Molten Salt Reactors Based on the Th-U3 Fuel Cycle

November 2001

The capture cross section of Thorium 232 and the fission and capture cross sections of Uranium 233 are such that fuel conversion, and even fuel breeding can be achieved both with fast neutrons and with thermalized neutrons in the Th-U3 fuel cycle. We report here on the work in progress at the Institut des Sciences Nucléaires of Grenoble (ISN) in view of evaluating the capabilities that a molten salt reactor operating with thermal neutrons could offer.

The condition for conversion is that the number of fissile nuclei created must be equal to the number of fissile nuclei consumed, i.e.

N^fert$\sigma^{{\textrm{fert}}}_{{\textrm{c}}}$ = N^fiss($\sigma^{{\textrm{fiss}}}_{{\textrm{f}}}$ + $\sigma^{{\textrm{fiss}}}_{{\textrm{c}}}$)

From this, the required fissile to fertile nuclei concentration can be inferred:

${\frac{{\textrm{N}^{\textrm{fiss}}}}{{\textrm{N}^{\textrm{fert}}}}}$ = ${\frac{{\sigma ^{\textrm{fert}}_{\textrm{c}}}}{{\sigma ^{\textrm{fiss}}_{\textrm{f}}+\sigma ^{\textrm{fiss}}_{\textrm{c}}}}}$

With a Th-U3 fuel, the cross section ratio ( ${\frac{{\sigma _{\textrm{c}}^{\textrm{fert}}}}{{\sigma _{\textrm{f}}^{\textrm{fiss}}+\sigma _{\textrm{c}}^{\textrm{fiss}}}}}$) is much better with thermal neutrons (approximately 1.6%) than with fast neutrons (approximately 10%), breeding can be achieved with less fissile matter and that is one of the driving reasons for this exploration.

A Brief History of Molten Salt Reactors

The present work rests on the results obtained in two evaluations done in the United-States, at the Oak Ridge National Laboratory. The first of these evaluations (1964-1969) is the Molten Salt Reactor Experiment (MSRE) an experiment which consisted in building a low power reactor and operating it with various fuels. The second evaluation, the Molten Salt Breeder Reactor (MSBR) was an extension of the MSRE, the aim being to evaluate, through computer simulations, how the MSRE could be extended to an industrial scale and to examine the possibility of operating the power reactor as a breeder with a thermal neutron spectrum. This reactor was never built but the simulation results proved full of promise. The project was stopped in 1976, in favor of a fast neutron reactor project. The MSRB project, then, was given up because of an investment option rather than because of any real technical difficulties. During the MSRExperiment, molten salt physical and chemical data were obtained in a real setting. The experiment confirmed that the issues of molten salt circulation and structure corrosion are not insurmountable, it showed that breeding could be obtained with a molten salt reactor, Thorium serving as the fertile nucleus.

The Model's Geometry

The core of the reactor model is cylindrical, it is a block of graphite through which holes are bored in which the molten salt circulates. The graphite serves as a neutron moderator, it is the solid structure of the reactor. The molten salt that circulates in the channels is both the fuel and the coolant, it contains the Thorium and Uranium needed to sustain the chain reaction.

The reactor dimensions are set by the power to be produced, and by the amount of heat that must be evacuated at each point. The simulated reactor yields 2500 MW thermal power, giving 1000 MW (1 GW) electric power: at the chosen operating temperature (salt temperature approximately 700C in the heat exchanger), the thermal efficiency rises to 40%.

The reactor core consists of a set of high density (2,3 g/cm³) graphite 15 cm. side hexagons, each traversed by a cylindrical 15 cm diameter channel in which the salt circulates. The set of hexagons forms a cylinder whose radius is 2.3 meters and whose height is 4.6 meters. Salt tanks are placed above and below the cylinder. The number of neutrons liable to escape from the core is minimized thanks to thick graphite reflectors placed above, below, and around the core.

Figure 1 : View of the reactor model

Salt Composition

The salt that was chosen is a fluoride. The carrier salt is a mixture of Lithium Fluoride (LiF) and of Beryllium Fluoride (BeF₂), or 4LiF-BeF₂, to which the fuel is added, that we will note here as HN¹, in a 12.5 % molar proportion, also a fluoride: (NL)F₄. Inevitably, the salt also contains fission products. This salt's fusion temperature is sufficiently low to ensure a nicely fluid salt at the reactor operating temperature, allowing fast salt circulation. The salt's physical and chemical properties are well known, thanks to the Oak Ridge experiment. A circulation speed on the order of 1 m/s in the channels lets the salt extract the fission generated heat from the channels.

The salt temperature at the core outlet is 700C. The salt passes briefly in the tank above the core, whose role is to facilitate salt circulation, and from there to the heat exchanger where the heat is communicated to the secondary circuit; the salt, cooled to approximately 600C, is fed to the tank below the core, from which it returns within the core.

Neutron Balance

The number of available neutrons in a converter reactor is:

N_d = $\nu$ - 2(1 + $\alpha$) - losses

with $\alpha$ = $\sigma_{{\textrm{c}}}^{}$/$\sigma_{{\textrm{f}}}^{}$ and $\nu$ the average number of neutrons per fission.

In the reactor considered, mean $\sigma_{{\textrm{f}}}^{}$ is 50 barns and mean $\sigma_{{\textrm{c}}}^{}$ 6 barns; thus, (1 + $\alpha$) is 1.12. If conversion requires 2.24 neutrons per fission and a fission yields 2.5 neutrons on average, 0.26 neutrons are left over per fission. If neutron losses are less than that, the number of available neutrons is larger than zero.

Reduced neutron losses (due to escaping neutrons and sterile captures) thus can allow for extra neutrons over and above those needed for conversion. These extra neutrons can be used either for breeding (in view of increasing the number of reactors in the field) or for the transmutation of radioactive isotopes, e.g. some of the fission products.

Some neutrons are liable to escape from the reactor. Its size determines such escapes. A reactor, if it were too small, would lose many neutrons. To minimize neutron escapes, the reactor size is chosen large enough for most of the neutrons to interact with the fuel before they can exit the core. Those that do escape, insofar as possible, are sent back into the core thanks to reflectors, thick graphite blocks that surround the reactor. Most of the neutrons that enter the reflector are sent back into the core if the reflector is thick enough and does not capture neutrons. Thanks to these reflectors, neutron escapes from this reactor model are practically inexistent.

Sterile captures in the graphite and the salt are small. The capture cross section of graphite is small for all the possible neutron energies in the reactor. This is mostly true also of the salt components. Sterile captures in the graphite and the salt carrier proper are as low as 0.1 neutrons per fission.

As for sterile captures in the fission products and in the Protactinium found in the Thorium 232 to Uranium 233 decay tree (Protactinium 233 has a 27 day half life), they have to be reduced and this can be done thanks to well designed and efficient on line fuel processing.

On Line Fuel Processing

As the salt is liquid, as it contains both the fuel and the fission products, it seems worthwhile to divert some of it to a processing unit in order to modify its chemical composition.

A first step consists in extracting the Uranium from the diverted salt sample and reinjecting it directly in the main salt circulation. The Uranium thus does not continue on to the other levels of the processing unit.

In a second stage, the Protactinium is extracted and let decay to Uranium 233 away from the neutron flux. The decay half life is 27 days. There are several good reasons for extracting the Protactinium:

  First, to keep a good neutron balance: Protactinium can capture neutrons before decaying to Uranium (the capture cross section has a resonance at 2 eV). A neutron capture on Protactinium wastes at least 2 neutrons².

  Second, to reduce the production of minor actinides insofar as a Uranium 235 nucleus resulting from the above process can, itself, be subject to fission-less neutron capture, and so on.

  Finally, to avoid the accumulation of fissile nuclei in the reactor in the event it is stopped (the Protactinium present in the salt would continue to decay to Uranium 233). Such an accumulation would increase the reactivity, because teh ratio of fissile to fertile nuclei would increase.

However, Protactinium extraction is useful only if the full salt volume can be run through the processing unit within a time that is small compared to Protactinium's 27 day half life. In the simulations, the total salt volume is run through the processing unit within 10 days. Note that, since discrete sampling is used, this does not imply that all salt molecules have been processed.

The extracted Protactinium sits for three months, long enough (a little over three half lifes) for most of it to decay to Uranium 233. The mixture of Uranium 233 and any remaining Protactinium is reinjected into the main salt circuit.

In the third stage, the fission products are extracted insofar as possible: they are liable to capture neutrons, in what are inevitably sterile captures since these cannot lead to a fission. If fission products were not extracted, the chain reaction would not be sustainable.

Finally, the Thorium that has been consumed has to be replaced, fresh Thorium is injected.

The rest of the salt remains unchanged, it is put back in the reactor as is, it will remain in the circuit indefinitely.

If the reactor is operated as a breeder, the excess Uranium (about 5%) must be removed. This can be done either at the end of the Protactinium decay, or directly during the first step, when Uranium 233 is removed from the salt sample and put back in the circuit.³.

In summary, the on line chemical processing takes the diverted salt sample through the following steps:

  Extract the Uranium and put it back in the reactor.

  Extract the Protactinium, let it sit for three months and, once it has decayed to Uranium, put it back in the circuit.

  Extract the fission products.

  Replace the Thorium consumed.

  If the reactor operates as a breeder, extract the excess Uranium.

An interesting feature is that all heavy nuclei remain in the reactor, except for those that leak out during processing: a few heavy nuclei (Uranium and transuranics) are taken along with the fission products when they are extracted; chemical reactions are never perfect. These losses are estimated to be about 10^-5 (one heavy nucleus taken along for each 100 000 fission product nuclei removed). Because they remain in the salt indefinitely, all heavy nuclei (other than those lost) will eventually fission. The actinides produced in the reactor are thus incinerated in the reactor. Excluding losses, the only radioactive nuclei that escape the reactor are the fission products.

Figure 2 : Schematic of the fuel processing system. TRU: Transuranics; MSR: Molten Salt Reactor; FP: fission products.

Radiotoxic losses can be further reduced with off line processing of the fission products: the transuranics and the Uranium that are taken with the fission products could be retrieved thanks to more thorough processing in a specialized unit, and put back in the reactor to be incinerated.

Summary of the Reactor Model Operated with 232Th/233U Fuel

The geometric characteristics that have been set for the reactor simulation are those given above, the power chosen is 2.5 GW thermal, or 1 GW electric, the mean salt temperature being 650C. The volume of Carbon is 70 % of the total core volume. The volume of salt is 46m³ with 20.5m³ in the core, 10.2m³ in the ``reservoirs'', 15.3m³ in the heat exchanger. The salt is liquid, its composition is: Lithium Fluoride (⁷LiF) with 70% of the atoms, Beryllium Fluoride (BeF₂) with 17,5% of the atoms, and heavy nuclei fluoride, mix of Thorium 232 and Uranium 233, ([NL]F₄) accounting for 12,5% of the atoms, plus a small quantity of fission products. It contains 70 metric tons of Thorium and one metric ton of Uranium. The reactor consumes 2.6 kg of Thorium per day, i.e. about one metric ton per year. The processing unit processes 46m³ of salt each 10 days; it extracts the Protactinium and the fission products. In these conditions, conversion is obtained, and excess neutrons are available. These excess neutrons can be used either for breeding or for the transmutation of fission products.

  Breeding - the reactor can generate approximately 5% more Uranium than is needed for conversion, i.e. it has a 25 year doubling time: after 25 years' operation, one metric ton of Uranium is available, the amount needed to start a new, similar, reactor.

  Transmutation - the transmutation of long lived fission products, such as ⁹⁹Tc whose half life is 211 000 years and ¹²⁹I whose half life is 15 700 000 years is considered. After a neutron capture, ⁹⁹Tc, the only Technetium isotope produced in the core, yields ¹⁰⁰Ru which is stable. Any neutron captures by ¹⁰⁰Ru and its successors will not produce long lived isotopes. As for Iodine, two of its isotopes are produced in the core: ¹²⁷I which is stable and whose capture cross section is small, and ¹²⁹I whose half life is large and that has a large capture cross section. After a neutron capture, ¹²⁹I yields ¹³⁰Xe which is stable and, being gazeous, escapes from the core. Any captures by ¹²⁷I yields ¹²⁸Xe which, too, is stable and gazepus, it escapes from the core. Thus, neutron captures on Iodine do not lead to any long lived radioactive products either. The Iodine that is to undergo transmutation can be mixed in with the salt. The Technetium is put in needles that are placed in the graphite block and are replaced after a few years' irradiation.

The radiotoxicity induced after 200 years operation and 1000 years cooling of the wastes - the time when the radiotoxicity of the fission products becomes constant because the medium lived products (those whose half life is on the order of 30 years) have decayed to stable nuclei - can be compared to the radiotoxicity induced by a PWR that generates the same power over the same time period. The radiotoxicity induced by a molten salt reactor is more than 10 000 times less than that due to a PWR similar to those presently operated in France, if only the heavy nuclei losses are considered in the MSR case. If the MSR residual inventory is added to the heavy nuclei leaked during processing, the radiotoxicity is 100 times less than that of the PWR. The residual inventory would have to be included in the event reactor operation were stopped after 200 years production, e.g. if renewable energies were ready to take over.

Figure 3 : Comparison of radiotoxicities: losses and inventory of several reactors. From left to rigt: MSR - molten salt reactor with Th-U fuel and thermalized neutrons; ADS - Th-U fast neutron reactor; U-Pu fast neutron reactor; PWR - U-Pu Pressurized Water Reactor with thermalized neutrons. The values are given for 200 years operation,1000 years after unloading, they are normalized to the same power yield.⁴

Transition Phase with Th-PU

As we know, Uranium 233 is not to be found in nature and, yet, a Thorium based molten salt reactor cannot be started without fissile matter. Could Plutonium from the PWRs be used for that purpose? If the answer is yes, two problems are solved: fissile matter is available to start the MSR; Plutonium, usually considered a waste, as with the PWR system which does not allow Plutonium multirecycling, can be incinerated.

From the simulations, it appears that the Plutonium taken from a PWR's spent UOX fuel can be used to start a Thorium based molten salt reactor. The spent fuel is let cool for five years after irradiation, the Plutonium is extracted and a Fluoride is made (PuF₃) which is mixed with the Thorium, in the molten salt. The ratio of heavy nuclei in the salt is set equal to that of a Th-U3 salt so as to reach the Th-U3 equilibrium situation as quickly as possible. In this hypothesis, approximately 4 metric tons of Plutonium are needed for the reactor to be critical. As the Plutonium is incinerated, it is replaced with Thorium. The Uranium 233, produced from the Thorium, progressively replaces the Plutonium as the fissile isotope. After 15 years of operation, the fuel composition is practically the same as that of a genuine Th-U3 reactor. It is somewhat different, however, because the irradiation of Plutonium generates more minor actinides than that of Uranium 233. These are progressively incinerated in the reactor but it takes more than a hundred years to reduce the minor actinides to the amount in a Th-U3 reactor that was started directly with Uranium 233. This difference also affects, obviously, the radiotoxicity of losses. With 200 years production and after 1000 years cooling, the radiotoxicity of the losses from a Th-U3 reactor started with Plutonium is about 4 times higher than in a reactor started directly with Uranium 233. This is not necessarily a major issue: at least the Plutonium has been incinerated.

The Strong Points of the Th-U3 Molten Salt Reactor

This reactor, with its on line fuel processing, is at least a converter. Thus,it burns 100% of its fuel.

Because fuel processing is done in situ, and continuously, only one inventory is needed (whereas, in a reactor with solid fuel, two inventories are necessary in order that the reactor continue to operate while its spent fuel is being processed). Moreover, the inventory is small: 70 metric tons Thorium 232, one metric ton Uranium 233 for a 1 GW electric power. Finally, on site processing reduces to a minimum the transport of radioactive materials.

If fuel processing includes Protactinium extraction, we have seen that excess neutrons are available. They can be used for one of two things:

  for breeding: the reactor can produce 5% excess Uranium 233, the doubling time is 25 years.

  For the transmutation of fission products such as ⁹⁹Tc and ¹²⁹I.

Figure 4 : Radiotoxicity for 200 years production, 1000 years after fuel unloading for a Th-U3 molten salt reactor. From left to right, radiotoxicity due to: LLFP - long lived fission products with and without transmutation of the Iodine and Technetium; actinide waste - actinides leaked during processing; final inventory - residual inventory; used Th232 - radiotoxicity of the amount of Thorium 232 that has been used.⁵

The Pending Issues

A number of assumptions have been made regarding the fuel processing chemistry aspects that will have to be validated experimentally. The extraction efficiency of the various fission products, how well and how fast the Protactinium can be extracted have to be verified. Our assumptions have not been made lightly, they rest on chemical data, but nothing can replace actual tests.

The computations simulating the behavior of the neutrons in the reactor rest on the availability of accurate cross sections. These have to be available for all neutron interactions, over a wide energy range, for a large number of nuclei: those of the salt, all the fission products, the graphite, Thorium, the other heavy nuclei. Numerous cross section measurement experiments have been done in various laboratories, others are being done currently. The experimental results serve as input parameters to models that are used for the elaboration of data bases. Many uncertainties remain: some experiments disagree on measurement results; the models used are different; finally, many measurements are still missing.

In addition, the simulations are assumed to be reliable, the programs used have been checked insofar as possible. However, the computer simulations will have to be validated experimentally.

The MSRE experiment has not encountered any corrosion problems due to the salt. It seems that graphite, which constitutes the core of the reactor, is resistant to that salt. But what about the fuel processing unit, the heat exchanger? In time, materials engineers will have to think up appropriate solutions. But this will be useful only after further progress is made, and if there is a real drive to build this type of reactor. At the present, we are exploring potentialities.

Reactor safety has not been examined in detail at this stage. All the simulations apply to a critical reactor but, at least during the testing phase of an experimental reactor, and perhaps definitively, it could be operated as a sub-critical reactor, associated to a proton accelerator to produce the missing neutrons via a spallation reaction. A sub-critical reactor can stay safe even if the effective multiplying coefficient, k_eff, changes, within limits.

Project for an Experiment

The computer simulations of neutron transport in the reactor considered yield promising results as to the feasibility of developing a molten salt reactor based on the Thorium-Uranium 233 fuel cycle capable of fuel conversion and even fuel breeding. But, before a prototype reactor is built, an experimental validation of neutron behavior in the various components of the reactor should be performed.

In order to simplify the problem, and to avoid measuring too many things at the same time, an external neutron source can be used to provide the neutrons. This option alleviates security constraints that would have to be applied in the presence of fissile matter. Since tiny quantities of fissile matter are involved (a few grams of Thorium are irradiated), the experiment can be set up in a university laboratory such as ISN. The other advantage of measuring the propagation of neutrons emitted by an external source instead of fissions is that the initial characteristics of the neutrons are well controlled, the time and location where they are created, and their initial energy are precisely determined. The initial energy of the neutrons will be 2.5 MeV, close to the mean energy of the neutrons released during a Uranium 233 fission. The neutrons will be bunched in pulses that should not be more than approximately 0.1 microseconds wide, with a pulse frequency that will be adjustable up to 10 kHz (10 000 pulses per second). About 10⁶ neutrons will be emitted with each pulse. A similar neutron generator was designed and built at ISN. The pulse width is narrower for this generator but it seems that it will be possible to meet the specification. Once created, the neutrons will be taken to the middle of a high density graphite block. Channels, drilled through the block, will accomodate the insertion, in the neutron flux, of the various bodies whose characteristics are to be measured. Sensors will also be placed in some of the channels, in order to measure the energy of the neutrons at a given location and their time of arrival relative to the time they were created, i.e. the pulse. The general idea is to measure the transport of individual neutrons in each of the bodies considered (Carbon, Fluorine, Beryllium, Lithium, Thorium, and so on). By comparing the measurement findings with the computer simulation results, it will be possible to correct any erroneous cross section values, and to validate the simulations.

This experiment will be done in collaboration with EdF, with other CNRS-IN2P3 laboratories (CENBG Bordeaux⁶, IPN Orsay⁷), and with laboratories from the chemical branch of the CNRS (ENSEEG⁸ and Paris VI) so that the chemical aspects of the reactor model can be verified as well. The device will operate with solid rather than molten salts but the salt composition will be the same as in the simulated reactor except for the fissile nuclei that will be left out. Many aspects of the chemical separation processes required for the molten salt reactor can be studied on this experimental set up. In particular, the extraction of Protactinium, whose importance has been stressed, can be tested. Also, heavy nuclei leaks during processing can be estimated and separation techniques improved in order to minimize these leaks.

... HN¹

HN: Heavy Nuclei - the fertile nuclei (Thorium 232), the fissile nuclei (Uranium 233 and, possibly, Plutonium or Uranium 235 at startup), any other actinides.

... neutrons²

A neutron capture on Protactinium 233 gives Protactinium 234 that decays to Uranium 234 through $\beta^{{-}}_{}$emission. Uranium 234 is not fissile. It will have to capture a neutron before it can become Uranium 235 which is fissile. Thus, 2 neutrons are needed instead of none at all to reach a fissile nucleus.

... circuit.³

In the latter case, the Uranium contains Uranium 232 which is easily identified thanks to its intense $\gamma$ decay. This has the additional advantage that it is difficult to handle, providing a real deterrent as far as proliferation is concerned.

... yield.⁴

Some of the data presented in this figure are due to Sylvain David, currently at IPN Orsay

... used.⁵

Some of the data presented in this figure are due to Sylvain David, currently at IPN Orsay

... Bordeaux⁶

Centre d'études nucléaires de Bordeaux Gradignan 

... Orsay⁷

Institut de Physique Nucléaire d'Orsay

... (ENSEEG⁸

Ecole Nationale Supérieure d'Electrochimie et d'Electrométallurgie de Grenoble

Last update: 11 June 2002