Sodium Coolant Chemistry -
Experience, Impressions and Perspectives

G. Periaswami, MCD

 

            Given below is an account of my impressions and observations on selected issues in the field of sodium chemistry gained through my personal work over a period of three decades.  These are only personal views and not the final word as they are likely to be subjective. 

 1. Development of Sodium Glove Box :

The first and foremost requirement for sodium chemistry work is a highly  pure inert atmosphere glove box for handling sodium metal.  Based on an AERE report,  a small evacuable glove box of 1000 l volume was designed and fabricated with help from AFD, BARC workshop.  Since it had to be evacuated, it was made of 6mm thick SS walls and evacuable caps for gloveports.  Argon was let in through a purification column of mixed bed (molecular sieves5A and BASF-R-3 11) copper catalyst).  The purity indicator was tungsten filament lamp with a hole made in the glass.  Reasonable purity was achieved in this box.  Technical help from Shri S. Srinivasan of RChD and Shri Betkarur of AFD was instrumental in giving shape to this box.  One big lesson learnt in the commissioning of the box was in leak testing of  such a big volume.  A 500 l/sec diffusion pump was used for evacuating it.  A thermocouple gauge was used for leak testing while spraying outside with hydrogen.   Since I had to arrange everything single handedly, I learnt a lot on commissioning a vacuum system and also in fabrication methods including brazing.

     After some experience with the said glove box a bigger box, with recirculation system was designed.  Evacuation of the box was found unnecessary.  So the box was designed with thinner walls but still with a cylindrical configuration to help in leak testing.  A diaphragm pump of 2 cfm capacity was initially used for argon recirculation.  This was latter replaced with a sealed blower from Rotron USA and the columns were redesigned for low pressure drop.  One important lesson learnt in this activity was that the gas-flow through the prolification bed should be from top to bottom as otherwise the bed would get fluidized generating  large quantities of dust.  The pressure control system consisting of Dwyer pressure switches and solenoid valves were found to be suitable.  It was also understood from this work that RTV sealant can be used very well as the gasket for fixing the Perspex view ports. The glove ports should have three “O” ring grooves and should be fitted with  ambidextrones butyl rubber gloves with the palms kept horizontally so that the gloves will wrap around the forearm while working.

A small sodium bath that  holds about 1.5 kg of sodium in liquid state (150oC) was kept inside the box to indicate the purity.  A shining surface staying for about half an our without coverage by oxide layer was assumed to indicate the sufficient purity.  This sytem is currently in use at RCL.  A further advancement of this design would be the inclusion of an automated regeneration cycle for the purification columns.

 2.Oxygen estimation in sodium:

Oxygen in sodium was estimated by separating the sodium oxide from bulk sodium by extraction using freshly distilled mercury.  The sodium oxide thus separated was then dissolved in water and estimated by titrimetry.  A novel vacuum  distillation set up  was constructed for Hg distillation.  One main problem in this amalgamation technique was the spillage of Hg which splits into doplets.  A mercury wetted copper ring was used for picking up mercury drops. However, the amalgamation process was found to be tedious. 

The vacuum distillation process developed as the alternative where the sample was heated using an RF induction ws found to be more reliable and convenient.  The  RF generator used should be a medium frequency (500 kHz) one and not a high frequency unit (2MHz) as otherwise the sodium vapour will ionize and flash. The glass outer tube should be coated with sodium before taking the sample for distillation. This will give a lower blank as contamination from surface sorbed gases is minimized.  Similarly the end point identification is better done with a thermocouple from top dipping into molten sodium.  Otherwise there is a possibility of over heating and the sodium oxide residue decomposing.   The system currently in use at RCL has a thermocouple inserted from bottom.  This has the advantage of a lower contamination from outside as the O-ring seal is avoided. But the end point detection should be done carefully.

 3.Carbon Estimation

Most of the carbon in sodium is in the form of particulates from the initial sodium charge.  Some amorphous carbon can come from any oil vapour introduced into hot sodium during operation.  The best method for carbon estimation in  sodium is the combustion of the residue after vacuum distillation of the sample contained in a quartz crucible where carbon is converted to CO2.  The combustion is done at 1100oC while passing purified oxygen and the CO2 is cryogenically trapped using on organic solid/liquid slurry.  Then it is measured by manometry.  A lot of precaution is taken while transferring the sodium sample from distillation vessel to the combustion tube for minimizing CO2 pick up from atmosphere.  Also, the sampling sodium for carbon analysis should be done carefully as there is a lot of carbon segregation while cooling the sample.  This aspect was checked by sampling  sodium both by flow-through method and overflowing sampling method in FBTR secondary circuit. Surprisingly, the values obtained by both the techniques were same contrary to the expectations.  Finally,  it was identified that the carbon load in FBTR sodium is mainly particulate which does not segregate while cooling. Only the dissolved carbon will segregate and this component is very low.  So analysis of carbon is reliable only for quality control of initial charge and not useful for detecting smaller oil-leaks into sodium.  Monitoring the cover gas for methane will be the best way to detect oil leak into sodium.  

4.Sodium Sampling  

Taking a representative sodium sample from flowing sodium is very important.  This is because most of the impurities dissolved in sodium have a highly temperature dependent solubility and can segregate very rapidly to the sampler wall and deposit in an irreversible manner while cooling the hot sodium. (However, this is not true for particulate carbon).  This requires that the entire sample held in a container should be analysed for getting the real value for oxygen and dissolved trace metals.  Among the most commonly used sampling methods,  only the overflow sampler meets the above requirement.  It is possible to use crucibles of different materials for getting samples for different impurities e.g. nickel crucible for oxygen, quartz for carbon and titanium for trace metals.  The entire sample can be taken for an anlysis thus avoiding problem of segregation.  But the process for sampling using this device is a bit tedious.  Sodium flushing has to be done carefully and the procedure has to be followed scrupulously, which is a big problem at the operator level.  Taking overflow samples from secondary loop of FBTR was a challenge.  For reasons not easily understood, there has been one problem or the other in most occasions in FBTR

On the otherhand, the tube sampler  used in French reactors and in FBTR are easy to operate.  These samples are reliable only for analysis for particulate carbon.  For oxygen and trace metals there is possibility of segregation and loss.  However, the French are using only tube samples till todate.

A low volume and compact overflow sampler was designed at RCL and put to use in FBTR secondary and EDG loops.  PFBR is supposed to have overflow samplers.  

5.Trace metal analysis  

Analysis of FBTR samples for trace metals using AAS and ICPMS have given values within specifications except for iron. The results for iron  used to give occasional high values.  This is easily explained on the basis of particulates circulating with sodium mostly in the form of gamma iron. This has been observed in other reactors also.  

6.Oxygen Monitors  

Monitoring for oxygen in sodium will help detect air leaks into primary system and steam leaks in secondary system.  This will avoid incidents similar to the airleak that occurred  into primary covergas in Super Phoenix which went undetected because there was no oxygen sensor in that reactor.  These sensors use an oxide ion conducting solid electrolytes configured into a galvanic cell where sodium with dissolved oxygen forms the sample electrode.  Owing to electronic conductivity consideration and compatibility with  sodium at 400oC, Yttria Doped Thoria (YDT) happens to be  the most  suitable electrolyte  for this purpose.  YDT procured from Zircoa in the form of 6mm dia and 120mm long tubes were used as the first oxygen sensors built at RCL.  The frequent breakage of these tubes at the freeze seal owing to the presence of  high temperature gradients rendered this configuration unsuitable.  Then YDT thimbles brazed to a low expansion alloy (Dilasil) obtained from GE were used and   successfully tested in loops.  They also exhibited high breakage tendencies due to the strain generated at the braze seal.  More over YDT sensors are costly ($5000/piece) and are not available in the open market.   This made us look for alternatives for YDT.  Calcia stabilized zirconia  (CSZ) earlier tried in US had been given up because CSZ reacts with sodium at 400oC and exhibits electronic conductivity on contact with sodium.  At RCL,  we felt that both these problems could be over come if CSZ is operated at lower temperatures.  LEDB of CSZ were measured down to 200oC for the first time employing a method developed at RCL (Na/Mg electrode).  The LEDB thus obtained showed that CSZ can be used for oxygen sensing in sodium if operated below 250oC.   Sensors were constructed using CSZ tubes and K, K2O reference electrode and tested in benchtop sodium loops. This experiment showed that CSZ works well as oxygen sensor though with the short life times (~3 months). After this period the emf started increasing because of N2ZrO3 formation at the electrolyte surface.

 This was sought to be overcome by converting the surface into CaZrO3.  This was done by reacting it with CaO  slurry coated over it  at 1400oC for 8 hours.  The secret was the addition of trace quantities of  Li2CO3 to CaCO3 as otherwise the coating generated had high resistance.  The coated zirconia tubes were assembled into meters with both K, K2O reference electrode and In, In2O3 reference electrode.  In, In2O3 electrode required a  soaking at 500oC for a while before assembling the sensor.  These sensors were tested in benchtop loop.  They gave a higher emf value but with a theoretical slope for cold trap temperature variation.  And they gave a better long term stability also. These sensors can be used for oxygen sensing in sodium but  individual sensors have to be calibrated.  The emf measuring system should have high input impedance  as the  cell resistnace is high.  The high zero error was attributed to the asymmetric nature of the cell.  There is a nagging doubt whether it is due to protonic conductivity of CaZrO3.  The only way to test this hypothesis was to vary the hydrogen concentration without charging oxygen valves.  We could not figure out an easy experiment to verify it.

The above are only arbitrary designs, not practically very sound.  The most successful designs world wide are the GE design based on YDT cup brazed to metal with In, In2O3 reference and the glass soldered YDT pellet  design from Rossendorf, Germany.  The development of the isopressed YDT was taken up in collaboration with CGCRI in the X plan.  YDT powder was produced at RCL based on a modified Pechini process where the combustion of the citrate is done in a controlled way.  The sintering process conditions were also standardized at RCL.  Isopressing the powder into crucibles as well as the final sintering are being done at CGCRI.

Yet the most promising design for YDT based sensor should be based on YDT rod.  The rod is easy to isopress and sinter.  This can also be easily glass soldered to Al2O3.  If an air reference electrode with RuO2  contact can be used, this sensor would give a stable emf though not theoretical.  This is similar to the Westinghouse design without the prolems of cracking.  This deserves an examination in XI plan period.

7.Oxygen in sodium

Oxygen is present in sodium as its oxide.  The  saturation solubility of oxygen in sodium has been well established and the most reliable equations are that of 1) Noden and 2) Eichelbergor.  The oxygen dissolved in sodium is assumed to obey Henry’s Law as it is a dilute solution. i.e ai = giCi.  This has not been experimentally confirmed till recetly.  The availability of an oxygen meter which can operate at 230oC and the possibility of changing the cold trap from 120 to 2300 C provided us with an easy means of verifying this.  The activity can be obtained from the meter output and the concentration from the cold trap temperature.  The activity thus measured was expected to be linear with respect to the oxygen concentration if Henry’s Law was obeyed.  But in actual measurements,  the activity was found to deviate negatively from Henry’s Law.  This was the first  such measurement ever reported in literature.  A formula for activity of oxygen in sodium with oxygen concentrationwith pure oxygen as the standard state was worked out.  This has a significant influence on the calculated  threshold oxygen concentrations for the formation of ternary oxides in sodium.  A Russian model (Heterophase model) which gives the activity of oxygen “a” as equal to (X/Xs)2 where X in the concentration and XS in the solubility was found to be more accurate in giving the oxygen activity which is inagreement with our measurements.  

8.Sensor for Hydrogen in Sodium  

Monitoring hydrogen in sodium is very crucial for detecting steam generator leaks. The most used device for this purpose is the ion pump based hydrogen detector. This uses a nickel membrane at about 450oC  with one side exposed to sodium and the other side to the high vacuum generated by an ion pump.  The ion-current of the ion pump is proportional to the hudrogen flux which in turn can be correlated to the hydrogen concentration in sodium. This is a device with quick response time and very good sensitivity.  But there is a problem with respect to the ion pump.  Special pumps with titanium anodes have to be used as otherwise the activation of the anode by the hydrogen pumping can lead to oscillation of the pump capacity and the ion-current.  Moreover the task of measuring a small current across high voltages can lead to discharge like peaks which makes them unreliable.  Even  the gasket  leaks can cause change in current and so frequent calibration becomes necessary.

The system used at FBTR makes use of a mass spectrometer to measure the dynamic pressure in addition to the ion-pump current.  This takes care of all the problems mentioned earlier.  But the whole system becomes a bit complicated and costly with the addition of MS.  There were many operational  problems at FBTR with this kind of system.

As an alternative,  a galvanic cell based hydrogen sensor was developed at RCL.  This was based on a  sensor reported by CG Smith of Berkelay Labs, UK .   This makes use of a hydride iron conducting solid electrolyte CaCl2-5%CaH2.  The reference electrode is Li, LiH slurry.  The electrodes and the electrolyte are separated by Armco iron membranes which is permeable to hydrogen.  The selection of Li, LiH as the reference electrode is dictated by the fact that this system has a hydrogen pressure similar to that in sodium at 450oC, so that there is no big gradient and migration of atomic hydrogen across the electrodes is kept to a minimum.  The choice of CaCl2-CaH2 is dictated by the zero volume change during the freezing of the electrolyte.

The sensor is configured such that the electrolyte is held in an iron cup into which another iron thimble containing Li, LiH is inserted.  The electrolyte is melted at ~800oC and the inner electrode is lowered into it.  The electrolyte is then allowed to freeze.  Because of the near zero volume change during freezing the electrolyte solidifies without  leaving a void.  The most important precaution to be taken is that the electrodes are electrically shorted while freezing.  Otherwise the electrolyte will solidify which an electrochemical gradient leading to a large zero error and drift.  The diaphragm thickness is also kept. uniform at ~0.5 mm to get a quick response.   The meter was constructed and tested in table top sodium  loops first.  Only the meters that gave stable and near theoretical values were taken for further use.  Others were rejected.

About four such meters were incorporated in the reactor, FBTR.  The first meter incorporated in the east loop performed very well for well over two years always giving immediate and reproducible response to cold trap changes and hydrogen injections.  The rest of the meters gave good outputs initially but drifted and became sluggish in response in a few months of operation. As a result more meters were constructed and tested in the lab and then placed in the reactor.  All these meters had varying outputs after some time though they gave good emf values initially.  The various reasons attributed to this  were:

1)     Formation of carbide in CaCl2 from carbon from iron and increased ionic conductivity.

2)     Diffusion of lithium through the iron thimble.

3)     Dissolution of iron into the electrolyte while casting the electrolyte while casting the electrolyte at ~800oC.

4)     Introduction of oxygen into electrolyte during fabrication.  

In order to reduce the possibility of carbide formation precautions were taken by decarburizing the steel by  hydrogen firing and then also by equilibrating in sodium containing calcium.

The lithium problem was sought to be overcame by taking a mixture MgO and CaH2 as the reference electrode.  This gives a slightly higher hydrogen pressure compared to Li, LiH and it is an original idea from Dr. R. Sridharan.

The third problem of dissolved iron in CaCl2 was tackled by reducing the casting temperature from 800oC to 700oC by taking a mixture of CaCl2-LiCl which has a lower eutectic temperature.  The alternative route of reducing the temperature as well as increasing the conductivity was tacked by taking CaBr2-CaH2 as the electrolyte which has a lower eutectic point and better conductivity.

Finally the problem was identified to be due to the oxygen contamination of the electrolyte while inserting the reference.  The development of a reference electrode introduction tool which helps move the reference electrode smoothly without loosening the teflon gland much.  Only after using this device, the meters constructed were found to give consistently good results.  To arrive at  this state,  nearly a hundred meters had to be constructed with different variables.  

It is obvious that the present  design having a Teflon gland for insulating the reference and sample electrodes needs improvement.  A better choice would be a ceramic to metal seal for separating the reference electrode from the body.  The reference capsule should have a conical bottom to help insert the reference into the electrolyte powder before welding the ceramic to metal seal. This also demands that the sensor electrode also should have a conical bottom to keep the electrolyte thickness uniform.  With all these modifications ECHM would have reached a stage where it can be used in PFBR.  CaCl2-CaH2 still remains the best electrolyte because of its zero volume change during freezing.  

9.Carbon Meter  

The electro chemical carbon meter (ECCM) for monitoring carbon in sodium is also a galvanic cell with a eutectic liquid of Li2CO3-Na2O3 as the electrolyte and pure graphite as reference electrode. This is based on a sensor developed by Hobdell of CEGB.  The improvement made here is in the reference electrode. Since the graphite  used in CEGB meter disintegrates with time and shortens the cell, a nickel sleeve was introduced in the carbon meter developed by Rajendran Pillai.  This also had long term stability problems.  Since the nickel got decarburized much faster than the graphite dissolving into it.   This problem was later overcome at RCL by including a wetting electrolyte between the graphite and the iron sleeve.  A small hole was drilled near about the top in the sleeve so that the electrolyte can get  in and fill the interface.  This concept gave better results.  But still a basic problem with this meter is the iron needles that form over a period of time. Also the meter can not be kept connected to the read out meter continuously since even a small measuring current can polarize the electrodes because of the low diffusivity of carbon. One important thing learnt in this meter  is the need for keeping the oxygen leakage through the Teflon gland very low.  Here also they will have to go for a CM seal.  A simple test for performance testing of this meter is to getter  the carbon in sodium using ss foil and noting  the maximum emf developed.  

10.Carbon in sodium  

One important experiment done using this meter was the measurement of the carbon activity in sodium for different temperatures while keeping the carbon concentration constant.  The activity variation with temperature for the same carbon concentration was more in agreement with the postulate that the carbon exists as a diatomic species  in sodium at the high carbon activity of about 0.8 as obtained with FBTR sodium.  This is at variance with the British results which showed that  carbon exists in sodium as methides.  That may be true at low carbon activities as their measurement were done with low carbon sodium.

One question that interests fast reactor operators is whether carburization of structural materials by any oil leak is serious enough to alter the mechanical strength.  Available data from the reactor operating experience and the experimental results reported indicate that it is not  serious.  The carburized layer formed discourages further diffusions and there is no deterioration in mechanical strength in the operating temperatures of the reactor.

One problem with Indian sodium is the high carbon activity and the carbon meter’s utility for detecting any oil  leakage is limited since the change in voltage output is low.  

11.Reaction of oil with sodium  

It  is quite interesting to know the way that coolant oil reacts with high temperature sodium (3~00oC)as oil  leak through shaft seal is a definite possibility in a fast reactor. Oil reacts with sodium to give amorphous carbon which is voluminous.  About 10% of carbon is released as methane in the initial cracking.  The amorphous carbon produced by oil-sodium reaction  is sticky and can clog any filter down stream.  With passage of time the graphitization of carbon increases and ends as  graphitic carbon.  So the best way to detect oil leakage in FBTR sodium is to look for methane in covergas.  

12. Graphite reaction with sodium  

Carbon solubility in sodium is low.  Graphite is stable in sodium.  So we expect  graphite rods exposed to sodium at high temperatures  be stable.  To verify this, experiments were done at RCL.  It was found that graphite swells well even at 500oC due to the intercalation of Na2O into sodium.  So the mechanical integrity of graphite exposed to sodium is suspect.   This means that any graphite to be used in sodium system should be doubly clad with steel as otherwise it can swell in the event of a leak.  A proposal to use graphite blocks between the core and the neutron detectors in PFBR was discouraged owing to this problem.  

13.Carbon meter for PFBR  

Since the PFBR sodium will most probably be having lower carbon level since sodium is going to be produced HWB using quality carbon anodes. Monitoring PFBR using electrochemical meter would would appear to be more meaningful.  Since the main source of hydrocarbon is the shaft coolant for the pump, the seal design for PFBR would be more uptodate.  Since it is a pool type reactor monitoring the covergas for methane using a fast cycle GC would be sufficient for detecting oil leaks.  Putting a carbon meter in the purification line may not be useful and necessary.  

14.Cold Trap Regeneration  

The main source of hydrogen for sodium is the corrosion generated hydrogen in the steam generator.  This enters the sodium almost  quantitatively in the steam generator.  This ends up in the secondary cold trap and the cold trap capacity gets used up fast..  As a result the secondary cold traps require periodic regeneration.  There are many approaches to this job.  The American approach was to raise the temperature to (>420oC) and convert it into hydroxide by reacting the hydrogen and oxygen.  The NaOH generated is in liquid form and sink as it has a higher density.  They use a catch pot to collect this at the bottom of the cold trap.  There is also the Japanese approach where argon flushing over the sodium after raising the sodium temperature.  Standardising the procedure for PFBR cold trap generation was taken up at RCL during X plan.  Based on the basic studies done on the Na-O-H system a reliable procedure for secondary cold trap generation must become available at the end of the plan.  This work should be continued and a regeneration procedure for PFBR cold traps should be evolved with utmost priority.  

15.Limitations of hydrogen sensor for steam generator leak detection  

In-sodium hydrogen sensors can measure the hydrogen dissolved in sodium as sodium hydride.  Almost 50% of the hydrogen contribution in the event of a steam leak is in the form of hydrogen bubbles.  The dissolution of this gas into sodium is temperature dependent.  The average size of these bubbles produced in the region of 2mm or so and good fraction of it can escape into the covergas  depending on the temperature.  This is because of the slow kinetics of hydrogen gas dissolution in sodium at low temperatures.  Under these conditions there is a fear that the in-sodium hydrogen sensors may not be able to detect   steam generator leaks at  low sodium temperatures as obtained during reactor startup.  In order to verify this, small amounts of water taken in a specially designed capsules equipped with fusible seals which melt at low temperatures (~250o C) and injected into sodium in RASCL. The in-sodium meters gave a good response (~75%) since 50% of the hydrogen content of the steam leak ends up a NaOH.  NaOH is highly soluble in sodium and helps the assimilation of the hydrogen  by the liquid metal.  So there is no possibility of in-sodium meter missing out any steam leak through the signal is not   quantitative  at low sodium temperatures (T<250oC).  

16.Cover gas hydrogen meter  

For unambiguous and positive detection of any steam generator leak at sodium temperatures below 3000C (as in reactor start up) it is helpful to measure  hydrogen in cover gas.  The cover gas hydrogen meter  developed by RCL for use in FBTR is based on the diffusion of hydrogen into a nickel capillary tube at  400o C.  The capillary is swept by argon carrier gas at a low flow rate (30 ml/min) and the argon is monitored for hydrogen using a thermal conductivity detector (TCD).  The length of the tube was optimized at 6 M so that the signal is maximum and flow independent.  This concept was chosen for FBTR since it was suitable to incorporate it as an add on and avoids problem introduced by sodium aerosols.  The TCD is a proven detector and commercially available.  The electronics and measuring system can be located away from sodium loops. This device was extensively checked at RCL and then at EDG loops for its reliable performance before incorporating it in the reactor.  It has performed well in the reactor, always responding quantitatively for hydrogen injections.

One drawback with this equipment is the possible reduction in sensitivity with the passage of time since the nickel can recrystallise.  But the degradation is less than the in-sodium meters of the same type as in PFR where the additional pick up of impurities from sodium by the nickel makes the deterioration a bit faster.

The cover gas hydrogen meter based on a nickel crucible and an ion-pump can give a faster response.  The system is also simpler.  But there are some practical problems such as heating the entire area of the nickel crucible uniformly which is difficult in the gas phase.  The ion-pump also has problems of activation by hydrogen and change in the capacity leading to base line shifts.  

17.Sodium leak detector for SGTF  

The  oil  heated  sodium  carrying tubes  in  SGTF  facility  can  develop  cracks  and  the sodium leaks have to be detected early before they can expand.  The logical way to detect these leaks is to look for Na2O aerosol in the flue gas.  A leak detection device based on a flame photometer detector was designed and developed by RCL.  The flue gas is sampled and let into a flame photometer and the sodium emission lines are monitored using a wave length filter and a photo tube. This detector is based on a proven principle and so it is suitable for plant use. The detector was extensively tested at the laboratory by aspirating sodium containing solutions.  This was finally incorporated in SGTF and tested by carrying out sodium fires in the oil burner area.  However, over the time,  some problems were encountered in the signals as the aerosols started depositing on the sampler walls.  Thus there is a need to keep the length of the sampler line as small as possible.  This modification in the sampler line is being carried out by EDG.  

18.Activity Transport  

Radionuclides are produced in the reactor core 1) by activation of reactor structural materials and 2) fission of the fuel. These radionuclides can get  released into sodium, get transported to components outside the core and deposit there leading to radiation build up in areas and components requiring repair and maintenance.  This will lead to prolonged shutdown and increased exposure to personnel during repairs.  Thus there is a need to prevent  such build ups either by minimizing their transport or by continuously removing the radionuclides from sodium.  Among the activation products,  Mn-54 has been found to be the major nuclide that gets transported to cold legs forming almost 90% of the radioactivity burden.  This is because Mn-54 is produced by (n, p) reaction of Fe-54 which is a  component of the structural materials. Mn-54 has a high diffusion coefficient in steel and a higher solubility in sodium.  If Mn-54 problem is solved, the activity transport problem is almost solved in fast reactors. Other activation products like Co-60, Cr-59, etc., deposit immediately downstream from the core and do not migrate much. It is true that the activity transport problem is less acute in pool type reactors.  However,  the contamination of IHX,  and the pump still remain.

Among the fission products that get released into the coolant on fuel-failure, Cs-137 is the only nuclide that gets quantitatively released.  Since this is highly soluble in sodium it gets mixed with the entire primary sodium.  In the hot leg regions Cs-137 can diffuse into the grain boundaries and complicate decontamination of components for repairs.  Cs-137 can also evaporate preferentially and get fixed in the sodium frosts in the primary covergas space.   This is found in the above sodium regions of the pump shaft as well.  Other fission products like Ce-144, Y-90, Ba/La-140 etc., are released only by a small fraction into sodium and deposit immediately downstream from the failure location.  So, Cs-137 is the only fission product that should b dealt with.

Among the impurities in sodium that can get activated Zn is the  most important.  Zn-65 has been observed in almost all the reactors. Ag-110m is another such nuclide. Sn-113  also has been found.  All these are highly soluble and are found throughout  sodium. But their levels never rise beyond µCi levels as the amounts of these impurities is small.  

19.Radionuclide traps  

Brehm and Coworkers in USA and Stamm in Germany have shown that pure nickel has a high sorbing capacity for Mn-54 at high temperatures(~500oC)  because of the formation of intermetallics at the surfaces.  A nickel foil kept at the fuel subassembly outlet at 500 plus temperatures can pick up the MN-54 released by the core materials in the upstream and act as a Mn-54 trap.  Even electroplating the upper sides of the fuel-pin can do this job.  The most  successful design concept is a thin nickel foil suitably dimpled so that when rolled it can form a cartridge with properly spaced layers through which sodium can flow easily when kept at the top of the fuel sub-assembly.  It is worth attempting this for PFBR also even though it is a pool type reactor and activity transport is not a big problem.  Even equipping a few fuel pin with MN-54 trap can ensure a cleaner primary system IHX and pumps).

Reticulated Vitreous Carbon in the form a porous foam supplied by Ergenics Inc of USA hs been found to be an excellent sorbent of Cs-137 by forming intercalation compounds when exposed to sodium at 200o C.  RVC is a high surface area material with a smooth surface that does not release carbon particles when sodium flows through it. It can take up Cs upto very high capacities and a small volume can clean up the entire sodium charge. Since this foam has a strategic value as it is used an insulator in the space reentry vehicles and in rocket noses, it is not freely available.  We have stocked enough for FBTR.  But for PFBR either a new source has to be identified or it has to be developed indigenously.  The two options should be pursued with equal intensity.

There are two designs of CS-137 trap.  In out of core concept used in western reactors  a small vessel filled with RVC is kept at 200oC in a purification loop outside the core.  This is usable all the time  even when the reactor is in operation.  The vessel should be shielded with lead and will be the last  component to be removed  from the reactor during decommissioning.  The amount of RVC required is quite small and the sodium cleanliness can be kept  very good.

The alternative in-core caesium trap involves a subassembly incorporating a known and adequate amount of RVC  which is inserted into the ore during reactor shutdown and when sodium temperature is low.  After the clean up this has to be removed and disposed off.  This is the concept  used in Russion reactors.  There is no extra pipeline or shielding required except that RVC used once can not be used again.  This calls for a larger amount of RVC to be stocked.  This is more desirable when the anticipated fuel pin failures are low.  This is the concept chosen for PFBR.

Other radionuclides that are amenable for trap purification by traps are Zn-65, Ag-110m and Hg-203.  These are highly soluble and can be removed by exposing sodium to copper turnings at a low temperature (~250o C) in a purification loop.

However, Mn-54 trap and Cs-137 trap are sufficient for reducing the activity of a fast reactor sodium.  

20.Fuel-coolant  reaction  

The mixed oxide fuel can react with sodium coolant in the event of fuel-sodium contact when the clad looses its tightness.  The reaction product is a low density material and its formation leads to increase in fuel volume.  This results in expansion of the defect and exchange of sodium in and out  of  the fuel. Ultimately this can lead to release of fuel into flowing sodium which is not desirable.  Also understanding of the fuel-coolant reaction is essential in operating the reactor till the failed fuel pin is located and removed.

The mixed oxide reacts with sodium to give a compound Na3(U, Pu) O4 where the ratio of U/Pu remain sense in the fuel and the product.  At equilibrium the oxide, Na3MO4 and the sodium are coexisting.  As per phase rule,  the  three phases would have same oxygen potential for all the three phases.  The reaction is an oxygen consuming reaction and can proceed forward only till the fuel can contribute oxygen, since  the amount of sodium entering the fuel pin initially is limited and so cannot contribute much oxygen.  The equilibrium O/M in the fuel and the oxygen in sodium have been measured experimentally.  This could also be calculated from the free energy of formation of Na3MO4.  These values are in agreement with each other.  Using this equilibrium oxygen potential it is possible to calculate the extent of the volume increase for the fuel of a given initial O/M.

However,  the presence of a temperature gradient gives rise to an O/M ratio distribution in the fuel from the centre to the periphery.  Also the Na3MO4 compound is not stable above ~1200oC giving rise to a dissociation isotherm.  As a result, the formation of the reaction product is limited to the outer 200µ of the fuel.  Moreover, the poor thermal conductivity of the reaction product and the reduced thermal conductivity of the reduced O/M of the fuel results in a higher temperature gradient.  This pushes the reaction product boundary to still further outside.  

21.Evolution of fuel failure  

The DND signal as a function of time gives a measure of the fuel surface exposed to sodium.  RBCB experiments in EBR II and experiments in SILOE reactor indicate that the evolution in general tapers off after reaching a shoulder.  This indicates that fuel-sodium reaction product acts as a barrier for further fuel reaction. This factor ensures that the fuel-coolant reaction is not catastrophic in nature and will not lead to extensive release of fuel.  Only in case of a few disturbed operations the fuel-failure tends to increase monotonically. 

The time available for the reactor operator from DND signal is noticed to fuel-removal DND threshold varies from 10 mm to a few hours depending on the burnup and clad swelling.  This is sufficient for locating the failed fuel and shutdown the reactor for its removal.  This calls for reliable fission gas detector, delayed neutron detector and failed fuel detection system.  

22. High plutonium fuels  

The oxygen available for the reaction increases with the plutonium content of the fuel as plutonium is less oxygen binding.  But this factor is counteracted by the poor thermal conductivity of the fuel which pushes the reaction front still further outside and limits the reaction product  formation.  Moreover the buffering action of the clad and the fuel-fission product system which always limits to O/M at the outer periphery to 1.98 also helps in reducing the severity of the reaction.  Thus in practical terms it is still possible to operate the reactor with high plutonium fuels even with slightly high O/M without any great risk.  

23. Failure at the storage positioner  

The fuels that fail in the core and have undergone the initial reaction before they are removed and moved over to storage position at the outer rings of the core behave very benignly.  There is no further opening of the fuel or release of the particulates.  But in the case of the fuels that fail after they are moved to the storage position the reaction induced swelling is severe since the fuel cannot creep.  This has resulted in the break up of some  fuels into pieces.  This factor again points to the importance of the plasticity of the fuel on power and its role in accommodating the volume increase due to  fuel-sodium reaction.  

24. Sodium removal from reactor components

Components taken out of the sodium system for repair and maintenance have to be cleaned off sodium as a first step.  That includes fuel pins removal from the reactor for reprocessing.  Different techniques have to be adopted for different components as the process requirements are slightly different.  

25. Sodium removal from fuel-subassembly  

The subassembly removed from the core has a large decay heat.  So this has to be continuously cooled by blowing nitrogen.  This gas jet is able to remove a good amount of sodium sticking to the fuel in between the fuel pins.  After the initial nitrogen flushing  period steam is injected into the nitrogen stream while continuously monitoring the hydrogen content of the exit gas.  This process is quite fast and highly suitable for fuel pins.  However, this process can not be used with components that are reusable.  

26. Water vapour nitrogen process

Components such as pumps, 1HX, valves etc., have to be repaired and reused.  The temperature of the components has to be kept below ~100oC during sodium removal in order to prevent caustic stress corrosion cracking.  This is done by passing nitrogen containing water vapour over the components kept at a low enough temperature in a closed vessel.  The water vapour/steam injection is done in a controlled manner.  The process is slow since the NaOH layer formed at the surface slows down the water supply to the sodium surface.  

27. Water vapour-CO2 process

Here CO2 moistened by passing through water kept at a slightly elevated temperature but below 90oC is passed over the component kept in a vessel.  The CO2 keep the pH of the layer low and prevents caustic stress corrosion cracking.  The NaHCO3 layer formed at the surface is porous and permits the sodium-water reaction to proceed without hindrance.  The reaction is monitored by measuring the hydrogen in the exhaust gas.  There is no fear of NaOH droplets dripping from the components and releasing NaOH aerosols and fouling the hydrogen sensor.  This is the process recommended for use in PFBR.

Other processes like water-vacuum process, autoclave process, and alcohol process have specific limited applications.  

28. Sodium disposal

When there is a need to dispose off large quantities of sodium the process of choice is the NOAH process.  Here the liquid sodium is pumped into 10 M NaOH where the sodium reacts with water in a smooth and a controlled manner.  Of course the hydrogen generated has to be exhausted out well.  The trick here is the design of the nozzle ejecting sodium into fine droplets.  NaOH solution is pumped around the nozzle to take the sodium away from the nozzle.  This prevents the attack of the nozzle metal by the hot alkali produced at the tip.  Of course the selection of the nozzle material is also to be carefully done.  The alkali has to be cooled using a heat exchanger.  

29. Sodium reuse

Reuse and disposal of primary and secondary sodium when the reactor is decommissioned is important.  The secondary sodium can be cold trapped for sufficient length of time to remove any tritium and allowed to cool for sufficient time (10 half lives) so that Na-24,  if any can completely diedown. Then it can be used for any chemical process.

The primary sodium can be first polished using 1) RVC for removing Cs-137 and Cs-134 and 2) copper trap for removing any activation products (Hg-203, Sb 113, In 129, Ag-110 m etc.,).  Then it can be allowed to cool for about 6 years so that Na-22 levels come down.  Then it can be transported in shielded tanks for other reactors within the campus.  If it has to be transported to far away places then a greater cooling time would be required (~12 years).  Anyhow it is not desirable to convert the Na to NaOH as the environmental impact is severe.  

30. Suggestion for the chemistry lab of PFBR  

PFBR should have an in-place chemistry laboratory which can carry out chemical quality control of sodium, water and cover gas.  There should be an instrument room, radiochemistry room and a counting room sufficiently manned to carry out all the routine analysis. An Atomic Absorption Spectrophotometer, pH meter, spectrophotometer, ion-selective electrodes and a quick cycle gas chromatograph are the essential instruments.  Specific high sensitivity analysis like ICP-MS can be done at Chemistry Group, IGCAR.  A sodium box, sodium distillation set up, active hoods are the facilities required at the Radiochemistry room.  Three chemists (from training school) should be recruited in advance and trained at Chemistry Group to get their doctorate eventually after posting at PFBR so that chemistry problem in PFBR could be tackled.

 31.Immediate problems to be attended

         1)     A process for regenerating PFBR cold trap should be developed.

2)     An indigeneous alternative to RVC should be developed.

3)     A sodium compatible high alumina concrete should be developed.

4)     Mn54 trap should be tested in FBTR.

5)     YDT based oxygen meter should be developed for use in PFBR.