G

M

 

n

    

a

 

.
I
|

 

 

 

 

     

) ———————

o= . 7
= m £Fy W &
o Smeretstt— i

00 F——et— & ;
—1 SRR

& a— o0

-~ e 3

& e N

i==

= 0
==

= S w0

= =r

xs

2 m

 

 

 

 
 

i Ch - i D e
e U?“:itm} \;»ta‘tes u? Am%n'\,a Avaitaoie from
"\

Y e
SVICS

;i@%d Virgima 221671

O P S~ - 4 L. g KL n 3
Price. Printed Copy $4 06, Muicrofiche $3.00

This 7 ep 1t was orgpared as an account of waork Q’)CflSOfes‘j oy the Urniteg Siates
Governm cither the United States nor the Dnergy Hasearch and Developmend

  

Admmiinistration/United States Nuclear Fogulatory Commission nor any of iher
employees, nor any of their contractors, subcontractors, oo ther emplovess, makes
any warranty, express orimphied, or assumes afl\/1893”!&1{}1;!{)’0(:"i‘:‘spOHS‘b!My fortnhe
accuracy, compiateness or useltulnoss of any information apparatus pmduct 01
proce5° dISC'OS\,d or represents that its use would naot '

    

 
ORNL/TM~-5782

Contract No. W-7405-eng~26

METALS AND CERAMICS DIVISION

The Corrosion Resistance of Type 316 Stainless Steel to LijzBeFy

J. R. Keiser, J. H. DeVan, and D. L. Manning

Date Published -~ April 1977

NOTICE This document contains information of a preliminary nature,
it is subject to revisiotvor-correction and therefore does not represent a
final report.

OAK RIDGE NATIONAL LABORATORY
Oak Ridge, Tennessee 37830
operated by
UNION CARBIDE CORPORATION
for the
ENERGY RESEARCH AND DEVELOPMENT ADMINISTRATION

OAK RIDGE NATICNAL LABORATORY LIBRARIES

AR

3 4456 0551078 3

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 
ABSTRACT . .
INTRODUCTION .

EXPERIMENTAL METHODS
EXPERIMENTAL RESULTS

CONCLUSIONS
ACKNOWLEDGMENTS

-

®

TABLE OF CONTENTS

iidi
 

THE CORROSTION RESTISTANCE OF TYPE 316 STAINLESS STEEL TO Li,BeF,

J. R. Keiser, J. H. DeVan, and D. L. Marming1

ABSTRACT

The corrosion rate of type 316 stainless steel in
molten LiF-BeF, (66-34 mole %) has been measured in a thermal-
convection loop operating with a maximum temperature of 650°C
and a temperature difference of 160°C. The corrosion rate
was correlated with the concentration of impurities in
the salt and with the fluoride ion oxidation potential as
determined by an on-line voltameter.

A corrosion rate of —10 um/year was observed initially
in the as-received salt. This rate decreased as reactions
with initial salt impurities went to completion.  Direct
addition of beryllium metal to the salt further reduced the
corrosion rate.

 

INTRODUCTION

The selection of the‘first»wall and blanket materials for the first-
generation TOKAMAK-type magnetic fusion reactors will bépdetermined by
a combination of neutronic, radiation damage, and chemical requirements.
The technological problems associated with fabrlcatlon, weldlng, and
adequate production of the refractory metals nloblum, vanadlum, dnd
molybdenum have resulted in type 316 stainless steel being glven strong
con51derat10n for the fusion reactor first-wall material. The choices
for a blanket capable of breedlng tritium are limited to llthlum or
a lithium-bearing compound, but‘this blanket material must also be
chemically compatible with theyfiist wall. Because the extent of corro-
sion of stainless steel by liquid lithium may impose a severe temperature

limit, strong consideration is being given to the use of a mixture of

 

1'Analytical Chemistry Division.
the salts LiF and BeF, for a blanket of a fusion reactor with a stainless
steel first wall. The purpose of this paper is to report on the corrosion
resistance of type 316 stainless steel to LijsBeF,.

There are potential problems associated with the use of LiF-BeF»
as a combined coolant-breeding material.? These include the effective
breeding and the efficient recovery of tritium, the effect on the salt
of induced electric fields and of transmutation products, and the compati-
bility of the salt with the component it contacts. The tritium breeding
ratio of this salt is not as high as is that of lithium, and some designs
would probably require the addition of a mneutron multiplier such as
beryllium.

Another problem could be the electric field induced in the conducting
salt when it flows through the intense magnetic fields required for plasma
containment. This potential difference occurring between the salt and
the pipe wall could be great enough to electrolyze the salt and thus
cause severe corrosion problems for the metallic container. Means of
reducing the effect of the induced field to a tolerable level have been

3 According to him, electrolytic corrosion could

suggested by Homeyer.
be minimized by avoiding, as much as possible, high fluid velocities
perpendicular to the magnetic field. Solution of this problem will
most probably have to be a result of engineering design changes rather
than through additions or modifications to the salt.

An additional concern will be the corrosive effect of the salt on
the stainless steel. As seen from the free energies of formation in
Table 1, LiF and BeF,; are much more stable than the fluorides of the
major constituents of stainless steel, Fe, Ni, and Cr. Consequently,
no significant reaction of the salt components with the metallic container
is expected; however, impurities in the salt may cause corrosion. Impuri-
ties expected to be present are HF and NiF,, and these will cause

reactions of the type:

 

’W. R. Grimes and Stanley Carter, "Molten Salt Blanket Fluids in
Controlled Fusion Reactors." pp. 16190 in The Chemistry of Fusion
Technology D. M. Gruen, ed. Plenan Press, New York, 1972.

W. G. Homeyer, "Thermal and Chemical Aspects of the Thermomiclear
Blanket Problem," Technical Report 435, M. I. T. Research Laboratory of
Electronics (1965).
Table 1. Formation Free Energy of Fluoride at 1000 K

 

Free Energy per Gram—~Atom of Fluorine~

 

 

Compound

(kJ) (kecal)
LiF —523 —125
BeF, —~448 —107
UF, 410 98
CrF, =314 ~75
FeF, ~280 ~67
HF —276 ‘~66
NiF, 230 —55
MoF¢ ~209 ~50

 

4c. F. Baes, Jr., "The Chemistry and Thermo-
dynamics of Molten Salt Reactor Fuels," Nucl. Met.
15, 624, P. Chiott; Ed., CONF-690801 (1969).

2 HF + M = Hy, + MF,, where M could be Fe or Cr ; (1)
MF, + Cr ¥ M + CrF, where M could be Ni or Fe . ‘ (2)

These reactions of impurities with the container will occur at a high
rate initially but will slow down as the impurities are consumed. As

long as no additional impurities are added to the system, after the first
few hundred hours these initial reactions are not expected to contribute
significantly to corrosion. The addition of a reductant to the salt could
eliminate the corrosive effects of the impurities by the reductant rather
than the wall reacting with the impurities. The transmutation products
formed by neutron reaction with the salt represent another significant
source of oxidants in the salt. All three elemental constituents of the
salt — lithium, beryllium, and fluorine — are expected to undergo trans-

mutation reactions.
The reactions of lithium,

3 -~
"LiF + n > He + |HF + n (3)
and
3
*Li¥ + n » He + [ HF , (4)

are, of course, desired because this is the means of producing the
tritium fuel. However, the TF produced can react with the stainless
steel container to form metals fluorides as shown in Eq. (1).

Transmutation of beryllium can occur in either of two ways:
4
BeF, + n » 2n + 2 ,He + 2F (5)
or
y 6
BeF, + n + ,He +  He + 2F . (6)

6
The ,He undergoes beta decay with an 0.8-sec half-life yielding for the

right side of Equation 6,
y 61 0 .
2He + "LiF + F

The fluorine produced in these beryllium transmutation reactions
can react with the stainless steel container, constituting another
source of corrosion.

The other transmutation of the salt componenis is that of fluorine:

19 16 .. 4
F +#n-> N + ,He . (7)

7
16~
The 7N undergoes beta decay with about a 7-sec half-life yielding 16O,
which can be expected to react with the containment metal causing addi-
tional corrosion. The net result of all three transmutation reactions

is the production of substances that will cause corrosion of the stainless
steel containment alloy. The extent of this corrosion would be very
significant and possibly intolerable. One means of at least reducing
and perhaps of eliminating this corrosion is to add a redoi buffer to
the salt. This buffer would need to be capable of (1) reducing the
TF to T, and a soluble fluoride, (2) reacting with the fluorine to
form a'stable, soluble fluoride, and (3) forming a stable oxide that
dissolves in salt and can be processed out of the salt.

One other problem associated with the use of Li;BeF, is the recovery
of tritium from the salt, The redox buffer required to keep the tritium
in the form of T, because of the corrosivity of TF with stainless steel
will probably make it easier to remove the tritium fromjthe salt.,

The potential problems associated with impurities, transmutation
products, and tritium recovery will almost certainly require a redox
buffer if a Li,BeFy, blanket is to be used with a stainless steel contain~
ment vessel. A great deal of experience with molten salt containing a
redox buffer has been accumulated at ORNL as a result of work for the

“ The fuel salt for this reactor

Molten Salt Breeder Reactor Program.
is composed of LiF and BeF, to which the fertile material ThF, and
fissile material UFy have been added. Because of reactions that occur
during both preparation and holding in the containment vessel, the
uranium is actually found to be present in both the three and four plus
oxidation states (UF; and UFy). Together these compounds make up
0.3 mole % of the fuel salt. This amount of material is sufficient to
ensure that addition of an oxidizing or reducing impurity to the salt
does not cause serious corrosion of many containment vessel materials.
Addition of an oxidizing impurity results in some U(III) being oxidized
to U(IV) and, conversely, addition of a reducing impurity results in
reduction of U(IV) to U(III).  Consequently, a small change in the
U(IV)/U(I11) ratio might occur, but because of this buffering action,
no drastic swings in the oxidation potential will take place.

The choice of a redox buffer for Li,BeF, in stainless steel will

depend on how "reducing' it is necessary to make the salt. One material

 

“J. W. Roger, Alloy Compatibility with LiF-BeF, Salts Containing
ThFy and UFy, ORNL/TM~4286, (December 1972).
that could make the salt sufficiently "reducing" to practically
eliminate the corrosion caused by impurities and transmutation products

is beryllium. Typical reactions could be:

Be + FeF, > Fe + BeF, , (8)
Be + 2HF - H, + BelF, , (9)
Be + 2F - BeF, . (10)

To determine the suitability of stainless steel for containing Li,BeF,,
corrosion measurements have been made with salt in the as-received
condition and in the reducing condition reached by adding beryllium to

it.

EXPERTMENTAL METHODS

The experimental assembly used for this corrosion work, called
a thermal-convection loop, is shown in Fig. 1. The loop portion of
the assembly is constructed of type 316 stainless steel tubing and
is heated on the bottom and one verticle side. Cooling the other two
sides causes the molten salt in the tubing to flow because of the
variation in density of the salt with temperature. The velocity of
the salt is about 1 m/min. Other important features of the loop are
the removable corrosion specimens and the accesses for insertion of
electrodes for voltammetry measurements and for addition of material
to the salt in the loop or removal of salt samples from the loop.
Sixteen corrosion specimens are used, with eight inserted in the hot
leg (heated vertical section) and eight in the cold leg (cooled
verticle section). These specimens are removed from the loop every
500 hr for weighing and examination. When the investigations to be
made with a particular set of specimens are completed, several of the
specimens are examined metallographically and, if warranted, with the

electron microprobe. These results reveal at which temperature weight
ORNL~-DWG 68-398TR3

   
   
 
     

|

i T NN
BALL VALVES

o ,L""":"Ebw‘

(2T TN

g - 3
N IR YRR -

7

e

»

         
  
   

- ‘\>>~CLAMSHELL
- HEATERS

=

30i

3

AANSULATION

e _;_;_‘\_

 

 

 

 

 

i

INCHES

 

1 \FREEZE
1 VALVES
FLUSH

TANK

 

 

 

  

oo x, T ¥ e T
o —~ =% - F oo o - %
B o e

=)

R

|
E»
P
]

Eo oo

 

SAMPLER

g
i

DUMP
TANK

 

 

 

Fig. 1. Schematic of Thermal-Convection

Loop.
shown is Q.76 m.

Scale length is 0.15 m.

Height
gains or losses have occurred, and the microprobe examination identifies
which components of the specimen material have been itransported.

One of the most significant improvements in thermal-convection
loop operation has been the application of voltammetry. Controlled-
potential voltammetry is a modern electrochemical technique, which
may be used to examine the nature of an electrochemical reaction in
different media, such as aqueous and nonaqueous solutions and high-
temperature salt systems. For measurements in molten Li,BeF,, the
electrodes consisted of the loop itself, which served as the counter
electrode, and two iridium electrodes, which were the quasi-reference
and working electrodes. These electrodes were typically 25-mm (1-in.)
lengths of 18-gage iridium wires welded to 3-mm~OD (1/8-in.) nickel
risers. The electrode area exposed to the melt was 10 to 20 mm?. The
voltammograms were recorded versus the iridium quasi-reference electrode
(Ir QRE), which is poised at the equilibrium potential (Eeq) of the
melt. In this manner relative changes in the oxidation potential of
the salt and of the concentration of certain impurities and corrosion
products can be measured on-line. The equilibrium potential of the
salt is a measure of the tendency of the melt to react with oxidizable
materials that it contacts. For melts that are not poised by a redox
buffer relative changes in the equilibrium potential are conveniently
observed by determining voltammetrically the potential difference (AF)
between the melt equilibrium potential and the cathodic limit of the
melt, which for LizBeF, is the reduction of beryllium(II). We define
the cathodic limit for these measurements as the potential at 80 mA
cell current. Other reference markers may also be used as long as the
same conditions are maintained. A decrease in AF, resulting in a
cathodic shift in Eeq for example, indicates that the melt has become
less oxidizing; and an increase in AF indicates that the opposite has
occurred, A voltammetric scan that shows the cathodic limit relative

to Eoq is illustrated in Fig. 2.
ORNL-DWG 77~3824

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

- AE -
80
60
<
E
}...
Z
& 40
@
-
o
20
A
o —
0 -0.2 -0.4 -06 -0.8 -1.0 -2

VOLTAGE (V)

Fig. 2. Voltammetric Scan to Final Limit of
Lio,BeF, in Loop NCL 31.

The reduction of Fe?t and cr?t impurities is illustrated by the

voltammogram shown in Fig. 3.

correspond to about 130 ppm Fe?¥ and 100 PPM cr2t,

The peak heights of the two waves

Fig. 4 shows the

results of a voltammetric scan made several weeks after that of Fig. 3.

The peak due to iron has disappeared, while the chromium peak is about

the same height. It is also interesting to note that the melt has

become less oxidizing as indicated by the cathodic shift in Eeq relative

to the peak potential of the chromium wave. The good agreement between

the voltammetric results and the results of the analyses of salt samples

indicates the usefulness of controlled-potential voltammetry for rapid

in situ measurements.
CURRENT {mA)

CURRENT (mA)

10

ORNL-DWG 77-8367

 

 

4 2+ o
Foo + 26~ Fe Cr + 28— Cr

   

 

 

 

 

 

 

 

 

~0.2 -0.4 ~0.6
APPLIED POTENTIAL (V)
Fig. 3. Reduction of Iromn and Chromium in Loop NCL 31.
ORNL-DWG 77-5368
+ 0.2 0O - .2
APPLIED POTENTIAL (V)
Fig. 4. Reduction of Chromium in Loop NCL 31.
11

EXPERIMENTAL RESULTS

Type 316 stainless steel specimens were first exposed in the
type 316 stainless steel thermal-convection loop containing as-received
Li,BeF, salt with a high impurity level (190 ppm Fe, 60 ppm Cr). The
circulating salt had a maximum temperature of 650°C and a minimum of
490°C. The specimens were examined after 500 and 1000 hr, and the
condition of the salt was monitored by means of voltammetry and salt
samples. Following this initial exposure, a beryllium rod was lowered
into the salt and a new set of type 316 stainless steel specimens was
inserted. These specimens were examined after 500, 1000, 1500, and
2000 hr in the salt. Meanwhile, monitoring of the condition of the
salt was continued.

Examination of the specimens after the first 500-hr exposure
showed that significant weight changes had occurred, most likely
because of reactions with impurities. Figure 5 shows the weight change

results for specimens near the highest and lowest temperatures. The

Y-137150

30 ORNL-DWG 76-3496

51020 -

20

     

® A5 RECEIVED" SALT
X AFTER BE ADDITION

!il!!ifti?t!

S10°C

WEIGHT CHANGE (mg)

850°C

i!ilii!t'ii

 

 

 

Q i 800 1060 1500 2000
EXPOSURE. TIME (ho)

Flg 5. Welght Change Results for Type 316 Stalnless Steel
Specimens Exposed to LiF-BeF; Salt in Both the As- Received ‘Relatively
Oxidizing Condition and in the Relatively Reducing Condition Created
by the Addition of Berylllum to the Salt.
12

weight lost by the hottest specimen during the first 500 hr, if we
assume that the material was lost uniformly from the surface,
corresponds to a corrosion rate of about 10 ym (0.4 mil) per year.
However, in previous work" with stainless steel in fluoride salts
containing UF, as an oxidizing species, corrosion did not result in
uniform removal of material; rather, a porous structure was found
(Fig. 6). The 1000-hr exposure of our specimens was not long enough
for extensive corrosion to occur, but Fig. 7 indicates that material
was not removed uniformly from the hotter specimen. Consequently, the
depth of attack can be expected to exceed that calculated for uniform
material removal. Futhermore, in the salt of a fusion reactor, impuri-
ties would be produced continuously and in significant quantities, and
these impurities would be more oxidizing than the impurities found

in this experimental salt. Thus, corrosion of stainless steel in a
fusion reactor could be a serious problem if no effort is made to

buffer the salt.

Y-~127785

{

i

600 700

{

MICRONS
100 X i
INCHES 0.020

i

0.010

 

i

100 200

1

 

 

Fig. 6. Type 304L Stainless Steel Specimen from Loop 1258
Exposed to LiF-BeF,~-ZrF,-ThF,~UF, (70-23-5-1-1 mole %) for 45,724
hr at 685°C.
13

Y-134722

 

 

    

650°C 510°C

Fig. 7. Type 316 Stainless Steel Specimens Exposed to Li;BeF, for
1004 hr.

To evaluate the effect of a buffer on the corrosion of stainless
steel in Li,BeF,, a piece of beryllium was inserted into the salt and
allowed to react. As shown in Fig. 8, the equilibrium potential shifted
cathodically, indicating the salt was made less oxidizing. Voltammetric
measurements indicated the FeF, concentration to be less than about
20 ppm. A new set of specimens was exposed to this reducing salt for
2000 hr. Examination of the specimens after 500 and 1000 hr showed
very small weight changes, near the limit of detection. As shown in
Fig. 5, the weight changes for the specimen temperatures considered
were much smaller for the specimens in the reducing salt during the
first 1000 hr than in the as~received salt.

Figure 5 shows that once the beryllium was removed from the salt
(during the first 500-hr specimen exposure to the reducing salt), the
specimen in the hottest position showed increasing weight loss as a
function of time. This most probably occurred because, once the source
of beryllium was removed, the species in the salt were no longer in
equilibrium and the salt became progressively more oxidizing. Thus,
to maintain a low correosion rate, beryllium would have to be added
continuously or at least periodically. If that were done the corrosion
rate of type 316 stainless steel in Li,BeF, salt could be kept within

a range acceptable for fusion reactor service.
14

ORNL~DWG 77 -3822

 

 

 

 

 

 

 

 

 

 

 

 

 

-4.15
®
A
I/
/ ® //

- ®/ ® A e Y
> \ 4 7
é '”‘.05 ® @ /
- p ® «
2
tf ® ,//'
g 1

-{,00 7
g ® ///
3 7

/“/ ®
-0.95 e
“4s
Be ADDITION
-0.90 :
O 20 40 60 80 100
TIME (days)

Fig. 8. Variation of the Limit for Li,BeF,
Salt in NCL 31.

CONCLUSTONS

The conclusions that can be drawn from this research are:

1. The oxidation potential of the salt can be controlled by
additions of a reducing metal.

2. Voltammetric techniques provide on-line measurement of the
oxidation potential of the salt and the concentration of impurities
in the salt.

3. High corrosion rates are encountered initially in salt with
a normal impurity concentration.

4. The corrosion rate of type 316 stainless steel in Li,BeF,
salt was reduced significantly by the direct addition of beryllium

metal to the salt.
15

ACKNOWLEDGMENTS

The authors would like to make note of the efforts of E. J. Lawrence
for operating the thermal-convection loop; J. R. DiStefano and J. L. Scott
for reviewing, S. Peterson for editing, and Gail Golliher for final
preparation of the manuscript for reproduction.
15.

17.
18—19.
20.
21.
22.
23.
24—26.
27.

70~71.

72.

13—74.

75.

17

INTERNAL DISTRIBUTION

Central Research Library
Document Reference Section
Laboratory Records Department
Laboratory Records, ORNL RC
ORNL Patent Office

R. G. Alsmiller, Jr.

E. §. Bettis

E. H. Bryant

D. Cannon

J. Clarke

R. E. Clausing

S. D. Clinton

J. E. Cunningham

R. A. Dandl

J. H. DeVan

A. P. Fraas

Fusion Energy Division Library
Fusion Energy Division Reports Office
T. A. Gabriel

M. L. Grossbeck

P. N. Haubenreich

M. R. Hill

J. A. Horak

PO ETY

EXTERNAL DISTRIBUTION

2847 .
48 .
49,
50.
51.
52.
53.
54,
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.

NHUUERTmoOoOOm NS OO

R. Keiser

G. Kelley
Krist

. Lawson
Manning

. McCoy
McCurdy

. McHargue
. Morgan
Postma
Roberts

W. Rosenthall
L. Scott

. E. Selle

» E. Shannon

. M. Slaughter
D. Steiner
W.C.T. Stoddart
J. 5. Watson

J. R, Weir, Jr.
W. M. Wells

R, Wernex

F. W. Wiffen

oy

> *

*

¥

oo Oy @

» *

ARGONNE NATIONAL LABORATORY, 9700 South Cass Avenue, Argonne, IL 60439

V. A. Maroni, CTR Program, 208 W1l5

D. L. Smith

CULHAM LABORATORY, Abingdon, Oxon, 0X14 3DB, United Kingdom

PRINCETON PLASMA PHYSICS LABORATORY, P.O. Box 4531, Princeton, NJ 08540

R. Mills
W. G. Price

PRINCETON UNIVERSITY, Chemical Engineering Department, Princeton,

NJ 08540

E. Johnson
18

76. WESTINGHOUSE ELECTRIC CORPORATION, Research and Development
Center, Pittsburg, PA 15235

R. E. Gold
77—-82. ERDA DIVISION OF MAGNETIC FUSION ENERGY, Washington, DC 20545

. Cohen
.C. Dalder
R. Finfgeld

C. Reuther, Jr.
M. Williams

M. Zwilsky

83—84. ERDA, OAK RIDGE OPERATIONS OFFICE, P.O. Box E, Oak Ridge, TN 37830

Director, Reactor Division
Director, Research and Technical Support Division

85-111. ERDA TECHNICAL INFORMATION CENTER, Office of Technical Information,
P.0. Box 62, Oak Ridge, TN 37830