The melting behaviour of plutonium dioxide: A laser-heating study

Journal of Nuclear Materials 416 (2011) 166–172

Contents lists available at ScienceDirect

Journal of Nuclear Materials journal homepage: www.elsevier.com/locate/jnucmat

The melting behaviour of plutonium dioxide: A laser-heating study F. De Bruycker, K. Boboridis, P. Pöml, R. Eloirdi, R.J.M. Konings, D. Manara ⇑ European Commission, Joint Research Centre, Institute for Transuranium Elements (ITU), P.O. Box 2340, 76125 Karlsruhe, Germany

a r t i c l e

i n f o

Article history: Available online 20 November 2010

a b s t r a c t In this work the melting behaviour of plutonium dioxide has been studied for the first time via laser heating. With this method, the short experiment duration combined with minimal contact between sample and containment minimized undesired side effects, such as sample reduction, vaporisation or reaction with the holder. The sample temperature was measured by fast pyrometry, and inflections in the recorded thermograms revealed the liquid-to-solid transition. This latter was also detected via a method based on the reflectance study of a low-power laser beam reflected by the sample surface. Multi-channel spectro-pyrometry was used to investigate, in parallel, the normal spectral emittance of the samples in the wavelength range between 550 and 920 nm at the investigated temperature. The present experimental melting temperature of (3017 ± 28) K is much higher than the values obtained in the past by traditional heating techniques. It is suggested that, in contrast to the present data, previous results were affected by extensive reaction of the plutonium dioxide samples with the containment and/or changes in the O/Pu ratio at high temperature. Ó 2010 Elsevier B.V. All rights reserved.

1. Introduction Plutonium dioxide is one of the components of mixed oxidebased nuclear fuels, normally in a mixture with uranium dioxide. This type of fuel is employed for the generation of nuclear energy for civil purposes with PuO2 concentrations typically varying between 0 and 30 mol%. Thus, many physico-chemical properties of low Pu-content (U, Pu) mixed oxides (MOX) have been thoroughly investigated since the 1950s. The difficulties inherent to the study of these compounds considerably increase with Pu-content and temperature, essentially due to the high oxygen potential and the reactivity of plutonium dioxide. Although pure PuO2 is generally not employed in industrial applications, its melting temperature is an important reference parameter for the comprehension of the UO2–PuO2 phase diagram at high temperature. Historically, this point was assessed to be near 2700 K in the beginning of the 1970s, essentially based on the most recent of a series of measurements performed by Lyon and Baily [1] and by Aitken and Evans [2] using the thermal arrest technique and induction heating of tungsten-encapsulated samples, and by Riley [3] who combined visual detection of melting with a variety of experimental setups, including flame melting under controlled atmosphere. However, these measurements were certainly affected by reaction of the sample with the tungsten crucible in the first case, and by reduction of the investigated specimen reported by Riley himself in the second. This latter issue

⇑ Corresponding author. E-mail address: [email protected] (D. Manara). 0022-3115/$ - see front matter Ó 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.jnucmat.2010.11.030

was also reported in the review by Lemire et al. [4]. Since then, no further experimental research was performed on this subject until the last decade, as summarised in the reviews of Carbajo et al. [5] and of Guéneau et al. [6]. In the period 2007–2009 Kato et al. [7–10] published a series of novel experimental studies on the (U, Pu) MOX fuel used for the Monju reactor, in which the commonly accepted value was questioned, and a considerably higher one (by about 200 K) was proposed. Kato et al. performed a traditional thermal arrest analysis, but used crucibles of different shapes and materials (tungsten, rhenium) to demonstrate that in Pu-rich MOX and, especially, in pure PuO2 samples, the interaction between sample and containment affected the apparent solidus and liquidus points well beyond the intrinsic uncertainty limits of the experimental method. Because of this, they were not able in the end to measure the melting point of pure PuO2 directly, as instead of the expected congruent melting they observed two distinct transition temperatures differing by about 150 K. They attributed this result to the high reactivity of liquid PuO2 towards rhenium, just as in the case of tungsten crucibles and all MOX with a PuO2-content higher than about 20 mol%. More specifically, they suspected that the high oxygen potential of PuO2 resulted in the oxidation of Re and the simultaneous reduction of the samples. Nonetheless, it was a big merit of Kato et al. to highlight this source of uncertainty, and also to perform a systematic study of the oxygen potential influence on experimental data. Their work has encouraged the current research. In this paper, a novel experimental study on the melting temperature of plutonium dioxide is presented. An experimental method developed in recent years at the European Commission’s JRC – ITU (Institute for Transuranium Elements) was employed,

167

F. De Bruycker et al. / Journal of Nuclear Materials 416 (2011) 166–172

combining rapid laser heating under quasi-containerless conditions and controlled atmosphere with fast pyrometry. With such a technique, the sample integrity towards reduction/oxidation, incongruent vaporisation and reaction with any containing material are limited during heating and cooling. Thus, the major experimental issues that affected previous results can be resolved yielding original results. 2. Experimental 2.1. Sample preparation A series of five samples of plutonium dioxide was prepared (Table 1) for this melting temperature investigation. The isotopic composition of the plutonium employed was checked by High Resolution Gamma Spectroscopy (HRGS) to be 93.4 wt.% 239Pu and 6.4 wt.% 240Pu. Traces (<0.2 wt.%) of 241Pu and 241Am stemming from b-decay of 241Pu were detected by Thermal Ionisation Mass Spectrometry (TIMS). The starting material was powder or beads obtained by gel-supported precipitation (SOL–GEL). Disk-shaped samples of 8–9 mm in diameter and about 4 mm in thickness were obtained using a bi-directional press. The samples were then sintered in an atmosphere of Ar + H2 with traces of H2O to obtain dense material (Table 1). In order to obtain the exact O/Pu = 2 stoichiometry, samples C and D were subjected to two consecutive heat treatments under a flow of air at 1423 K for 8 h in both cases. Already after the first heat treatment the measured weight gain of the samples corresponded to stoichiometric PuO2 separately measured by thermo-gravimetry (TG). Since, in addition, no further change was noticed after annealing the samples for a second time, it was considered that PuO2.00 stoichiometry had been reached. Similar heat treatments under air were performed on the remaining samples E–G. A lattice parameter of 0.5396(1) Å was measured by X-ray diffraction (XRD) on sample G, corresponding to the literature value for PuO2 [11]. Selected samples were examined after the laser experiments by TG and XRD. The micro-structure, homogeneity, and purity of specimens taken from both molten and non-molten sample areas were analysed by Electron Probe Micro-Analysis (EPMA) and Scanning Electron Microscopy (SEM). 2.2. Material characterization techniques The structure and chemical composition of PuO2 samples were characterized by X-ray diffraction (XRD) and Electron Probe MicroAnalysis (EPMA) before and after the laser-heating experiments. Xray diffraction (XRD) was performed using a BrukerÒ D8 advanced diffractometer (Cu Ka1 radiation) with a 2h range of 10–120° using 0.009° steps with 2 s of count time per step at operating conditions of 40 kV to 40 mA. The XRD instrument was equipped with a LynxeyeÒ 3° linear position sensitive detector. Secondary electron (SE) and backscattered electron (BSE) images were recorded on a PhilipsÒ XL40 scanning electron microscope operated at 20 kV installed in a glove box. Quantitative analysis and X-ray element maps were acquired using a shielded

CamecaÒ SX100R electron microprobe operated at 20 kV and 20 nA. 2.3. Measurement method A schematic of the experimental setup employed in the current research is shown in Fig. 1. The samples were mounted in a controlled-atmosphere cell, which was placed inside an a-shielding glove box. Physical contact to their mount was limited by using three radially arranged graphite screws to hold them in place. The samples were laser-heated to a temperature above the melting transition from which they rapidly cooled and re-solidified when the laser was switched off. Short heating pulses of 60–200 ms were employed in order to minimize the dwelling time at very high temperature while still maintaining local thermal equilibrium. Furthermore, quasi-containerless conditions were obtained by directly heating only a limited area (about 5 mm in diameter) on the sample surface. Thus, the molten part was contained by the outer still solid portions of the sample. Problems such as sample vaporisation and interaction with its containment, typical at these temperatures and particularly in the liquid state, were greatly reduced or completely avoided by these two principles of the measurement technique, namely its high speed and containerless character. The melting experiments were performed under air slightly above atmospheric pressure. A PC-controlled Nd:YAG continuous laser radiating at 1064 nm was used to heat the samples. The laser-pulse profile was programmable, allowing the heating and cooling rates to be optimized. Simple trapezoidal pulse shapes were used to reach the desired temperature before switching the laser off. They consisted of an initial power ramp, to reduce the thermal shock to the samples, followed by a constant-power plateau. The plateau power ranged from 450 to 810 W. The sample surface temperature during heating and subsequent cooling was measured by fast pyrometry, as detailed in the next section. The melting temperature of PuO2 was determined from the cooling part of the recorded thermograms, by locating the thermal arrest indicating sample solidification. As is often the case in rapid laser-heating experiments [12], melting was not observable during heating. This was a consequence of the surface-heating nature of the current technique, the samples being essentially opaque at the laser wavelength, combined with relatively slow heat diffusion into the bulk. In other words, a thin surface layer was quickly driven through the melting transition, with only little associated latent heat of fusion and therefore no noticeable temperature arrest, and then onto higher temperatures in the liquid state before the deposited laser energy diffused deeper into the material leading to an increase in the molten layer thickness. In addition to locating the characteristic freezing arrest upon cooling, an additional method was applied in some cases to facilitate the observation of the melting/freezing transition onset. This so-called ‘‘reflected light signal (RLS)’’-method relies on recording the intensity of a lower-power 488 nm Ar+ probe-laser which is reflected off the sample surface and detected by a pyrometer tuned at the same wavelength. Abrupt changes in the reflectivity of the sample may indicate phase transitions. In particular, a characteristic noise-like structure appeared in the RLS upon melting, due to

Table 1 Summary of the preparation of the PuO2 samples used in the present investigation. Sample

Starting material

Sintering (gas/temperature (K)/time (h))

Annealing (gas/temperature (K)/time (h))

Composition

C D E F G

Powder Powder Sol gel beads Sol gel beads Sol gel beads

Ar + 2% Ar + 2% Ar + 2% Ar + 2% Ar + 2%

Air/1423/8 Two times Air/1423/8 Two times Air/1073/24 Two times Air/1073/24 Two times Air/1073/24 Two times

PuO2.00±0.01 PuO2.00±0.01 PuO2.00±0.01 PuO2.00±0.01 PuO2.00±0.01

H2 + 1500 ppm H2 + 1500 ppm H2 + 2000 ppm H2 + 2000 ppm H2 + 2000 ppm

H2O/1923/8 H2O/1923/8 H2O/1923/8 H2O/1923/8 H2O/1923/8

168

F. De Bruycker et al. / Journal of Nuclear Materials 416 (2011) 166–172

Two Channel Pyrometer

Pressurized vessel

λ = 488 nm and 652 nm Time resolution ≈ 10µs

Sample

Heating laser Probe laser

Glove box

λ = 1064 nm Power = 4.5 kW

λ = 488 nm Power = 1 W

α - Shielding

Spectropyrometer 256 λ from 488 nm to 1011 nm Time resolution ≈ 1 ms Fig. 1. The experimental laser heating + fast pyrometry setup employed in this work. The two-channel fast pyrometer is used to measure both the radiance temperature at 652 nm and the reflected probe-laser beam at 488 nm. The spectro-pyrometer is used for spectral analysis yielding emissivity and true temperature.

Fast pyrometer thermogram (ε = 0.83)

Temperature (K)

3600

3200

Melting point

2800

2400 Laser power

2000 0

20

40

60

3000 2800 2600 Solidification point 2400 2200 2000 1800 1600 1400 RLS first derivative 1200 (arbitrary units) 1000 800 600 400 200 0 80 100

Laser Power (W)

A pyrometer utilizing a fast logarithmic amplifier (settling time of about 10 ls to 1% of log output) and operating at 652 nm was used to measure the surface radiance temperature of the sample at the centre of the laser-heated area. Its nominal measurement spot was 0.5 mm in diameter. It was calibrated against a standard tungsten-ribbon lamp in the range of 1800–2500 K, ensuring traceability to the International Temperature Scale of 1990 [14]. Beyond this temperature, the validity of the calibration, as well as the quality of the optical windows and the alignment, were tested by mea-

Time (ms)

4000 3800 3600 3400

PuO sample C

2 Laser profiles Thermograms ---------------------------------------------test 2 test 3 test 4 test 5 test 6 test 7 --------------------------------------------------

1400 1200 1000

3200

3022 K 800

3000

600

2800 2600

400

2400 200

2200 2000

Fig. 2. Thermogram, laser power profile and first derivative of the reflected light signal (RLS) recorded in a laser melting experiment on the PuO2 sample F. Onset of melting on the sample surface can be observed with the help of the RLS technique. A clear thermal arrest due to solidification is visible on the cooling stage, at a temperature corresponding (within uncertainty) to the melting point. The dotted circles around the melting and solidification points indicate the uncertainties on the exact time and temperatures at which melting and solidification occurred.

Laser Power (W)

2.4. Pyrometry

suring in situ the solidification radiance temperatures of molybdenum and tungsten. These temperatures observed with the current technique were in a better than ±0.5% agreement with the recommended values (2530 K for Mo and 3207 K for W at 653 nm [15]). In addition, a spectro-pyrometer, based on a linear array of 256 photodiodes was used to record the thermally emitted sample radiance in the range from 488 to 1011 nm. This instrument allows a more complete spectral analysis, whereby its main disadvantage is in the poorer time resolution (one spectrum per millisecond at best) [13]. Due to low signal-to-noise ratio, moreover, only the range of 550–920 nm was useful. The photodiode at 649 nm was calibrated up to 2500 K using the tungsten-ribbon lamp and this calibration was transferred to a tubular-cavity variable-temperature graphite blackbody-furnace capable of reaching 3500 K. The remaining photodiodes were then calibrated with this blackbody, allowing a conversion of output signal to spectral radiance over the entire useful wavelength range. The measured radiance spectra were fitted by least-squares regression to Planck’s distribution law, modified by a wavelength- and temperature-dependent function assumed to represent the (near-) normal spectral emittance

Temperature (K)

vibrations of the sample liquid surface, which disappeared again upon freezing. Further details of the laser-heating setup can be found elsewhere [13]. Fig. 2 reports a typical heating/melting/cooling solidification cycle performed during a melting point measurement on sample F. A full thermogram (sample temperature vs. time) is plotted along with the RLS first derivative curve and the heating laser power profile.

0

20

40

60

80

100

120

140

0 160

time (ms) Fig. 3. Thermograms recorded in successive laser-heating experiments performed on sample C. The observed solidification temperature, 3022 K, is very close to the mean value of all measurements performed on five different samples (3017 ± 28) K. The dash-dotted lines indicate the laser power-vs.-time profiles.

169

F. De Bruycker et al. / Journal of Nuclear Materials 416 (2011) 166–172

(NSE) e(k, T) of PuO2. The simplest possible choice for this NSE function near the freezing plateau, namely that of greybody behaviour (wavelength-independent NSE), already resulted in a good fit and was therefore adopted. This choice was also supported by recently published ab initio calculations of the optical properties of PuO2, albeit at low temperature [16]. The best estimate of this greybody-NSE, obtained from the fitting procedure, was then used to convert the radiance temperature measured by the monochromatic pyrometer to true temperature. Because of space constraints, the spectro-pyrometer was only employed in the measurement series on one sample (sample E). However, all other parameters being the same, significant sample-to-sample variations in the thermally emitted spectral radiance were considered to be unlikely. All measurements were actually taken on the cooling part of the thermograms on already molten samples. At that point, the sample surface was assumed to be equally smooth due to surface tension, as supported by the good repeatability of the measured radiance temperatures upon freezing.

Greybody behaviour readily suggested the application of another, particularly simple, method to estimate true temperature, T, and NSE, e, from radiance temperature measurements, Tr, at selected wavelengths, k. The so-called ‘‘extrapolation to zero wavelength’’ [17] is based on an equation easily derived from Wien’s law, which is accurate to better than 1% for kT < 3100 lm K:

1 1 k ¼  ln e; T r T c2

where c2 = 14388 lm K is the second radiation constant. If e is independent of wavelength or at least constant at the wavelengths at which Tr is measured, then plotting 1/Tr as a function of k will result in a straight line that, when extrapolated to zero wavelength, intersects the inverse-temperature axis at 1/T. In practice, the radiance temperatures measured both on liquid PuO2 and during the freezing arrest by 120 photodiodes of the spectro-pyrometer, selected to span its useful wavelength range, exhibited this linear behaviour. The values computed for T and e by means of this extrapolation to 3200

3200 test 6 test 7

3100 3050

test 2 test 3

3150

Temperature (K)

Temperature (K)

3150

3022 K

3000 2950

3100 3050

3017 K

3000 2950 2900

2900 2850

ð1Þ

Sample C 100

102

Sample D

2850 104

106

108

96

110

98

100

102

Time (ms) 3200

108

110

test 1 test 2

3150

Temperature (K)

Temperature (K)

3100

2 3 4 5

3050

3013 K 3000 2950 2900

3100 3050

3027 K

3000 2950 2900

Sample E 100

Sample F

2850

105

110

115

120

125

60

61

62

Time (ms)

63

64

65

66

Time (ms)

3200

3040

Temperature (K)

test 3

3150

Temperature (K)

106

3200 test test test test

3150

2850

104

Time (ms)

3100 3050

3004 K

3000 2950

3030

D

F F

C C 3020

E D

3010

E E

E

G

3000 2900 2850

Sample G 61

62

63

2990 64

65

Time (ms)

66

67

68

1

2

3

4

5

6

7

8

9 10 11

Experiment

Fig. 4. Thermal arrests observed in the cooling stage of thermograms recorded on five PuO2 samples (C–G). The horizontal dashed lines show the accepted solidification temperature for each sample. A summary of all 11 measurements is shown in the lower right plot with the dashed line indicating the overall mean value for the melting and solidification temperature of PuO2 (3017 K). In this plot, the indicated ±28 K uncertainty band combines the current data dispersion with uncertainties on the sample emissivity and the temperature scale definition.

170

F. De Bruycker et al. / Journal of Nuclear Materials 416 (2011) 166–172

zero wavelength agreed well with those deduced from the fitting procedure. This, of course, was not surprising since the two methods were physically equivalent. Their agreement constitutes nonetheless a good corroboration of the obtained results, because they are based upon slightly different numerical assumptions, whereby poor numerical stability of solutions is known to be a weak point of the least-square regression to Planck’s distribution law [18].

1.0 3600 0.9

3400 3200

0.8

3000

3. Results

2800

To obtain a pronounced freezing arrest, it was necessary to melt enough matter by using sufficiently energetic laser-pulses. The dwelling time at the highest temperatures had to be as short as possible to minimize vaporisation, while at the same time care had to be taken not to destroy the samples by thermal shocks due to excessively high heating rates. In order to optimize the pulse parameters, sample C was heated repeatedly with gradually increasing laser power. The thermograms recorded in these successive tests are shown in Fig. 3. It can be seen that at low laser power solidification upon cooling was merely indicated by an inflection. By contrast, increasing the power, and consequently the peak surface temperature by about 180 K, led to the formation of more liquid mass, hence to very clear and reproducible freezing arrests. Similar settings were then used in subsequent experiments, allowing a reduction of the number of laser shots per sample, thereby further reducing the risk of cracking and breaking the specimens. The shape of the thermograms and particularly of the freezing arrests varied slightly depending on the sample morphology (cracks) and its evolution under rapid heating, the occurrence of supercooling, or simply the pyrometer alignment relative to the laser spot. Fig. 4 summarises all experiments that were taken into account for the determination of the PuO2 melting temperature. The overall repeatability was excellent. Sample E shows a prominent example of freezing arrests obtained upon recalescence after supercooling. When solidification was not accompanied by a clear horizontal plateau, as was the case for sample D, the freezing temperature was determined from the early and more reproducible part of the arrest, as also suggested by the measurement sequence on sample C (Fig. 3). Application of the RLS technique on sample F is depicted in Fig. 2. The strength of this technique is in detecting a phase transformation when it is not evident in the thermogram. In this particular case, the melting transition, which is not accompanied by a thermal arrest as pointed out in the previous section, is easy to spot by the initiation of the noise-like structure, characteristic of a liquid surface, which is even more apparent when plotting the time derivative of the RLS, as was done in this graph. As already described, the NSE of PuO2 was obtained from an analysis of radiance data measured with the multi-wavelength spectro-pyrometer. One such data set is shown in Fig. 5. The two insets depict two of the radiance spectra, one of the liquid sample and one during solidification, as well as the respective least-squares fits based on greybody behaviour. A similar radiance spectrum was recorded and fitted to obtain each of the full (temperature) and empty (NSE) circles plotted in the figure. Because of the limited time resolution (2 ms per point in this case), the temperature and NSE values obtained with this spectro-pyrometer, are influenced by heating and cooling rates, therefore the fitting is only meaningful where the temperature does not vary too rapidly. Moreover, the analysis was more accurate in liquid plutonia at high temperature, where the temperature was more stable and the signal/noise ratio higher. Finally, this instrument was essentially used to estimate the NSE only. Although temperature values obtained from the recorded radiance spectra were always within 2% of those yielded by the fast pyrometer at 652 nm with fixed NSE, only the latter

2600

0.7

0.6

2400 50

60

70

80

90

0.5 100 110 120 130 140 150

Fig. 5. Thermogram recorded on the PuO2 sample E, including the spectral emittance analysis based on regression to Planck’s radiance law. The two insets show example spectra recorded and fitted in liquid and freezing PuO2, respectively. In these plots, the radiance Lk is divided by the first radiation constant c1 for the sake of simplicity. The main thermogram was obtained using a constant emittance of 0.83.

were retained for the reported melting/freezing temperature and error analysis. The mean value of the NSE of liquid and solid PuO2 near the freezing transition was 0.83 ± 0.05 throughout the analysed wavelength range and for five successive runs on the same sample. This value was used in the conversion of all radiance temperatures measured by the monochromatic pyrometer at 652 nm to true temperatures. It should be noted that the ‘‘systematic’’ uncertainty due to the greybody assumption is difficult to estimate within the current method. Nonetheless, the presented results are in excellent agreement with those calculated ab initio from low temperature electronic properties of PuO2 [16]. A total of eleven measurements of the PuO2 melting temperature on five different samples yielded a mean value of (3017 ± 28) K (Table 2). Expanded uncertainties reported here are estimates with a coverage factor k = 2 (two-standard deviations) [19]. Although a direct emittance measurement in the future will probably reduce the uncertainty, a dramatic change in the melting temperature value is rather unlikely, considering that a change by 10% of the NSE at this wavelength and temperature would trans-

Table 2 The quoted measurements of PuO2 melting point compared to the result obtained in this work. Authors

Ref.

PuO2 melting temperature (K)

Year

Type of work

Lyon and Baily Aitken and Evans Riley Carbajo et al. Guéneau et al. Kato et al.

[1]

2663 ± 20

1967

[2]

2718 ± 15

1968

[3] [5] [6] [7–10]

2673 ± 20 2701 ± 35 2660 2843a

1970 2001 2008 2008

This work

–

3017 ± 28

2010

Furnace heating in W crucibles Furnace heating in W crucibles Oxidative furnace Review paper Review paper Furnace heating in W or Re crucibles with different shapes. Laser heating under containerless conditions and controlled atmosphere

a Large uncertainty, impossible to estimate due to extensive reaction between PuO2 samples and Re crucible.

171

F. De Bruycker et al. / Journal of Nuclear Materials 416 (2011) 166–172

late into a corresponding change in true temperature of less than 1.5%, as can easily be verified using Eq. (1). 4. Discussion The current result is in line with the recent conclusion of Kato et al. [7,8] that the previous investigations [1–3] yielded too low transition temperatures because of sample-containment interaction during the heat treatments. Yet, Kato proposed a temperature of melting still 170 K lower than that found in this work (Table 2). This could be due to underestimation of the interaction between the PuO2-rich MOX samples and the rhenium capsule employed during the experiments. These interactions could result from not only the contamination of the plutonium dioxide samples with metal coming from the holder, but also, and probably more effectively, the chemical diffusion of oxygen from the samples into the crucible, as the authors themselves pointed out. Moreover, it should be noted that most researchers performed this sort of measurements with pyrometers calibrated against the melting points of metals recommended as secondary references. However, even during the measurement of these reference melting points reaction between the standard samples and their containment could have occurred, thus affecting the entire calibration procedure and, hence, the temperature scale definition. The main goal of the current experimental technique is to neutralize all these possible error sources. In addition, a series of post-melting analyses were performed to make sure that the measured thermal arrests corresponded to the melting point of stoichiometric PuO2 and that no detectable composition changes occurred at high temperature because of incongruent vaporisation or segregation. The samples were controlled after each laser experiment. No indication of vaporisation was noticed; the molten/refrozen material looked shiny and well confined in the centre of the sample. The appearance of the non-molten part did not change upon the heating/cooling process. No signs of any (already unlikely) interaction between the molten part and the holding screws were observed. Furthermore, the repeatability of the measurements both for each sample and for the overall five samples (originating from different batches) indicated good chemical and structural stability of the samples under the laser heat treatment. The micro-structure of a molten and re-solidified sample was studied by SEM. From the micrograph shown in Fig. 6 it is likely

that the top 30 lm (zones 1 and 2) of the sample were molten. Zone 1 is characterized by the presence of some porosity and columnar grains, whereas zone 2 appears highly dense and porosity-free. Zone 3 represents the unaltered starting material (zone 3). However, the origin of the formation of the two distinct zones in the molten layer is not clear. A possible explanation is that cooling of the melt starts from two sides, i.e. the sample surface and from the bulk. These two solidification fronts move towards each other and meet at the interface between zone 1 and 2. It is important to stress again that the entire molten matter was surrounded by the bulk so that no contamination was possible. The molten matter was partially detachable from the bulk. XRD analysis was performed on both a piece of fresh PuO2 taken from the bulk and a molten/refrozen part. The resulting lattice parameters could be compared and, using the analysis reported by Gardner et al. [11], the respective O/Pu ratios before and after melting could be determined. Fig. 7 shows the two diffractograms obtained on sample G before and after melting and freezing. One can clearly observe that both pieces had the same crystal structure, fluoritelike face centred cubic. It was necessary to magnify some of the XRD pattern features at high diffraction angles to see some slight differences (inset in Fig. 7). The fcc lattice parameter obtained for the unmolten material was 0.5396(1) Å. The diffractogram of the molten part displayed peaks slightly broader and twinned, possibly corresponding to the presence of two very similar phases with lattice parameters of 5.4018(1) Å and 5.3989(1) Å. In reality, the molten part being only 15 mg in weight, it is very likely that it was not removed and isolated perfectly, so that non-negligible traces of unmolten material were analysed by XRD together with the frozen one. Taking into account the uncertainties, calculated from the equipment precision and the fit of the diffractogram, the measured lattice parameters based on the correlation curve reported in [11] corresponded in both cases to stoichiometric plutonium dioxide PuO2.00±0.01. This uncertainty on the composition is largely negligible in terms of observable effects on the melting temperature. EPMA was also performed on molten and refrozen plutonium dioxide samples in order to check their composition and homogeneity. This kind of characterization, although affected by a larger uncertainty, corroborated the XRD results. It can therefore be assured that the PuO2 samples investigated were, within the experimental uncertainty, stoichiometric and pure before and after being subjected to laser heating and melting. This implies, moreover, that stoichiometric plutonium dioxide melts quasi-congruently, at least within the limits of the current approach, as any substantial composition difference between solid and liquid would have resulted

Counts / 1000 (-)

60

Counts / 1000 (-)

50 40 30

8 6 4 2 0 115

116

117

118

2

20 10 0

Fig. 6. SEM images of a cross-section through a partly molten PuO2 pellet. The extent of the molten zone is indicated by the two bars in the overview image. The location of the magnified image is marked by the arrow in the inset. Three different zones were observed: (zone 1) a top surface zone where re-solidified grain shapes are visible, (zone 2) that shows no porosity, and (zone 3) corresponding to the original unheated bulk. A sharp interface can be seen between the three zones.

0

20

40

60

80

100

120

2 Fig. 7. XRD comparison of the bulk and molten/frozen part of sample G. Inset: magnification of XRD comparison of the bulk and molten/frozen part of sample G at high diffraction angles.

172

F. De Bruycker et al. / Journal of Nuclear Materials 416 (2011) 166–172

in segregation in the refrozen material, as was never observed in the current samples. This conclusion allows using Richard’s rule [20] extended to oxides according to [21] in order to estimate the melting enthalpy of PuO2. According to such a statistical-thermodynamic approach, the molar entropy of fusion DsM for a congruently melting compound is, with fair approximation:

DsM  nR;

ð2Þ

where n is the number of atoms in the formula unit and R = 8.314 J K1 mol1 the ideal gas constant. The molar melting enthalpy DhM is obtained as

DhM ¼ DSM  T M ;

ð3Þ

TM being the melting temperature. Using the current values one can estimate 1

DhM ðPuO2 Þ  75 kJ mol : Of course this value is only indicative for the molar melting enthalpy of PuO2. Its uncertainty is larger than the uncertainty limits of the current measurements, as Eq. (2) is only valid for ideal liquids where molecular interactions, surface tension and intrinsic defects are neglected. However, these uncertainty sources should not affect the proposed latent heat by more than ±10%, by comparison with the values already estimated in the literature: 70.3 kJ mol1 [21] and 67 ± 10 kJ mol1 [22]. Moreover, the current value is close to those of other material systems very similar to PuO2 (e.g., uranium dioxide) in which more data about the melting enthalpy are available [23]. In principle, a value for the latent heat of melting could be estimated also from the length of the experimental solidification thermal arrests reported in Figs. 2–5. To this purpose, the melted mass and the energy supplied to the sample by the heating laser beam need to be known with sufficient accuracy in each experiment, and the heat losses have to be well estimated too. In practice, the uncertainty with which all these quantities (in particular the melted mass and the heat losses) can be estimated in the current experiments is much too large to allow any quantitative determination of the latent heat by such an approach, and the simpler use of Richard’s rule yields more meaningful results. As a further conclusion, it can be remarked that this is not the first time that the current method yields results in disagreement with older ones about very high temperature phase transitions. It happened, for example, in the case of the melting behaviour of hyperstoichiometric uranium dioxide [24] and magnesium oxide [25]. On the other hand, the current method already confirmed earlier results for material systems (such as U–C, Zr–C, and Zr–O) that are known to be chemically more stable and less volatile at high temperature [26,27].

The current method seems suitable to a further extension of the current investigation to hypo-stoichiometric plutonium dioxide (PuO2x) and to the binary UO2–PuO2 phase diagram, both systems being of great interest for the nuclear energy industry. The current result is also a motivation to revisit the melting behaviour of other transuranium dioxides. Acknowledgements The Authors are indebted to J. Somers, M. Rini, S. Stohr, C. Boshoven, M. Ernstberger, G. Pagliosa, D. Bouexière and F. Sarli (JRC-ITU) for the help with the sample preparation and characterization. The Authors also wish to thank P. Raison, O. Benes, P. Gotcu, M. J. Welland (JRC – ITU), C. Guéneau (CEA – Saclay), M. Malki (CNRS-Orléans) and B. Sundman (INSTN – Saclay) for the useful discussions. This research has been partly financed by the Seventh Framework Programme of the European Commission through the FBRIDGE project (Grant Agreement No. 211690). References [1] [2] [3] [4]

[5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17]

[18]

[19]

[20]

5. Conclusion The melting/freezing temperature of stoichiometric plutonium dioxide has been studied for the first time by fast laser heating and multi-wavelength pyrometry. The transition temperatures obtained by the current technique are in disagreement with those previously proposed on the basis of more traditional measurements. These latter were most likely affected by extensive reaction between samples and containment, an issue entirely avoided with the present method. Based on the present investigation, the following temperature is recommended for the melting/freezing point of stoichiometric PuO2: (3017 ± 28) K.

[21] [22]

[23] [24] [25] [26] [27]

W.L. Lyon, W.E. Baily, J. Nucl. Mater. 22 (1967) 332–339. E.A. Aitken, S.K. Evans, General Electric, GEAP-5672, 1968. B. Riley, Sci. Ceram. 5 (1970) 83–109. R.J. Lemire, J. Fuger, H. Nitsche, P.E. Potter, M.H. Rand, J. Rydberg, K. Spahiu, J.C. Sullivan, W.J. Ullman, P. Vitorge, H. Wanner, OECD-NEA, Chemical Thermodynamics Series Chemical Thermodynamics of Neptunium and Plutonium, vol. 4, Elsevier, North-Holland, 2001, p. 333. J.J. Carbajo, G.L. Yoder, S.G. Popov, V.K. Ivanov, J. Nucl. Mater. 299 (2001) 181– 198. C. Guéneau, C. Chatillon, B. Sundman, J. Nucl. Mater. 378 (2008) 257–272. M. Kato, K. Morimoto, H. Sugata, K. Konashi, M. Kashimura, T. Abe, J. Alloys Compd. 452 (2008) 48–53. M. Kato, K. Morimoto, H. Sugata, K. Konashi, M. Kashimura, T. Abe, J. Nucl. Mater. 373 (2008) 237–245. M. Kato, K. Morimoto, S. Nakamichi, H. Sugata, K. Konashi, M. Kashimura, T. Abe, Trans. Atom. Energy Soc. Jpn 7 (4) (2008) 420–428 (in Japanese). M. Kato, Melting temperatures of oxide fuel for fast reactor, in: Proceedings of ICAPP ‘09 Tokyo, Japan, Paper 9451, May 10–14, 2009. E.L. Gardner, T.L. Markin, R.S. Street, J. Inorg. Nucl. Chem. 27 (1965) 541–551. M.J. Welland, W.T. Thompson, B.J. Lewis, D. Manara, J. Nucl. Mater. 385 (2009) 358–363. D. Manara, M. Sheindlin, W. Heinz, C. Ronchi, Rev. Sci. Instrum. 79 (2008) 113901–113908. H. Preston-Thomas, Metrologia 27 (1990) 3–10 (Erratum in Metrologia 27 (1990) 107). R.E. Bedford, G. Bonnier, H. Maas, F. Pavese, Metrologia 33 (1996) 133–154. H. Shi, M. Chu, P. Zhang, J. Nucl. Mater. 400 (2010) 151–156. A.S. Tenney, Radiation ratio thermometry, in: D.P. DeWitt, G.D. Nutter (Eds.), Theory and Practice of Radiation Thermometry, Wiley, New York, 1988, pp. 489–492. G. Neuer, L. Fiessler, M. Groll, E. Schreiber, in: J.F. Schooley (Ed.), Temperature: Its Measurement and Control in Science and Industry, vol. 6, AIP, New York, 1992, pp. 787–789. Joint Committee for Guides in Metrology, Evaluation of Measurement Data – Guide to the Expression of Uncertainty in Measurement, JCGM 100:2008, Corrected version 2010. P. Papon, J. Leblond, P.H.E. Meijer, Physique des Transitions de Phases, Dunod, Paris, 1999, p. 116. L.F. Epstein, J. Nucl. Mater. 22 (1967) 340–349. V.P. Glushko, L.V. Gurvich, G.A. Bergman, I.V. Veits, V.A. Medvedev, G.A. Khachkuruzov, V.S. Yungman, Thermodynamic Properties of Individual Substances, vol. 4, Nauka, Moscow, USSR, 1982. J.K. Fink, J. Nucl. Mater. 279 (2000) 1–18. D. Manara, C. Ronchi, M. Sheindlin, M. Lewis, M. Brykin, J. Nucl. Mater. 342 (2005) 148–163. C. Ronchi, M. Sheindlin, Int. J. Thermophys. 23 (2002) 293–305. C.A. Utton, F. De Bruycker, K. Boboridis, R. Jardin, H. Noel, C. Guéneau, D. Manara, J. Nucl. Mater. 385 (2009) 443–448. H.F. Jackson, D.D. Jayaseelan, W.E. Lee, M.J. Reece, F. Inam, D. Manara, C. Perinetti- Casoni, F. De Bruycker, K. Boboridis, Int. J. Appl. Ceram. Technol. 7 (2010) 316–326.