- Email: [email protected]
Plasma spray deposition of yttrium oxide on graphite, coating characterization and interaction with molten uranium
Available online at www.sciencedirect.com
ScienceDirect Journal of the European Ceramic Society 35 (2015) 787–794
Plasma spray deposition of yttrium oxide on graphite, coating characterization and interaction with molten uranium Y. Chakravarthy a , Subhankar Bhandari a , Vandana Chaturvedi a , A. Pragatheeswaran b , A. Nagraj a , T.K. Thiyagarajan a , P.V. Ananthapadmanaban a,∗ , A.K. Das a a
Laser & Plasma Technology Division, Bhabha Atomic Research Centre, Mumbai 400 085, India b Department of Physics, Karunya University, Coimbatore 641 114, India
Received 27 March 2014; received in revised form 4 September 2014; accepted 8 September 2014 Available online 23 September 2014
Abstract Yttrium oxide coatings on graphite substrates were prepared by atmospheric plasma spray technique. Temperature and velocity of the yttrium oxide particles during the plasma spray process were measured by ‘spraywatch’ diagnostics system. Coatings were characterized for phase composition, microstructure and thermal stability. Dense adherent coatings could be deposited at 24 kW power. Corrosion behaviour of the coatings in molten uranium was studied by using differential thermal analysis (DTA) up to 1473 K. The results showed that the coatings offered sufficient protection to graphite against corrosion by molten uranium. © 2014 Elsevier Ltd. All rights reserved. Keywords: Plasma spray deposition; Yttrium oxide coating; Corrosion resistant coatings
1. Introduction Plasma sprayed ceramic coatings are extensively used for thermal barrier, corrosion barrier and wear resistant applications.1–5 High density graphite is used for processing uranium and its alloys. Since molten uranium is not chemically compatible with graphite, a protective ceramic coating should be provided on the crucible to prevent the reaction between uranium and carbon. Although many refractory oxides, carbides and nitrides have been suggested for this purpose,6–8 yttrium oxide (Y2 O3 ) is the most preferred coating material. Yttrium oxide has high melting point (about 2700 K) and is chemically stable in many reactive environments.9,10 Besides, it does not react with uranium and many reactive metals. By virtue of these reasons, yttrium oxide is used as a protective coating on graphite crucibles and metallic moulds used for processing uranium and its alloys.11–14
∗
Corresponding author. Tel.: +91 22 25595107; fax: +91 22 25505151. E-mail address: [email protected] (P.V. Ananthapadmanaban).
http://dx.doi.org/10.1016/j.jeurceramsoc.2014.09.012 0955-2219/© 2014 Elsevier Ltd. All rights reserved.
Tournier et al.15 studied the reaction of sintered yttrium oxide with molten uranium metal. These authors found that yttria reacted with molten uranium above 1675 K to form uranium oxide and sub-stoichiometric Y2 O3−x . The dissolved oxygen in uranium, after the experiment, was reported to be 1.2 × 10−3 mole fraction. Yasushi et al.10 melted uranium in sintered yttria crucible and reported a 100 m thick layer of UO2 after 96 h of operation at 1675 K. Padmanabhan et al.16 reported that yttrium oxide was chemically stable against attack by uranium up to 3000 K from thermodynamic calculations. The authors also studied the reaction of yttrium oxide with molten uranium and found that there was no reaction between the molten metal and the oxide. Alangi et al.17 reported that yttrium oxide coating offered sufficient protection to tantalum substrate against corrosion by molten uranium. The present study focuses on plasma spray deposition of yttrium oxide on graphite substrates and characterization of the coatings. Yttrium oxide coating was applied directly over graphite without any bond coat. The coatings were characterized for phase composition and microstructure. Temperature and velocity of yttrium oxide particles in the plasma jet were measured by ‘spraywatch’ diagnostics system. Differential thermal
788
Y. Chakravarthy et al. / Journal of the European Ceramic Society 35 (2015) 787–794
Fig. 1. SEM image of Y2 O3 raw powder.
analysis (DTA) from room temperature to 1475 K was used to investigate the reaction between yttrium oxide and uranium. 2. Experimental methods 2.1. Plasma spray deposition Chemically pure yttrium oxide powder (Grade: MEC2100P1 by M/s MEC Pvt. Ltd., Jodhpur, India) with particle size in the range of 15–40 m was used for plasma spray deposition. The scanning electron microscope (SEM) photograph of the raw powder is shown in Fig. 1. The SEM image of a single particle is also shown alongside. It is evident from the figures that the powder consists of agglomerates of spherical particles. A 40 kW DC arc plasma spray system, developed in our laboratory, was used for plasma spray deposition. The torch consists of a tungsten cathode (10 mm in diameter) with a conical tip and a copper nozzle, which acts as the anode. The anode is 40 mm in length and the nozzle diameter is 7 mm. The electrodes are cooled by water and a Teflon insulator separates the electrodes. A mixture of argon and nitrogen was used as the plasma gas, which was injected into the torch through an opening in the insulator segment. Input power to the plasma torch was varied from 16 kW to 30 kW by controlling the gas flow and arc current. The flow rate of argon gas was varied from 25 to 30 standard litres per minute (SLPM). Trial experiments had shown that erosion rate of the nozzle was very high for gas flow rate of 20 SLPM and below. Therefore, experiments were carried out only at 25 and 30 SLPM of primary gas flow rate. Based on a large number of trial experiments and operational experience, the standoff distance between the torch and substrate was kept constant at 100 mm. Yttrium oxide powder was stored in a powder feeder and injected into the plasma jet by means of a carrier gas (Ar) through a side port located 2 mm before the exit of the plasma torch nozzle.
Table 1 Typical operating parameters for plasma spray coating of Y2 O3 . Operating parameter
Experimental value
Input power Arc voltage Arc current Plasma gas Plasma gas (N2 ) Powder carrier gas Powder feed rate Particle size Torch-base distance
16–24 kW 40 V 400–600 A 30 LPM 3 LPM 10–12 LPM ∼15 g/min 15–40 m 100 mm
The range of operating parameters for coating process is given in Table 1. Spray deposition was carried out on graphite substrates, which were thermally etched by the plasma flame for about 1 min before starting the coating process. This was done to remove any organic binder and to increase the surface roughness of the substrate. The SEM image of the etched graphite surface is shown in Fig. 2. It is evident from the figure that the etched surface has better interlocking sites. The weight loss observed during thermal etching was small and found to be about 8.5 mg/cm2 . Plasma spray coating was carried out on cylindrical graphite specimens of 20 mm diameter and 10 mm thickness and rectangular specimens of 25 mm × 20 mm × 10 mm. Cubic graphite specimens (4 × 4 × 4 mm3 ), coated on all sides with yttrium oxide, were used for differential thermal analysis (DTA) experiments. The coated cubic sample and a button of uranium metal were loaded in the DTA apparatus and the experiment was carried out from room temperature (RT) to 1475 K at a heating rate of about 25 ◦ C/min. The specimen was held at 1475 K for about 1 h and then cooled to room temperature. Microstructure and energy dispersive X-ray (EDX) analysis of the coated sample after the DTA experiment was also carried out.
Fig. 2. SEM microstructure of graphite surface (a) before thermal etching and (b) after etching.
Y. Chakravarthy et al. / Journal of the European Ceramic Society 35 (2015) 787–794
Thermal stability of the coatings was studied by heating the samples under vacuum up to 1475 K at a heating rate of 300 ◦ C/h. Details of the vacuum furnace used have been described elsewhere.17 The samples were held at 1475 K for 2 h and then allowed to cool to room temperature under vacuum. The heating, holding and cooling cycle was repeated and the sample was examined periodically after each cycle. The experiment was not continued after 40 cycles.
Table 2 Inflight particle temperature and velocity of Y2 O3 powder during spraying. Standoff distance (mm) 60
80
2.2. Coating characterization 100
Temperature and velocity of the yttrium oxide particles in the plasma jet during the coating process was measured by the ‘SprayWatch2i’ diagnostics system (Oseir Ltd., Tampere, Finland). The ‘SprayWatch’ uses an optical camera based system to image the plasma spray process. Particle velocity and temperature are measured by time-of-flight method and two-colour pyrometer respectively. Particle trajectory was first obtained by high speed photography and then the SprayWatch2i diagnostics was used to image the particle stream for a short duration of time of about 5–10 ms, which correspond to 20–60 pixels on the CCD detector. The distance travelled by the particle during the exposure time was calculated using a special image-processing algorithm and the velocity of the particle was computed. Temperature of the particles along the plasma jet axis was measured at 60, 80 and 100 mm away from the exit of the plasma torch nozzle. Temperature measurement was not possible up to 40 mm from the nozzle exit due to the brightness of plasma jet, which eclipses the particles. Carl Zeiss model EVO 40 scanning electron microscope (SEM) was used for analyzing the microstructure of the coatings. The samples were cross-sectioned by a precision diamond cutting machine, coated with gold (∼10 nm) and the interface was observed under SEM. Elemental analysis, before and after DTA experiment was carried out using Bruker Quantax 2000 EDX analyser attached to the electron microscope. The phase composition of the coatings was determined by X-ray diffraction technique using Rigaku Miniflex-II X-ray diffraction unit. X-ray diffraction patterns were recorded using nickel-filtered copper K-␣ radiation in θ–2θ geometry. The adhesion between the coating and the substrate was measured by standard tensile adhesion tester. Graphite specimen holder of standard size as per ASTM C633 was not used directly because of failure of the threading in the graphite block during testing. So, a graphite cylindrical disc of 25 mm diameter and 5 mm thickness was used as the test specimen. The disc was coated with yttria on one side and then sandwiched between two specimen holders made of stainless steel (25 mm in diameter and 25 mm in length) by using commercially available glue araldite (standard epoxy adhesive made by Huntsman Advanced Materials). After applying the glue, the specimens were allowed to cure for 24 h at room temperature. Adhesion tests were carried out only on those specimens deposited at 24 kW. The bond strength of araldite on stainless steel specimens was separately determined and found to be 20 MPa.
789
Power (kW)
Temperature (K) (standard deviation)
Velocity (m/s)
16 20 24 16 20 24 16 20 24
2605 (±21) 2816 (±22) 2893 (±25) 2780 (±21) 3000 (±20) 3123 (±28) 2675 (±19) 2953 (±18) 3083 (±23)
96 (±3) 106 (±4) 121 (±2) 101 (±2) 113 (±3) 128 (±3) 97 (±3) 108 (±2) 124 (±3)
3. Results and discussion 3.1. Particle temperature and velocity in the plasma jet Table 2 shows the temperature and velocity of yttrium oxide particles at different locations along the plasma jet axis. Temperature and velocity of the particles were measured at different plasma power levels. Temperature and velocity values reported are the average of 10 readings. Temperature of particles could be measured within about 1% accuracy and the velocity could be measured within about 3% accuracy.18 It is evident from the table that particle temperature and velocity at a given location along the plasma jet axis increase with plasma power. This behaviour is expected in view of the increase in plasma temperature with increase in plasma power. The increased plasma temperature leads to enhanced heat transfer and momentum transfer from the plasma to the particle. Consequently, the temperature and velocity of the particles increase with plasma power. It is also seen from the results that, at given plasma power, the particle temperature and velocity increase initially and then fall off. It was observed that at 16 kW, the particle temperature at 60 mm from the torch exit was about 2605 K. The temperature then increased to about 2750 K at 80 mm, about 50 K higher than the melting temperature of yttrium oxide. Beyond 80 mm, the temperature of the particle started decreasing and reached a value of about 2675 K at 100 mm. A similar trend was observed at all power levels. However, at 20 kW and 24 kW, it was observed that the particle temperature was above the melting point at all locations. The particle velocity also followed a similar trend with plasma power and distance. These results are consistent with the theoretical results obtained by Thiyagarajan et al.19 3.2. Phase, microstructure and thermal stability Fig. 3 summarizes the X-ray diffraction results. X-ray diffraction pattern of the coated sample was seen to be similar to that of the starting material, consisting of a single homogeneous cubic phase of Y2 O3 corresponding to JCPDS file 41-1105 Presence of the meta-stable monoclinic phase, as reported by Gourlaouen et al.,20 could not be identified. Fig. 4 shows the cross sectional microstructures of yttrium oxide coatings deposited at various plasma power levels. The
790
Y. Chakravarthy et al. / Journal of the European Ceramic Society 35 (2015) 787–794
Fig. 3. XRD analysis of Y2 O3 raw powder and coating.
micro-structural features show molten/partially molten lamellae with scattered porosity, characteristic of plasma sprayed coatings. The effect of plasma power on the microstructure is evident from the figure. SEM photograph of yttrium oxide deposited at 16 kW shows high percentage of porosity. The microstructure also shows molten particles, which did not have sufficient velocity to spread out. Surface porosity showed pore sizes ranging from 2 to 10 m. It is also seen from the microstructures that, as the plasma power is increased from 16 kW to 20 kW, porosity is reduced and the interface adhesion is improved. The microstructure of plasma deposited yttrium oxide at 24 kW and gas flow rate of 42 SLPM of argon + 3 SLPM of nitrogen gas, shows fully molten grains and flattened lamellae. Progressive reduction in porosity and better interlocking of the coating with the substrate surface with increasing plasma power are evident from the cross sectional microstructures. The plasma temperature increases with increase in torch power and consequently the particle temperature and velocity also increase.19 This leads to better spreading and flow of the molten particles into the surface irregularities of the substrate resulting in decreased porosity and better interface bonding. These findings are in complete
agreement with the measured particle temperature and velocity. The average coating thickness of the coatings was found to be about 300 m, as observed under SEM. The adhesion strength of yttria coating (deposited at 24 kW) on graphite substrate was found to be around 3 MPa, which is the average of five readings. Examination of the test samples after tensile adhesion testing showed that coating failure occurred at the coating–graphite interface. Photographs of the samples after adhesion test are shown in Fig. 5. Examination of the samples after thermal treatment showed that the coatings retained their integrity even after repeated thermal treatments. SEM observation (Fig. 6) of the interface after 40 cycles of thermal heating experiment did not show any crack indicating the stability of the coating–substrate system at highly reducing conditions. However, it was observed that the colour of the coating has changed to dull white indicating sub-stoichiometry of Y2 O3 with oxygen deficiency.21,22 X-ray diffraction analysis of the coating after thermal treatment, given in Fig. 7, showed only cubic yttrium oxide identical with that of the starting material.
Fig. 4. SEM at coating/substrate interface (inserted images are at higher magnification).
Y. Chakravarthy et al. / Journal of the European Ceramic Society 35 (2015) 787–794
791
Fig. 5. (a) Samples before adhesion test and (b) after testing.
Fig. 6. Coating/substrate interface (a) before heating and (b) after heat treatment.
3.3. Interaction with molten uranium Results of differential thermal analysis (DTA) of the reaction couple consisting of yttria coated graphite and uranium are summarized in Fig. 8. The first endothermic peak observed at about 955 K is attributed to the transition of the ␣-phase of uranium to the -phase of uranium. The endothermic peak observed at about 1075 K is assigned to the phase transition of the -phase to the ␥-phase of uranium. The subsequent endothermic peak occurring at about 1410 K corresponds to the melting of uranium. During the cooling cycle, all the above peaks appear as exothermic peaks. No other peak attributable to any reaction between
Y2 O3 –U, U–C or Y2 O3 –C was observed. This confirms that yttrium oxide does not react with uranium metal and carbon. Another important point that emerges is that yttrium oxide coating offers adequate protection against infiltration of molten uranium through the yttrium oxide layer into the graphite substrate. Visual examination of the yttrium oxide-coated graphite specimens after the DTA experiment showed that the coating–substrate integrity was intact and no cracks were observed. The coating surface did not show any uranium particle sticking on the surface. This is due to the low wettability of Y2 O3 towards molten uranium.23 Similar results were found by Koger et al.11 after 30 min of yttria–uranium interaction at 1575 K. Penetration of liquid uranium into these pores is dependent on the surface tension and the contact angle. The rate of capillary penetration of liquid metal into the pores is given by the Washburn equation: dh/dt = (1/8μh){p − ρgh}r 2
Fig. 7. XRD patterns of coating surface after thermal treatment.
Fig. 8. DTA curve between uranium and ytrria coated graphite couple.
(1)
792
Y. Chakravarthy et al. / Journal of the European Ceramic Society 35 (2015) 787–794
Fig. 9. EDX spectra of coating surface (a) before experiment and (b) after experiment.
In the above equation, ‘ρgh’ is the static pressure exerted by the liquid, ‘μ’ is the viscosity of the liquid, ‘ρ’ is the density of the liquid, ‘r’ is the capillary radius, ‘h’ is the distance of penetration of liquid into the capillary, ‘g’ is the acceleration due to gravity and ‘p’ is the capillary pressure which is given by: p = (1/D)4σ cos θ
(2)
In Eq. (2), ‘D’ is the pore diameter, ‘σ’ is the surface tension of liquid uranium and ‘θ’ is the contact angle between liquid uranium and yttrium oxide, all in consistent units. The surface tension (σ) of liquid uranium between 1408 K and 1850 K is given by Cahill and Kirshenbaum,24 according to the following relation: σ (in dynes/cm) = 1747 − 0.14T
(3)
Considering a maximum pore size of 10 m on the surface, 1.54 N/m as the surface tension (1473 K) of liquid uranium and
contact angle of 106◦ , the pressure required for liquid uranium to penetrate the pore is found to be 0.169 MPa. Assuming a pool of 5 mm liquid uranium in the DTA sample holder, the total hydraulic head available is only about 930 Pa, which is well below the pressure required to force uranium liquid into the pore. This explains the observation of uranium not penetrating into the coating. The EDX elemental spectra of the coating surface before and after DTA experiment are shown in Fig. 9. The EDX spectra before and after the DTA experiment are observed to be the same. The only elements that could be detected were Y, O and C. Uranium concentration was seen to be below the detectable limit (1 at.%). The EDX line spectra of the coating–substrate interface after DTA, shown in Fig. 10, also gave similar results. The above results confirmed that uranium has not penetrated through the pores on the surface. Similar findings were reported by Kim et al.,12 who studied the reaction of yttria with U–10wt.%Zr alloy
Fig. 10. EDX line spectrum of coating surface (a) before DTA and (b) after DTA.
Y. Chakravarthy et al. / Journal of the European Ceramic Society 35 (2015) 787–794
793
Fig. 11. (a) EDX elemental mapping of coating surface before DTA. (b) EDX elemental mapping of coating surface after DTA.
at 1873 K for 15 min. These authors carried out EDX analysis and confirmed that yttria did not react with uranium. Penetration of uranium vapour through the pores/inter-splat voids would have resulted in the accumulation of uranium by condensation through all the interconnected pores. However, during the SEM/EDX analysis, uranium concentration present in the coating on the graphite substrate was found to be below the detection limit of the instrument. This further confirms the absence of interconnected porosity in the coatings. Elemental mapping of Y, O and U on the surface of the sample after the DTA experiment, shown in Fig. 11(a) and (b), revealed no trace of uranium. Uranium concentration in all the cases observed was found to be below the detection limit. The variation in the signal counts across the sample surface is due to the surface roughness of the coating. The above results clearly show that there was no chemical reaction between molten uranium and yttrium oxide. The results also confirm the efficacy of plasma sprayed yttrium oxide coating in protecting the graphite substrate from interaction with uranium. Therefore, yttrium oxide coatings on graphite can be used as an effective barrier coating against molten uranium. It is clear from the results discussed in the foregoing paragraphs that yttrium oxide and uranium do not chemically react. The results also show no degradation of the coating and that the integrity of the coating is retained even after repeated thermal cycling. The results also rule out the possibility of infiltration of molten uranium through the yttrium oxide layer into the graphite substrate and consequent interaction with carbon. The most significant point that emerges from the study is that plasma deposited yttrium oxide coating offers adequate protection to the underlying graphite substrate against corrosion by molten uranium.
4. Conclusion Thermally stable and adherent coatings of yttrium oxide on graphite substrates were prepared by plasma spray technique. The cubic phase of Y2 O3 was retained after plasma spray deposition and thermal treatment. Micro-structural features of the coatings deposited at different plasma power levels could be correlated with the temperature and velocity of yttrium oxide particles measured during the deposition process. The coatings were found to retain their structural integrity even after repeated thermal cycles. Differential thermal analysis and subsequent SEM and EDX analysis showed that the coating was impervious to penetration of molten uranium and offered adequate corrosion protection to the graphite substrate. Acknowledgement The authors express their gratitude to Dr. T.R.G. Kutty, Radio Metallurgy Division, BARC, Mumbai for his help in recording the DTA analysis. The authors thank Dr. L.M. Gantayet, Director, Beam Technology Development Group, BARC for his constant encouragement and support. References 1. Heimann RB. Plasma spray coating principle and applications. 2nd ed. Weinheim, Germany: Wiley-VCH; 1996. 2. Pawlowski L. The science and engineering of thermal spray coatings. 2nd ed. West Sussex, England: Wiley; 1995. 3. Fauchais P. Understanding plasma spraying. J Phys D: Appl Phys 2004;37:R86–108. 4. Cao XQ, Vassen R, Stoever D. Ceramic materials for thermal barrier coatings. J Eur Ceram Soc 2004;24:1–10.
794
Y. Chakravarthy et al. / Journal of the European Ceramic Society 35 (2015) 787–794
5. Celika E, Ozdemir I, Avci E, Tsunekawa Y. Corrosion behavior of plasma sprayed coatings. Surf Coat Technol 2005;193:297–302. 6. Lee SH, Cho CH, Lee YS, Lee HS, Kim. J-G. Chemical reactivity of oxide materials with uranium and uranium trichloride. Kor J Chem Eng 2010;27(6):1786–90. 7. Kim KH, Lee CT, Lee CB, Fielding RS, Kennedy JR. Characterization of ceramic plasma-sprayed coatings and interaction studies between U–Zr fuel and ceramic coated interface at an elevated temperature. Thin Solid Films 2011;519:6969–73. 8. Holcombe CE, Banker JG. Uranium/ceramic oxide and carbon/ceramic oxide interaction studies. Metall Trans B 1978;9(2):317–9. 9. Pfdender E. Thermal plasma technology, where do we stand and where are we going. Plasma Chem Plasma Process 1999;19(1):1–31. 10. Hoshino Y, Ota K, Yamagami S, Endo K. Corrosion behaviors of materials used in uranium metal production. In: Electrochemical society proceedings, Proc. of the tenth international symposium on molten salts, vol. 96(7). 1996. p. 222–9. 11. Koger JW, Holcombe CE, Banker JG. Coatings on graphite crucibles used in melting uranium. Thin Solid Films 1976;39:297–303. 12. Kim YH, Kim HT, Woo YM, Kim KH, Lee CB, Fielding RS. Interaction studies of ceramic vacuum plasma spraying for the melting crucible material. Nucl Eng Technol 2013;45(5):683–8. 13. Massalski TB, Murrey JL, Bennet LH, Baker H. Binary alloy phase diagram. Ohio: American Society for Metals; 1986. 14. Condon JB, Holcombe CE. Solubility of yttrium in uranium at 1700 K. J Less-Common Metals 1977;55:297–8. 15. Tournier C, Lorrain B, Le Guyadec F, Coudurier L, Eustathopoulos N. Kinetics of interfacial reactions in molten U/solid Y2 O3 system. J Nucl Mater 1998;254:215–20.
16. Padmanabhan PVA, Ramanathan S, Sreekumar KP, Satpute RU, Kutty TRG, Gonal MR, et al. Synthesis of thermal spray grade yttrium oxide powder and its application for plasma spray deposition. Mater Chem Phys 2007;106:416–21. 17. Alangi N, Mukherjee J, Anupama P, Verma MK, Chakravarthy Y, Padmanabhan PVA, et al. Liquid uranium corrosion studies of protective yttria coatings on tantalum substrates. J Nucl Mater 2011;410: 39–45. 18. Kanta AF, Montavon G, Berndt CC, Planche MP, Coddet C. Intelligent system for prediction and control: application in plasma spray process. Expert Syst Appl 2011;38:260–71. 19. Thiyagarajan TK, Ananthapadmanabhan PV, Sreekumar KP, Chakravarthy Y, Das AK, Gantayet LM, et al. Experimental and simulation approach to plasma spray deposition of yttrium oxide. Surf Eng 2012;28(9): 646–56. 20. Gourlaouen V, Schnedecker G, Lejus AM, Boncoeur M, Collongues R. Metastable phases in yttrium oxide plasma spray deposits and their effect on coating properties. Mater Res Bull 1993;28:415–25. 21. Nakamichi M, Kawamura H, Terai T, Tanaka S. Characterization of Y2 O3 coating for liquid blanket. J Nucl Mater 1997;248: 165–9. 22. Cheng X-R, Dai H-Y, Qi Z-M, Wang Y-Y, Zhang G-B. Local structure of oxygen-deficient yttrium oxide. Chin Phys C 2013;37:098002. 23. Tournier C, Lorrian B, Le Guyadec F, Eustathopoulos N. The influence of non-stochiometry of oxides on their wettability by liquid metals. J Mater Sci Lett 1996;15:485–9. 24. Cahill JA, Kirshenbaum AD. The surface tension of liquid uranium from its melting point, 1406–1850 K. J Inorg Nucl Chem 1965;27(1): 73–6.
ScienceDirect Journal of the European Ceramic Society 35 (2015) 787–794
Plasma spray deposition of yttrium oxide on graphite, coating characterization and interaction with molten uranium Y. Chakravarthy a , Subhankar Bhandari a , Vandana Chaturvedi a , A. Pragatheeswaran b , A. Nagraj a , T.K. Thiyagarajan a , P.V. Ananthapadmanaban a,∗ , A.K. Das a a
Laser & Plasma Technology Division, Bhabha Atomic Research Centre, Mumbai 400 085, India b Department of Physics, Karunya University, Coimbatore 641 114, India
Received 27 March 2014; received in revised form 4 September 2014; accepted 8 September 2014 Available online 23 September 2014
Abstract Yttrium oxide coatings on graphite substrates were prepared by atmospheric plasma spray technique. Temperature and velocity of the yttrium oxide particles during the plasma spray process were measured by ‘spraywatch’ diagnostics system. Coatings were characterized for phase composition, microstructure and thermal stability. Dense adherent coatings could be deposited at 24 kW power. Corrosion behaviour of the coatings in molten uranium was studied by using differential thermal analysis (DTA) up to 1473 K. The results showed that the coatings offered sufficient protection to graphite against corrosion by molten uranium. © 2014 Elsevier Ltd. All rights reserved. Keywords: Plasma spray deposition; Yttrium oxide coating; Corrosion resistant coatings
1. Introduction Plasma sprayed ceramic coatings are extensively used for thermal barrier, corrosion barrier and wear resistant applications.1–5 High density graphite is used for processing uranium and its alloys. Since molten uranium is not chemically compatible with graphite, a protective ceramic coating should be provided on the crucible to prevent the reaction between uranium and carbon. Although many refractory oxides, carbides and nitrides have been suggested for this purpose,6–8 yttrium oxide (Y2 O3 ) is the most preferred coating material. Yttrium oxide has high melting point (about 2700 K) and is chemically stable in many reactive environments.9,10 Besides, it does not react with uranium and many reactive metals. By virtue of these reasons, yttrium oxide is used as a protective coating on graphite crucibles and metallic moulds used for processing uranium and its alloys.11–14
∗
Corresponding author. Tel.: +91 22 25595107; fax: +91 22 25505151. E-mail address: [email protected] (P.V. Ananthapadmanaban).
http://dx.doi.org/10.1016/j.jeurceramsoc.2014.09.012 0955-2219/© 2014 Elsevier Ltd. All rights reserved.
Tournier et al.15 studied the reaction of sintered yttrium oxide with molten uranium metal. These authors found that yttria reacted with molten uranium above 1675 K to form uranium oxide and sub-stoichiometric Y2 O3−x . The dissolved oxygen in uranium, after the experiment, was reported to be 1.2 × 10−3 mole fraction. Yasushi et al.10 melted uranium in sintered yttria crucible and reported a 100 m thick layer of UO2 after 96 h of operation at 1675 K. Padmanabhan et al.16 reported that yttrium oxide was chemically stable against attack by uranium up to 3000 K from thermodynamic calculations. The authors also studied the reaction of yttrium oxide with molten uranium and found that there was no reaction between the molten metal and the oxide. Alangi et al.17 reported that yttrium oxide coating offered sufficient protection to tantalum substrate against corrosion by molten uranium. The present study focuses on plasma spray deposition of yttrium oxide on graphite substrates and characterization of the coatings. Yttrium oxide coating was applied directly over graphite without any bond coat. The coatings were characterized for phase composition and microstructure. Temperature and velocity of yttrium oxide particles in the plasma jet were measured by ‘spraywatch’ diagnostics system. Differential thermal
788
Y. Chakravarthy et al. / Journal of the European Ceramic Society 35 (2015) 787–794
Fig. 1. SEM image of Y2 O3 raw powder.
analysis (DTA) from room temperature to 1475 K was used to investigate the reaction between yttrium oxide and uranium. 2. Experimental methods 2.1. Plasma spray deposition Chemically pure yttrium oxide powder (Grade: MEC2100P1 by M/s MEC Pvt. Ltd., Jodhpur, India) with particle size in the range of 15–40 m was used for plasma spray deposition. The scanning electron microscope (SEM) photograph of the raw powder is shown in Fig. 1. The SEM image of a single particle is also shown alongside. It is evident from the figures that the powder consists of agglomerates of spherical particles. A 40 kW DC arc plasma spray system, developed in our laboratory, was used for plasma spray deposition. The torch consists of a tungsten cathode (10 mm in diameter) with a conical tip and a copper nozzle, which acts as the anode. The anode is 40 mm in length and the nozzle diameter is 7 mm. The electrodes are cooled by water and a Teflon insulator separates the electrodes. A mixture of argon and nitrogen was used as the plasma gas, which was injected into the torch through an opening in the insulator segment. Input power to the plasma torch was varied from 16 kW to 30 kW by controlling the gas flow and arc current. The flow rate of argon gas was varied from 25 to 30 standard litres per minute (SLPM). Trial experiments had shown that erosion rate of the nozzle was very high for gas flow rate of 20 SLPM and below. Therefore, experiments were carried out only at 25 and 30 SLPM of primary gas flow rate. Based on a large number of trial experiments and operational experience, the standoff distance between the torch and substrate was kept constant at 100 mm. Yttrium oxide powder was stored in a powder feeder and injected into the plasma jet by means of a carrier gas (Ar) through a side port located 2 mm before the exit of the plasma torch nozzle.
Table 1 Typical operating parameters for plasma spray coating of Y2 O3 . Operating parameter
Experimental value
Input power Arc voltage Arc current Plasma gas Plasma gas (N2 ) Powder carrier gas Powder feed rate Particle size Torch-base distance
16–24 kW 40 V 400–600 A 30 LPM 3 LPM 10–12 LPM ∼15 g/min 15–40 m 100 mm
The range of operating parameters for coating process is given in Table 1. Spray deposition was carried out on graphite substrates, which were thermally etched by the plasma flame for about 1 min before starting the coating process. This was done to remove any organic binder and to increase the surface roughness of the substrate. The SEM image of the etched graphite surface is shown in Fig. 2. It is evident from the figure that the etched surface has better interlocking sites. The weight loss observed during thermal etching was small and found to be about 8.5 mg/cm2 . Plasma spray coating was carried out on cylindrical graphite specimens of 20 mm diameter and 10 mm thickness and rectangular specimens of 25 mm × 20 mm × 10 mm. Cubic graphite specimens (4 × 4 × 4 mm3 ), coated on all sides with yttrium oxide, were used for differential thermal analysis (DTA) experiments. The coated cubic sample and a button of uranium metal were loaded in the DTA apparatus and the experiment was carried out from room temperature (RT) to 1475 K at a heating rate of about 25 ◦ C/min. The specimen was held at 1475 K for about 1 h and then cooled to room temperature. Microstructure and energy dispersive X-ray (EDX) analysis of the coated sample after the DTA experiment was also carried out.
Fig. 2. SEM microstructure of graphite surface (a) before thermal etching and (b) after etching.
Y. Chakravarthy et al. / Journal of the European Ceramic Society 35 (2015) 787–794
Thermal stability of the coatings was studied by heating the samples under vacuum up to 1475 K at a heating rate of 300 ◦ C/h. Details of the vacuum furnace used have been described elsewhere.17 The samples were held at 1475 K for 2 h and then allowed to cool to room temperature under vacuum. The heating, holding and cooling cycle was repeated and the sample was examined periodically after each cycle. The experiment was not continued after 40 cycles.
Table 2 Inflight particle temperature and velocity of Y2 O3 powder during spraying. Standoff distance (mm) 60
80
2.2. Coating characterization 100
Temperature and velocity of the yttrium oxide particles in the plasma jet during the coating process was measured by the ‘SprayWatch2i’ diagnostics system (Oseir Ltd., Tampere, Finland). The ‘SprayWatch’ uses an optical camera based system to image the plasma spray process. Particle velocity and temperature are measured by time-of-flight method and two-colour pyrometer respectively. Particle trajectory was first obtained by high speed photography and then the SprayWatch2i diagnostics was used to image the particle stream for a short duration of time of about 5–10 ms, which correspond to 20–60 pixels on the CCD detector. The distance travelled by the particle during the exposure time was calculated using a special image-processing algorithm and the velocity of the particle was computed. Temperature of the particles along the plasma jet axis was measured at 60, 80 and 100 mm away from the exit of the plasma torch nozzle. Temperature measurement was not possible up to 40 mm from the nozzle exit due to the brightness of plasma jet, which eclipses the particles. Carl Zeiss model EVO 40 scanning electron microscope (SEM) was used for analyzing the microstructure of the coatings. The samples were cross-sectioned by a precision diamond cutting machine, coated with gold (∼10 nm) and the interface was observed under SEM. Elemental analysis, before and after DTA experiment was carried out using Bruker Quantax 2000 EDX analyser attached to the electron microscope. The phase composition of the coatings was determined by X-ray diffraction technique using Rigaku Miniflex-II X-ray diffraction unit. X-ray diffraction patterns were recorded using nickel-filtered copper K-␣ radiation in θ–2θ geometry. The adhesion between the coating and the substrate was measured by standard tensile adhesion tester. Graphite specimen holder of standard size as per ASTM C633 was not used directly because of failure of the threading in the graphite block during testing. So, a graphite cylindrical disc of 25 mm diameter and 5 mm thickness was used as the test specimen. The disc was coated with yttria on one side and then sandwiched between two specimen holders made of stainless steel (25 mm in diameter and 25 mm in length) by using commercially available glue araldite (standard epoxy adhesive made by Huntsman Advanced Materials). After applying the glue, the specimens were allowed to cure for 24 h at room temperature. Adhesion tests were carried out only on those specimens deposited at 24 kW. The bond strength of araldite on stainless steel specimens was separately determined and found to be 20 MPa.
789
Power (kW)
Temperature (K) (standard deviation)
Velocity (m/s)
16 20 24 16 20 24 16 20 24
2605 (±21) 2816 (±22) 2893 (±25) 2780 (±21) 3000 (±20) 3123 (±28) 2675 (±19) 2953 (±18) 3083 (±23)
96 (±3) 106 (±4) 121 (±2) 101 (±2) 113 (±3) 128 (±3) 97 (±3) 108 (±2) 124 (±3)
3. Results and discussion 3.1. Particle temperature and velocity in the plasma jet Table 2 shows the temperature and velocity of yttrium oxide particles at different locations along the plasma jet axis. Temperature and velocity of the particles were measured at different plasma power levels. Temperature and velocity values reported are the average of 10 readings. Temperature of particles could be measured within about 1% accuracy and the velocity could be measured within about 3% accuracy.18 It is evident from the table that particle temperature and velocity at a given location along the plasma jet axis increase with plasma power. This behaviour is expected in view of the increase in plasma temperature with increase in plasma power. The increased plasma temperature leads to enhanced heat transfer and momentum transfer from the plasma to the particle. Consequently, the temperature and velocity of the particles increase with plasma power. It is also seen from the results that, at given plasma power, the particle temperature and velocity increase initially and then fall off. It was observed that at 16 kW, the particle temperature at 60 mm from the torch exit was about 2605 K. The temperature then increased to about 2750 K at 80 mm, about 50 K higher than the melting temperature of yttrium oxide. Beyond 80 mm, the temperature of the particle started decreasing and reached a value of about 2675 K at 100 mm. A similar trend was observed at all power levels. However, at 20 kW and 24 kW, it was observed that the particle temperature was above the melting point at all locations. The particle velocity also followed a similar trend with plasma power and distance. These results are consistent with the theoretical results obtained by Thiyagarajan et al.19 3.2. Phase, microstructure and thermal stability Fig. 3 summarizes the X-ray diffraction results. X-ray diffraction pattern of the coated sample was seen to be similar to that of the starting material, consisting of a single homogeneous cubic phase of Y2 O3 corresponding to JCPDS file 41-1105 Presence of the meta-stable monoclinic phase, as reported by Gourlaouen et al.,20 could not be identified. Fig. 4 shows the cross sectional microstructures of yttrium oxide coatings deposited at various plasma power levels. The
790
Y. Chakravarthy et al. / Journal of the European Ceramic Society 35 (2015) 787–794
Fig. 3. XRD analysis of Y2 O3 raw powder and coating.
micro-structural features show molten/partially molten lamellae with scattered porosity, characteristic of plasma sprayed coatings. The effect of plasma power on the microstructure is evident from the figure. SEM photograph of yttrium oxide deposited at 16 kW shows high percentage of porosity. The microstructure also shows molten particles, which did not have sufficient velocity to spread out. Surface porosity showed pore sizes ranging from 2 to 10 m. It is also seen from the microstructures that, as the plasma power is increased from 16 kW to 20 kW, porosity is reduced and the interface adhesion is improved. The microstructure of plasma deposited yttrium oxide at 24 kW and gas flow rate of 42 SLPM of argon + 3 SLPM of nitrogen gas, shows fully molten grains and flattened lamellae. Progressive reduction in porosity and better interlocking of the coating with the substrate surface with increasing plasma power are evident from the cross sectional microstructures. The plasma temperature increases with increase in torch power and consequently the particle temperature and velocity also increase.19 This leads to better spreading and flow of the molten particles into the surface irregularities of the substrate resulting in decreased porosity and better interface bonding. These findings are in complete
agreement with the measured particle temperature and velocity. The average coating thickness of the coatings was found to be about 300 m, as observed under SEM. The adhesion strength of yttria coating (deposited at 24 kW) on graphite substrate was found to be around 3 MPa, which is the average of five readings. Examination of the test samples after tensile adhesion testing showed that coating failure occurred at the coating–graphite interface. Photographs of the samples after adhesion test are shown in Fig. 5. Examination of the samples after thermal treatment showed that the coatings retained their integrity even after repeated thermal treatments. SEM observation (Fig. 6) of the interface after 40 cycles of thermal heating experiment did not show any crack indicating the stability of the coating–substrate system at highly reducing conditions. However, it was observed that the colour of the coating has changed to dull white indicating sub-stoichiometry of Y2 O3 with oxygen deficiency.21,22 X-ray diffraction analysis of the coating after thermal treatment, given in Fig. 7, showed only cubic yttrium oxide identical with that of the starting material.
Fig. 4. SEM at coating/substrate interface (inserted images are at higher magnification).
Y. Chakravarthy et al. / Journal of the European Ceramic Society 35 (2015) 787–794
791
Fig. 5. (a) Samples before adhesion test and (b) after testing.
Fig. 6. Coating/substrate interface (a) before heating and (b) after heat treatment.
3.3. Interaction with molten uranium Results of differential thermal analysis (DTA) of the reaction couple consisting of yttria coated graphite and uranium are summarized in Fig. 8. The first endothermic peak observed at about 955 K is attributed to the transition of the ␣-phase of uranium to the -phase of uranium. The endothermic peak observed at about 1075 K is assigned to the phase transition of the -phase to the ␥-phase of uranium. The subsequent endothermic peak occurring at about 1410 K corresponds to the melting of uranium. During the cooling cycle, all the above peaks appear as exothermic peaks. No other peak attributable to any reaction between
Y2 O3 –U, U–C or Y2 O3 –C was observed. This confirms that yttrium oxide does not react with uranium metal and carbon. Another important point that emerges is that yttrium oxide coating offers adequate protection against infiltration of molten uranium through the yttrium oxide layer into the graphite substrate. Visual examination of the yttrium oxide-coated graphite specimens after the DTA experiment showed that the coating–substrate integrity was intact and no cracks were observed. The coating surface did not show any uranium particle sticking on the surface. This is due to the low wettability of Y2 O3 towards molten uranium.23 Similar results were found by Koger et al.11 after 30 min of yttria–uranium interaction at 1575 K. Penetration of liquid uranium into these pores is dependent on the surface tension and the contact angle. The rate of capillary penetration of liquid metal into the pores is given by the Washburn equation: dh/dt = (1/8μh){p − ρgh}r 2
Fig. 7. XRD patterns of coating surface after thermal treatment.
Fig. 8. DTA curve between uranium and ytrria coated graphite couple.
(1)
792
Y. Chakravarthy et al. / Journal of the European Ceramic Society 35 (2015) 787–794
Fig. 9. EDX spectra of coating surface (a) before experiment and (b) after experiment.
In the above equation, ‘ρgh’ is the static pressure exerted by the liquid, ‘μ’ is the viscosity of the liquid, ‘ρ’ is the density of the liquid, ‘r’ is the capillary radius, ‘h’ is the distance of penetration of liquid into the capillary, ‘g’ is the acceleration due to gravity and ‘p’ is the capillary pressure which is given by: p = (1/D)4σ cos θ
(2)
In Eq. (2), ‘D’ is the pore diameter, ‘σ’ is the surface tension of liquid uranium and ‘θ’ is the contact angle between liquid uranium and yttrium oxide, all in consistent units. The surface tension (σ) of liquid uranium between 1408 K and 1850 K is given by Cahill and Kirshenbaum,24 according to the following relation: σ (in dynes/cm) = 1747 − 0.14T
(3)
Considering a maximum pore size of 10 m on the surface, 1.54 N/m as the surface tension (1473 K) of liquid uranium and
contact angle of 106◦ , the pressure required for liquid uranium to penetrate the pore is found to be 0.169 MPa. Assuming a pool of 5 mm liquid uranium in the DTA sample holder, the total hydraulic head available is only about 930 Pa, which is well below the pressure required to force uranium liquid into the pore. This explains the observation of uranium not penetrating into the coating. The EDX elemental spectra of the coating surface before and after DTA experiment are shown in Fig. 9. The EDX spectra before and after the DTA experiment are observed to be the same. The only elements that could be detected were Y, O and C. Uranium concentration was seen to be below the detectable limit (1 at.%). The EDX line spectra of the coating–substrate interface after DTA, shown in Fig. 10, also gave similar results. The above results confirmed that uranium has not penetrated through the pores on the surface. Similar findings were reported by Kim et al.,12 who studied the reaction of yttria with U–10wt.%Zr alloy
Fig. 10. EDX line spectrum of coating surface (a) before DTA and (b) after DTA.
Y. Chakravarthy et al. / Journal of the European Ceramic Society 35 (2015) 787–794
793
Fig. 11. (a) EDX elemental mapping of coating surface before DTA. (b) EDX elemental mapping of coating surface after DTA.
at 1873 K for 15 min. These authors carried out EDX analysis and confirmed that yttria did not react with uranium. Penetration of uranium vapour through the pores/inter-splat voids would have resulted in the accumulation of uranium by condensation through all the interconnected pores. However, during the SEM/EDX analysis, uranium concentration present in the coating on the graphite substrate was found to be below the detection limit of the instrument. This further confirms the absence of interconnected porosity in the coatings. Elemental mapping of Y, O and U on the surface of the sample after the DTA experiment, shown in Fig. 11(a) and (b), revealed no trace of uranium. Uranium concentration in all the cases observed was found to be below the detection limit. The variation in the signal counts across the sample surface is due to the surface roughness of the coating. The above results clearly show that there was no chemical reaction between molten uranium and yttrium oxide. The results also confirm the efficacy of plasma sprayed yttrium oxide coating in protecting the graphite substrate from interaction with uranium. Therefore, yttrium oxide coatings on graphite can be used as an effective barrier coating against molten uranium. It is clear from the results discussed in the foregoing paragraphs that yttrium oxide and uranium do not chemically react. The results also show no degradation of the coating and that the integrity of the coating is retained even after repeated thermal cycling. The results also rule out the possibility of infiltration of molten uranium through the yttrium oxide layer into the graphite substrate and consequent interaction with carbon. The most significant point that emerges from the study is that plasma deposited yttrium oxide coating offers adequate protection to the underlying graphite substrate against corrosion by molten uranium.
4. Conclusion Thermally stable and adherent coatings of yttrium oxide on graphite substrates were prepared by plasma spray technique. The cubic phase of Y2 O3 was retained after plasma spray deposition and thermal treatment. Micro-structural features of the coatings deposited at different plasma power levels could be correlated with the temperature and velocity of yttrium oxide particles measured during the deposition process. The coatings were found to retain their structural integrity even after repeated thermal cycles. Differential thermal analysis and subsequent SEM and EDX analysis showed that the coating was impervious to penetration of molten uranium and offered adequate corrosion protection to the graphite substrate. Acknowledgement The authors express their gratitude to Dr. T.R.G. Kutty, Radio Metallurgy Division, BARC, Mumbai for his help in recording the DTA analysis. The authors thank Dr. L.M. Gantayet, Director, Beam Technology Development Group, BARC for his constant encouragement and support. References 1. Heimann RB. Plasma spray coating principle and applications. 2nd ed. Weinheim, Germany: Wiley-VCH; 1996. 2. Pawlowski L. The science and engineering of thermal spray coatings. 2nd ed. West Sussex, England: Wiley; 1995. 3. Fauchais P. Understanding plasma spraying. J Phys D: Appl Phys 2004;37:R86–108. 4. Cao XQ, Vassen R, Stoever D. Ceramic materials for thermal barrier coatings. J Eur Ceram Soc 2004;24:1–10.
794
Y. Chakravarthy et al. / Journal of the European Ceramic Society 35 (2015) 787–794
5. Celika E, Ozdemir I, Avci E, Tsunekawa Y. Corrosion behavior of plasma sprayed coatings. Surf Coat Technol 2005;193:297–302. 6. Lee SH, Cho CH, Lee YS, Lee HS, Kim. J-G. Chemical reactivity of oxide materials with uranium and uranium trichloride. Kor J Chem Eng 2010;27(6):1786–90. 7. Kim KH, Lee CT, Lee CB, Fielding RS, Kennedy JR. Characterization of ceramic plasma-sprayed coatings and interaction studies between U–Zr fuel and ceramic coated interface at an elevated temperature. Thin Solid Films 2011;519:6969–73. 8. Holcombe CE, Banker JG. Uranium/ceramic oxide and carbon/ceramic oxide interaction studies. Metall Trans B 1978;9(2):317–9. 9. Pfdender E. Thermal plasma technology, where do we stand and where are we going. Plasma Chem Plasma Process 1999;19(1):1–31. 10. Hoshino Y, Ota K, Yamagami S, Endo K. Corrosion behaviors of materials used in uranium metal production. In: Electrochemical society proceedings, Proc. of the tenth international symposium on molten salts, vol. 96(7). 1996. p. 222–9. 11. Koger JW, Holcombe CE, Banker JG. Coatings on graphite crucibles used in melting uranium. Thin Solid Films 1976;39:297–303. 12. Kim YH, Kim HT, Woo YM, Kim KH, Lee CB, Fielding RS. Interaction studies of ceramic vacuum plasma spraying for the melting crucible material. Nucl Eng Technol 2013;45(5):683–8. 13. Massalski TB, Murrey JL, Bennet LH, Baker H. Binary alloy phase diagram. Ohio: American Society for Metals; 1986. 14. Condon JB, Holcombe CE. Solubility of yttrium in uranium at 1700 K. J Less-Common Metals 1977;55:297–8. 15. Tournier C, Lorrain B, Le Guyadec F, Coudurier L, Eustathopoulos N. Kinetics of interfacial reactions in molten U/solid Y2 O3 system. J Nucl Mater 1998;254:215–20.
16. Padmanabhan PVA, Ramanathan S, Sreekumar KP, Satpute RU, Kutty TRG, Gonal MR, et al. Synthesis of thermal spray grade yttrium oxide powder and its application for plasma spray deposition. Mater Chem Phys 2007;106:416–21. 17. Alangi N, Mukherjee J, Anupama P, Verma MK, Chakravarthy Y, Padmanabhan PVA, et al. Liquid uranium corrosion studies of protective yttria coatings on tantalum substrates. J Nucl Mater 2011;410: 39–45. 18. Kanta AF, Montavon G, Berndt CC, Planche MP, Coddet C. Intelligent system for prediction and control: application in plasma spray process. Expert Syst Appl 2011;38:260–71. 19. Thiyagarajan TK, Ananthapadmanabhan PV, Sreekumar KP, Chakravarthy Y, Das AK, Gantayet LM, et al. Experimental and simulation approach to plasma spray deposition of yttrium oxide. Surf Eng 2012;28(9): 646–56. 20. Gourlaouen V, Schnedecker G, Lejus AM, Boncoeur M, Collongues R. Metastable phases in yttrium oxide plasma spray deposits and their effect on coating properties. Mater Res Bull 1993;28:415–25. 21. Nakamichi M, Kawamura H, Terai T, Tanaka S. Characterization of Y2 O3 coating for liquid blanket. J Nucl Mater 1997;248: 165–9. 22. Cheng X-R, Dai H-Y, Qi Z-M, Wang Y-Y, Zhang G-B. Local structure of oxygen-deficient yttrium oxide. Chin Phys C 2013;37:098002. 23. Tournier C, Lorrian B, Le Guyadec F, Eustathopoulos N. The influence of non-stochiometry of oxides on their wettability by liquid metals. J Mater Sci Lett 1996;15:485–9. 24. Cahill JA, Kirshenbaum AD. The surface tension of liquid uranium from its melting point, 1406–1850 K. J Inorg Nucl Chem 1965;27(1): 73–6.