Isotopically enriched nitrides for nuclear power

(1)

Isotopically enriched nitrides for nuclear power

Simo Saarinen June 10, 2012

(2)

(3)

“Technological advance is an inherently iterative process.

One does not simply take sand from the beach and pro- duce a Dataprobe. We use crude tools to fashion better tools, and then our better tools to fashion more precise tools, and so on. Each minor refinement is a step in the process, and all of the steps must be taken.”

Chairman Sheng-ji Yang, “Looking God in the Eye”

– Sid Meier’s Alpha Centauri, 1999

3

(4)

(5)

Abstract

The goal of this project was to create uranium nitride powders from uranium metal with minimal loss of nitrogen. Uranium metal was first hydrided in a flowing mixed hydrogen-argon atmosphere, then dehydrided in pure argon, and finally nitrided in a pure nitrogen atmosphere. Measurements showed a near- perfect uptake of nitrogen once the reaction was under way. The resulting powders were clearly shown to be uranium sesquinitride.

5

(6)

(7)

Acknowledgements

I’m not very good at thanking people. That doesn’t mean I’m not grateful for everything everyone’s done, even though it may sometimes seem that way – I’m simply bad at expressing it. Because of this, these acknowledgements will be short. I would like to thank everyone I’ve been working with in the fuel lab these last few months: Pertti, Kyle, Carl and Patrik, for good company and good help. I would also like to thank Mikael, for good supervision and excellent comments and advice. And finally, I’d like to thank all the friends I’ve made and studied with – you know who you are, and you should know I probably wouldn’t have gotten this far without you.

Thank you.

//Simo.S.

7

(8)

(9)

Introduction

Nitride fuels are the primary choice of fuel for the Lead-Cooled Fast Reactor due to their advantages over common oxide fuels in the more demanding environment of a fast reactor.[1] However, they have at least one major drawback: ¹⁴N changes into ¹⁴C through the ¹⁴N(n,p)¹⁴C reaction, which has a sufficiently large likelihood of occuring in a fast neutron spectrum to be an issue.[2][3]

Naturally, there are ways to deal with this. One is to remove the¹⁴C during reprocessing as a barium carbonate slurry and then storing it separately as intermediate level waste after cementation.[4][5] Whether this is the best solution has been questioned[6] and will be even more so should nitride fuels be used at industrial levels.

Another solution would be to use nitride fuels enriched in ¹⁵N, though this will be fairly expensive, since the fuel will have to be enriched to over 99.9 %.[7] In order to minimise costs, a fuel fabrication route which wastes a minimal amount of ¹⁵N will have to be devised. In this project, a simple method with very high efficiency has been tested and found satisfactory for laboratory-scale production with, in this simple masters-student’s humble opinion, potential for industrial application in the future.

But before discussing this further, a closer look at nitride fuels in general is in order:

1.1 Nitride fuels

1.1.1 Importance

UN has a higher thermal conductivity than UO2 as well as a higher melting point,[8] making higher operating temperatures possible. UN also has a smaller amount of light atoms, causing less moderation in the fuel which

11

(12)

UN α-U2N3 β-U2N3 UN2 ref.

Structure f.c.c. NaCl b.c.c. Mn2O3 h.c.p. La2O3 f.c.c. CaF2 [10]

Stoichiometry (1) 1.54-1.75 1.45-1.49 1.75-(2) [12]

Lattice parameter [Å] 4.880 10.678-10.580 5.31 [13]

Density [g/cm³] 14.32 11.24 [13]

Table 1.1: Uranium nitride compounds

allows for a better neutron economy in a fast neutron spectrum. Nitrides have the smallest amount of fission gas release during operation out of most fuels considered today for use in fast reactors,[2] and have been shown to achieve high burnups. Furthermore, UN and PuN are isostructural, and thus mutually soluble[9] - a particularly important trait when used in breeder reactors, since their fuels will contain larger amounts of plutonium than current light water reactor fuels.

These properties might be interesting in common light water reactors, but are unlikely to be significant enough to replace current oxide fuels. Rather, the true strengths of nitrides are shown when looking at fuels for the transmutation or burning of higher actinides, fuels for fast neutron reactors, and space reactors.

For the transmutation or burning of higher actinides, the high safety afforded by the nitride fuels is their main strength; for fast reactors, the low amount of light atoms in the fuel; and for space reactors, the higher density of the fuel makes it possible to have a smaller reactor, lowering the launch weight.

1.1.2 Uranium Nitrides

There are four different compounds of uranium nitride: UN, α-U2N3, β-U2N3

and UN₂,[10] though some sources report that stoichiometric UN₂ is practically impossible to produce, except for under extremely high pressures.[11][12]

See table 1.1

Phase diagram The equilibrium phase diagram prepared from various literary resources by Hiroaki TAGAWA[12] is shown in fig. 1.1. As can be seen, at lower temperatures the phase is mainly α-U₂N₃ in solid solution with either UN or UN2. At about 800^◦C the β-U2N3phase appears, existing mainly in solid solution with either UN or α-U₂N₃ before finally dissociating at higher temperatures into UN and N2.

(13)

1.1. NITRIDE FUELS 13

Figure 1.1: Equilibrium phase diagram at 1 atm. nitrogen pressure

(14)

(a) UN – NaCl (b) β-U2N3 – La2O3

(c) α-U2N3– Mn2O3 (d) UN2 – CaF2

Figure 1.2: The crystal structures of the uranium nitride phases [14]

(15)

1.1. NITRIDE FUELS 15

UN UO2 Ref.

Crystal structure NaCl CaF2 [8]

U density [g/cm³] 13.53 9.6 [8]

Lattice constant [Å] 4.889 5.470 [8]

Melting point [^◦C] 2850±30 2750±40 [8]

Thermal conductance at 200 ^◦C . . . 18 5.9 [8]

. . . and at 1000^◦C [W/m-K] 26 2.9 [8]

Expected centreline temperature [K] 1000 2350 [16]

Table 1.2: Comparison between UN and UO₂

Crystal structures and stoichiometries It is interesting to note that there is no clear discontinuity when going from the α-U2N3-phase to the UN₂-phase. Instead, the two phases exist in solid solution at higher stoichiometries, the UN₂crystallising directly in the α-U₂N₃-lattice. In fact, the

Mn2O3-structure of α-U2N3can be seen as a distorted CaF2-structure.[15][11][13][10]

One should also note that there is significantly less information on β- U2N3, most likely owing to its narrow band of stoichiometry, and in many older resources it is not even mentioned.

UN – UO₂ comparison Seeing as how UO₂ is the most commonly used fuel today, a comparison highlighting the advantages of UN is in order.

As can be seen in table 1.2, UN has a higher uranium density and thermal conductance, with a lower centerline temperature at expected working conditions, than UO₂. The melting point of UN is less clear, however: it has been reported as low as 2650^◦C[17] and as high as 2850 ^◦C.[8][12] To add to the confusion, it is rarely reported at any specified nitrogen pressure, except for in a report by Hiroaki TAGAWA[12], where it is clearly stated that the melting point of UN in 2.5 atm of N₂ is 2850^◦C.

In any case, the melting point of UN is similar to that of UO2. With this in mind, it is interesting to note that the expected centerline temperature of UO2 is more than 1300^◦C higher than that of UN.

A comparison of the thermal conductivity of UN (at 100% density) and UO₂ (at 95% density), based on the work of Ross et al.[18] and Ronchi et al.[19], respectively, is presented in fig. 1.3. It clearly shows two great differences between the fuels: not only is the thermal conductivity in UN 3-10 times greater than in UO₂, but in UO₂ the thermal conductivity also decreases with increasing temperature, for a large part, whereas in UN it increases with increasing temperature.

This means that as UN-fuel gets hotter, it begins to transfer its heat more

(16)

Figure 1.3: Thermal conductivity for UN and UO₂

efficiently to the coolant, making the fuel heat up slower. This naturally makes it more unlikely to have a core meltdown in a UN-fuelled reactor, as any unplanned temperature increases will be slower and, due to the higher thermal conductivity of UN-fuel, have a lesser effect on the fuel itself. The higher thermal conductivity also allows the UN-fuel to operate at higher temperatures, as the fuel centreline temperature will be lower than in UO2- fuel at the same fuel surface temperature, due to the higher efficiency of heat transfer.

1.1.3 Fabrication routes

While there is currently no standardised method for producing uranium nitrides, the two most common methods are listed below:

Carbothermic reduction Also called “carbothermic nitriding”, this method consists of mixing UO2 with carbon powder and pressing the mixture into briquettes. It is then heated in vacuum to 1600 ^◦C which transforms the UO2 into UC through the following reaction[20][21]:

UO₂ + 3C ⇀↽ UC + 2CO

The uranium carbide is then subjected to a flow of 94-6 % N2-H2mixture, which transforms the UC into UN through the following reaction:

UC + ¹₂N2 ⇀↽ UN + C

(17)

1.1. NITRIDE FUELS 17 This also evacuates the excess carbon by transforming it into CH4through the following reaction:

C + 2H₂ ⇀↽ CH4

Although if a sufficiently large surplus of carbon powder is used initially, it may react with the hydrogen and nitrogen to form hydrogen cyanide[22]:

UC + ¹₂H2 + N2 ⇀↽ UN + HCN

This method is generally considered as the one which will be used in industrial fabrication of uranium nitrides.[23] It has some problems with the purity of the fabricated nitrides, but it has been shown capable of creating ni- trides with <500 ppm C and <200 ppm O contamination, as well as a weight fraction of 5.50% nitrogen. The weight fraction of nitrogen in stoichiometric UN is 5.55%. [20]

A slight variation, possibly more modern, on this method involves greater mixing of the powders but no pressing into briquettes. They are then immediately heated in a nitrogen-hydrogen atmosphere.[24]

Hydriding/Nitriding The hydriding/nitriding route is well-suited for laboratory-scale production, since the nitrides can be made very pure, but this requires high-purity uranium metal as starting material.[21] Most commonly, the uranium metal is cleaned with acid to remove surface oxidation[21][25]

then hydrided by heating in a hydrogen atmosphere to 220-230^◦C[23][21][17]:

2U + 3H2 ⇀↽ 2UH3

This has the advantage that UH3, having a much lower density (or in- versely, much larger molar volume) than uranium metal, falls or spalls off as a powder, making the following steps in the process much faster.[26]

The uranium hydride is then heated in vacuum or an inert atmosphere to about 400 ^◦C in order to dehydride the sample, decomposing it into pure uranium powder and hydrogen gas by reversing the previous reaction.[21][17]

In some cases this step is ignored, as it is not strictly necessary to dehydride before nitriding.[27]

Finally, the uranium powder is nitrided by heating it in a nitrogen atmosphere at up to 800 ^◦C,[21][17][25] turning it into uranium sesquinitride:

4U + 3N2 ⇀↽ 2U2N3

(18)

This reaction is very exothermic and has been claimed to potentially melt the uranium metal if performed in an uncontrolled fashion.[28] It proceeds at most temperatures above 250 ^◦C[17][21][25] but is fastest around 500- 600 ^◦C.[17] At 800 ^◦C the reaction begins to slow down, as some of the sesquinitride starts to decompose.[17]

The stoichiometry of the sesquinitride produced depends on temperature and nitrogen pressure, where lower temperature and higher pressure will produce hyper-stoichiometric α-U₂N₃ and conversely, higher temperature and lower pressure will produce near-stoichiometric α-U₂N₃ or even hypo- stoichiometric β-U2N3.

Finally, the sesquinitride can easily be decomposed to the mononitride in an inert atmosphere or vacuum at temperatures above 1000 ^◦C, though the exact value used varies in the litterature, from 1050^◦C[17] through 1120

◦C[21] and 1350^◦C (in 1 atm N₂)[12] to 1400 ^◦C.[25]

NOTE: Notable differences between the method above and the one used in this project are:

First – the uranium metal is mechanically polished rather than cleaned in acid, in order to reduce oxidation. This is very efficient at removing surface oxidation, but creates a large volume of uranium-contaminated waste.

Second – all of the literature referenced above have placed the uranium in boats of a resistant material and controlled the atmosphere through pressure.

In this project, the uranium has been placed on a silica filter creating a fluidised bed effect, and the atmosphere has been controlled through flow meters/controllers. It is generally preferable to perform nitriding reactions in a flowing gas rather than a static atmosphere.[29]

1.2 Previous work at this lab

Some work on this subject has already been done at the Reactor Physics’

Fuel lab:

In 2009, Pertti Malkki performed the first investigations into manufacturing methods for mixed uranium-zirconium nitrides, as described in his thesis “Establishing a nuclear fuel laboratory: Investigating manufacturing methods for a mixed uranium-zirconium nitride”.[30]

In 2010, Miquel Torres Oliver looked into the production of mixed uranium- zirconium nitride powders from metallic uranium-zirconium alloys, as described in his thesis “Production of Mixed U-Zr Nitride Nuclear Fuel Pow- ders from Metallic U-Zr Alloys”. Although the process proved difficult, he

(19)

1.2. PREVIOUS WORK AT THIS LAB 19 successfully alloyed uranium and zirconium and then synthesised a uranium- zirconium nitride in solid solution.[31]

In 2011, overlapping somewhat with Oliver’s work, Tobias Hollmer not only built the gas system and it’s software, SCraM - still in use today at the lab - but also wrote his thesis titled “Manufacturing methods for (U-Zr)N- fuels”. The protocol he set up for synthesising uranium nitrides was the basis for all the experiments performed in this project.[32]

(20)

(21)

Chapter 2

Equipment and method

2.1 Equipment

2.1.1 Overview

The samples are prepared and placed in a quartz tube (of appropriate size) on a filter in the centre of the tube. A thermocouple, connected to the computer, is wound around the tube on the outside, so that it is as close to the centre of the sample as possible. The quartz tube is placed in the furnace and attached to the gas system (fig. 2.1a).

The gas system consists of two lines, connected to three different inlets of gas: one line for hydrogen, and one for nitrogen or argon.

The inlets are connected to the gas lines through separate pressure reg- ulators that decrease the gas pressures from 15 bar to approximately 4 bar.

They are then passed through manual SHO-RATE flow meters/controllers.

After this point, the nitrogen and argon lines are connected through a common Swagelok T-couple.

NOTE: This method of connecting the nitrogen and argon lines is not ideal and was initially installed in order to achieve the argon-nitrogen mixture used in the first experiment, UA120214, and then never reversed. It is not a major problem however, since the entire system is always flushed through the glovebox before nitriding in order to have a pure nitrogen atmosphere.

The gas lines then pass through a series of gas cleaners (fig. 2.1b and fig. 2.1e). The nitrogen/argon line is first passed through a tube of silica gel, then a tube of heated copper turnings, at about 550 ^◦C. After this, the line is passed through a set of three tubes in sequence, filled with first silica gel again, then magnesium perchlorate, and finally ascarite. The fourth and

21

(22)

final step consists of a tube of heated magnesium turnings, at about 450^◦C.

At this point, the nitrogen/argon gas line diverges in three directions: one to the glove-box, one to the sintering furnace, and one to the synthesis furnace.

The hydrogen line contains fewer gas cleaners, consisting only of three tubes in sequence, filled with silica gel, magnesium perchlorate and ascarite respectively, exactly like the corresponding part on the nitrogen/argon line.

After these steps both gas lines pass through individual Bronkhorst flow controllers (fig. 2.1d), connected to the computer, before being mixed in the gas mixer, which consists of a cylinder filled with two distinct sizes of glass balls (fig. 2.1c).

After the gas mixer, the entire system is finally connected to the quartz tube in the furnace.

After the furnace, the quartz tube is connected to a Bronkhorst flow meter that is connected to the computer. Then, the gas is passed through a pair of mechanical filters comprising a large reservoir of outflow gas, to avoid oxygen contamination from the atmosphere in case there is an under-pressure in the quartz-tube, before finally being evacuated into the atmosphere.

2.1.2 Flushing the gas system and dead volume

An astute reader may have noticed the three valves beneath the heated magnesium gas cleaner in fig. 2.1b and 2.1e. These valves go to the glove box, the sintering furnace and the synthesis furnace, respectively. This set- up has two distinct advantages: first, every part of the lab using inert gases can use the same gas cleaning system; second, it allows for flushing a large portion of the gas system.

Flushing the gas system is performed simply by opening the air lock on the glove box to the atmosphere while the desired gas is flowing, and keeping it open for at least one minute. This ensures that the entire gas system, up to the point where the valve is, will be filled with the correct gas.

Unfortunately, since the valve could not be placed immediately before the sample inside the furnace, there is a small ‘dead volume’ of the previous gas between the sample and the valve. This dead volume needs to be bypassed or pushed out before the nitriding reaction (or any other reaction with gas, for that matter) can take place. This is why the nitrogen uptakes plotted in chapter 4 are not at 100% from the beginning.

Curiously, the dead volume measured in all the production experiments was less than the 0.2454 l expected for the quartz tube alone (having inner diameter of 25 mm, height of 1000 mm, with the sample placed on the filter at 500 mm), excluding the piping from the valve, through the gas mixer, to

(23)

2.1. EQUIPMENT 23

(a) The furnace, a quartz tube with filter, and insulation

(b) The heated gas cleaners

(c) The gas mixer (d) A flowcontroller

(e) The gas system, underside

Figure 2.1: The various parts of the synthesis gas system

(24)

(a) Static, no gas flow (b) Active

Figure 2.2: The fluidised bed effect, simplified

the quartz tube. There is no clear explanation for this. The dead volume in the analysis experiments, using a smaller quartz tube, was considered to be of negligible importance.

2.1.3 Fluidised bed effect

An important part of the set-up and procedure used in this project is the fluidised bed effect. When the uranium metal reacts with hydrogen to form a powder of uranium hydride, the powder falls or spalls off of the metal and lands on the silica filter in the quartz tube. This filter acts as a fluidised bed distributor: it is too fine to let the powder fall through, yet porous enough to allow the reactant gas to pass (fig. 2.2a).

As the gas goes through the filter, it becomes slightly more turbulent, thus mixing with the powder to a greater extent, increasing the speed and efficiency of the reaction. But the real advantage of a fluidised bed effect becomes apparent when the gas flow is large enough to lift the powder particles off of the filter, greatly mixing them and effectively turning the powder’s behaviour from static to fluid (fig. 2.2b). The result of this is the reactant gas having a much larger activity, as the powder particles are effectivey separated and continuously mixed while “suspended in mid-air”.

One risk associated with this is the possibility of the uranium powders escaping the quartz tube, if the gas flow is too large. In order to prevent this, two identical filters of glass wool are inserted at the top of the tube.

The second filter is installed as defence-in-depth, a second layer of contain- ment – during the course of this project it was never observed to have been contaminated, whereas the first one always was. The first of these two filters is slightly visible, uncontaminated, in fig. 2.1a.

(25)

2.2. SYNTHESIS PROCESS 25

(a) SCraM (b) Manual calibration program

Figure 2.3: The synthesis software

2.1.4 Software

The program controlling the synthesis system and logging the data is called SCraM, short for “Synthesis Control and Monitoring”, made by Tobias Hollmer.

It is written in DASYLab from National Instruments, and using this program it is possible to set up the entire synthesis process, by programming the temperatures and gas flows, and leave it to run by itself without manual interference, except for when switching between nitrogen and argon – at least in theory.

Unfortunately, the program has been growing ever more erratic as of late and will occasionally crash for no apparent reason, with various effects. As the program still functions mostly well enough, and there are plans for a major overhaul of the software, computer and data acquisition hardware, no monumental effort has been made to find and correct these issues.

NOTE: To make things easier on the software during synthesis, the data logged is only calibrated for nitrogen gas. This means that the gas flows logged are only accurate during the nitriding step. Fortunately, this is the only area of interest, so it is not an issue. To get the correct flows under argon, the data needs to be adjusted by a factor 0.7141, which is deceptively similar, but wholly unrelated to, the mass ratio between nitrogen and argon gas.

2.2 Synthesis process

The standard process for nitriding uranium in this lab, developed by Tobias Hollmer,[32] can be seen in table 2.1.

The main differences between this process and the one adapted for this project is the addition of a dehydriding step before the nitriding, by heating to 600 ^◦C under argon before nitriding at that temperature. The result is

(26)

Gas 1, [l/min]

Gas 2, [l/min]

Temp [^◦C]

Ramp [^◦C/min]

Time [s]

Time [h]

Step

Ar, 0.4 H₂, 0.5 20 0 Hydr.

300 20 840

300 0 300

200 -5 1200

225 5 300

225 0 18000

20 -10 1230

6.08

N2, 0.5 H2, 0 225 10 1230 Nitr.

225 0 1200

500 10 1650

500 0 1200

1.47

Ar, 0.5 1150 15 2600 DeNitr.

1150 0 18000

20 -15 4520

6.98 Total time [h] 14.52 Table 2.1: Tobias’ process for synthesis of UN

(27)

2.3. METHOD 27 that a pure uranium metal powder is nitrided, rather than a uranium hydride powder. Also, the denitriding step has been omitted entirely, making it possible to perform experiments in one day rather than two.

2.2.1 XRD

For the analysis of our powders we use a SIEMENS D5000 X-Ray Diffrac- tometer. The powders are mixed in paraffin oil before measurement, in order to prevent oxidisation. The oil has a negligible effect on the measurement.[33]

2.3 Method

2.3.1 Uptake

The monitoring of the reaction process inside the quartz tube is performed by the three Bronkhorst flow controllers/meters, measuring in l/min. The flow controllers on each gas line, before the gas mixer, measure and control the inlet flows into the quartz tube. The flow meter after the furnace measures the outlet flow. Thus the volume of gas consumed or released in the reaction can be calculated as follows:

V˙_out− ( ˙VH2,in+ ˙V_N₂_/Ar,in) = ˙V_reaction

Clearly, if ˙V_reactionis positive, there is a net gas release from the sample.

If ˙Vreaction is negative, there is a net gas consumption in the sample.

NOTE: When plotting the uptake of nitrogen, it is worthwhile to take the absolute value of ˙Vreaction, as this is intuitively easier to understand. This is what has been done for the results in chapter 4.

From this, the amount of gas molgasconsumed or released in the reaction can easily be calculated:

molgas = V˙reaction· tduration

24.061

where 24.061 is the molar volume of an ideal gas at 20 ^◦C, as calculated from pV = nRT . It is assumed the gas has cooled to this temperature when it reaches the flow meters.

Once the background distortion from the flow meters has been taken into account, the stoichiometry of the produced powder can be estimated.

(28)

2.3.2 Weight fraction and stoichiometry

The predicted stoichiometry of the product is calculated by first finding the weight fraction of nitrogen in the product by using the measured uptake and the weight of the product:

wf = 2· molN2· MN

m_product

where wf is the nitrogen weight fraction, the factor 2 takes into account that there are two moles of nitrogen atoms for each mole of nitrogen gas, and M_N is the atomic weight of nitrogen.

This then gives the calculated stoichiometry x of uranium sesquinitride, U Nx, by solving the following equation:

wf = x· MN

MU + x· MN

which becomes:

x = wf · MU

M_N − wf · MN

where MN and MU are the atomic weights of nitrogen and uranium, respectively.

This estimate of the stoichiometry is far from accurate, but it does give a good indication of whether or not a sesquinitride has been produced. One of the problems is that the product mass is less than it should be, as some of the powder always sticks to the walls of the quartz tube and cannot be removed. The presence of oxides mitigates this somewhat, but adds yet another uncertain variable. Also, the measured uptake obviously depends on the accuracy of the flow meters.

(29)

Chapter 3

Experiments

3.1 Analytical batches

These experiments were performed in a 15 mm quartz tube with small discs (about 10 mm diameter and 2 mm thickness) of natural uranium. The surfaces of the discs were clearly oxidised. The masses of the samples varied between 4.6 - 5.8 grams.

UA120214 The first experiment was quite different from all the following ones: all the steps were divided into separate runs, mechanical polishing of the discs was not used, and the nitriding was not performed in pure nitrogen.

Instead, an admixture of approximately half-argon half-nitrogen was used in order to avoid the possibility of a runaway reaction, seeing as how it was the first experiment nitriding uranium metal powder, rather than uranium hydride, and a very rapid reaction could not be excluded – this due to the lack of the negative feedback achieved when hydrogen is replaced by nitrogen, thus decreasing the concentration (and activity) of nitrogen in the immediate vicinity of the sample.

Two discs were cleaned first in 3M nitric acid, then in methanol, with no visible effect. They were then hydrided in the quartz tube under a 0.3 l/min flow of half-argon, half-hydrogen mixture. The temperature was increased to 300 ^◦C at 20 ^◦C/min, then held for 5 minutes, then decreased to 200^◦C at -5^◦C/min, then immediately increased to 225 ^◦C at 5^◦C/min, then held for 5 hours, then finally cooled to 20^◦C at -10^◦C/min.

The dehydriding was performed under a 0.3 l/min flow of argon. The temperature was increased to 600^◦C at 10^◦C/min, then held for 2 minutes, then allowed to cool naturally to room temperature.

The Nitriding was performed under a 0.3 l/min flow of manually cali- 29

(30)

Figure 3.1: UA120214 – Hydriding

brated half-argon half-nitrogen mixture. The temperature was increased to 225 ^◦C at 10 ^◦C/min, then held for 20 minutes, then increased to 500 ^◦C at 10 ^◦C/min, then held for 20 minutes, then finally cooled to 20^◦C at -10

◦C/min.

Procedure and comments The hydriding process is presented in fig.

3.1. The “waves” in the measured temperature are caused by the thickness of the thermocouple causing it to react slowly when the furnace heats up or cools down. The discontinuity at ∼60 min. is due to operation of the glovebox. Similar waves and discontinuities will not be commented on in the parts following this one. The hydriding appears to have been completed at

∼110 min.

The dehydriding process is presented in fig. 3.2. The dehydriding reaction begins at∼25 min. at 250 ^◦C and appears to have been completed at

∼45 min. at 500^◦C.

The nitriding process is presented in fig. 3.3. The temperature plateau at 225^◦C was obviously unnecessary and was not repeated in future procedures.

The nitriding reaction begins at ∼65 min. at 450-500 ^◦C. There is a spike in the measured temperature at ∼75 min., probably due to the exothermic nitriding-reaction speeding up while the temperature was programmed to increase, causing it to overshoot somewhat. As the program tries to correct this, there is a corresponding temperature drop after the spike, before it reaches an equilibrium.

The nitriding process was considered benign enough to warrant attempt- ing future nitriding under pure nitrogen instead of the nitrogen-argon ad-

(31)

3.1. ANALYTICAL BATCHES 31

Figure 3.2: UA120214 – Dehydriding

Figure 3.3: UA120214 – Nitriding

(32)

mixture used here.

UA120301 Two whole discs and a broken quarter-segment of one, clearly oxidised on the surface, were manually polished, then hydrided in the quartz tube under a 0.3 l/min flow of half-argon, half-hydrogen mixture. The temperature was increased to 300 ^◦C at 20 ^◦C/min, then held for 5 minutes, then decreased to 200 ^◦C at -5^◦C/min, then immediately increased to 230

◦C at 5^◦C/min then held for 1 hour.

The dehydriding was performed under a 0.3 l/min flow of argon while increasing the temperature to 600 ^◦C at 10 ^◦C/min. The gas system was then flushed in nitrogen.

The nitriding was performed under a 0.04 l/min flow of pure nitrogen for 12 minutes 51 seconds. The temperature was held at 600 ^◦C. The gas system was then flushed again, this time in argon.

Finally, the cooling was performed under a 0.3 l/min flow of argon. The temperature was decreased to 20^◦C at -10 ^◦C/min.

Unfortunately, the sample was not completely nitrided, so the process had to be repeated as follows: the sample was heated under a 0.3 l/min flow of argon while the temperature was increased to 600^◦C at 10 ^◦C/min. The gas system was then flushed in nitrogen.

The nitriding was performed under a 0.04 l/min flow of pure nitrogen for 20 minutes. The temperature was held at 600^◦C. The gas system was then flushed again, this time in argon.

Finally, the cooling was performed under a 0.3 l/min flow of argon. The temperature was decreased to 20^◦C at -10 ^◦C/min.

Procedure and comments The primary synthesis process is presented in fig. 3.4. The hydriding appears to have been completed just in time, at ∼100 min. before the dehydriding step was initiated. The dehy- driding reaction begins at∼110 min. at 300 ^◦C and appears to have been completed at∼130 min. at 500^◦C. The nitriding is begun at∼150 min. and stopped at∼165 min., before having finished. There is a very slight drop in temperature at∼165 min. coinciding with the gas flow stopping, ending the exothermic nitriding reaction.

The continued nitriding process is presented in fig. 3.5. Nothing unexpected is noted. The nitriding is begun at ∼60 min., appears to have been completed at∼70 min., and stopped at ∼80 min. There is a slight drop in temperature at∼85 min., probably caused by removing the insulation.

(33)

Figure 3.4: UA120301 – Primary synthesis

Figure 3.5: UA120301 – Continued nitriding

(34)

Figure 3.6: UP120321 – First hydriding

3.2 Production batches

These experiments were performed in a 25 mm quartz tube with small sticks (about 3 mm diameter and up to 100 mm length) of natural uranium. The surfaces of the sticks were clearly oxidised. The total masses of the samples varied between 27 - 51 grams.

UP120321 Two sticks, clearly oxidised on the surface, were manually polished, then hydrided in the quartz tube under a total flow of 0.9 l/min: 0.4 l/min argon and 0.5 l/min hydrogen. The temperature was increased to 300

◦C at 20 ^◦C/min. Sometime during this temperature ramp, the software controlling the process crashed in an unfortunate manner and the synthesis had to be interrupted.

The hydriding step intended above was repeated: the temperature was increased to 300^◦C at 20^◦C/min, then held for 5 minutes, then decreased to 200^◦C at -5^◦C/min, then immediately increased to 230^◦C at 5^◦C/min then held for 2 hours 35 minutes. The sample was then allowed to cool naturally under a 0.5 l/min flow of argon.

The dehydriding was performed under a 0.5 l/min flow of argon. The temperature was increased to 600^◦C at 10^◦C/min, then held for 30 minutes.

The gas system was then flushed in nitrogen.

The nitriding was performed under a 0.09 l/min flow of pure nitrogen for 45 minutes. The gas system was then flushed again, this time in argon.

Finally, the sample was allowed to cool naturally to room temperature under a 0.5 l/min flow of argon.

(35)

3.2. PRODUCTION BATCHES 35

Figure 3.7: UP120321 – Continued hydriding

Figure 3.8: UP120321 – Dehydriding and nitriding

(36)

Procedure and comments The first hydriding process is presented in fig. 3.6. The lack of “waves” in the measured temperature is due to changing to a thermocouple that is thinner and thus reacting faster to the changes in temperature. The steeper slope of the temperature beginning at∼8 min.

was caused by the furnace having been left throttled at a maximum power of 100V instead of the usual 150-200V until then. The fact that the program had already crashed at this point was not made evident until later. At∼18 min. the temperature was observed to be over the programmed 300^◦C, and the furnace was manually shut down. The large drop in temperature at∼120 min. is due to removal of the insulation.

The continued hydriding process is presented in fig. 3.7. Once again,

“waves” in the measured temperature are observed, though now mainly when holding the temperature at a certain level. This is most likely caused by the software regulating the furnace. Due to the small uptake of hydrogen, it is difficult to say when hydriding begins or ends, though it can safely be assumed to have finished sometime during the process. At∼62 min. a large outflow of gas is recorded, coinciding with skipping the dehydriding-nitriding parts of the original program and going straight to cooling under pure argon.

This is not entirely explained. Discontinuities are always seen when changing gases or steps in the program, but rarely this wide. At ∼72 min. the insulation is removed, causing a large drop in temperature. The insulation also prevents a certain convection through the furnace – once removed, the convection causes the fluctuations observed. Similar fluctuations will not be commented on in the parts following this one.

The dehydriding and nitriding process is presented in fig. 3.8. The

“bump” in measured temperature and corresponding outflow were caused by the insulation being forgotten until ∼31 min. and then added. This also caused the temperature fluctuations up to that point. Dehydriding starts after ∼25 min. at 250-300 ^◦C and is finished at ∼60 min. at 600 ^◦C. The nitriding is begun at∼90 min. and stopped at ∼135 min. The exponential decrease in temperature after this point is caused by natural cooling after removal of the insulation.

UP120404 Two sticks, clearly oxidised on the surface, were manually polished, then hydrided in the quartz tube under a total flow of 0.9 l/min: 0.4 l/min argon and 0.5 l/min hydrogen. The temperature was increased to 300

◦C at 20 ^◦C/min, then held for 5 minutes, then decreased to 200 ^◦C at -5

◦C/min, then immediately increased to 230^◦C at 5^◦C/min, then held for 2 hours 35 minutes.

The dehydriding was performed under a 0.5 l/min flow of argon. The

(37)

Figure 3.9: UP120404 – Complete synthesis

temperature was increased to 600^◦C at 10^◦C/min, then held for 30 minutes.

The gas system was then flushed in nitrogen.

The nitriding was performed at a constant temperature of 600 ^◦C but the inflow of pure nitrogen was varied. The flow was increased from 0.04 to 0.1 l/min over 6 minutes, then held for 14 minutes, then decreased to 0.05 l/min over 10 minutes, then held for 10 minutes, before finally being stopped completely. The gas system was then flushed again, this time in argon.

Finally, the cooling was performed under a 0.5 l/min flow of argon. The temperature was decreased to 20 ^◦C at -10^◦C/min.

Procedure and comments The complete synthesis process is presented in fig. 3.9. The shapes of the measured temperature and measured flow difference, especially at ∼45 min., clearly indicate that the behaviour of the hydriding process depends on the behaviour of the temperature. The temperature appears to not have reached the prescribed 200 ^◦C, possibly due to the insulation. The hydriding appears to have been most efficient at the lowest temperature achieved, ∼210 ^◦C. The behaviour of the dehydriding process also appears to depend on the behaviour of the temperature. It is unclear why the “waves” in the measured temperature have reappeared, as the thermocouple should have been the same one as in UP120321. The nitriding was performed with gradual increase and decrease of gas flow, for no other reason than curiosity. As it had no noticeable effect, it was never repeated. The drops in temperature at∼320 min. and ∼340 min. are caused by stepwise removal of the insulation.

(38)

Figure 3.10: UP120412 – Hydriding

UP120412 Three sticks, clearly oxidised on the surface, were manually polished, then hydrided in the quartz tube under a total flow of 0.9 l/min:

0.4 l/min argon and 0.5 l/min hydrogen. The temperature was increased to 300 ^◦C at 20^◦C/min, then held for 5 minutes, then decreased to 200 ^◦C at -5 ^◦C/min, then immediately increased to 230 ^◦C at 5 ^◦C/min, then held for a total of 3 hours. During this step, the software controlling the process crashed, but in a fortunate manner, so the step could be finished.

After restarting the software, the dehydriding was performed under a 0.5 l/min flow of argon. The temperature was increased to 600^◦C at 10^◦C/min, then held for 30 minutes. The gas system was then flushed in nitrogen.

The nitriding was performed under a 0.1 l/min flow of pure nitrogen for 1 hour. The gas system was then flushed again, this time in argon.

Finally, the cooling was performed under a 0.5 l/min flow of argon. The temperature was decreased 20^◦C at -10^◦C/min.

Procedure and comments The hydriding process is presented in fig.

3.10. The same temperature-dependant behaviour in the hydriding process as was seen in UP120404 is still present. Clearly the hydriding temperature could be adjusted. Nothing else worth noting, until the program-crash at

∼165 min. after which the furnace was controlled manually before being allowed to cool naturally.

The dehydriding and nitriding process is presented in fig. 3.11. There is odd behaviour in the measured temperature at∼55 min., ∼95 min. and

∼155 min. There are no clear explanations for these discrepancies.

(39)

Figure 3.11: UP120412 – Dehydriding and nitriding

(40)

(41)

Chapter 4

Results

4.1 Analytical batches

UA120214

Nitrogen uptake (fig. 4.1) The “waves” apparent in the uptake of nitrogen correspond to similar “waves” in the measured temperature, as discussed in chapter 3. There is a strange drop in nitrogen uptake in the middle of the peak, also coinciding with a peak in the temperature. The peak and drop in temperature are most likely caused by the thermocouple and software reacting slowly to the increase in temperature caused by the exothermic nature of the reaction, as discussed in chapter 3.

The shape of the nitriding uptake itself, however, is most likely caused by a more subtle mechanism: as the uranium powder reacts rapidly with the nitrogen in the atmosphere closest to it, the composition of the atmosphere shifts from a nitrogen-argon admixture to a predominantly argon atmosphere, thus slowing down the reaction. This mainly-argon atmosphere is quickly flushed out of the reaction zone and the reaction picks up speed again. As there was only a small amount of powder, the reaction completed before this effect could be repeated.

Because the exothermic reaction was not as extreme as might have been feared, it was decided that the next experiment would use a pure nitrogen atmosphere under a very low flow.

XRD analysis (fig. 4.2) and stoichiometry Despite performing all XRD-measurements in the same manner, this first one was by far the worst one. It is difficult to see in fig. 4.2 but there are clear indications that the main components of the powder are U2N3, U3O8 and UO3. The

41

(42)

Figure 4.1: UA120214 – Nitrogen uptake

Figure 4.2: UA120214

(43)

Figure 4.3: UA120301 – Nitrogen uptake, first nitriding

relatively high level of oxidation was expected as the original uranium metal was already heavily oxidised, and while care was taken not to expose the powders to oxygen, some oxidation was unavoidable after the synthesis.

As this first synthesis used a mixed argon-nitrogen atmosphere, the flow meters cannot be relied upon to measure completely accurate flows, since they were calibrated for a pure nitrogen atmosphere. Therefore, the nitrogen weight fraction and stoichiometry cannot be properly calculated.

UA120301

Nitrogen uptake, first nitriding (fig. 4.3) There are two plateaus visible in the nitrogen uptake. The first one, beginning at ∼151 min., is most likely caused by inaccuracy in the flow meters along with the geometry of the quartz tube and can therefore be considered a ‘zero-level’ for flow measurements, as no nitrogen could possibly have reached the filter at that point already. The second one, at slightly before ∼152 min., is when the nitrogen begins to reach and react with the uranium powder. As there is still some argon left over in the dead volume before the powder, the reaction is not fully under way until at ∼156 min., when the entire dead volume has been filled with nitrogen. From this point until the reaction is stopped at ∼163 min. there is no outflow from the quartz tube. This means that all of the nitrogen sent into the tube is reacting with the uranium powder.

The reaction is clearly still under way when the nitriding process is halted by stopping the gas flow. Because of this, a second nitriding process was performed.

(44)

Figure 4.4: UA120301 – Nitrogen uptake, second nitriding

Nitrogen uptake, second nitriding (fig. 4.4) Nothing unexpected happens. The two plateaus are seen, at ∼62 and ∼63 min. respectively.

However, the reaction never reaches full uptake, and at ∼67 min. the ni- triding appears to have peaked and at ∼71 min. it appears to have been completed. After this, there is still some nitrogen uptake, as evidenced by the slow downslope of the uptake curve and the fact that it is not yet at the

‘zero level’ of the first plateau. This is to be expected, since α-U₂N₃ has a stoichiometry between 1.54-1.75 N/U.

XRD analysis (fig. 4.5) and stoichiometry This time the XRD results were much better and the peaks can easier be seen to correspond to those of U2N3, U3O8 and UO3.

The nitrogen weight fraction was calculated to be 9.15 % which corre- sponds to a stoichiometry of 1.71 N/U.

4.2 Production batches

UP120321

Nitrogen uptake (fig. 4.6) and stoichiometry The first plateau in nitrogen uptake is barely visible this time, but the second one is very clear, beginning at ∼92 min. At ∼95 min. the nitrogen uptake reaches 100%

of nitrogen inflow. This continues until ∼127 min. after which nitriding appears to have been completed and nitrogen uptake decreases to the earlier noted slow downslope, turning the sesquinitride hyperstoichiometric.

(45)

Figure 4.5: UA120301

Figure 4.6: UP120321 – Nitrogen uptake

(46)

Figure 4.7: UP120321

XRD analysis (fig. 4.7) Nothing particular worth commenting. The peaks are clear and correspond to those of U2N3, U3O8 and UO3.

UP120404

Nitrogen uptake (fig. 4.8) Two plateaus in the nitrogen uptake are clearly visible, the first one at∼264 min. and the other one at ∼266 min.

No unexpected behaviour can be seen. Nitrogen uptake appears to have been completed at∼297 min. There is a slight downslope towards the end, indicating hyperstoichiometric nitrogen uptake. The gradual increase and decrease of nitrogen flow had no real effect on the uptake of nitrogen and so was omitted from future procedures.

XRD analysis (fig. 4.9) and stoichiometry Nothing particular worth commenting. The peaks are clear and correspond to those of U₂N₃, U3O8 and UO3.

(47)

Figure 4.9: UP120404

(48)

UP120412

Nitrogen uptake (fig. 4.10) a The first plateau is again barely visible, but the second one is clearly present at ∼92 min. At ∼95 min. the nitrogen uptake reaches 100%. At∼132 min. the nitriding appears to have been completed. There is a clear downslope towards the end, indicating hyperstoichiometric nitrogen uptake.

XRD analysis (fig. 4.11) and stoichiometry Nothing particular worth commenting. The peaks are clear and correspond to those of U₂N₃, U₃O₈ and UO₃.

(49)

Figure 4.11: UP120412

(50)

(51)

Chapter 5

Summary and conclusions

In this project, a semi-novel method for producing uranium nitride from uranium metal was tested. In this process, the metal is placed in a fluidised bed reactor where it is first hydrided, then dehydrided, before finally being nitrided. The process is monitored by the temperature of the reaction vessel, and the inflow and outflow of gas.

The process worked very well, showing 100% nitrogen uptake for large parts of the reaction. The two parts not showing 100% nitrogen uptake were the beginning and the tail-end. The poor uptake in the beginning is effectively a measurement artefact – it is simply caused by the dead volume between the uranium powder and the gas inlet. It can most likely be made arbitrarily small by designing a more compact gas system. But this might not even be necessary, as all nitrogen reaching the powder should react with it immediately. This could be examined by performing the same experiments with a gas chromatographer.

The tail-end of the reaction is where any waste of nitrogen takes place.

This waste can be minimised by cutting the inflow of nitrogen at the right moment, and using larger amounts of uranium. Also, a simple recycling system could practically eliminate any nitrogen waste, as the main element it might be mixed with is argon, which is inert, though there may also be some residual hydrogen if the dehydriding is poorly done.

The procedure developed over the course of these experiments is presented in table 5.1 and is a recommended starting point for any further development using this synthesis system. It may, however, be worthwhile changing the hydriding step somewhat, as it might prove more advantageous to use a slightly lower temperature (since the thermocouple is on the outside and does not perfectly reflect the inside temperature), and the step is unnec- essarily long right now. Naturally, the process should be adjusted according

51

(52)

to the amount of uranium metal used, especially for batches above 50-60 grams.

The powders produced in this project were examined by XRD-analysis.

All were shown to contain U2N3 as well as U3O8 and UO3. No traces of uranium metal, uranium hydride or other uranium nitrides were found. As can be seen from fig. 5.1 and 5.2, the powders were all very similar – almost identical.

Gas 1, [l/min]

Gas 2, [l/min]

Temp [^◦C]

Ramp [^◦C/min]

Time [s]

Time [h]

Step

Ar, 0.4 H2, 0.5 20 0 Hydr.

300 20 840

300 0 300

200 -5 1200

230 5 360

230 0 10800

3.75

Ar, 0.5 H₂, 0 Dehydr.

600 10 2220

600 0 1800

1.12

N₂, 0.1 Nitr.

600 0 3600

1

Ar, 0.5 Cool.

20 -10 3480

0.97 Total time [h] 6.84 Table 5.1: Synthesis process

(53)

53

Figure 5.1: UA120214 and UA120301, superimposed

Figure 5.2: UP120321, UP120404 and UP120412, superimposed

(54)

(55)

Bibliography

[1] A technology roadmap for generation IV nuclear energy systems. Issued by the U.S. DOE Nuclear Energy Research Advisory Committee and the Generation IV International Forum, 2002.

[2] J. Wallenius. Transmutation of nuclear waste. Leadcold books and games, 2011.

[3] N. M. Stepennova Rogozkin, B. D. and A. A. Proshkin. Mononitride fuel for fast reactors. Atomic Energy 95, no. 3, 2003.

[4] S. T. Arm and E. J. Butcher. Treatment and disposal of process wastes arising from the recycle of spent nuclear fuel in the usa - 9094. In WM2009 Conference, Phoenix, AZ, 2009.

[5] Vilkamo O. et al. Effluent release options from nuclear installations.

Technical report, Nuclear Energy Agency Organisation for Economic Co-operation and Development, 2003.

[6] B. Morley. Best practicable means (BPM) and as low as reasonably practicable (ALARP) in action at sellafield. Journal of Radiological Protection, 2004.

[7] Nomura M. Suzuki T. Fujii Y. Ding, X. High enrichment of 15N by chromatographic chemical process. Journal of Chromatography A 1201, 65-68., 2008.

[8] Dell R.M. Allbutt, M. Chemical aspects of nitride, phosphide and sul- phide fuels. Journal of Nuclear Materials 24, 1-20., 1967.

[9] Frost B.R. Mustelier J. Bagley K. Crittenden G. Van Dievoet J. Kittel, J. History of fast reactor fuel development. Journal of Nuclear Materials 204, 1-13., 1993.

55

(56)

[10] Serizawa H. Katsura, M. Formation of a nitrogen-rich α-U2N3+x phase by the reaction of uranium with a stream of ammonia. Journal of Alloys and Compounds 187, 389-399., 1992.

[11] Miyake M. Serizawa H. Katsura, M. Some problems in nonstoichiometry of α-uranium sesquinitride. Journal of Alloys and Compounds 193, 101- 103., 1993.

[12] T. Hiroaki. Phase relations and thermodynamic properties of the uranium-nitrogen system. Journal of Nuclear Materials 51, 78-89., 1974.

[13] Baenziger N.C. Wilson A.S. McDonald R.A. Rundle, R.E. The struc- tures of the carbides, nitrides and oxides of uranium. J. Am. Chem.

Soc. 70, 99-105., 1948.

[14] Pictures taken from the crystal lattice structures web page, http://cst- www.nrl.navy.mil/lattice/, provided by the center for computational materials science of the united states naval research laboratory.

[15] Yeamans C.B. Sattelberger A.P. Hartmann T. Cerefice G.S. Czerwin- ski K.R. Silva, G.W.C. Reaction sequence and kinetics of uranium nitride decomposition. Inorg. Chem. 48, 10635-10642., 2009.

[16] J. Somers. Minor actinide bearing fuels: Fabrication and irradiation experience in europe. Energy Procedia 7, 169-176., 2011.

[17] Davies T.J. Evans, P.E. Uranium nitrides. Journal of Nuclear Materials 10, 43-55., 1963.

[18] El-Genk M.S. Matthews R.B. Ross, S.B. Thermal conductivity correla- tion for uranium nitride fuel between 10 and 1923 K. Journal of Nuclear Materials 151, 318-326., 1988.

[19] Sheindlin-M. Musella M. Hyland G.J. Ronchi, C. Thermal conductivity of uranium dioxide up to 2900 K from simultaneous measurement of the heat capacity and thermal diffusivity. Journal of Applied Physics 85, 776-789., 1999.

[20] Chidester-K.M. Hoth C.W. Mason R.E. Petty R.L. Matthews, R.B.

Fabrication and testing of uranium nitride fuel for space power reactors.

Journal of Nuclear Materials 151, 345., 1988.

(57)

BIBLIOGRAPHY 57 [21] Fielding-R.S. Porter D.L. Meyer M.K. Makenas B.J. Burkes, D.E. A us perspective on fast reactor fuel fabrication technology and experience.

part II: Ceramic fuels. Journal of Nuclear Materials 393, 1-11., 2009.

[22] Shirasu-Y. Minato K. Serizawa H. Ogawa, T. Thermodynamics of car- bothermic synthesis of actinide mononitrides. Journal of Nuclear Ma- terials 247, 151-157., 1997.

[23] Stepennova-N.M. Bergman G.A. Proshkin A.A. Rogozkin, B.D. Ther- mochemical stability, radiation testing, fabrication, and reprocessing of mononitride fuel. Atomic Energy 95, 835-844., 2003.

[24] M. Streit. Fabrication and Characterisation of (Pu,Zr)N Fuels. PhD thesis, Swiss Federal Institute of Technology Zurich, 2004.

[25] P.E. Doherty C.H. McMurtry K.M. Taylor, C.A. Lenie. Synthesis and fabrication of refractory uranium compounds, fourth quarterly report.

Technical report, U.S. Atomic Energy Commission, Oak Ridge Ten- nessee, 1960.

[26] Smith C.S. Burke, J.E. The formation of uranium hydride. J. Am.

Chem. Soc. 69, 2500-2502., 1947.

[27] H. Tagawa. Equilibrium nitrogen pressures and thermodynamic proper- ties of uranium sesquinitride. Journal of nuclear materials 41, 313-319, 1971.

[28] Tennery V.J. Potter, R.A. Preparation of uranium nitride. United States Patent Appl. No.: 474,284, 1976.

[29] Laurent Y. Maunaye M. Marchand R. Lang, J. Nitrides-structures and crystal growth. Progress in Crystal Growth and Characterization 2, 207-235., 1981.

[30] P. Malkki. Establishing a nuclear fuel laboratory: Investigating manufacturing methods for a mixed uranium-zirconium nitride. Master’s thesis, Stockholm University Department of Physics, 2009.

[31] M.T. Oliver. Production of mixed U-Zr nitride nuclear fuel powders from metallic U-Zr alloys. Master’s thesis, Royal Institute of Technology Reactor Physics Department, 2010.

[32] T. Hollmer. Manufacturing methods for (U-Zr)N-fuels. Master’s thesis, Royal Institute of Technology Reactor Physics Department, 2011.

(58)

[33] P. Andersson and C-M. Arvhult. Metallic residues after hydriding zirconium. Bachelor’s degree thesis, Royal Institute of Technology Reactor Physics Department.

Isotopically enriched nitrides for nuclear power