NSE_moltenFluorides.txt

NUCLEAR SCIENCE AND ENGINEERING: 2, 797-803 (1957)

Molten Fluorides as Power Reactor Fuels?

R. C. BRIANT? AND ALVIN M. WEINBERG
Oak Ridge National Laboratory,” P.O. Box X, Oak Ridge, Tennessee
Received June 13, 1957

Molten fluorides of uranium, thorium, plutonium, and other elements potentially have wide
applicability as fuels for power reactors. Because of their low vapor pressure they can be used in
very high-temperature but low-pressure liquid-fuel reactors. In addition, they possess great
chemical flexibility—the molten-salt principle can be applied to burners, thorium-uranium thermal
breeders, plutonium-uranium converters, and possibly even to fast plutonium breeders. Because of
the very high thermal efficiency obtainable in reactors using molten salt fuel, the fuel cost in a
simple burner using enriched **°U is of the order of 2-3 mills/k Wh

A high-temperature reactor using molten uranium salts (Aircraft Reactor Experiment) was operated
for a short time at the Oak Ridge National Laboratory. The reactor was of the circulating-fuel type,
with a BeO moderator. The maximum outlet temperature achieved was greater than 1500°F (1100
K). It is believed that with further development the ARE could be a prototype for an economical
uranium burner.

Two very different schools of reactor design have emerged since the first reactors were built.
One approach, exemplified by solid fuel reactors, holds that a reactor is basically a mechanical
plant; the ultimate rationalization is to be sought in simplifying the heat transfer machinery. The
other approach, exemplified by liquid fuel reactors, holds that a reactor is basically a chemical
plant; the ultimate rationalization is to be sought in simplifying the handling and reprocessing of
fuel.

At the Oak Ridge National Laboratory we have chosen to explore the second approach to
reactor development. The ORNL aqueous homogeneous reactor is the best-known embodiment of
the liquid reactor approach; in a sense it represents the natural culmination of the aqueous reactor
systems.

The two outstanding advantages of aqueous solutions of uranium as possible reactor fluids are:
first, that they exploit D,O, the best of all moderators; and second, that because of the wide range
of solubility of UO,SO4 in D,0O, there are many possible embodiments of the aqueous
homogeneous idea. There are, however, disadvantages to the aqueous systems: first, because of
the high vapor pressure of water, aqueous homogeneous power reactors, like all aqueous power
reactors, require high pressure to achieve rather modest thermal efficiencies; second, there is no
practical thorium compound which is soluble in water—the blanket of the two-region aqueous
homogeneous reactor is a slurry, not a solution; and third, the mixture of radiolytic D, and O,
produced, because it is potentially explosive, requires very complicated equipment to deal with it
properly.

Thus it has long been recognized that a liquid fuel which did not require high pressure, in which
thorium or its compounds could dissolve, and which did not decompose under radiation would
indeed be a major invention for the reactor art. One fluid which satisfies the requirement of low

 

! Presented by Dr. Weinberg at the American Nuclear Society Meeting in Pittsburgh, June 10, 1957.
? Dr. Briant died April 25, 1954.
3 Operated by the Union Carbide Nuclear Company for the US Atomic Energy Commission.
pressure and no gas production is the uranium-bismuth eutectic which is the basis for the
Brookhaven-Babcock and Wilcox LMFR. The U-Bi system is, however, rather limited in the
concentrations of uranium which can be dissolved; moreover, thorium 1s not soluble in bismuth.

TABLE I
PHYSICAL PROPERTIES OF VARIOUS SALT MIXTURES

 

 

B Mel.ting Density Thermal Viscosity Them.la.l Specific Prandil
Composition, mole % Point, at 973 K expansion at973 K  conductivity, Heat,
K kg/m’ per K kg/(m*s)  W/(m*K) Kl/(kg*K) ~Mumber
53.5 NaF - 40 ZrF, - 6.5 UF, 813 3270 3.36e-4 0.0057 2.1 1.00 2.74
71 LiF - 16 BeF, - 12 ThF, - 798 3250 2.52e-4 0.0071 1.55
1 UF,
67 LiF - 30.5 BeF, - 2.5 UF, 737 2100 1.90e-4 0.0055 2.4
Water (at 20°C) 0.0010 0.6 4.2 7

 

At the Oak Ridge National Laboratory we have been investigating another class of fluids
which satisfies all three of the requirements for a desirable fluid fuel: large range of uranium and
thorium solubility, low pressure, and no radiolytic gas production. These fluids, first suggested by
R. C. Briant, are molten mixtures of UF, and ThF, with fluorides of the alkali metals, beryllium,
or zirconium. In order to assess better the possibilities of molten fluoride reactors, ORNL in 1954
constructed and operated a high-temperature, molten-fluoride, circulating-fuel reactor with a BeO
moderator and an outlet temperature which ranged above 1500°F (1100 K). The papers which
will follow are a description of this reactor. Since the work was supported by the Aircraft
Reactors Branch of the U. S, Atomic Energy Commission, the reactor was called the Aircraft
Reactor Experiment (ARE).

The nuclear properties of the fluoride salts are in many respects very good. The capture cross
section of fluorine at thermal energy is only 0.009 b, and its resonance integral is small.
Unfortunately fluorine is not a very good moderator; the critical mass of a system moderated
entirely by fluorine is estimated to be several hundred pounds. Consequently, many embodiments
of fused salt reactors use some solid moderator, such as graphite, Be, or BeO. The physical
properties of several possible molten fluoride mixtures are compared with those of water in Table
I

It may be seen from Table I that the thermal conductivity, heat capacity, viscosity, and Prandtl
number of the salts are of the same order of magnitude as the corresponding quantities for water.
The higher viscosity of the salts 1s somewhat compensated for by higher thermal conductivity.
Hence, there 1s no decisive difference between water and molten fluorides from the heat transfer
viewpoint. The main disadvantage of these fluids is their very high melting point—any reactor
which uses them must be completely equipped with external heating devices. However, the high
melting point also has an advantage: the reactor can be designed in such a way that, in case of an
accident, the spilled salt would freeze and occlude radioactive fission fragments. This is an
important safety consideration.

From the chemical standpoint, the molten fluorides have two great advantages: their stability,
and the wide range of uranium and thorium (and, one hopes, plutonium) which can be dissolved in
them. Thus the fused salt reactors can be made the basis of at least three classes of reactors, each
of which may include numerous embodiments: the straight “*°U or *’Pu burner, the one- or two-
region thermal breeder, and the fast reactor.

STRAIGHT **U OR *Pu BURNER
The straight >°U or plutonium burner is the simplest embodiment. The great advantage that
accrues to a really high-temperature liquid-fuel burner is that such a device offers the possibility
of burning *’U with good thermal efficiency and without refabrication costs. With U at
$17/gram and a liquid reactor operating at 35% thermal efficiency, the calculated fuel cost is
between 2 and 3 mills’kWh. As for burning plutonium in a liquid fluoride system, there is
insufficient knowledge of the phase diagrams to be certain that this is feasible. Our general
knowledge of these fluoride systems suggests that enough plutonium fluoride will be soluble in
one of the melts to make plutonium burning feasible.

ONE- OR TWO-REGION THERMAL BREEDER

Since ThE, is very soluble in fluoride melts, a regenerative system based on the “*U-Th cycle
appears to be entirely possible. The neutron economy will, however, not be as good as in a D,O
system, mainly because Li, Be, and F are not as good moderators as 1s deuterium. Nevertheless,
calculations which have been made at ORNL suggest strongly that a “>>U-Th system in which the
breeding ratio is very nearly unity is feasible.

FAST REACTORS

A fast reactor based on the fluoride system is almost out of the question: fluorine is too good a
moderator. Nevertheless, a reactor fueled by a melt of pure plutonium fluoride or **°U fluoride
does have a very energetic neutron spectrum. The trouble is that the fuel is extremely
concentrated; the critical mass, plus the external holdup, would be prohibitively high. At first
sight a more likely possibility is the molten chloride system; chlorine is a poorer moderator than
fluorine, and therefore it would be possible to keep the neutron spectrum very fast in a molten
chloride reactor. However, because of the very high (n,p) cross section of *>Cl, only the isotope
37C1 would be tolerable in such a reactor. This requires a difficult isotope separation and makes a
fast reactor based on molten chlorides rather unlikely.

PROBLEMS OF MOLTEN FLUORIDE REACTORS

Circulating-fluoride reactors share, with all fluid reactors, the difficulties in equipment
maintenance and the rigid requirement of leak tightness and component reliability. There are,
however, three difficulties that affect the molten-fluoride systems particularly and, incidentally,
also the liquid bismuth systems.

The first difficulty has to do with corrosion. The Aircraft Reactor Equipment was run only for
a short time. Before the fused salts can be considered as the basis for a large-scale, long-term
central power reactor system, it will be necessary to develop metals which will withstand
prolonged attack by fluorides. Nevertheless, our considerable experience with these systems
strongly suggests that this corrosion problem is a tractable one.

A second basic difficulty of the fluorides 1s that, as compared with the aqueous systems, they
tend to require higher fuel concentration for criticality. This means that the power output per
kilogram of fuel in the salt system is relatively low; i.e., the material efficiency tends to be low.
This, of course, is balanced by the higher thermal efficiency of these systems; the high inventory
also decreases the number of reprocessing cycles required for a given number of megawatt days.
The third difficulty results from the high melting point already mentioned. The necessity of
heating the system to above the melting point of the salt before the salt is introduced, and the
necessity of preventing freeze-up during operation, resulted in a complex system of heaters and
thermocouples in the ARE, as will be evident from the following papers. Nevertheless, this
difficulty was successfully overcome and in future production reactors the complexity will be
much less, partly because of the experience already gained, and partly because such a future
reactor will not be called upon to furnish as much experimental information as was the ARE.

A MOLTEN FLUORIDE LARGE-SCALE POWER PLANT

In order to see just what the potentialities of fluoride fuels for power plants might be, several
design studies have been made of power plants based on fluoride fuels. The most recent, and
most exhaustive, such study has been made by a group at ORNL directed by H.G. MacPherson. 1
should like to give a very brief description of the reactor which they have studied.

The reactor, Fig. 1, 1s a two-region homogeneous reactor with a core approximately 6 ft in
diameter and a blanket 2 ft thick. Moderation is provided by the salt, so there is no need for
moderator or other structure inside the reactor. The core, with its volume of 113 ft’ (3.2 m’) is
capable of generating 600 MW of heat at a power density in the core of 187 W/ecm®. The net
power generation is approximately 240 MW,

The basic core salt is a mixture of about 70% 7LiF and 30% BeF,. Additions of thorium
fluoride can be made if desired, and enough *°UF, is added to make it critical. The blanket
consists of the eutectic of LiF and ThF, or mixtures of it with the basic core salt. The melting
point of the core i1s 867°F (737 K) and that of the blanket salt is 1080°F (855 K) or lower.

Both the core fuel and the blanket salt are circulated to external heat exchangers, six in parallel
for the core and two in parallel for the blanket. The beat is transferred by intermediate fluids from
these heat exchangers to the boilers, superheaters, and reheaters. The heat transfer system is
designed so that with a fuel temperature of 1200°F (920 K), a steam temperature of 1000°F (810
K) at 1800 psi (125 bar) can be achieved.

After careful consideration of the problem of control of the reactor, it has been decided that
there 1s no need for any control rods. Reactor control is automatically maintained by the negative
temperature coefficient, as demonstrated by the ARE and reported in the following papers.
Uranium fluoride fuel or thorium fluoride poison will be added for shim control.

The nuclear calculation involved a multi-group, multi-region program using the latest
available cross sections. These included higher parasitic absorption cross sections for U and
33U than had previously been used, and the resonance-energy absorption cross sections for fission
products proposed by Greebler, Hurwitz, and Storm'. Thorium cross sections were estimated by
reconciling the observed resonance integral with the observed resonances and by taking Doppler
broadening and self-shielding into consideration. These various assumptions gave more
pessimistic estimates of the neutron economy than previous studies had assumed, which should be
borne in mind when making comparisons.

The starting fuel is 2°U, and fresh *°U is added as needed to compensate for burn-up and
buildup of fission-product poisons. When the core fuel is processed, the unburned uranium is
returned to the core along with any “°U available from the blanket. With this system, no uranium
is returned to the vendor, and eventually the even-numbered isotopes of uranium build up to
it
PUMR QUTLET

  

FLUID LEVEL

 

SECTION

Figure 1: Reference Design Reactor

equilibrium values. The acceptance of this poison handicaps the neutron economy, but the ability
to use uranium with its equilibrium content of even-isotope poison is an advantage for the molten
salt system.

The processing of the fuel is by the volatility process”. Uranium is removed as UFg, and is
returned in fresh salt to the reactor. The used salt containing the fission products and **Np is
stored for later recovery; however, in the calculation of power cost no credit is taken for recovery,
but instead a charge for storage is made.

The blanket is also processed by the volatility process; the uranium is removed but the
thorium-bearing salt is returned to the blanket. Not enough fission products build up in the
blanket during the course of the reactor life to warrant discarding the blanket fluid. With this
simple fuel cycle and a chemical plant based on the capacity and cost of the current ORNL
volatility pilot plant, a total fuel cycle cost of 2.5 mills/kWh is calculated. This fuel cycle cost is
equivalent to the sum of the items of chemical processing, fuel element refabrication, inventory
charge, and fuel cost for a conventional solid-fuel-element reactor. The estimate 1s based on a
$17/g cost of “**U, an inventory charge of 4%, and an 80 % load factor.

The reactor hardware can accommodate a number of different loadings to achieve different
performances. The above fuel cycle cost 1s obtained with no thorium in the core, an equilibrium
state inventory of 650 kg (about 3 years' burnup) of combined >°U and U, an effective breeding
ratio of only 0.56, and a core processing cycle of 9 months. If 2°U could be purchased at the
same price as = U, the fuel cycle costs could be reduced to 1.4 mills/kWh, with an equilibrium
breeding ratio of 0.83. The acceptance of higher inventories could lead to high breeding ratios
and somewhat lower costs.

Summarizing, we believe that high-temperature, molten-fluoride, circulating-fuel reactors are
adaptable to a variety of embodiments because of the advantages of high temperatures at low
pressures, the wider range of solubility of uranium and thorium compounds in these salts, and the
radiation damage resistance of these fluorides. Our confidence has been greatly enhanced by the
actual operation of a molten-fluoride reactor which will be described in the following papers.

REFERENCES

1. P. GREEBLER, H. HURWITZ, JR., AND M.L. STORM, Nuclear Science and Engineering 2,
334 (1957).

2. G. I. CATHERS, "Fluoride Volatility Process for High Alloy Fuels," presented at
Symposium on Chemical Processing, Brussels, Belgium, May 20-25, 1957; ORNL CF 57-
4-95 (1957).