BRAZING
MELTING POINT FLOWABILITY METHOD OF
TEMPERATURE

BRAZING ALLOY (°F) 2°F;U ON INCONEL APPLICATION
95 5% Cu-4 5% Be 1590 1750 Poor Powder
92% Cu-8% S1 1530 1700 Poor Powder
75% Cu-25% Sn 1470 1600 Good Powder
Low-melting-point Nicrobraz 1800 1850 Good Powder
82% Au-~18% Na 1800 1850 Excellent Sheet
90% Ag-10% Cu 1600 1800 Poor Sheet
90% Au-10% Co 1830 1870 Good Sheet
95% Ag-5% Ge 1650 1700 Poor Sheet
90% Cu-10% Ge 1800 1870 Good Sheet

142
fin is then edgebrazed in a hydrogen
atmosphere. Another method to be
investigated for covering these exposed
copper edges is the preplacing of a
fine wire of brazing alloy in the
crevice formed by chemically etching
away a small amount of copper in nitric
acid. Tt is expected that this latter
technique, if successful, will prevent
fin distortion, which is believed to
be due to unavoidable nonuniform
wetting of the fin periphery by
brazing alloy applied as a slurry

EVALUATION TESTS OF BRAZING ALLOYS

P. Patriarca V. G. Lane
G. M. Slaughter C. E. Schubert
Metallurgy Division

Corrosion of Brazing Alloys by
Fluorides. In a further attempt to

11.3.

Fig.
hr at 1500°F in NaF-KF-LiF-UF, (10.9-43.5-44.5-1.1 mole %).

regia. 100x.

PERIOD ENDING MARCH 10, 1953

lower the melting point of the 60% Pd-
40% Ni brazing alloy, 5% germanium was
added. This alloy can be consistently
brazed at 2150°F, which is in the
brazing range of the more common
Nicrobraz alloy. Static corrosion
tests on brazed joints in the molten
fluoride salts indicate that there is
only slight attack by molten NaF-KF-
LiF-UF, (10.9-43.5-44.5-1.1 mole %),
as shown in Fig. 11.3, whereas severe
attack is present on the sample
immersed in NaF-ZrF,-UF, (46-50-4
mole %), Fig. 11.4.

Effect of Brazing Time on Joint
Strength. A review of results of
recent and previous experiments con-
ducted to evaluate the effect of time
at the brazing temperature on the
joint strength of Nicrobrazed Inconel
joints shows that the effect of time

UNCLASSIFIED
Y-8629

.0t

INCHES

(100X)

o
&

Inconel Joint Brazed with a 60% Pd-35% Ni-5% Ge Alloy After 100

Etched with aqua

143

P eSS

ANP PROJECT QUARTERLY PROGRESS REPORT

Inconel Joint Brazed with
at 1500°F in NaF-ZrF,-UF, (46-50-4 mole

Fig. 11.4.

is not so important as the previously
reported limited data indicated. The
room temperature strengths of these
joints are generally much less than
the strength of the base Inconel and
do not vary appreciably with time at
temperature. Only occasionally does a
joint exhibit high tensile strength at
room temperature, and the previous
results were apparently distorted by
these infrequent high values. A
summary of the data is given in Table
11.2; four tensile bars were tested
for each brazing time.

It is expected that the brazing
time will have a greater effect with
stainless steel, because the width of
the diffusion zone is generally larger.
The composition of Inconel is similar

144

UNCLASSIFIED 1
¥-8701

a 60% Pd-35% Ni-3% Ge Alloy After 100 hr
%). Etched with aqua regia. 100X,

to that of Nicrobraz, with respect to
nickel, chromium, and iron contents,
and therefore diffusion may be some-
what hindered by the lack of large
concentration gradients. The tensile
strengths of Nicrobrazed Inconel joints
at 1500°F have all been consistently
high, as previously reported, and the
fractures often occur in the base
metal.

Effect of Spacing on Brazed Joint
Strength. An investigation was con-
ducted to determine the effect of
joint spacing on the short-time,
room-temperature, tensile strength of
Nicrobrazed Inconel joints. Four
butt-brazed tensile bars were prepared
for each joint spacing of 0.005, 0,010,
0.015, and 0.020 inch. The results of
PERIOD ENDING MARCH 10, 1953

TABLE 11 2 EFFECT OF TIME AT BRAZING TEMPERATURE ON STRENGTH OF
NICROBRAZED INCONEL JOINTS
TIME AT 2150°F AVEBRAGE ROOM- TEMPERATURE TENSILE STRENGTH OF JOINT EFFICIENCY

(min) BRAZED JOINTS (psi) (0 252-in specimen) (%)
5 34,900 40
10 40, 100* 46
20 32,900 38
30 31, 400 36

Includes one value of 69,800 psi

this i1nvestigation indicate that joint
spacing 1s not a critical factor 1in
the room-temperature tensile strength
of Nicrobrazed Inconel joints, at
least within the ranges 1investigated.
This range should cover nearly all
applications, since the effect of a
shrink fit would tend tobe lost during
the furnace brazing operation, and a
spacing of over 0.020 in. would result
1n extremely poor fitup.

Strength of Brazed Joints with
Various Base Metals Butt-braze
tensile data on the 73.5% Ni1-16.5% Cr-
10 0% S1 alloy indicate that the base-
metal composition may be a very
1mportant factor in determining the
subsequent tensile strengths of
brazed joints., The average of the
room-temperature tensile strengths of
the Inconel joints was 33,700 psa.
Similar tests on brazed type 316
stainless steel showed an average
tensile strength of 64,400 psi, which
1s nearly double the value recorded
for Inconel. The elevated-temperature
tests for both base metals gave
excellent results. Joint efficiencies
of 99% were obtained for Inconel, and
efficiencies of 98% were obtained for
type 316 stainless steel,

Nicrobrazed joints on stainless
steel also had higher tensile strengths
than Nicrobrazed joints on Inconel,
The average, room-temperature, tensile
strength for stainless steel joints

was 68,800 psi, whereas the corres-
ponding value for Inconel was 34,900
psi. The 1500°F tensile tests again
gave evidence of joint efficiencies
approaching 100% for both base metals.

A summary of a major portion of the
butt-braze tensile data obtained thus
far 1s presented 1n Table 11.3. The
value listed for the room-temperature
strength of Nicrobrazed Inconel joints
1s that resulting from a series of
tests inwhich some specimens fractured
at values nearly equal to the tensile
strength of Inconel., Hence, this
average value 1s somewhat higher than
the values shown in Table 11.2, 1n
which no data were recorded that indi-
cated such high trends. Also the
values of joint efficiency for stain-
less steel joints are only approximate
because the 0.252-1n.-di1a stainless
steel tensile bars have not yet been
tested after they were subjected to
the brazing cycle.

CREEP-RUPTURE TESTS OF STRUCTURAL
METALS
R. B. Oliver J. W. Woods
D. A. Douglas C. W. Weaver
Metallurgy Division

Preoxidized Inconel 1n Argon. It
was previously reported(?) that the
environment surrounding the specimen

(2)H B Oliver, D A Douglaa, K W Reber,
J W VWoods, and C W Weaver, ANP Quar Prog
Rep Dec 10, 1952, OBNL-1439, p 159

145
ANP PROJECT QUARTERLY PROGRESS REPORT

TABLE 11.3 BUTT-BRAZE TENSILE DATA FOR SEVERAL HIGH- TEMPERATURE
BRAZING ALLOYS
TEMPERATURE TENSILE JOINT
BRAZING ALLOY BASE METAL (°F) STRENGTH | EFFICIENCY
(ps1) (%)
Nicrobraz Inconel Room 47,000 54
1500 24,500 100
60% Pd-40% Na Inconel Room 87,600 100
1500 21,000 93
Type 316 stainless steel Room 75, 400 92
1500 22,400 88
73 5% N1-16 5% Cr-10 0% S1 | Inconel Room 33,700 39
1500 23,100 99
Type 316 stainless steel Room 64,400 79
1500 25,000 98
15% Ag-20% Pd-5% Mn Inconel Room 55,900 64
1500 19,700 85

during test would have a significant
effect on the creep rate and rupture
life. Results obtained during the
past quarter continue to indicate that
the properties are best when the
Inconel 1s tested in air, poorest when
tested i1n a hydrogen atmosphere, and
intermediate when tested in argon. The
rupture life observed 1n tests 1in a
hydrogen atmosphere was of the order
of one-tenth the life observed when
tested i1n air, this observation holds
for stresses of 4000 psi or less. One
hypothesis regarding the long rupture
life 1n air i1nvolved a strengthening
action by the oxide film formed during
the early part of the test., To test
this possibility, aseries of specimens
was heated in air for 200 hr at 815°C
to form such a film.

The preoxidized specimens were then
tested at 815°C 1n argon, stressed to
2500, 3500, and 4500 psi. In the
three tests, the rupture times were
far short of the rupture times 1n aair,
and were very close to the life when
tested i1n argon., It 1s to be noted

146

that the elongation during test was
much greater than that observed 1n
other environments, and there was no
evidence of cracking i1n the failm.
Figure 11.5 shows the strain vs., time
relationships of the bright and the
preoxidized specimens tested 1n argon,
These results could mean either that
nitrogen rather than oxygen was the
controlling factor or, as 1s more
probable, that the oxide film formed
stress rises when 1t cracked and that
this action, 1n turn, promoted the
integranular cracking that ultimately
led to failure.

Peened Inconel 1n Argon Specimens
of Inconel have been heavily peened
with fine steel shot and are now being
tested i1n argon. This mechanical
working of the specimen surface has
increased the rupture life in com-
parison with the life of specimens
that did not receive this treatment.
Also, the creep rate for the peened
specimen 1s much lower than that
observed for other specimens at the
same stress 1n any atmosphere. It 1s
PERIOD ENDING MARCH 10, 1953

UNCLASSIFIED
DWG 18659

[ 7301

/‘ 1%
]

z
5 © 2 s
2 © g o /
4 &
S O A J/
= O bl
i % 7
B_‘t ! / o)
za A
— / "3‘3 /
" S e
S ,,//’
<& A
3 4 —

NN
\\ \\\\\

0
0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500
TIME (hr)
Fig. 11 3., Strain vs. Time Curves for Control and Preoxidized Inconel Specl-

mens at 1500°F 1n Argon.

improbable that this effect 1s the
result of residual compressive stress
induced by the peening action, since
such stresses would recover at a
temperature lower than the test temper-
ature, It 1s hoped that metallographic
examination will furnish a clue for
explaining the 1mproved properties
resulting from the peening.

Effect of Inconel Grain Size on
Time to Rupture As a general state-
ment, fine-grained Inconel (0.105 mm
1n diameter) 1s more sensitive than
coarse-grained Inconel (0.250 mm in
diameter) to the several test environ-
ments, Table 11.4 tabulates the
comparative responses to these environ-
ments at a representative stress
level.

Effect of Environment on Strain vs
Time Curves for Inconel A fine-
grained specimen 1s being tested that
was run in argon for the fairst 160 hr
and then alternately 1n hydrogen and
argon atmospheres. Each atmosphere
was maintained for about 100 hours.
When the hydrogen was introduced, the
strain rate was accelerated, and when
the argon was re-introduced, the creep
rate during the argon cycle remained
approximately the same as the last
creep rate i1n hydrogen. These results
appear to indicate that the effect of
the hydrogen atmosphere 1s one of a
permanent nature. Metallographic
studies do not clearly reveal what the
effect 1s, however, the microstructure
1indicates the possible removal or

147
ANP PROJECT QUARTERLY PROGRESS REPORT

TABLE 11 4 EFFECT OF GRAIN SIZE ON TENSILE STRENGTH OF INCONEL

TEST CONDITIONS

TIME TO RUPTURE AT 3500 psi AND 815°C (hr)

Tested 1n air

Peened and tested in argon
Tested 1n argon

Preoxidized and tested in argon

Tested 1n molten NaF-ZrF4-UF4
(46-50-4 mole %)

Tested in hydrogen

Fine-grained Inconel Coarse-grained Inconel
~2500 ~2500
835
730 500
730 -
415 470
250 600

solution of carbides and oxides and a
much more freauent i1ncidence of
intergranular cracking when the tests
are conducted 1n a hydrogen atmosphere.

The strain vs. time relationships
of fine-grained Inconel tested 1in the
several environments at 815°C and
stressed to 3500 psi are shown 1in
Fig. 11.6. The results are representa-
tive of several tests i1n each environ-
ment The absence of a linear, second-
stage section in the strain vs time
curves might indicate that the inter-
granular cracking characteristic of
high-temperature failure 1s 1nitiated
very early i1n the test life. The
various test conditions may either
hasten or retard the formation and
propogation of these i1ntergranular
cracks and, hence, alter the creep-
rupture properties.

The program of testing in molten
fluoride mixtures continues, but the
¢ .curacy of the results 1s clouded by
temperature control difficulties and
by the effects
procedure,

of the descaling
The first series of tests
was preceded by a sodium descaling
operation. The rupture times were
very short, and the microstructures
1indicate a stress-corrosion type of
failure. Subsequent tests were pre-
ceded by a fluoride descaling process.
The rupture times were closer to those
observed 1n argon, and the micro-

structures revealed subsurface voids

148

or hole formation and very laittle
intergranular cracking, Only one test
of coarse-grained Inconel 1n fluorides
showed a longer ruptu-e life than the
best l1fe observed for the fine-grained
specimens at the same stress.

Type 316 Stainless Steel Tube Burst
Tests 1n Argon The tube-burst tests
run to date were for type 316 stain-
less steel tubes loaded internally
with argon under pressure and held in
an atmosphere of pure argon, A speci-
men so loaded has a multiaxial stress
system, and the results indicate that
the test life under these conditions
will be much less than that found for
the same material stressed i1n tension
only at a stress equal to the maximum
stress 1mposed on the tube.

FABRICATION OF CONTROL AND SAFETY RODS
E. S. Bomar H. TInouye
J. H. Coobs R. W. Johnson

Metallurgy Division
A. Levy
Pratt and Whitney Aircraft Division

Control Rods for the G-E R-1 Reactor.
A study concerning the adaptation of
the 1ron-cemented boron carbide compo-
si1tion used for the ARE safety rod
slugs to the control rods for the G-E
R-1 reactor was undertaken at the
request of the General Electric Co.
In brief, the specifications for these
previous progress report(s) included
some results of tests conducted 1n
argon purified by titanium sponge.
The work during this period has in-
cluded 100-hr tests i1in vacuum at 800
and 1000°C. This work will also
indicate the pressures tolerable in
subsequent heat treatments. The data
obtained thus far are given in Table
11.6. Future work 1s to include the
effect of recirculating the purified
gas over titanium sponge and dryaing
agents by using a pump 1n a closed
system.

Oxidation. Tests have been com-
pleted on the oxidation of columbium
in air containing 18.7 mm water vapor.
This 1nvestigation was undertaken after
a run made at 400°C 1n air that
inadvertently contained water vapor
resulted i1n a particularly rapid rate
of oxidation. A constant humidity was
obtained by passing the air through a
saturated solution of NH4C1 at 25°C,
the moisture content checked to withan
0 5% of the saturation value when the
water absorbed i1n anhydrone was
weighed.

The results at 400°C have been
checked, and they show that the presence
of moisture 1ncreases the oxidation
rate by a factor of about 400 at the
end of 10 hours. At 600°C, a reverse
effect was found i1n that the oxidation

(3)5. S Bomar and H Inouye, ANP Quar
Rep Dec 10, 1952, ORNL-1439, p 157

Prog

PERIOD ENDING MARCH 10, 1953

rate decreased significantly (by a
factor of 2). It was also apparent
that the oxi1dation rate was not linear.
Several breaks in the curve of a plot
of weight gain vs. time have been
observed and duplicated. At 800,
1000, and 1200°C, the oxidation rates
1n moist air are no different from
those 1n dry aar.

X-ray data have been obtained for
the specimens oxidized in air, and
thus far the oxides formed are all
modifications of Cb,0,, of which there
are three,(%) Inconsistencies have
been observed for the oxidation rates
at 800°C. In these runs, flat sheet
oxi1dized to a white scale on the
edges, but the other areas formed a
rather dense black oxide. A 0.040-1n,
wire became entirely white with oxide.
Above 800°C, the oxidation rate showed
a slight deviation from linearity,.
This may have been the result of
partial protection of the underlying
metal by the thick oxide or a decrease
1n the surface area. The results at
800, 1000, and 1200°C are being checked
by oxidizing columbium rod that 1s
clad on the circumference with Inconel.
It 1s hoped that the i1nconsistencies
observed at 800°C can be eliminated by
this method, since 1t 1s believed that
the i1nconsistencies were caused by the
surface areas not being constant.

(4)6. Brauer, Z anorg u allgem <chen, 248,
1-21 (1941)

TABLE 11.6. RESULTS OF 100-hr TESTS OF COLUMBIUM IN VACUUM
TEMPERATURE | PRESSURE | CHANGE IN rEIGHT HARDNESS CHANGE VPN REMARKS
(°C) () (g/cm®) (10-kg load)
800 10 +0 0046 +271 Brittle, slightly tarnished
01 +0 0006 +22 Bright
0 01 -0.0002 -16 Bright
1000 01 +0 00051 +17 Slightly tarnished
0 05 +0 00037 +9 Slightly tarnished
0 01 0 00005 ~40 Braght

151
ANP PROJECT QUARTERLY PROGRESS REPORT

FABRICATION OF METALS

E. S. Bomar, Jr. H. Inouye
J. H. Coobs R. W. Johnson
Metallurgy Division

A. Levy
Pratt and Whitney Aircraft Division

High-Conductivity Metals for
Radiator Fins A program has been
outlined for the evaluation of coated
metals suitable for radiator fin ma-
terials, methods of assembly, and
effects of exposure to high tempera-
tures. The coated metals must have
good thermal conductivity, as well as
the required high-temperature corrosion
resistance. The materials required
for this work have been ordered, and
the detailed experimental work must
await their arraval.

Small quantities
simi1lar to those on order have been
prepared and submitted to the Welding
Group for evaluation of brazing tech-
niques. These materials include
chromium-plated copper, Inconel- and
type 310 stainless-steel-clad copper,
and Inconel- and type 310 stainless-
steel-clad silver.

Oxygen-free, high-conductivaity
copper chromium-plated to various
thicknesses was tested in flowing air
at 815°C for times up to 30 hours. It
was found that the chromium oxide
formed on the surface spalled and did
not form a protective film. Thus, the
possibility of using chromium-plated
copper for high-conductivity fins 1s
not very promising.

Solid-Phase Bonding. Two series of
samples of solid-phase bonded metals
have been prepared and are to be
studied. One series of samples will
be used for evaluating the effect of
time on the extent of bonding, the
other set will be used for measuring
the effectiveness of electrodeposited
chromium 1n bonding molybdenum to
other metals.

Chromium was selected as a trial
bonding medium because of 1ts un-
limited solubilaity with molybdenum.

of materials

152

Runs have been made with chromium-
plated molybdenum vs. nickel, 1iron,
stainless steel, and Inconel at 1150
and 1250°C. One run has been made to
check the self-welding of molybdenum
at 1500°C.

Extrusion of High-Purity Tubing.
This phase of development 1s a result
of the present need for high-puraity
Inconel tubing - and subsequently for
other high-temperature alloys - for
corrosion testing to determine the
effect of i1mpurities on corrosion
resistance. The primary problem to
date 1n the extrusion process has been
one of lubrication, as evidenced by
several attempts to produce the Inconel
tubing. None of the extrusions has
been entirely successful, because
either the mandrel broke or the tube
ruptured.

Molten glass was selected as the
lubricant for the extrusion process,
which 1s a modification of Sejournet-
Ugine process, because the unit
extrusion pressures requlired are lower
with glass than with graphite or o1l.
Another factor favoring molten glass
1s the reduced possibility of pickup
of carbon and sulfur. As an 1llus-
tration, a solid rod was extruded
successfully by using Necrolaine,
which has a graphite base, and analysas
1indicated an 1ncrease 1n carbon and
sulfur from case to extruded form, as
follows 0.01% carbon as cast, 0,03%
carbon extruded, 0.007% sul fur as
cast, 0.017% sulfur extruded.

For the extrusions made thus far,
the extrusion billet was heated 1n
molten lead glass. Unfortunately, a
portion of the lead compound present
in the glass was reduced, and as a
result metallic lead precipitated 1n
the bath. The mechanism of the re-
duction 1s not yet known, however, the
glass was held i1n a stainless steel
container that was seated i1n and
extended above the top of a gas-fired
pot furnace. The prescribed extrusion
temperature for Inconel 1s 2300°F, but
with the present heating arrangement,
2100°F 1s the maximum temperature
obtainable., These factors, no doubt,
contributed to or caused the extrusion
failures.

One of the ultimate objectaives ais
to obtain tubing that has no inclusions
resulting from the addition of mal-
leabilizing elements such as magnesium,
titanium, manganese, and aluminum. To
determine whichmalleabilizing elements
are necessary, 1ngots were vacuum cast
with (1) no malleabilizing elements,
(2) 0.05% magnesium (nominal),
(3) 0.25% titanium, as a Ti1-Al-Mn
master alloy, and 0 05% magnesium.
Difficulty was experienced 1n attempting
to add magnesium in the form of a
master alloy to Inconel melts under
vacuum because of the high vapor
pressure of magnesium.

These 1ngots were successfully
extruded by using rock wool insulation
in the pressӎontainer and glass wool
(both pyrex and soft glass) on the hot
extrusion billet, This procedure 1s
essentially the same as that employed

and

by the International Nickel Co. The
extrusion conditions were

Temperature 2225°F
Die 45°V
Extrusion time 1 to 3 sec
Extrusion pressure ~50 ts1
Extrusion ratio 13.7

Surface slightly roughened

PERIOD ENDING MARCH 10, 1953

Additional work 1s to include the
extrusion of Inconel tubing under
similar conditions and the extrusion
of the austenitic stainless steels,
such as type 310 stainless steel.
Future melts will be made 1n zirconia
crucibles to eliminate magnesium
prckup.

HOT-PRESSED PUMP SEALS

E. S. Bomar, Jr. J. H. Coobs
R. W. Johnson
Metallurgy Division

A. Levy
Pratt and Whitney Aircraft Divasion

A series of exploratory runs has
been made onmaterials being considered
for test seal rings. The rings are to
be composed of Ag, Cu, 95% Ag-5% Cu
alloy, or 18-8 stainless steel with
14 vol % MoS,. Test specimens 1/2 in.
in diameter and 3/8 in. long were
prepared by hot pressing at 2500 psa
in a graphite die, with the results
shown i1n Table 11.7.

The first runs on the copper,
si1lver, and Ag-Cu alloy compositions
with MoS, demonstrated that these
materials may be readily fabricated by
hot pressing. Run 5 shows the behavior
of a coarse grade of silver (-60 + 100
mesh), which 1s on hand 1n quantity,
Bun 6 shows that this coarse powder

TABLE 11.7. EXPLORATORY PREPARATIONS OF VYARIOUS SEAL RING COMPOSITIONS
PRESSING
RUN NO MATERIAL TEMPERATURE DENSITY
(°C) (% theoretical)
1 Cu (*C" grade) 760 92 8
2 Cu (“C"” grade) 910 95 3
3 Ag (precipitated) 655 95 7
4 95% Ag (ppt )-5% Cu (“C"” grade) 830 95 2
5 Ag (-60 + 100 mesh) 930 94 0
6 18-8 stainless steel® 1010 78 0

‘Type 304 stainless steel, -355 mesh

153
ANP PROJECT QUARTERLY PROGRESS REPORT

may be satisfactory but that it is
somewhat more difficult to consolidate.
An initial run made with the stainless
steel-MoS, composition did not con-
solidate sufficiently., Further runs
are planned at somewhat high tempera-
tures, with molybdenum or Al,0; used
as die liners if necessary.

CERAMICS

L. M. Doney J. R. Johnson
S. D. Fulkerson A. J. Taylor
G. D. White
Metallurgy Division

Ceramic Coatings for Shielding. A
lead borate enamel was applied by
flame-spraying to a 5-ft by 5-ft by
7/8-in. steel plate, but the coating
was imperfect on part of the surface.
The thickness of the plate prevents
the spraying of a smooth, glass coat-
ing, even when the plate is initially
preheated to 1000°F. A larger flame-
spray unit would probably work, but
with the available unit, a maximum
steel thickness of about 1/8 in. is
indicated.

The coating can be successfully
applied by the standard enameling
technique of dusting the powdered
glass on the red hot metal. An
initial metal temperature of 1400°F is
adequate for this coating.

Development of Cermet Fuel Elements.
A Cr-Al,0, solid fuel element of
possible use in nuclear-powered air-
craft has been fabricated. These
cermet elements would be for use in a
reactor system that resembled a disk
heat exchanger. Disks of Cr-Al,0,
containing an enclosed “sandwich” ring
of UO, are strung on 1/4-in,-dia
stainless steel tubes through which a
coolant passes. These disks give up
their heat to an air stream passing
over their surfaces. The coolant
keeps the steel tubes cool enough to
maintain adequate strength, resist
oxidation, and provide some moderation
in the reactor core.

154

The Cr-Al1,0; cermet has excellent
oxidation resistance up to 2200°F and
adequate heat-shock resistance in the
form proposed. However, radiation
damage and fission product retention
are unknown. The disks have been
fabricated with a 0.020-in. thickness
and a l-in. outside diameter. A
photomicrograph of a Cr-Al,0, disk
section containing 8 wt % U0, is shown
in Fig., 11.7.

A silicon-silicon carbide material
has been proposed as a potentially
useful solid aircraft reactor fuel
element, and experiments have been
carried out to fabricate such elements.
A process similar to that developed by
the Carborundum Co. for impregnating
carbon is used, Cross-shaped elements

0.020 in.

11.7.

Fig.
Cermet Disk Containing 8 wt % UO,.

Section of Cr-Al,0,
have been fabricated and are being
tested to obtain physical property
measurements. Radiation damage studies
and simulated reactor service tests
have been outlined.

Reduction of the Porosity of ARE
Beryllium Oxide. A series of experi-
ments was 1nitlated 1n an attempt to
reduce the porosity and to 1ncrease
the corrosion resistance of the
hexagonal beryllium oxide blocks that
are to be used for the ARE. Impregnation
of porous beryllium oxide test pieces
with various materials was attempted,
and preliminary vapor deposition
studies were begun., Samples of the
impregnated beryllium and the other
specimens will be subjected to cor-
rosion tests 1n sodium.

In the first i1mpregnation, the
beryllium oxide test pieces (1/2 by
1/4 by 1/4 1n.) were 1mmersed 1n
molten hydrated beryllium nitrate at
135°C for 48 hours. The test pieces
were then placed 1n a small porcelaain
crucible, covered with the beryllium
nitrate, and heated until evolution of
NO, was complete. Then, after the
powdered beryllium oxide had been
scraped from the surfaces, the test
pieces were heated to 1550°C, held for
15 min, and allowed to cool to room
temperature. This procedure caused a
slight permanent increase in the weight
of the test pieces but no measurable
reduction 1n porosity.

The second i1mpregnation experiment
was carried out under similar con-
ditions, but magnesium nitrate was
used i1nstead of beryllium nitrate.
This impregnation resulted 1in an
appreciable permanent weight gain and
in a reduction of the porosaity to
about 35% of 1ts original value.

The third method tried was successful
1in reducing the porosity of the test
pieces to an almost negligible amount,
In this method, a mixture of 37,5%
AlF, and 62.5% CaF, was prefused to
give a glass. The glass was then
pulverized and suspended 1n water to
give a very heavy suspension. The

PERIOD ENDING MARCH 10, 1953

small (1/2 by 1/4 by 1/4 1n.) beryllium
ox1de test pieces were dipped into
thi1s suspension and drained, and the
adhering coating was dried. These
pieces were then heated to 1550°C,
held for 15 min at this temperature,
and allowed to cool 1n the furnace.
This treatment resulted in the coating
soaking 1nto the porous pileces and
completely filling the pores.

A test was performed to determine
whether the method outlined above for
the AlF;-CaF, mixture would be feasible
when used on one of the ARE-s1ize
blocks. One of the half-blocks of
beryllium oxide, which had been shown
to be porous by both i1mmersion 1in a
fuschin-dye—alcohol solution and by
the water absorption method, was
coated with the AlF,-CaF, mixture.
The coated block was then heated at
750°C per hour to 1550°C, held at
1550°C for about 45 min, and cooled
overnight 1n the furnace. The result
was a nonporous block, except for
cracks that, although present in the
block before the heat treatment,
now much enlarged.

To determine the cause for the
enlargement of the cracks, two, full-
si1ze, hexagonal blocks of beryllium
oxide were heated at 750°C per hour to
1550°C and held for 1 hour. These
blocks had no applied coating, and
hence any effects could be attributed
to the heat treatment. There were
several minute cracks and a zone of
weakness perpendicular to the axais
across the middle of each block before
the heat treatment was carried out.

were

However, these blocks are typical -
nearly all blocks on hand have this
zone of weakness. After the heat
treatment, both blocks had broken
along the zone of weakness, and one
had also split axially from one side
into the core hole. It 1s to be
concluded therefore that the block
can be broken by heating at 750°C per
hour and that the impregnation treat-
ment 1s not the cause of the cracking
of the pieces.

155
12.

HEAT TRANSFER AND PHYSICAL PROPERTIES

H. F. Poppendiek

Reactor Experimental Engineering Division

Some preliminary thermal conduc-
tivity measurements on molten sodium
hydroxide were made, and results
indicate that the conductivity 1in-
creases from about 0.4 to about 0.8
Btu/hr*ft*°F over the temperature
range of 650 to 1000°F, The enthalpies
and heat capacities of strontium
hydroxide and NaF-ZrF,-UF, (50-46-4
mole %) have been obtained. The heat
capacity of strontium hydroxide was
found to be 0.32 * 0.03 cal/g*°C over
the temperature range of 570 to 900°C,
the heat capacity of NaF-ZrF,-UF,
(503$§-4 mole %) was found to be
0.31 £ 0.03 cal/g*°C over the tempera-
ture range of 550 to 850°C. Some
preliminary vliscosity measurements on
molten sodium hydroxide were made that
varied from 4.6 cp at 340°C to 2.5 cp
at 448°C. Vapor pressure measurements
made on the enriched ARE fuel were
much lower than the theoretical value
determined for the pure components
(NaF, ZrF,, and UF,), which indicates
the existence of complex compounding
of 1ons at high temperatures. A graph
of the temperature variation of the
Prandtl moduli of several fluorade
mixtures, as well as some liquad
metals, water, and sodium hydrox-
1de, has been prepared, the Prandtl
modulil of the fluoride mixtures are
between approximately 1 and 10 i1n the
high-temperature range in which air-
craft reactors are to operate.

An analysis has been completed
that can be used to minimize the
weight of a radiator for a given power
output by optimizing the fin spacaing,
fin thickness, tube diameter, and tube
spacing for plate-fin radiators
operating in the laminar flow region.
An apparatus has been designed for
determining fluid velocaity profiles
1n the annular spaces found 1n reactor
systems. Temperature distributions

air,

1n thermal entrance regions of cair-
culating fuel reactor systems are
being determined mathematically by
the finite difference method. An
apparatus for experimental study of
the thermal structure ina circulating-
fuel system similar to the ARE con-
figuration has been assembled.

THERMAL CONDUCTIVITY OF LIQUIDS

W. D. Powers S. J. Claiborne
R. M. Burnett
Reactor Experimental Engineering
Division

Three different methods are now
being used to determine the thermal
conductivityof liquids at high tempera-
tures, Two methods are based on
steady-state flow of heat downward
through a thin section of liquid, and
the third method i1nvolves transient
heat flow from an electrically heated
tube to the surrounding test liquad.
These methods have been described in
previous quarterly reports. At
present, the thermal conductivity of
molten sodium hydroxide 1s being
determined. Preliminary results
indicate that the conductivity of
sodium hydroxide increases from about
0.4 to about 0.8 Btu/hr-ft*°F over
the temperature range of 650 to
1000 °F.

After additional thermal conduc-
tivity measurements on sodium hydroxaide
have been completed, some of the
other hydroxides will be investigated.

HEAT CAPACITY OF LIQUIDS
W. D. Powers G. C. Blalock

Reactor Experimental Engineeraing
Divisien

The enthalpy and the heat capacaty
of strontium hydroxide and NaF-ZrF,-UF,
(50-46-4 mole %) have been determined

157
ANP PROJECT QUARTERLY PROGRESS REPORT

with Bunsen 1ce calorimeters for

strontium hydroxide over the tempera-
ture range 570 to 900°C, (1)

0327T-19 ,
0.32 £ 0 03

Hy (liquid) - Hye. (solid)

c
P

for ARE fuel mixture NaF-ZrF4-UF4

(50-46-4 mole %) over the temperature
range 550 to 850°C, (2)

Hy (liquid) - Hyop (solad)

031 T-9,
031 + 003,

c
P

(I)W D Powers and G C Blalock, Heat Capacity
of Strontium Hydroxide, ORNL CF-53-2-84 (Feb 9,

1953)

(2)W D Powers and G. C Blalock, Heat Capacity
of Fused Salt Mixture No 30, ORNL CF-53-2-56
(Feb 6, 1953)

where H 1s the enthalpy in cal/g, T
1s the temperature in °C, and ¢, 1s
the heat capacity in cal/g*°C.

Shown 1n Fig. 12,1 are a few of
the many enthalpy-temperature points
for the ARE fuel mixture that were
used to establish the above enthalpy
and heat capacity relationships.

VISCOSITY AND DENSITY OF
ALKALI HYDROXIDES

R. F. Redmond S. I.
T. N. Jones
Reactor Experimental Engineering
Division

Cohen

A preliminary studyof the viscosity
of sodium hydroxide was made by using

A
DWG 18661
280
AJm’,/””/’;
240 —0-0
200 //zflfflf//é;’
o)
b o]
2 160 00‘47/,/’
3 o
3 :
g |8
T 120 ;
35 |
j~=—MELTING POINT ~ 520°C
ol _J
- w-
80 -=%
40
0
400 500 600 700 800 900
TEMPERATURE (°C)
Fig 12 1 Enthalpy vs Temperature for Fluoride Fuel, NaF-ZrFq-UF4 (50-46-4
mole %).

158
a modified Brookfield viscometer to
check values reported i1n the litera-

ture. The results, which agree with
those 1n the literature to withain
about 10%, are as follows
TEMPERATURE (°C) VISCOSITY (cp)
340 3 4 55
399 5 2 95
448 0 2 45

A drybox that will contain facilaties
for measuring densities and viscosities
has been designed and 1s being fabrai-
cated. A new device will be employed
that relates density to the difference
1n pressure requlred to bubble gas
through outlets placed at two levels
1n the material. A horizontal capillary
tube viscometer 1s being designed and
should soon be ready for construction.
This device for obtaining absolute
measurements will operate under more
nearly 1sothermal conditions than some
of the previous viscometers,

VAPOR PRESSURES OF FLUORIDES
R. E. Moore

Materials Chemaistry Division

Vapor pressure studies of the
fluoride salt mixtures and their pure
components of interest i1n connection
with the Aircraft Reactor Experaiment
were continued during the past quarter.
The method and apparatus originally
described by Rodebush and Dixon(®’ and
discussed i1n previous reports¢*:%?
were used for these measurements.

The vapor pressure of the enriched
fuel mixture NaF-ZrF,-UF, (50-25-25
mole %) was found to increase from 5
to 63 mm Hg over the temperature range

of 848 to 1096°C and 1s represented
by the equation

log P (mm Hg) = -6906/T (°K) + 6.844 .
The heat of vaporization is 32 kcal/mole

(3)W H Rodebush and A, L Dixon, Phys Rev
26, 851 (1925)

)B E Moore and C J Barton, ANP Quar
Prog Rep Sept 10, 1951, ORNL-1154, p 136

(S)H E Moore, ANP Quar Prog Rep Dec 10,
1951, ORNL-1170, p 126

PERIOD ENDING MARCH 10, 1953

and the calculated boiling point 1is
1470°C.

If the mixture 1s regarded as being
composed of the compounds zirconium
tetrafluoride, uranium tetrafluorade,
and sodium fluoride, the 1deal vapor
pressure at 970°C, a temperature
arbitrarily chosen for comparison
purposes, 1s calculated from Raoult’s
law to be 518 mm Hg. The actual vapor
pressure of 19 mm 1s a very large
negative deviation from ideality. This
result 1s 1n agreement with results
obtained for other zirconium tetra-
fluoride-bearing mixtures and has been
interpreted as an indication that
zirconium tetrafluoride and uranium
tetrafluoride exi1st i1n the fused
mixture 1n the form of complex 1ons.
However, 1t 1s 1nteresting to note
that when the mixture 1s considered
tobe composed of NaZrF,; and NaUFg (the
vapor pressure of NaUF, at 970°C 1s
negligible), the 1deal pressure 1s
calculated to be 33 mm. The negative
deviation 1s still present, but to a
much less extent. Jt 1s possible that
further complexing of the zirconium
tetrafluoride 1s accomplished by
adding NaUF; to NaZrF.;. Ions such as
NaZrF, may exist in these melts,

Preliminary vapor pressure data for
zirconium tetrafluoride and NaF-ZrF,-
UF, (50-46-4 mole %), the composition
chosen for the ARE fuel, were re-
ported previously.{®) The vapor
pressure equation for zirconium tetra-
fluoride, obtained from the latest
data by the method of least squares,
1s
log P (mm Hg) = -10935.6/T (°K)

t+ 12,113 ,
from which the heat of sublimation,
50 kcal/mole, and the sublimation
temperature, 912°C, were calculated.
Similarly, the data for NaF-ZrF,-UF,
(50-46-4 mole %) are represented by
log P (mm Hg) = -7551/T (°K) + 7.888 .
The heat of vaporization, 35 kcal/mole,

(6)R. E Moore, ANP Quar
1952, ORNL-1375, p 147

Prog Rep Sept 10,

159
ANP PROJECT QUARTERLY PROGRESS REPORT

and the boiling point, 1235°C,
calculated from this equation.

were

PRANDTL MODULI OF VARIOUS MATERIALS

The Prandtl number 1s one of the
1mportant moduli that characterize
convective heat transfer. In the case
of turbulent flow, convective heat
transfer may be divided 1nto three
classes, depending upon the nature of
the fluaid the liquid metals are 1in
the Prandt]l modulus range of about
0.003 to 0.06, the ordinary fluids
are 1n the Prandtl modulus range of
about 0.5 to a hundred, very viscous
fluids have Prandtl numbers of several
hundred and above. Each of these
different classes of fluids 1s defined
by a different turbulent convection
heat transfer relationship. It 1s of
1interest to note the class 1n whaich
the fluoride mixtures fall. Figure
12,2 presents a graph of Prandtl
modulus vs temperature for the liquad
metals, water, air, and sodium hydrox-
1de, as well as for three fluoraide
mi1xtures., In the high-temperature
range 1n which aircraft reactors are
to operate, the fluorides have Prandtl
moduli that are in the range of approxi-
mately 1 to 10 and thus should have
the heat transfer characteristics of
ordinary fluaids.

SPECIFIC REACTOR HEAT TRANSFER PROBLEMS
W. S. Farmer

Reactor Experimental Engineering
Division

The problem of designing a minimum-
weight air radiator for the turbojet
engines of a nuclear-powered aircraft
has been given further study during the
past quarter. An analysis was made(”?
which i1nvolved minimizing the weight
of an air radiator for a given power
output by optimizing the fin spacing,
fin thickness, tube diameter, and
tube spacing for plate-fin radiators

(7)W S Farmer, Mintaun Weight Analysis for

an Air Radiator, OBNL CF-53-1-111 (to be 1ssued)

160

operating in the laminar flow region.
Difficulties due to poor braze contact
between the fins and tubes have recently
been experienced in the fabrication
of plate-fin radiators. Therefore
attention 1s now being given to other
types of fins, such as circular disk
fins, that can be fabricated so that
the difficulties can be obviated. A
general analysis 1s being made of the
performance of radiators that embody
several different types of extended
surfaces and shapes.

Because the air radiator 1s to be
operated at high temperatures with
large temperature differences existing
between the wall temperature and the
air temperature, the problem of non-
1sothermal heat transfer assumes con-
siderable i1mportance. Since efforts
to correlate the performance of several
air radiators tested by the Experai-
mental Engineering Group with existing
theories for nonisothermal heat
transfer have not been too successful,
the possibility of developing new
analytical solutions is being examined.

TURBULENT CONVECTION IN ANNULI

J. 0. Bradfute J. I. Lang
Reactor Experimental Engineering
Division

A critical review of all the
1sothermal velocity distribution data
known to ex1st(®+%:19) for annulus
systems was completed, and the data
are judged to be either partly or
wholly unsatisfactory because suffi-
cient care 1n making the different
experimental measurements was appar-
ently not exercised. The need for
good experimental velocity data an
developing mathematical, turbulent-
flow, heat-transfer relations for

(B)J. G. Koudsen and D. L. Katz, Proceedings
of the Midwestern Conference on Fluid Dynanmics,
First Conference, May 12-13, 1950, p 175

(g)v Mikrjukov, J Tech Phys (U,S S R ) 4,
961 (1937).

)R R Rothfua, C.
Senecal, Ind Eng Chen

C Monrad,
42, 2511 (1950)

and B E
PERIOD ENDING MARCH 10, 1953

L
DWG 18662
{00 LY { I N
\\ NaF - ZrFs-UF4, 50-46-4 mole % =3
\ . Sy
\"Hao \\‘ \
\ NoOH— \\ N
NaF-KF-LiF-UFg, 10 9-43 5-44 5-11 mole %\
| I I N
| 1 \
NoF-KF-LiF,115-42 0-46 5 mole % ——-\
10
AIR
w
s |
| .
e |
(a]
Q
= 01
wd
-
g Lt
<{
[« o g T—
o
-y
\— Pb-Bi (EUTECTIC)
HX\\(—N K (EUTECTIC)
\‘ ‘..-_‘\
. Na
~ .
\\
0 001
0 200 400 600 800 1000 1200 1400

TEMPERATURE (°F)

Fig. 12.2. Prandt]l Modulus vs. Temperature for Some Fluoride Mixtures and Other
Fluids.

161
ANP PROJECT QUARTERLY PROGRESS REPORT

annuli (vw¥a, the heat-momentum transfer
analogy) has prompted a modest program
for determining such data.

An apparatus has been designed
(Fig. 12.3) that will consist of a
4-1n., vertical, brass pipe about 20 ft
long which will be the outer annulus
wall, smaller diameter tubes which
wi1ll be the 1inner annulus walls, a
6-1n., lucite, test section 1inserted
in the outer pipe, a high-intensity
light source of short duration, a
camera, and an air blower. As air 1is
pumped through the annulus, photographs
will be taken of the colloidal dust
particles in the air 1lluminated by
the vertical plane of light coming
through the lucite. The positions of
the particles scattering the light
can be determined by superimposing a
previously determined photograph of

UNCLASSIFIED
DWG {8663

958 ft-2 1n FLOOR

CORE-—s\\:
-

4 —in BRASS PIPE——m

. f . S —

LUCITE TEST
SECTION—— L m
L

40 1n
938 {t—-0in FLOOR

-

FLOWMETER
/SURGE CHAMBER

) FAN

OTOR

Fig 12.3 Schematic Diagram of the
Annulus Apparatus for Measuring
Velocity Distributions

162

a grid system,. By stroboscopic
flashing of the light, successive
positions of several particles can be
obtained, from known time intervals
between flashes and the positions of
the particles, the velocity of the
air can be measured as a functaion of
radial position. A modified version
of this annulus system can be used to
study the hydrodynamic behavior of
circulating-fuel reactors.

CIRCULATING-FUEL HEAT TRANSFER

H. F. Poppendiek G. Winn
Reactor Experimental Engineerang
Division

Thermal and hydrodynamic boundary
layers are usually not established
in short-duct circulating-fuel systems,
such as the Fireball, for example.
These systems are being studied
mathematically. For some specific
systems, the differential equations
describing the boundary-value problem
have been transformed to finaite
difference equations and evaluated
numerically. Recent evaluations have
been made with the aid of high-speed
computing machines,

An apparatus that will yield expera-
mental 1nformation on the thermal
structure i1n circulating-fuel systems,
1in both the entrance region and the
established flow region, was briefly
described previously.¢(!!) Thais
apparatus, which has been fabricated
and assembled (shown schematically
in Fig. 12.4), consists of a closed
loop that cairculates a sulfuric acad
solution, a volume heat source 1is
generated within the circulating acad
by means of an alternating current.
Several orifice meters have been
calibrated, and the power supply that
1s to generate the volume heat source
and the thermocouples are to be
installed next.

(ll)H F Poppendiek and G Winn, ANP Quar
Prog Rep Dec 10, 1952, ORNL-1439, p 185,

-
COOLANT SUMP —

~=——HEAT EXCHANGER

PERIOD ENDING MARCH 10, 1953

UNCLASSIFIED
DWG 18664

ELECTRODES

/UM
MIXING WELLS

COOLANT PUMP —=] ‘ 1
ACID
SUMP
Fig 12 4.

HIGH TEMPERATURE REACTOR
COOLANT STUD IES

H. F. Poppendiek J. I. Lang
Reactor Experimental Engineering
Division

The effectiveness of some of the
pertinent high-temperature coolants,
particularly flinak, sodium hydroxaide,
bismuth, sodium, and air, 1s beaing
investigated. Comparisons of the
circulation velocities, radial tempera-
ture differences, axial temperature

e \_

! SURGE Tfij:$;7 ORIFICE

—clo—

MANOMETER -

~—ACID PUMP

Schematic Diagram of the Circulating-Fuel Heat Transfer System.

differences, pressure drops, and
pumping powers requlired to achieve
a series of different cooling tasks
are being made. Because of the large
differences in physical properties of
the various coolants being studied,
both laminar and turbulent flow are
encountered. At times, a significant
fraction of the coolant duct may lie
in a thermal and hydrodynamic entrance

region, hence, entrance region heat
transfer 1s also i1nvolved 1in these
studies.

163
13.

RADIATION DAMAGE

D. S. Billington, Solid State Division
A. J. Miller, ANP Division

Radiation-damage studies of struc-
tural materials, fuels, and other
coolants exposed in the ORNL graphite
reactor and the LITR have continued.
Two Inconel capsules containing the
ARE zirconium-bearing fuel were
1irradiated in the MTR and examined.
In both cases, various portions of the
capsule walls attained temperatures
well above 1500°F. There were inda-
cations of changes i1n the fuel com-
position, and corrosion occurred 1n
excess of that i1n out-of-reactor tests
at 1500°F. A third irradiation 1s
now 1n progress in the MTR, and, with
a more elaborate control system, an
attempt 1s being made to hold the
maximum wall temperature at 1500°F,.
Additional 1n-reactor cantilever
creep measurements on Inconel in an
air atmosphere were made 1in both the
graphite reactor and the LITR. In
confirmation of previous observations,
1t appeared that the irradiation had
little effect on the creep strength.

These and other radiation-damage
studies are described. Additional
information 1s contained 1n the
Solid State Division Quarterly Progress
Report for the Period Ending February
19, 1953, ORBRNL-1506.

IRRADIATION OF FUSED MATERIALS

G. W. Keilholt:z M. T. Robinson
J. G. Morgan D. D. Davies
H. E. Robertson A. Racht

C. C. Webster W. J. Sturm

P. R. Klein W. R. Willis

M. J. Feldman
Solid State Division
R. J. Jones R. L. Knight

Electromagnetic Research Division

Thermal-flux measurements of fused
salts were made 1n the testing facilaty
in the MTR. The power density in the

ARE type of fuel was found to be 2700
watts/cm?®, rather than the 1900
watts/cm® predicted.

The wall temperature of the first
capsule of the ARE type of fuel
irradiated in the MIR was detected and
controlled by a single thermocouple.
Examination of the surface 1n contact
with the fuel disclosed that at least
some portions of the wall had been
heated to temperatures far above 1500°F
during the 116 hr of airradiation. In
some sections, the Inconel grain size
had been greatly 1ncreased and corrosion
penetration of up to 10 to 12 mils
occasionally occurred. In the regions
of normal grain size, the wall showed
uniform, dense, 1intergranular corrosion
that averaged about 2 mils in depth,
and there were subsurface voids 2 to
4 mils deep.

Fhe second MTR fuel capsule was
irradiated for 325 hr and the wall
temperature was detectedat two points.
The control thermocouple held the
capsule wall locally at 1500°F, but
the second thermocouple, several
milliliters away, recorded temperatures
as high as 1740°F. Examination of the
Inconel 1n this case showed only
dense, 1intergranular corrosion that
averaged about 2 mils in depth.

Sealed Inconel capsules that had
been prepared for irradiation in the
MTR were x rayed, and the photograph
showed that the fuel was not solidly
packed i1nto the 100-m11-ID fuel
chamber. Rather than occupying the
bottom of the chamber, the fuel was
scattered over the length, and thus
there were numerous voids in what had
been assumed to be asolid fuel column.
Considerable study was given to the
filling and temperature control
techniques. A thard capsule, which 1is
currently being irradiated in the MTR,

165
ANP PROJECT QUARTERLY PROGRESS REPORT

was filled by an improved method, and
1t was observed with x rays that the
fuel formed a nearly continuous
100-m1l-d1a columnin the lower 1/2 1n.
of the fuel chamber. A rod attached
to the cap of the capsule was fitted
tightly 1nto the fuel chamber to
press on the top of the fuel to help
hold 1t i1n place. Three thermocouples
were spotted on the outside chamber
walls to record the temperatures in
various locations, and they were
instrumented so that the temperature
could be regulated by means of the
hottest thermocouple. During the
course of the experiment the middle
thermocouple has been damaged, and
the temperature 1s now controlled by
the top and bottom thermocouples.
Chemical analyses and analyses for
the fission product Cs!3®’ were carried
out on the two fuel samples irradiated
1in the MTR  Since there 1s some doubt
as to how thoroughly the fuel was
dissolved during the course of the
analytical work, there 1s some doubt
as to the validity of the analytical
results. Work 1s under way to improve
the methods of dissolution of the fuel.
However, a trend that warrants mention
was i1ndicated by the analytical re-
sults. There was indication that the
fuel which could be readily melted
out of the capsule was more deficient
1n uranium than could be accounted
for by burnup. This deficiency in
uranium did not occur 1in the controls
heated on the bench at 1500°F for an
equivalent number of hours. In addi-
tion, when the Cs'!37 was counted 1in
the fuel that had been melted out,
there was less detected than would
be expected from the thermal-flux
measurements, In one case, the
discrepancy 1n Cs'37 corresponded
fairly closely with the discrepancy
in uranium content, 33% and 28%, and
led to the supposition that the
uranium had been segregated during the
early portion of the 1rradiation
period However, the shortage an
cesium might have been due to escape

166

of xenon 1nto the capsule free space
and subsequent deposition of cesium
on the walls

If 1t 1s assumed there are no con-
vection currents i1n the capsule to
relieve the temperature gradients,
the central portion of the fuel 1s
calculated to be 1200°F hotter than
the fuel at the wall, which the
thermocouples try to hold at 1500°F.
Whether the apparent uranium segre-
gation was due to whatever temperature
gradients may have existed, the general
overheating of the capsule, faulty
analyses, or to radiation-damage 1s
a question that requires further
investigation. In 1rradiations
carried out several months ago with
the Y-12 Cyclotron and by North
American Aviation{(!) on the Berkeley
Cyclotron, some evidences of uranium
segregation in the various fuels were
also noted, but the question of un-
certain thermal history exists for
these cases too,

Before opening, the first capsule
irradiated 1n the MTR was heated to
well above the normal melting point
of the fuel and 1ts freezing point
was compared with that of unirradiated
fuel 1n a similar capsule by means of
a differential thermocouple arrange-
The melting point of the
irradiated fuel appeared to be 23°F
higher, but the difference was within
the range of experimental error. In
a similar experiment, a sample of
fuel 1rradiated for 140 hr in the
LITR at 140 watts/cm® appeared to have
an 11°F lower melting point than 1ts
corresponding unirradiated bench
tests

Only a small amount of work 1is
now being done on radiation damage
with the Y-12 Cyclotron. This work
included an i1nvestigation of methods
for helium-cooling targets containing
fused fluorides, the study has been
discontinued. Also, some work 1s

ment.

“)\V V Goddell et al , Cyclotron Irradiation
of Fused Fluorides i1n Inconel at Elevated Tempera-
tures, NAA-SR-208 (Jan 26, 1953)
being carried out to determine the
usefulness of proton bombardment as
a method for studying hydrogen em-
brittlement 1in metals.

IN-REACTOR CIRCULATING LOOPS

(L‘glsman R. M. Carroll
W. W, Parkinson C. D. Bauman

J. B. Trice C. Ellis

A. S. Olson W. E. Brundage
M. T. Morgan F. M. Blacksher

Solid State Division

The loop previously reported(?) as
having circulated sodium 1n the LITR
"for a week was examined to determine
the cause of stoppage of flow. It
was found that the difficulty was due
to a leak 1n the Inconel tubing that
caused a lowering of the sodium level
in the surge tank. Additional experi-
ments were carried out with sodium
loops 1in the graphite reactor while
1improvements were being made 1in the
design of the LITR loop. Development
continued on a pump for use with an
in-reactor fluoride-fuel-circulating
system.

CREEP UNDER IRRADIATION
W. W. Davais J. C. Wilson
J. C. Zukas

Solid State Division

Several new tests have been com-
pleted on annealed Inconel at 1500°F
with stresses of 3000 and 4000 psi in
air in both the LITR and the graphite
reactor. The spread i1n the data and
some variations 1in experimental condi-
tions allow no quantitative estimates
to be made of the effect of neutron
bombardment on creep, but 1t can
safely be said that the effect 1s not
great enough to cause material change
1n the design stress values for these
materials under these experimeéntal
conditions. The next work to be

(2)ANP Quar
1439, p 187

Prog Rep Dec 10, 1952, ORNL-

PERIOD ENDING MARCH 10, 1953

performed will be to assess the effect
of the atmosphere in which the specimen
1s tested, since this 1s known to be
an important factor in the creep
strength of Incomel.(3)

In Fig. 13.1, the 4000-psi1 data
show that less creep was experienced
under irradiation than in the single
bench test. The odd shape of curve
X-1l 1s believed to have been caused
by the loading weight resting against
the microformer during the early
part of the test. The difference
between the LITR curves may have
resulted because the test shown by
curve L-2 received 24 hrof irradiation
before the furnace was turned on and
the specimen was stressed.

The total strains in each of two,
3000-ps1, graphite reactor tests, and
one LITR test, shown in Fig. 13.2,
are in good agreement with the values
for similar tests reported earlier. (%)
On the basis of the single bench test
to which all the in-reactor curves
are compared, 1t 1is not possible to
say whether more or less creep resulted
from irradiation. At the end of the
quarter, two tests were operating in
the LITR in a helium atmosphere and
duplicate bench tests were about to
be started.

Annealed Inconel and type 347
stainless steel bars were exposed for
a one-week period in hole HB-3 of the
LITR 1n a gradient furnace to determine
the highest temperature at which
neutron bombardment would cause an
increase 1n hardness of the metals.
The temperature gradient was from
1700 to 1150°F over 4.5-1in. specimens.
Rockwell B hardness measurements were
made after irradiation and compared
to the 1nitial hardness and to the
hardness of control bars subjected to
the same time and temperature history
as the 1n-reactor specimens. The
Inconel was hardened slaightly by

(B)ANP Quar Prog Bep Dec 10,
1439, Fig 14 1, p 188

(4)Soltd State Quar Prog Rep Nov 10, 1952,
ORNL- 1429, Fag 2, p 7

1952, ORNL-

167
ANP PROJECT QUARTERLY PROGRESS REPORT

) DWG 18!07

0 24 , |
ANNEALED 2 hr AT {700°F /\/
3 020 Vi
g B
& A
=2 ‘/,f
H o016
o / L
o /
s
N l
m 012 f//);- Cr””’,,v
o
L_

L; / X-2 )}_/
L)
—
= 008 O
=
<[
O

0 04 ;/ X -1 A BENCH TEST (B)

J HOLE HB-3 LITR({L-f, L-2)
HOLE 15 ORNL GRAPHITE REACTOR
(X-1, X-2) | |
0
o) 200 400 600
TIME (hr)

Fig. 13.1 In-Reactor Cantilever Creep Tests on Inconel 1n Air at 1500°F and

4000 ps1.

bombardment at temperatures as high as
1550°F., The stainless steel showed
an 1ncrease 1n hardness only at
temperatures below 1400°F, Earlier
work on type 347 stainless steel(5)
indicated that below 1400°F the creep
rate of the material was reduced by
neutron bombardment, whereas data
taken at 1500°F indicated that in-
creased creep resulted during irradai-

(S)Soltd State Quar
OBRNL-1301, Fig 2, p 6

Prog Rep May 10, 1952,

168

ation. A qualitative correspondence
could be expected 1n Inconel, and
the hardness data may help to explain
why no substantial effects of irrada-
ation have yet been observed. Because
the steep gradients in the experiment
reported here do not permit accurate
determination of temperature, hardness
blocks wi1ll be placed alongside the
creep specimens 1n future tests to
enable more precise determination of
irradiation hardening conditions at
several temperatures.
PERIOD ENDING MARCH 10, 1953

SSD-A-706
DWG 18808

O t2 l
ANNEALED 2 hr AT 1700° F

o
o

0 08 //

/A/

A BENCH TEST
HOLE HB-3 LITR
® HOLE 20 ORNL GRAPHITE REACTOR

CANTILEVER BEAM EXTENSION (%)

O

002

0 200 400 600
TIME (hr)

Firg 13.2 In-Reactor Cantilever Creep Tests on Inconel in Arr at 1500°F and
3000 pst.

169
s

oy
14.

ANALYTICAL STUDIES OF REACTOR MATERIALS

C. D. Susano, Analytical Chemistry Division

J. M. Warde, Metallurgy Division

The work on reactor fuel components
during the past quarter has produced a
satisfactory method for the determi-
nation of zirconium 1n the presence of
uranium. A method was also developed
by which the reduction products,
uranium trifluoride and zirconium
metal, may be determined 1n the presence
of one another. These reduced species
may be faormed upon excessive addition
of a corrosion i1nhibitor such as a
hydride or NaK to the ARE fuel mixture,

Study has also been concentrated on
developing methods for the determi-
nation of contaminants in the fluoraide
fuel mixtures. An apparatus has been
developed and tested in which traces
of metallic oxides in fluoride mixtures
may be determined. The determination
of trace guantities of hydrogen fluoride
1n helium 1s expedient during the final
stages of the fuel preparation process,
and a method for rapidly accomplishing
this at the site of the fuel preparation
has been developed.

A study 1s being made of the compounds
formed between fluoride melts and
their containers. Residues of these
corrosion products are being examined
analytically and spectrographically.

The systems UCl,-NaCl and UCl,-KCl
have been examined petrographically to
determine compound and eutectic compo-
sitions. Optical data are included
for the compounds 1n these systems.,
Phases present in the NaF-ZrF, -UF,
system have been studied petro-
graphically and by x-ray diffraction
in an effort to define the ternary and
pseudobinary phase regions and to
1dent1fy the compounds therein, Data
for the compositions studied to date
are included,

The bulk of work performed by the
service groups during the past quarter
involved chemical analyses and petro-

examination of fluorade
In addition, chemical
analyses were performed on a group of
test sections from the beryllium oxide
vs., NaK tests, and oxide determinations
were made on a number of chloride and
fluoride samples.

graphic
mixtures.,

CHEMICAL ANALYSIS OF REACTOR FUELS
AND CONTAMINANTS

J. C. White
Analytical Chemistry Division

A new volumetric method for the
determination of zirconium was de-
veloped. This method consists of
dissolving the zirconium salt of
p-chloromandelic acid 1n a nonagqueous
medium with excess sodium methylate
and back titrating with acetic acaid.
The reaction involved 1sstoichiometrac,
hence, no empirical factors are in-
volved.

A method for the determination of
uranium trifluoride and zirconium
metal, which are formed upon the
addition of zirconium hydride or other
simi1lar corrosion inhibitors, was
developed. These determinations can
be made i1n the presence of each other
because of the great differential
reaction rates of these two materials
with dilute hydrofluoric acid reagent.

The reactions of bromine trifluoride
with metallic oxides are being studied
1in an effort to measure trace amounts
of oxide i1n fluoride mixtures and also
to determine the completeness of
fluorination of zirconium oxide during
the preparation of the fluoride.

A method for the determination of
traces of hydrogen fluoride i1n helium
sweepings of fuel preparations was
developed in which the gas 1s trapped
in boric acid solution. The 1ncrease
in conductivity of the solution caused

171
ANP PROJECT QUARTERLY PROGRESS REPORT

by fluoboric acid formed by the
reaction between HF and boric acid 1is
measured and related as a function of
the concentration of hydrogen fluoride.
Concentrations of the order of 1 X 10°°
M HF can be determined in this manner,

The dissolution of Inconel con-
tainers, which have been corroded by
fluoride reactor fuels, was attempted
electrolytically in such a manner as
to separate metals from compounds. The

Zr(p-C1C4H,CHOHCOO), + 4NaOCH, —>

amount of the data collected 1s 1in-
sufficient to permit drawing any
definite conclusions.

Zirconium (C, K. Talbott, J. M.
McCown, Analytical Chemistry Division).
Of the several procedures explored for
the determination of zirconium, (!’ the
simplest and most accurate 1s a volu-
metric method employing mandelic acid,
This reagent 1s especially good for
determining zirconium 1n the presence
of uranium. It was decided to further
investigate reagents of the mandelaic
acid type, and a volumetric procedure
was developed employing p-bromomandelic
and p-chloromandelic acids. These
acids precipltate zirconium 1n acidic

solution 1n the form of zirconium
tetra p-bromo(chloro)mandelate, which

1s more easily filtered than the zair-
conium mandelate, (It 1s possible to
then dry this salt and determine zir-
conium gravimetrically,(2+3) but the
volumetric method 1s simpler to
execute, )

The salt dissolves 1n alkali with
subsequent hydrolysis of zirconium.
However, when an alkali 1s used 1n a
nonaqueous medium to dissolve the
salt, hydrolysis does not occur, On
this basis, a volumetric method for

(I)ANP Quar Prog Rep Dec 10,
1439, p 191

:z’n B Hahn, Anal Ches 23, 1259 (1951)

3)

R E OQOesper and J L. Klingenberg, Anal
Chea 21, 1509 (1949)

1952, ORNL-

172

determining zirconium has been de-
veloped 1n which the p-halomandelate
salt 1s dissolved 1n excess standard
sodium methoxide solution containing
3 parts benzene and 1 part methanol,
and the excess base 1s titrated with
standard acetic acid 1n a similar
benzene-methanol solution with thymol
blue used as the i1ndicator. The
equation for the reaction that takes
place 1s

4Na p-CI1C4H,CHOHCOO + Zr(OCH,), .

which appears to be

eliminates the need
for the use of empirical factors. The
Iimited availability of the reagent
has made necessary the synthesis of
p-chloromandelic acid. A method
proposed by Jenkins(*) has been used
for this preparation, a yield of 50%
1s obtained.

Reduction Products (W. J. BRoss,
Analytical Chemistry Division). The
addition of zirconium hydride to
inhibit corrosion i1n fluoride reactor
fuels has led to a study of the re-
duction products that appear as a
result of this addition, It 1s
postulated that some uranium tetra-
fluoride will be reduced to the tri-
fluoride and that zirconium metal will
be formed. A procedure in which 1t 1s
possible to determine both these
specles, uranium trifluoride and zir-

The reaction,
stoichiometric,

conium metal, 1n the same sample of
reactor fuel has been developed. The
procedure 1s based on the vast dif-
ferences 1n the rates of the following
reactions with 0.2 ¥ HF

Zr + 6HF —> H,ZrFg + 2H,, fast, (1)

2UF; + 2HF —> 2UF, + H,, slow, (2)
Reaction 1 proceeds quantitatively at

room temperature, whereas reaction 2

(4)g
(1931)

Jenkins, J Aa Chea Soc 53, 2341
exhibits no tendency to react under
this condition for some time. A study
of the rate of reaction 2 1s incomplete,
however, upon heating for 1hr at 90°C,
only 5.2% of the theoretical amount of
hydrogen 1s evolved. These results
1llustrate the stability of uranium
trifluoride i1n this medium. When a
mixture of zirconium and uranium
trifluoride was treated with 0,2 M HF,
the volume of hydrogen liberated was
equivalent to the amount of zirconium
present

Zr + 4HC1—> ZrCl, + 2H,,. slow,
4UF, + 4iCl —> 3UF, + UCl,
+ 2H2, fast.
Zirconium liberates 4.9% of the

theoretical volume of hydrogen upon
heating for 1 hr at 90°C in 9.6 M HCI.
Uranium trifluoride reacts gquanti-
tatively, as shown by Manning¢(%) 1in
his method for the determination of
uranium trifluoride. When a mixture
of the two 1s treated with 9.6 M HC1,
hydrogen equivalent to the total zair-
conium and uranium trifluoride present
1s liberated 1n approximately 20 min
at 90°C. This reaction can be ex-
plained on the basis of formation of a
complex 10on by zirconium with the
fluoride 10ns that are liberated by
hydrochloric acid attack on the uranium
trifluoraide.

A two-step oxidation reaction 1s,
1n effect, used to determine zirconium
and uranium trifluoride 1n the presence
of each other. A solution of 0.2 M
HF 1s added to the mixture and the
volume of hydrogen evolved from this
reaction, conducted at room tempera-
ture, 1s calculated as zirconium. A
solution of 9.6 M HCl 1s then added,
with boric acid, to complex excess
fluoride 10ons, and the volume of
hydrogen evolved upon heating 1s
calculated as uranium traifluoraide.

(S)D L Manning, W. K Miller, and R Rowan,
Jr , Methods of Determination of Uranium Tri-
fluoride, ORNL-1279 (May 25, 1952)

PERIOD ENDING MARCH 10, 1953

Preliminary results have been ex-
tremely promising. The effect of
large excesses of sodium, zirconium,
and uranium tetrafluorides on these
reactions 1s being studied.

oxides (J. E. Lee, Jr., Analytaical
Chemistry Division). Metallic oxides
react with bromine trifluoride(®)
according to the reaction

3M02 + 4BrF; —> 3MF, + 2Br, + 30, ,

where M represents a quadrivalent
cation. The metallic oxide content
can be determined by measuraing the
amount of oxygen evolved, hence, this
reaction 1s of potential interest 1in
determining the oxide content of
fluoride mixtures.

A schematic diagram of the apparatus
constructed for use 1n studying thais
type of reaction 1s shown in Fig. 14.1.
The reactor portion 1s designed so that
reaction pressures of up to the order
of 1 tsi1 can be permitted. The
remainder of the system consists of a
reagent transfer arrangement, in whaich
transfer may be made under vacuum, and
a high-vacuum system with the necessary
instrumentation for measuring the
oxygen liberated during the reaction.

Several samples of zirconium oxide
have been fluorinated with chromium
tri1fluoride 1n this egquipment at
temperatures ranging from 200 to 300°C.
Analyses of the reaction residues
1ndicate that the experimental con-
ditions employed thus far have resulted
1n fluorinations that are at least 80%
complete, Efforts are being made to
obtain quantitative fluorination by
means of such promoters as uranium
dioxaide,.

Hydrogen Fluoride (D. L. Manning,
Analytical Chemistry Division). A
possible explanation of the corrosive-
ness of reactor fuels that have been
treated with hydrogen fluoride 1s the
presence of the gas entrapped in the
solidified melt. In order to remove
this gas from the molten fuel, an

(6)1'[ J Emeleus and A A Woolf, J Chen
p 164-168 (1950)

Soc ,

173
ANP PROJECT QUARTERLY PROGRESS REPORT

RELIEF

GAGE
AN

~a—— REACTOR

Fig. 14.1.

outgassing-with-helium step has been
included 1n the fuel preparation pro-
cedure. It 1s desired to test the
helium off-gas for hydrogen fluoraide
and to monitor the gas stream. The
most attractive means of attacking
this problem appears to be the measure-
ment of the conductaivity of the

174

UNCLASSIFIED

DWG 18665
VACUUM
<+ SYSTEM
s —
s—TRAP
REAGENT

Apparatus for Studying Reactions with Bromine Trifluoride

scrubbing solutions through whaich the
gases are passed, This type of
analysis 1s rapid and has the added
advantage that 1t may be made at the
site of the experiment. Several
scrubbing solutions were studied, and
the most promising were calcium hy-
droxide and boric acid.
Calcium hydroxide reacts with hy-
drogen fluoride 1n the following manner

Ca** + 20H" + 2HF ——> CaF, + 2H,0

and reduces the conductivity of the
solution by removaing hydroxyl 1ons.
Approximately 100 pg of hydrogen
fluoride reduced the conductivity of
100 ml of 7 x 10°* N calcium hydroxide
solution by 7 X 10°® mho. A more
sensitive determination 1s possible 1f
boric acid 1s used as the scrubbing
solution. The reaction that occurs
between hydrogen fluoride and boric
acid 1s

H,BO, + 4HF —> H* + BF; + 3H,0 ,

1n which a strong acid (fluoboric) 1s
formed, therefore the highly mobile
hydrogen 1on 1s introduced 1nto the
system and the conductaivity 1s in-
creased. The concentration of the
boric acid solution should be very
dilute, 0.1 wt % or less, so that the
conductivity of the scrubbing solution
will be extremely low, of the order of
2 x 10°% mho. The conductivity of
100 ml of 3 X 10°3 M boric acid 1s
increased 13.5 X 10°% mho by the
addition of 100 ngof hydrogen fluoride
— a change nearly twice that produced
by the same amount of hydrogen fluoride
1n calcium hydroxide solution., These
results show that conductivity measure-
ments are extremely sensitive and well

PERIOD ENDING MARCH 10, 1953

suited for the purpose of determining
hydrogen fluorade.

Corrosion Products (D, L. Manning,
Analytical Chemistry Division). Ex-
periments have been conducted to
ascertain the effects of molten
fluoride salt mixtures on Inconel and
other metallic containers. In these
experiments, the metal container that
has been exposed to fluorides 1s made
the anode 1n a sulfuric acid-citric
acid bath, Since the metals exist as
elements, they are electrolytically
dissolved and thereby separated from
the compounds present, A similar metal
tube 1s dissolved for comparative
purposes. The residues from the anodic
dissolution have been analyzed chemi-
cally and spectrographically. The
analytical methods used to date have
been unreliable because of flaking of
metal particles during dissolution.
Refinements i1n the electrode system
are being made to eliminate this
difficulty. Spectrographic data have
been collected for Inconel specimens
that were subjected to the standard
100-hr static corrosion test with
NaF-ZrF,-UF,, with and without addition
of zirconium hydride. The results are
given i1n Table 14.1.

The amount of the data collected 1s
insufficient topermit drawing definite
conclusions. Several points of
The

interest are evident, however.

TABLE 14.1. SPECTROGRAPHIC ANALYSES OF RESIDUES FROM ANODIC DISSOLUTION OF
INCONEL USED IN STATIC CORROSION TESTS OF FLUORIDE REACTOR FUELS
COMPOSITION OF RESIDUE FROM DI SSOLUTION OF INCONEL (%)
REACTOR FUEL

Al Co Sh T1 Zr U
NaF-ZrF,-UF4 t ZrH, 01 01 >20 0 2 10 ND*

0 08 10 0 2 06 ND
NaF-ZrF,-UF, 01 20 <01 ND

0 08 01 10 0 2 <01 ND
Blank 0 08 01 03 0.2 <01

.None detected

175
ANP PROJECT QUARTERLY PROGRESS REPORT

si1licon content i1n the residue from
corroded Inconel 1s high and i1ndicates
that practically all the silicon in
the fuel 1s deposited on the walls of
the container, the addition of zair-
conium hydride results 1n the deposition
of some zirconium on the container
walls, and deposition of uranium does
not occur 1n either case,

PREPARATION OF URANIUM TETRACHLORIDE

W. J. Ross
Analytical Chemistry Division

The preparation of several kilograms
of uranium tetrachloride to be used in
phase studies by the ANP Chemistry
Group has been undertaken. The
tetrachloride 1s being prepared by
chlorinating UO; with hexachloro-
propene, according to the directions
given by Pitt et al ¢7)

PETROGRAPHIC EXAMINATION OF FLUORIDES

G. D. White, Metallurgy Division
T. N. McVay, Consultant

Routine petrographic examinations
of more than 800 samples of fluoraide
mixtures were made 1n connection with
fuel i1nvestigations. Petrographic
examinations were made of compositions
1n the UC14-NaCl and UC1, -KCl systems.
The data obtained from these exami-
nations were correlated with the
thermal data to locate compound and
eutectic compositions,

UC1 ,-NaCl System In the UC1,-NaCl
binary system, two compounds were
found 2NaCl-UCl,, and a compound with
more than 50% UC1,, possibly NaCl-2UCl1,.
The latter compound seemed to have an
incongruent melting point, since
compositions on the high UCl, side of
the binary always contain considerable
amounts of free UC],.

Optical Data.
Na,UCI,

Pale green to colorless

(7)3 M Paitt et al , The Preparation of TCI4
with Hexachloropropylene, C-2 350 3 (July 27,
1945)

176

Uniaxial negative (tetragonal or hexagonal)
Refractive indices

E, 1 652

0, 1 664

NaU,Cl,
Yellow-green
Biaxial negative (probably orthorhombic)
with small optic angle
Refractive indices
a, 1 790
¥, 1 850

ucl,-KCl system. The UCI, -KCl
system produced three compounds
2KC1-UCl,, KC1-UCl,, and an 1ncongru-
ently melting compound in the 67% UCI,
region of the binary. There were two
forms that crystallized in the compo-
sition range of 30% to 35% UCl,. They
were distinguished by their differences
1in birefringence and interference
figures.

Optical Data
K,UCl
1 Pale green to colorless
Uniaxi1al negative
Refractive indices
E, 1 644
0, 1 654
2 Pale green to colorless
Biraxial positive, small optic angle
Refractive indices
a, 1 636
v, 1 640
KUCI ¢
Pleochroism
X, grey
Z, blue-green
Biaxial positive, small optic angle

Refractive 1indices

a, 1 692

B, 1 705

v, 1759
KU,Cl

Yell ow-green
Biax1al positive, with very small optic angle
Refractive i1ndices

a, 1 740
B, 1 809
v, 1 820
X-RAY DIFFRACTION STUDIES

P. A. Agron
Materials Chemistry Division

The determination of the phases
present 1n solidified melts of various
compositions 1n the NaF-ZrF,-UF, system
has been helpful i1n elucidating the
paths of crystallization. The regions
of this ternary system that are rich
1n NaF have been examined during the
past quarter. Some progress has been
made here 1n definition of the ternary
and pseudobinary phase regions.

The petrographic and x-ray dif-
fraction 1dentification of the phases
found i1n the solid, upon melting given
compositions, are listed in Table 14.2.
The phases have been grouped according
to the similarities present in the
respective solidified melts, thus A-1
and A-2 represent one ternary area,
B-1 and B-2, the adjacent ternary
area, etc, This division i1into ternary
areas 1s 1n fair agreement with the
1sothermal contour plots (to be
published)., However, the anomalous
situation of four phases 1n samples
B-1 and D-2 obviously requires further
investigation, Additional compositions
of these ternary areas should also be
1nvestigated before mapping the phase
boundaries.

The examination of a number of
fused solids along the composition line
joining NaUF, and Na,ZrF, appears to
1indicate a pseudobinary behavior.
Approximately 90% of the x-ray daf-
fraction pattern can be assigned to
these two phases. The cooling curves
of a series of these compositions (to
be published) i1ndicate a binary
eutectic at about 22.5 mole % UF,.

The behavior of salts along the
composition line between NaUF_; and
NaZrF, 1s rather complex beyond 6
mole % UF, content, ®) Further study
along this composition line, with the

(B)P A Agron, X-ray Studies of Phase Segre-
gation in Pseudo NaZrF_-NalUF_ Binary as a Function
of Heat Treatment in Radiation Stud:ies, ORNL
CF-53-1-332 (Jan., 20, 1953)

PERIOD ENDING MARCH 10, 1953

use of liquid sampling and quenching
technigues, wi1ill be required to ex-
plain this area of the phase diagram.

SERVICE CHEMICAL ANALYSES

H. P. House S. A. Reed
A. F. Roemer, Jr.
Analytical Chemistry Division

Work in the ANP Analytical Chemistry
Laboratory during the period consisted
primarily of the analysis of alkala-
fluoride eutectic mixtures and zir-
conium fuels. Although there was a
slight decrease in the total number of
samples received during the quarter,
there was a significant i1ncrease 1n
the number of miscellaneous materials
recelved. Important among these was a
group of samples that were tested for
beryllium content. Included 1in the
group were large steel sections of
test rigs used to study the compati-
bility of NaK and beryllium oxide.
The specimens were first treated with
methanol to remove excess NaK and then
leached with hot, 50% sodium hydroxide
solutions to dissolve the beryllium
ox1de adhering to the surfaces of the
specimens. The beryllium content of
the alcoholic and caustic solutions
was determined colorimetrically, wath
p-nitrobenzeneazo-orcinol as the
chromogenic reagent,

A number of uranium tetrachloride
samples were analyzed for UO, and
UOCl,. Separation of UO, from the
salt was effected by refluxing portions
of the material with dilute ammonium
oxalate solution. The 1nsoluble UO2
was removed by filtration, dissolved
1in acid, and estimated quantitatively
by potentiometric titration with ferrac
sulfate. The UOCl,, along with any
UO,, was separated from the UCl, by
refluxing with anhydrous methyl
acetate. The 1nsoluble residue was
dissolved 1n dilute nitric acid, and
the UOCl, was calculated from a
volumetric determination of the
chloride content,

177
8L1

TABLE 14.2.

TERNARY PHASE REGIONS IN THE NaF-ZrF,-UF, SYSTEMS

CHEMICAL PETROGRAPHIC EXAMINATION!®) X_RAY SPECTROMETER DIFFRACTION ANALYSIS
SAWPLE | cowposITION EXPERIMENTAL CONDITIONS!®)
(mole %) Phases Refractive Index Intensaty of Strongest Line Phaaes
Al 55 NaF Salts fused under helium ges Nan(U)Fs 1 504 90 Nal(Zr)F,
20 UF, and starred until frozen NayZrF, 1 43 40 Na,ZrF,
25 ZrF, uo, (1 te 2%} 35 NaZr(U)F,
A-2 55 NaF Salts fused under helaum gas NaZr{(U)F, 1 504 80 Nal(Zr)F,
25 UF, and stirred untal frozen Nazszfi 1 43 50 N‘zz'Fs
20 ZrF, vo, (1 to 2%) 35 NaZr (U}F,
B1 62 5 NaF Salts equalibrated in closed NnBZr(U)FT 1 39 75 NaSZr(U)FT
18 75 UF, nickel container and cooled NazUFs 1 49 75 NuU(Zr)FS
slowly without stirring (e}
18 75 ZrF, Na,Zr(U}F, 1 43 20 By Na,UF ¢! (also ¥ NayUF,)
Ne,ZrF, 143 25 Na, ZrF,
B-2 62 5 NaF Salta fused under helium and NaZr(U)Fg 1 504 60 NalU(Zr}F,
20 0 UF, stirred untal frozen Na3Zr{U)F, 1 386 30 NayZr(U)F,
17 5 ZrF, Na,ZrF 1 43 30 Na,ZrF,
cl 66 5 NaF Saltas fused under helaium and B85 NaUF; and Nal{Zr)Fq
22 5 UF, stirred until frozen 40 ¥ Na,UFg
(also ‘Bl and a NazUFG)
11 0 ZrF, 30 NayZr(U)F,
cz2 60 0 NaF Salts fused under helium and NaZr(U)Fg (also 1 50 90 NaUFg
35 0 UF stirred unt1l frozen lower index 40 Na,UF
4 material) By NagUF
(also ¥ =and B; NayUFy)
5 02rF, 20 NeyZr(U)F,;
c3 67 5 NaF Salts fused under helium and Green phase 146t 0 0] 90 B3 NayUF,
27 5 UF, stirred untal frozen U0, (~2%) 15 NayZr(U)F,
50 ZrF, 10 NaU(Zr)F,
D-1 72 5 NaF Selte equilibrated in closed NayZr(U}F; (also 139 to 1 40 90 NayZr{U)Fy
13 15 U nickel container and then an internal green 3 Na,UF
Fu cooled slowly without starring phase) s '8’ nzUFs
13 75 ZrF, (slso ¥ and a Na,UF,)
10 NasUF,
D-2 72 5 NeF Salts fused under helium and NaaU(Zr)F., 1 413 90 NagU{Zr)F,
stirred until frozen
25 0 UF, NaF 20 Na,UF,
2 5 ZrFy U0g (1 to 2%} 12 NaF
5 Nay2c{U}F,
E 1 75 NaF Salts fused under helium and NajyZr(U)F; 106 NayZr(U)F; and NajU(Zr)F,
starred until frozen
15 UF, NaF
10 ZcFg U0, (~2%) 10 NoF
E 2 82 5 NaF Salts equilibrated in closed NayZr (U)F, 139 to 1 40 90 NeayZr (U)Fy
nickel conteiner and then
8 15 UF, cooled slowly without stirring NoF 75 NagU(zr)Fy
9 75 ZrF, 12 NoF
(a)

(&)

Samples prepaered by C J Barton's group

Petrographic examination by G D White, Metallurgy Division

(c

)fi3-N02UF6 18 a hexagonal form that has not been previously reported,

LU0d3Y SSHYO0Ud ATHALUVAD IJAL0Ud ANV
PERIOD ENDING MARCH 10, 1953

TABLE 14.3. SUMMARY OF SERVICE ANALYSES

NUMBER OF NUMBER OF
SAMPLES DETERMINATIONS

REPORTED REPORTED
ANP Reactor Chemistry Group 508 4477
ANP Experimental Engineering Group 216 2424
ADP Process Improvement Group 41 97
Heat Transfer and Physical Properties Group 6 86
Electromagnetic Research Division 2 15
General Electric Company 11 114
Total 184 7213

Comparative tests of phenylarsonic
acid and mandelic acid as precipitants
for zirconium were completed during
this period. Spectrographic exama-
nation of zirconium oxide residues
resulting from the agnition of pre-
cipitates of phenylarsonic acad
revealed the presence of significant
amounts of arsenic, antimony, and
1iron. Zirconium oxide from mandelic
acid precipitations was found to con-
tain only traces of silicon. The use
of mandelic acid as the precipitant
resul ted 1n much better separations
from interfering 1ons and, i1naddition,
a considerable saving 1n time. By the
use of this reagent during the latter
part of the gquarter, a marked reduction
in the number of duplicate determa-
nations was effected, and a consider-
able i1ncrease 1n the accuracy of this
determination was attalned.

Of the 784 samples reported during
the period, 65% were submitted by the
ANP Reactor Chemistry Group and 27% by
the Experimental Engineering Group.
Over 7000 determinations were made
during the quarter, a fourth of which
were made on nonroutine samples.
Summaries of the ANP analytaical
chemistry work for the gquarter are

given 1in Tables 14.3 and 14.4.

TABLE 14 4. BACKLOG SUMMARY
Samples on hand, 11-10-52 108
Number of samples received 808
Total number of samples 916
Number of samples reported 784
Backlog, 2-10-53 132

179
Part 1V

APPENDIX
REPORT NO

CF-53-1-111
CF-53-1-148

CF-53-2-159

CF-52-12-153

CF-53-1-84

CF-53-1-260
CF-53-1-276
ORNL-1461

CF-53-1-64

CF-53-1-267

CF-53-1-317
Y-881
CF-53-2-50

CF-53-2-99

ORNL- 1493

CF-52-12-109

(no number)

MM- 54

CF-53-1-83

CF-53-2-79
ORNL-1463

15. LIST OF REPORTS ISSUED DURING THE QUARTER

TITLE OF REPORT

I General Design
Minimum Weight Analysis for an Air Radiator

Component Tests Recommended to Provide Sound
Basis for Design of Fluoride Fuel Reactors
for Tactical Aircraft

ARE Design Data

AUTHOR(S)

S Farmer

A P Fraas

W B. Cottrell

II Experimental Engineering

An Improved AC Electromagnetic Pump Cell
ARE Regulating Rod

Moore Pressure Transmitter Test Summary
Components of Fluoride Systems

A Simple Electromagnetic Flowmeter for Liquad
Metals

III Reactor Physics

Delayed Neutron Damping of Non-Linear Reactor
Oscillations

Delayed Neutron Activity in a Circulating
Fuel Reactor

Delayed Neutron Activity i1n the ARE Fuel Element
A Graphite Moderated Critical Assembly

The Attenuation of Capture Gammas 1n a Plane
Limited Medium of Finite Thickness

Heating by Fast Neutrons in a Barytes Concrete
Shield

General Method of Reactor Analysis Used by ANP
Physics Group

IV  Metallurgy and Ceramlcs

Spot Welding of Stainless Clad Fuel Elements

Formability and Weldability of Vapor-Deposited
Mol ybdenum Final Report

Progress Report The FlashWelding of Molybdenum
Part I - Temperature Distribution During the
Flashing Cycle

A High-Temperature Cermet Fuel Element
Ceramic Fuel Element Radiation Tests

Methods of Fabrication of Control and Safety
Elements for the ARE and HRE Reactors

.Nunber assigned by ANP Reports Office

M E LaVerne

E R Mann

S H Hanauer

P W Taylor

W B Cottrell
R M Carroll

W K Ergen

H L F Enlund
H L F Enlund
E L Zimmerman
F Abernathy

H L F Enlund
F Abernathy

H L F Enlund
C B Mills

G M Slaughter

Massachusetts
Institute of
Technol ogy

Rensselaer
Polytechnic
Institute

R Johnson
Johnson

R
H Coobs
S Bomar

o BN R

DATE ISSUED

1-31-53
1-16-53

2-18-53

12.7-52
1-9-53

1-22-53
1-27-53

to be 1ssued

1-7-53

1-27-53

1-27-53
12-7-52
2-6-53

2-11-53

to be 1ssued

12-9-52
10-31-52

12-5-52

1-9-53
2-9-53

to be 1ssued

183
ANP PROJECT QUARTERLY PROGRESS REPORT

REPORT NO

ORNL-1491

CF-52-11-205

CF-52-12-124

Y-B4-59

CF-53-1-233

CF-53-1-248

CF-53-2-56

CF-53-2-84

ORNL-915

CF-52-11-129
CF-52-12-209
CF-53-1-279
ORNL-1438

ORNL-1471
ORNL-1479

ORNL-1430

Y-B31-390

ORNL- 1453

Y-B32-104

184

TITLE OF REPORT

Corrosion by Molten Fluoraides

AUTHOR(S)

. S Raichardson

D C Vreeland

¥ D Manly

V Heat Transfer and Physical Properties

Estimates of Heat and Momentum Transfer
Characteristics of the Two Fluoride Coolants

(L1F-48 M%, BeF-52 M%) and (NaF-10 M%, KF-46

M%, ZrF,-44 MZ)

Generalized Velocity Dastribution for Turbulent
Flow 1n Annula

Selected Physical Properties of Potassium and
Potassium Hydroxide in the Temperature Range
100 to 1000°C A Literature Search

Measurement of the Thermal Conductivity of
Fluoride Mixtures No 14 and No 30

Remarks on the Falling-Ball Viscometer
Heat Capacity of Fused Salt Mixture No 30

Heat Capacity of Sr(OH),

Forced Convection Heat Transfer i1n Thermal
Entrance Regions

vVl  Shielding
Materials Research for Mobile Reactor Shieldaing
Thermal Neutron Dose at the Crew Compartment
The Tower Shielding Facilaity

Determination of the Power of the Bulk Shielding
Reactor Part 2
Shield Optimization

A Proton Recoil Type Fast Neutron Spectrometer

VII. Chemistry

Modifications of the Dimethylglyoxime Method
for the Colorimetric Determination of Nickel
Based on the Use of Potassium Persulfate as

the Oxidant

Analytical Chemistry-ANP Program Quarterly
Progress Report for Period Ending November
25, 1952

Dissolution of NaK

Minutes of Committee Meeting for the Coordination

of Hydroxide Research, Fourth Meeting,
December 10 and 11, 1952

H F Poppendiek

W B Harraison

J 0O Bradfute

R V Bailey

R L Curtas

S J Claiborne

R F Redmond

S I Kaplan

W D Powers

G C Blalock

W D Powers

G C Blalock

W B Harraison

E P Blizard

J L Meem

E P Blaizard

E B Johnson

J L. Meem

E P Blizard

R G Cochran

K M Henry

M L Druschel

0O Menas

R BRowan, Jr

Analytaical
Chemistry
Division

J C White

C K Talbott

L J Brady

F Kertesz

DATE ISSUED

to be 1ssued

11-29-52

12-19-52

9-23-52

1-8-53

1-14-53

2-6~53

2-9-53

to be 1ssued

11-18-52

1-27-53

to be 1ssued

to be 1ssued

to be 1ssued

11-3-52

11-20-52

12-2-52

1-7-53
-y

REPORT NO

ORNL-1376

ORNL- 1490

ORNL-1495

Y-B32-103
CF-52-12-178
CF-53-1-129
ORNL-1476

ORNL- 1500

Y-B31-403

ORNL-1439

CF-53-2-126
CF-53-2-246

PERIOD ENDING MARCH 10, 1953

TITLE OF REPORT

Mass Spectrometer Investigation of UF,

General Information Concerning Fluorides, A

Literature Search, Suppl to ORNL-1252
General Information Concerning Hydroxides, A
Literature Search, Suppl to ORNL-1291

Possible Coolants for Solid Fuel Reactors
Hydroxide Systems
Fused Salt Compositions

An Indirect Colorimetric Method for the Deter-
mination of Uranium

Recovery of Uranium as a Single Product from
Fluorides

VIII Miscellaneous

A Guide for the Safe Handling of Molten Fluorides
and Hydroxides

ANP Project Quarterly Progress Report for Period
Ending December 10, 1952

Objective and Status of ORNL-ANP Reactor Program
ANP Information Meeting of February 18, 1953

AUTHOR(S)

L O Gilpatraick
Russell Baldock

J R Sates

M E Lee

M E Lee

W R Graimes
M D Banus

C J Barton
J C White

D L Manning
C F Coleman

Reactor Components
Safety Committee

W B Cottrell

R C Briant
W B Cottrell

DATE ISSUED

8-29-52

to be 1ssued

to be 1ssued

11-25-52
12-23-52
1-15-53
1-13-53

to be 1ssued

1-12-53

1-15-53

2-13-53
3-2-53

185
-1

CHART OF THE TECHNICAL ORGANIZATION OF THE AIRCRAFT NUCLEAR PROPULSION PROJECT
AT THE OAK RIDGE NATIONAL LABORATORY

MARCH ,1953

Thi maseri | contai | formerien slfacting the aetlan-

14 hense [ the United Stata withl the maenl g §

the pronegelw Til 18 USC.,

T94 the monami o or  vel Hon of whi h i eny
% an umeutiori od porson | pechibiled by lew

ANP PROJECT DIRECTOR
R C. Briont RD
D Hilyor * Sec ANP
ASSOCIATE DIRECTOR FOR ARE
H Buc RD
ASSISTANT DIRECTOR
FOR COORDINATION
A J Miller ANP
TECHNICAL ASSISTANT
¥ B. Cottrell ANP
J M Ciser* ANP
P Horman * Sec.  ANP
ANP LIBRARY
ASSISTANT TO DIRECTOR Burlding 9704 |
W Cordwall RD
L i Cook ANP M. Brown RD
P Horman  Sec. ANP E Certer RD
E Webster RD
i
STAFF ASSISTANT FOR PHYSICS A 1 Maller ANP STAFF ASSISTANT FOR DESIGN ARE PROJECT ENGINEER 1 H Buck RD, STAFF ASSISTANT FOR METALLURGY STAFF ASSISTANT FOR CHEMISTRY
W K Ergen ANP D Hulyer * Sec ANP A P Froos ANP E S Betis ANP P Hormen Sec ANP R C Briom RD W D Monly " ¥ R Grimos WC
SHIELDING RESEARCH REACTOR PHYSICS RADIATION DAMAGE GEMERAL DESIGN ARE OPERATIONS EXPERIAENTAL ENGINEERING HEAT TRANSFER AND PHYSICAL METALLURGY CHEMWISTRY
uilding 4500 Building 9704 } Building 3025 Building 5704 § Building 7503 Building 9201 3 PROIEE“RIZIES g%fimcn Bunlding 2000 Building 9733 3
1lding
E ff"',j:fil § W K. Ergen ANP D S Billingron 55 A P Fraas ANP E S Botnis ANP H W Sovege ANP H F Poppendisk REE ¥ D Menly M W R Grimes MC
LID TANK F H Abemath ANP R. G Affel ANP G M Adomson M J Q. Brodiute REE E S. Bomar M P A Agron [
Bu1d g 3000 R B Boveyos AN £ D Bevemrn % R W Bussard | ANP A Crsty Eu 1P Bloksly MC § L Caherr REE W H Budges W C J Barten ue
10 Flynn p C. 8 Mlls ANP W E Brundoge S H. L F Enfynd [ G W Eckerd EM G D Brody M W S Farmer REE J ¥ Cathcort L F F Biankenship NC
F G Prohammor* ANP R M Cerroll 55 B E Hill Yz H. L F Enlund* P 4 Cisar ANP D C. Hamiton REE J H. Caobs " C M Biood MC
T V Blosser P R Willioms Sec ANP A F Cohon 5 M. E LaVerna ANP B L Greonstroat ANP ¥ G Cobb ANP H ¥ Hoffman REE D A Dougles M R A. Bolomey WC
G. T Chapman P C P Coughlen ANP J 1 Lai REE LD r ] F P Bood MC
I W Mller P W W Dovis 5 WL Anp B Holuchan & P F E Lysch REE E E Heft u L M. Bratche MC
2L Srome GE COMPUTERS M. J Feldman sS ¥ L Scont ANP H. Homyg GE R E Engberg I;R, E b REE £ Hoffman " v : atcher
F N oo A Forbes ANP N E Hinkle ss H ) Stumpf ANP J' P Jackson EM W C George 2 L D Falmer REE H lnouye B v Colemon s
atsen . Teagoris ANP G ¥ Koilhaltx 58 A, G Johnson GE A G Grindell AHP D Powers A Cuneo
K J Kelly =" CONSUL TANTS ER * I R. Helten ANP T Sefton Sec. REE R B Oliver M D D Davies 11
TECHNICIANS P R Klem 55 E B Perrn ANP W R Huntley ANP TECHNICIANS P Patriorco M J E Eargon
W 3 Moynord P 3 G Morgon ed A H Fox Union Callege D Seart ANP E M Less ANP G. M. Sloughter M E E Ketchen MC
J N Mocay P T bor 5 R L Mexwall Univ Tem H L Wy LRP R E MacPherson ANP C.G Blalock REE G P Smth M R P Metcolf NC
gan G, F Wislicenus Johns Hopkins Univ LA ANP R M Burnent REE D C Vrestand I RE . MC
BULK BUELOWG REACTOR - g":' s§ 4 H Myld Reuction Mators Inc § B it e R M. Mazon EM S J Cluibor REE 3 W Woods M L G. Overholsar uC
Bu 19 3 301D CRITICAL EXPERIMENTS A E porgnsen 3 E Wrachhusan EW ¥ B McDonald ANP T N Josss REE J Thomas Sec u H.S. Powers MC
[ Building S113 HE R;enm 3 M, Milligan, Sec. ANP G. N, Neasle uC 3 Lomes REE J D Redman NC
4 L beom P W T Rebwson 55 ¥ R Oshomn ANP G M ¥um REE TECHNICIANS R. I Sheil uC
R G Cochron P A D Callhan P 0 S:sman 3 CONSUL TANTS G. F Petersen ANP G D Brady W B J Stum MC
M. P Haydon [ F W Smh oy ¥ R Chombers Unv T D F Solmon ANP J T Eost M R E Thoma Jr MC
K. M. Hanry P — R C Keen P ¥ J Stwm b 1 W Walker Unre Tern P G. Smith ANP G i Gonzaler " L E Topol MC
H E Hungerford P 4 J Lo P L C Templeten s$ A L Sow ANP L L Hall M R.E Trober MC
E B Johnson p J W Noaks P 1B Trice 55 P ¥ Toylo ANP J D Hudson M D E Coldwell Sec. MC
J K Leshe P D VP Wilioms P C C. Wobster 55 J . Trummel ANP Building 9766 R. W Johnson M
G M MeC p E L Zimmerman p W C, Tunnell ANP ™ ¥ G Lare M CORRGHION LABORATORY
ommon W R Willis sS
W Pagle cy B.5 Davia Sec Yn 1 C Wlson 55 D R Ward ANP 1M Warde " B, McNobb " By ldung 9764
E Rolor cy 3 C Zukas I REACTOR CONTROL J F Warner* Y12 J E Pope "] F Kertesz MC
R Bullard Sec P Budlding 4300 G D Whitman ANP C E Cums* M C. E Schubort " H I Buttrom NC
TECHNICIANS e D Alexonder Sec ANP L & Doney M L R Trorter " C R Coh
TECHNICIANS E P Epler* D Storay Record Clerk  ANF S D Fulkereon* ] C K. Thomas M E Moodows
K Honeycuti P C Ells §S D Homis Stanc. ANP J R Johnaon* M C W Weaver M F A Knox MC
J L Hll P 6 MEV VAN DE GRAAFF CROSS SECTIONS F M. Blocksher 5§ A E G Bates* Ic A J Taylar® M MY Smih NC
D J Kby P Budlding 5500 T E Cale* Ic TECHHNICIANS G D White* M METALLOGRAPHY
R M. Simmons P HB Willard P 861 CYCLOTRON IRRADIATIDNS F P Groen* P — 1S Addison ANE TECHNICIAN R S Crouse M CONSULTANTS
J K Bau* p Building 92012 $ H Honauer* P G 5. Chilton AN J A Goff M T K Roche [ J . Carter
TOWER SHIELDING FACILITY ’ ER P 4 M Coburn ANP e T R P Gibb Willioms Coll
J D Kington* P R § Lwvingston* ER n T E Crobt ANP TECHNICIANS \ clloge
C E Chiford P P H Statson p J B Ruble* P obtres D G H:ll Duke Univ
— A L Boch® ER 1 J Stone* P J R Croley ANP CONSULTANTS WD Allen M
T ¥ Blosser P ENGINEER o ER L C. Ookes* P J Didiake MC T M, McVay Umy Al N M, Archioy M CONTRACTORS
J Y Estabrock P R ¥ Lamohere* P R J Jones C § Walker* RP F A Doss ANP T Ot Sren E R Boyd M Baitell i
L B Holland P phere R L K‘“fh' E: J R Duckworth ANP T S Shevlin Ohio Srate Univ D H Brawor o tolle Memecrial Institvte
M K Hullings P TECHNICIAN F H Nedl T L Grogory ANP W H. Farmer " 'JR W Clegg
¥ T Hawton Ic F M, Grizzell ANP R. L, Fitzgerald* M on
RADIATION $PECTROSCOPY RESEARCH CONSULTANTS B L Johnzan ANP J C Gowsr " R J[:m'
FC Mo H. G MacPherson Mational Carbon Co  Inc J W Kingsley ANP E. P Griggs M H, A, Pray
renschein P L P Smith Cotnall Unsv DE M:Eny ANP R M ¥ollace M ond others
T A Lave P D F Weekes Texos A& M G E Milts ANP Metal Hydrides |
F J Mucksnthaler P J 1 Porsons ANP CONSULTANTS l:.D B;n.‘ ne
M, A, Redden ANP N J Gront MIT v
msonfl:f:: Aavsts F J Schofer ANP J L Gregg, Cornell Univ and athers
F H Mooy NOTE  Thiz chart shows only the lines of technrcal coordination of the ANP project R. G, Wiley ANP 5 'F: ;‘;:P“ :g: Universaty of Arkansas
A Simon P The various sndividuals ond groups of paople hated are engoged mither wholly o port CONSULTANTS EC '"::: Univ Alo :’ ¥ Gugoneff
H E Stern cv tima on resoorch ond design which ts coordinotad for tho benofit of the ANP propct in J Smathers
L Gordon Sec P J F Boiley Univ Tenn CONTRACTORS ond oihers
the manner indicated on the chart Eoch group however s dlso responmible 1o its ¥ R Ctombers Umyy Tean.
CONSUL TANT Diviscon Director for the detoiled progross of 115 research and for admimistrative motters J F Haines Boldwm-Lima Hemilton Corp
H A Bethe Comell Uniy ¥ K Stawr Univ Tom. F G Tomall
he!
CONTRACTOR The key to the abbraviations used 13 givon balow It should be ooted that several out and athers AMALYTICAL CHERISTRY
Nucloor Development Associotes Inc 2ide argamizations hove parsonnal participating in the GRNL—ANP progrom. Battelle Memonic! Institute Butlding 9733 2
{AEC Controctor)
H Goldstain AC Analytical Chomistry Division R K Porke C.D Suaano AC
R Aronaon ANP Aurcraft Nuclear Propulsion Division H Sall:
- C Chemistry Divizion Z Schoheald H P House* AC
cy Consolidated Vultee Aircraht Carp. and others i-? ?{'d » :g
EM Enginwering and bowdenonce Division Gerity Michigan Ca, J C. White AC
ER Eloc Ressorch . gy e d others
romagnetic 1eision G. Geoff an s o aC
GE Ganeral Eleciric Ca and others B. Young,
tc Instrumemdation and Cortrols Division CONSUL TANT
Massachysatts Inahivite of Tochnol
LRP Long Range Reactor Planning Geoup 7 Walff oar H H Willord Univ Much.
M Metollurgy Division and ofhers
NC Materials Chermatry Division SPECTROGRAPHIC ANALYSS
P Physics D vision Renzssloer Folytechnic nshitute IR IlhN“Blrild'nm9‘1‘.!-1 o
P Pratt and Wnztney Asrcrott Division E F Nippos AN v . st
RO Resvarch Derector s Doportmen: and othars and ;Th‘:rl
REE Reoctor Exparimental Enginesring Divizien Superior Tube Co.
MASS SPECTROMETRY
v el = EiT
olid Jtate Divikion
Y12 Corbide and Corbon Chemscals Co. {¥ 12 Site} Commonwealth Enginsering Ca, of Ohio C. R. Boldeck 1]
*Port tuer M. J Hileo

i87