- Email: [email protected]
The influence of Ar+ sputtering on the hydriding behavior of uranium
Journal of Nuclear Materials 494 (2017) 55e60
Contents lists available at ScienceDirect
Journal of Nuclear Materials journal homepage: www.elsevier.com/locate/jnucmat
The influence of Arþ sputtering on the hydriding behavior of uranium Ren Bin a, Guangfeng Zhang a, Hefei Ji b, Lizhu Luo a, Peng Shi b, Xiaolin Wang c, * a
Science and Technology on Surface Physics and Chemistry Laboratory, PO Box 9-35, Jiangyou 621908, Sichuan, China Institute of Materials, China Academy of Engineering Physics, PO Box 9-13, Mianyang 621900, Sichuan, China c China Academy of Engineering Physics, Mianyang 621900, Sichuan, China b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 22 December 2016 Received in revised form 7 July 2017 Accepted 7 July 2017 Available online 8 July 2017
The influence of superficial defects (induced by surface sputtering) on uranium oxidation and hydriding behaviors has been studied. Depleted uranium surface was etched by Arþ sputtering, the etched surface morphology was investigated by scanning electron microscope. The initial oxidation kinetics of sputtered uranium surface and mechanical polished uranium surface in ambient atmosphere were characterized using spectroscopic ellipsometry. As-polished and sputtered samples, as well as sequence sputtered samples ambient oxidized for different time, were compared in hydriding kinetics to investigate the coinfluence of surface oxidation and superficial defects induced by Arþ ion bombardment. The morphologic characteristics of hydride sites in a few grains were also revealed and enhanced by sputtering, showing a network of fractures for fast hydride transportation. © 2017 Published by Elsevier B.V.
Keywords: Depleted uranium Hydrogen corrosion Induction time Arþ sputtering
1. Introduction Uranium is an important material in military and energy industries. However, the storage and usage of uranium or uranium alloys components are at risk due to the high reactivity between uranium and ambient oxygen, water and hydrogen. Hydrogen corrosion is one of the most serious corrosion problems for damaging the integrity or decaying the mechanic properties of uranium. Due to the remarkable density mismatch between U (19.05 g/cm3) and UH3 (10.92 g/cm3), when constrained by metal the growth of hydride precipitates would cause a 75% expansion in lattice, force the surrounding metal to plastic deform or even fracture, and inevitably cause damage or failure to components made of uranium or uranium alloys. What's worse, the pyrophoric uranium hydride is dangerous if exposed to ambient and also threatening the environment. Therefore, research on uraniumhydrogen (U-H) reaction is a persisting focus since Manhattan Project in 1940s, covering many aspects like reaction kinetics, metallography and surface chemical analysis, etc. Fruitful achievements have provided us a lot of fundamental understandings and many methods are developed to reduce the surface hydrogen corrosion. However, the initial hydriding kinetics of U-H reaction,
* Corresponding author. E-mail address: [email protected] (X. Wang). http://dx.doi.org/10.1016/j.jnucmat.2017.07.017 0022-3115/© 2017 Published by Elsevier B.V.
especially for the key factors of hydride nucleation, is still ambiguous. One critical problem is the discrepancies in nucleation preference with just a little difference in experiment setup, or even the same setup would yield very different results. A popular idea is that clean (oxide free) uranium surface would react with pure hydrogen rapidly and entirely in an ideal assumption, the localization of hydride nucleation is attributed to the discontinuity in the semiprotective surface oxide film. Glascott et al. have proposed a model for the initial reaction assuming an enhanced transportation through thin areas on surface oxide [1], and their recent report has confirmed that the abundant grain boundaries in nm-sized polycrystalline air grown oxide contribute to the high diffusion coefficient in gaseous diffusion model [2]. Furtherly, Bloch et al. [3e5] proposed that not only surface oxide, but also the layer beneath the oxide which is saturated of water, hydroxyl clusters or organic absorbents, as well as strain defects and residual stress induced by mechanical processing, have critical influence in the induction time and localization for hydriding nucleation. The combination of surface oxide layer and underlying mechanically modified layer is called surface passivation layer (SPL). The characteristics of SPL changes with different mechanical processing methods and parameters, may consequently causes the inconsistencies in the hydriding behavior reported in different studies. To clarify the role of SPL playing in hydriding kinetics and nucleation behavior, preparing a surface without SPL, or making it uniform and known, is essential to reduce the inconsistencies and
56
R. Bin et al. / Journal of Nuclear Materials 494 (2017) 55e60
yield better understandings. Several pretreatments had been adopted to achieve this aim. The most common one is vacuum heat-treatment prior to hydriding experiment [6,7]. This pretreatment shows an ‘activation’ to the uranium surface, namely shortening the induction period and accelerating the nucleation, by removing chemisorbed water or hydroxyl group and release active sites for hydrogen's accumulation. This pretreatment is commonly adopted by researchers before conducting U-H reaction, and the consistency is generally better. However, with this method the oxide on uranium surface was not removed or even thickened by reaction of uranium and absorbed water or residual oxygen during heat treatment, the uniformity is also not improved. Another method is electropolishing [8]. Surface oxide and strained layer induced by mechanical polishing is electrochemically etched and the surface becomes uniform (not considering the inclusions) and covered by a protective oxide layer. However, the electropolishing process is usually conducted in aqueous solution so the absorbed water or hydroxyl are not negligible, and the newly formed oxide layer will conceal some characteristics which are important in U-H reaction, for example an intermediate layer of oxide along carbidemetal interface is formed under the driven of applied electric field, thus the preferred nucleation sites around inclusions would be blocked [9]. Despite of previous efforts, a more appealing approach is highly desired for making a uniform surface to deeply understand its reaction behavior. The technique of Arþ ion sputtering is used customarily for the processing and analysis of solid surfaces. The bombardment of Arþ ions with kinetic energy ‘physically’ removes the surface atoms layer by layer. Zhang et al. reported that the Arþ ion beam sputtering has the fine polishing effect on uranium metal surface [10]. Ion beams could also reproducibly generate atomic level defects and nanometer-scale vacancy clusters in a controlled manner [11], and induces changes in the composition and chemical state [12]. Compared with heat treatment and electropolishing, Arþ sputtering can almost avoid the adsorbed water, hydroxyl or other impurities. For the reasons mentioned above, it would be attractive to apply this approach to modify the metallic uranium surface. In this paper, uranium surface was modified via Arþ ion sputtering technique. The effect of sputtering on the morphology of uranium surface was investigated, as well as the oxidation and hydriding behavior, comparing with the samples just mechanical polished and vacuum heat treated. The difference between sputtered and mechanical polished samples in oxidation and hydriding kinetics could be referred as the effect of surface layer modified by mechanical processing, and the role of metal substrate plays in hydride nucleation was also partially revealed. 2. Experimental 2.1. Sample preparation Depleted uranium (DU) coupons (4 5 mm 1.5 mm, with impurities carbon <100 wppm, nitrogen <50 wppm, calcium 70 wppm, tungsten 90 wppm, others < 120 wppm) were cut from an ingot prepared by arc melting. Each sample was polished, rinsed with distilled water and alcohol, and then ultrasonic rinsed in acetone for 5 min. The sputtering was conducted with a high-flux ion beam sputtering system, in which samples were etched with Arþ ion beam of 600 eV, 24 mA for 30 min. The base vacuum and working argon pressure were 1 104 Pa and 2.2 102 Pa, respectively. 2.2. Oxidation Following preparation, the oxidation of sputter DU was
conducted in ambient air at a temperature of 25 C and RH 40%e 60%. The refraction spectra on oxidized surface were measured with spectroscopic ellipsometer (model Woollam M2000). Oxide thickness is obtained by fitting each spectrum with Lorentz model, and the value derived from first spectrum was subtracted as an assumption that the initial thickness at the first measurement is zero. Actually, there is an intrinsic oxide layer because the samples had been exposed to air for about 30 min before the spectrometer was calibrated and the tests began. Sequence of thickness data was obtained periodically and in situ. Note that the beam spot size is in mm scale thus the oxide thickness measured is a statistical average of many grains. 2.3. Hydriding Reactions of hydrogen and DU specimen polished or Arþ sputtered were carried out in a visual reaction chamber. A vacuum heat pretreatment was conducted by gradually heating to 160 C (about 5 C/min), kept for 1 h and then naturally cooled down to 70 C. Before and during heat-treatment the chamber was evacuated with turbo-pump to make sure the pressure was below 0.1 Pa. After the temperature stabilized, high purity hydrogen (99.999% and purified with a LaNi5 bed) was introduced to reaction chamber, making the initial pressure around 1 105 Pa and began the reaction. The pressure during reaction was closely observed and recorded, by about 2% dropping of the initial pressure, or the macroscopic nuclei appeared on sample surface, the chamber was immediately evacuated again and the reaction was halted. The microscopic morphologies before and after U-H reaction were observed with scanning electron microscope (model Helios Nanolab 600i). After hydriding experiment uranium surface was covered by a thick oxide layer due to the long exposure in ambient. To reveal the distribution of hydride nucleation sites and the morphology of hydride underlying the oxide layer, sample surfaces were gently polished, and etched to remove the oxide layer formed due to the exposure after reaction and clean the surface. 3. Results and discussion 3.1. Surface morphology of sputtered DU It has been reported that due to the extra low solubility of hydrogen in UO2 lattice, integrated UO2 layer can act as a good physical barrier for hydrogen diffusion theoretically. However, in real conditions, hydrogen always penetrates the oxide layer through its defects including dislocations, impurity atoms and vacancy clusters, formed intrinsically or extrinsically, and their assemblies like grain boundaries and inclusion-substrate interfaces. In addition, the solubility of hydrogen in uranium metal are also very low and far below the critical concentration to form UH3 [13], it's not likely to nucleate in a defect-free metal lattice [14]. The preferred nucleation sites are located where a lot of defects exist thus capable to accumulate supersaturate H atoms. Hence, it's essential to understand the relationship between preferred hydride nucleation sites and surface layer, which for several microns in depth is modified by surface processing while preparing. To avoid the influence of surface processing, the surface layer was removed completely by sputtering for sufficient long time, and keep the process in vacuum to avoid the ambient agents. Fig. 1 shows the micrographs of DU surface with or without etching. The surface only mechanical polished shows a flat and homogenous surface except for some inclusions and scratches, lacking microstructure details like grain boundaries (Fig. 1a). After etching for 30 min the difference in height for atomic planes of grains with different orientations are clearly observed, showing a
R. Bin et al. / Journal of Nuclear Materials 494 (2017) 55e60
57
Fig. 1. Micrographs of DU surface with different sputtering time. (a) As-polished; (b) 30 min sputtering. The square inclusions shown in are uranium nitride (see in supplementary material).
preferential sputtering by difference in sputter rate on grains with different orientations (Fig. 1b). The surface micrographs contain more details than on as-polished surface, similar to the effect of electrochemical polishing. The roughness is in nanoscale and varies on different grains. Inclusions on the surface are mainly nitride and oxide examined by EDS, they have changed little in height and roughness after being sputtered, reveals that the sputter rate on carbide and nitride are much lower than on metal uranium due to a higher hardness. 3.2. Oxidation In U-H reaction, oxide layer on uranium surface is unavoidable and always playing an important role [15,16]. It's necessary for us to investigate the oxidation behavior of sputtered uranium before conducting hydriding reaction, on this purpose the spectroscopic ellipsometry was applied. According to literature [17e19], in the early stage of uranium oxidization the diffusion of oxygen in UO2 lattice is the rate determining step at moderate temperature and moisture. The relationship between the oxide thickness s and the exposure time t is parabolic
s ¼ a þ kt 1=2
(1)
where k is the rate constant. The intercept a is none-zero due to three reasons: the error of fitting, the kinetics of very early stage of oxidation does not obey the time square-root law or the recorded t is not exactly the time oxidation begins. There is about 30 min prior to the beginning of the spectroscopic ellipsometry experiments, thus an initial oxidation on uranium surface is already formed due to the high reactivity of uranium and oxygen. The growth of oxide thicknesses on sputtered surface, as well as that of as-polished, is shown in Fig. 2. The fluctuation of these two curves is due to the drift in environmental temperature and moisture. As shown in Fig. 2, both curves can be split into two segments and linear fitted separately. On as-polished surface, the parabolic rate constant k is 3.66 nm h1/2 at the initial stage, and for the stage that the oxidation is deep than 15 nm, k is 2.20 nm h1/2. As for sputtered surface, the rate constant k for the initial stage and deep oxidation stage is 6.43 nm h1/2 and 1.93 nm h1/2, respectively. Given that the rate constant k is related to the diffusion barrier at a certain temperature, the fitting result indicates that the activation energy for oxygen diffusing in the oxide layer within 15 nm is much lower than the oxide layer deeper. On sputtered surface k is
Fig. 2. The oxidation on as-polished and sputtered uranium surfaces. Both curves are divided into two regions and fitted following equation (1), the y-axis represents the thickness of surface oxide, and the x-axis are square root of time to reveal the linear relationship between s and t1/2.
markedly higher than on the as-polished surface, which implies that a significant decrease in activation energy for uranium oxidation is caused by sputtering. The difference between the surface layer of as-polished and sputtered surface with oxide thickness less than 15 nm, is mainly density of defects. In close packed metal, especially for uranium with a large atomic radius, migrating through interstitials and vacancies is the dominating diffusion mechanism. In this mechanism, the density of vacancies is critical for the diffusion barrier. On sputtered surface the defects are much denser in a very thin (15 nm) surface layer due to heavy dose Arþ bombardment. In this layer, the saturated solubility, as well as chemical potential of oxygen atoms are higher than in the surface layer of as-polished sample, thus it shows a fast oxidation in the initial stage. Yet on as-polished surface the density of defects is also slightly higher than in intrinsic bulk due to the damage of mechanic processing, hence the oxidation rate is also higher than the inner layer. In the depth of sputtered or as-polished sample, the density of vacancies is very close to intrinsic level and the diffusion barrier is higher than sputtered and mechanical modified layer.
58
R. Bin et al. / Journal of Nuclear Materials 494 (2017) 55e60
Fig. 3. The pressure drop of hydrogen in reaction with sputtered and polished surface. The initial pressures are normalized to 1.0 105 Pa.
3.3. Hydriding Both as-polished and sputtered samples are ambient exposed for less than 10 min before they are put into reaction chamber and evacuated, hence the oxide layer developed on sample surface is relatively thin. The hydrogen consumption curves for both aspolished and Arþ sputtered specimen is illustrated in Fig. 3, which difference reveals the influence of surface defective layer. On as-polished sample, the reaction is delayed and the hydrogen pressure does not drop immediately after the initial hydrogen exposure, this delay on pressure-time curve is referred as the induction period. On sputtered uranium surface the induction period is negligible, and the consumption of hydrogen begins immediately after the initial exposure of hydrogen. Induction period are usually regarded as the time consumed for hydrogen to penetrate surface oxide layer, reach the metal-oxide interface and then exceed the critical concentration for nucleation. It is commonly accepted that
induction period is mainly due to the diffusion of hydrogen in ‘adhesive’ oxide layer on uranium surface, supported by the similarity in activation energy of hydriding reaction by Arrhenius fitting of the nucleation rate constants [20] and the diffusion barrier of hydrogen in single crystal uranium dioxide [21]. The absence of induction period is mainly due to the removal of residual surface oxide and introduction of abundant superficial defects. According to the case of oxidation discussed in section 3.2, for a sputtered sample with a short period exposure to air, the defects induced by the bombardment of Arþ ions would distribute in both the surface oxide layer and the top of underlying metal. These defects act as enhanced diffusion paths and hydrogen accumulation sites. The time consumed for hydrogen to penetrate the surface oxide layer and exceeds the critical concentration are greatly truncated, comparing to a regular oxidized surface. After induction period hydrogen pressure are dropped rapidly for both samples. The reaction rate for sputtered sample is slightly higher than as-polished one, probably this is also due to the superficial defects. The hydriding induction periods and consumption rates for sputtered sample, which are of different oxidation states before UH reaction, are listed in Table 1. Here the induction period is defined as the period for first 0.1 kPa drop of hydrogen pressure while the hydrogen consumption rate is the slope of the linear region in hydrogen consumption curve. The prolonging of induction period with increasing oxide thickness is observed for sputtered samples with longer ambient exposure, meanwhile the hydrogen consumption rates in linear region are progressively slowdown. For the sputtered samples exposed to ambient for 0.5 h and 2 h, there is no evident induction period observed because of the enhanced hydrogen diffusion and accumulation in defective layer, as discussed above. Stored in ambient air for 12 h or longer, the induction period appeared again, indicates that the defective layer is completely oxidized and the enhancement by defects vanishes. The relationship between surface defective layer induced by Arþ bombardment, surface oxide layer and induction time for U-H reaction is shown in Fig. 4 schematically. With short exposure (Fig. 4a), the surface layer with abundant defects is not completely oxidized and still provides the benefits for hydrogen diffusion and
Table 1 Induction time and hydrogen consumption rates on sputtered uranium surface with different oxidizing time. Exposure time (h)
Oxide thicknessa (nm)
Induction period (s)
Hydrogen consumption rate (Pa/s)
0.5 2 12 24 42 118
3.0 7.9 16.4 19.9 20.5 31.6b
51 36 245 290 405 461
8.85 6.85 6.56 5.66 4.82 4.00
a b
The oxide thicknesses are read from spectroscopic ellipsometry data. Oxide thickness for 118 h exposed surface is evaluated by parabolic fitting and extrapolating of spectroscopic ellipsometry data.
Fig. 4. Scheme of hydride nucleation on sputtered surface on situation that (a) the oxide layer is shallower than the defective layer, or (b) the oxidation exceeds the defective layer.
R. Bin et al. / Journal of Nuclear Materials 494 (2017) 55e60
accumulation, thus the induction time is eliminated or greatly shortened. As the oxidation exceeds the defective region (Fig. 4b), the integrated oxide layer formed underlying the defective layer act as a barrier for hydrogen diffusion, and its thickness would affect how soon hydride nucleation at oxide-metal interface would begin. In addition, the compression induced by surface oxide layer would also suppress the expansion of underlying nuclei, reduce the growth rate of hydride too, that is another reason for the slowdown with increasing oxide thickness. 3.4. Hydride morphology According to Owen et al. [3,22], the hydride nuclei can be classified into two families (or four types according to Arkush et al. [23]): One is high in nucleation density and low in growth rate blister-like nucleus with size limited to several microns. Another is much lower in density but growing much faster with an unlimited final size. In present work, the latter type of nucleus is focus on. Fig. 5 shows the microscopy of hydride nuclei on as-polished and sputtered sample surface. As shown in Fig. 5a, the hydride nucleation on polishing modified surface occurs densely and the hydrides expand laterally beneath surface layer, their growth are suppressed by overlying surface layer. As hydride nuclei grow, the stress induced by hydride expansion causes the crack of surface
59
layer, exposing a porous hydride surface and then the reaction is enhanced by the direct contact of hydrogen and bulk uranium. On sputtered surface the compression is less intense due to the integrity of surface layer is damaged by superficial defects, so the surface cracks easily and the hydrogen attack exhibits a ‘explosive’ look (Fig. 5b). It's natural that strong compression of thick, integrated and adhesive oxide layer has suppressed the lateral growth and expansion of uranium hydride, thus the hydride is more likely to develop inward. When the suppression is strong enough, if there is no possibility to accommodate the hydride newly formed, or the substrate is hard to deform and release the stress induced by hydride expansion, the hydride growth would be inhibited before grown up. Conversely, if there are ways to accommodate hydride growth like grain boundaries or interfaces between inclusions and substrate, or release of stress induced by hydride expansion (structure deformation like twinning and fracturing), hydride nuclei growth would persist and the final size is much larger than in the former case. From this point, the classification of hydride nuclei evaluating with the difference of size and growing behavior is in correspondence with not only the integration of surface oxide layer but also the properties of metal substrate. Fig. 6 shows the hydride sites on an as-polished sample, with the surface slightly polished and etched again after hydriding experiment. The sputtering process is very helpful not only to
Fig. 5. Surface morphologies of hydride corrosion sites on (a) as-polished surface and (b) sputtered surface.
Fig. 6. The preferential hydriding on as-polished surface. (a) The network of fractures; (b) details of the fractures that connect the hydride sites. Remind that (a) and (b) are polished and sputtered again after hydrogen attack.
60
R. Bin et al. / Journal of Nuclear Materials 494 (2017) 55e60
enhance the detailed surface structures like grain boundaries and fracture edges, but also revealed the trace of hydride propagation in the metal bulk. The hydride would oxidize immediately after exposure to ambient. Compares to uranium substrate, uranium oxide is higher in hardness hence the sputtering rate is relatively lower and the oxide would protrude from the sputtered surface. In Fig. 6 the oxide clusters exist in the cracks around hydride craters, which implies that the cracking occurred during U-H reaction, induced by expansion of hydride, and have formed a network among individual hydride sites, provide ‘highways’ for hydrogen or hydride to transport from one site to another. It occurs only in a few grains, as we observed the hydriding sites are sparse and inhomogeneous. However, for these specific grains the corrosion is serious and the integrity is almost destroyed. If the attack persists, the mechanical and structure properties of the whole sample would be progressively ruined. Remind that the network is beneath the surface layer and only a few reaction sites are visible without polishing and sputtering, so actually what it really happens would be worse than what it shows from the surface. 4. Conclusion Surface passivation layers on uranium are removed by Arþ sputtering and the oxidizing and hydriding behaviors of sputtered surface are investigated. The surface layer modified by mechanical polishing is removed by sputtering. The kinetics of initial oxidation on sputtered surface is parabolic indicates a diffusion controlled mechanism. The bombardment of Arþ introduces abundant superficial defects, which act as enhanced oxygen diffusion paths, thus the parabolic oxidizing scheme exhibits a transition in rate constant at the oxide thickness exceeds 15 nm. In hydriding kinetics of sputtered sample, the diffusion and accumulation is also enhanced while the initial oxide thickness is less than the thickness of defective layer, hence no evident induction period observed; when the oxidation is deeper than the defective layer, the diffusion of hydrogen is impeded by the integrated oxide lattice thus the induction period appears again and prolongs with the increase of oxide thickness. As compared in morphology, the oxide layer not only act as the diffusion barrier to hydrogen, but also inhibits the growth of underlying hydride. By polishing and sputtering it becomes visible that hydrogen attack mainly occurs in a few grains densely and the hydride sites are connected with a network of fractures, which provides pathways for hydride propagation. Acknowledgements The authors acknowledge Yongqiang Zhong, Xiandong Meng, Chao Lu, and Dongli Zou for their assistance in experimental measurement, and Xiaofang Wang for her valuable advice. This work is supported by Foundations for Development of Science and Technology of China Academy of Engineering Physics (2015B0307069), National Natural Science Foundation of China (11404295) and Major Scientific Equipment Development Project of China (2012YQ130125).
Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jnucmat.2017.07.017. References [1] J. Glascott, A model for the initiation of reaction sites during the uraniumehydrogen reaction assuming enhanced hydrogen transport through thin areas of surface oxide, Philos. Mag. 94 (3) (2014) 221e241. [2] C.P. Jones, J.R. Petherbridge, S.A. Davis, J.A. Jones, T.B. Scott, The crystallographic structure of the air-grown oxide on depleted uranium metal, Corros. Sci. 111 (2016) 486e493. [3] J. Bloch, M.H. Mintz, Kinetics and mechanisms of metal hydrides formationda review, J. Alloys Compd. 253e254 (6) (1997) 529e541. [4] C.P. Jones, T.B. Scott, J.R. Petherbridge, J. Glascott, A surface science study of the initial stages of hydrogen corrosion on uranium metal and the role played by grain microstructure, Solid State Ionics 231 (231) (2013) 81e86. [5] C.A. Stitt, C. Paraskevoulakos, N.J. Harker, C.P. Jones, T.B. Scott, The effects of metal surface geometry on the formation of uranium hydride, Corros. Sci. 98 (2015) 63e71. [6] A. Danon, J.E. Koresh, M.H. Mintz, Temperature programmed desorption characterization of oxidized uranium Surfaces: relation to some gasuranium reactions, Langmuir 15 (18) (1999) 5913e5920. [7] J. Bloch, M.H. Mintz, The effect of thermal annealing on the hydriding kinetics of uranium, J. Less Common Met. 166 (2) (1990) 241e251. [8] M.A. Hill, R.K. Schulze, J.F. Bingert, R.D. Field, R.J. Mccabe, P.A. Papin, Filiformmode hydride corrosion of uranium surfaces, J. Nucl. Mater. 442 (1e3) (2013) 106e115. [9] N.J. Harker, T.B. Scott, C.P. Jones, J.R. Petherbridge, J. Glascott, Altering the hydriding behaviour of uranium metal by induced oxide penetration around carbo-nitride inclusions, Solid State Ionics 241 (21) (2013) 46e52. [10] H. Zhang, X. Lv, D. Ren, R. Li, X. Lai, T. Zhao, Influence of Arþ ion beam sputter on surface roughness of uranium film, Rare Metal Mater. Eng. 39 (5) (2010) 889e891. [11] P. Karmakar, G.F. Liu, J.A. Yarmoff, Sputtering-induced vacancy cluster formation on TiO2(110), Phys. Rev. B 76 (19) (2007) 3410. [12] N.Q. Lam, G.K. Leaf, H. Wiedersich, Sputter-induced surface composition changes in alloys, J. Nucl. Mater. 88 (2e3) (1980) 289e298. [13] A. Loui, The Hydrogen Corrosion of Uranium: Identification of Underlying Causes and Proposed Mitigation Strategies, LLNL Technical Report, 2012. [14] C.D. Taylor, T. Lookman, R.S. Lillard, Ab initio calculations of the uraniumehydrogen system: thermodynamics, hydrogen saturation of a-U and phase-transformation to UH3, Acta Mater. 58 (3) (2010) 1045e1055. [15] D.F. Teter, R.J. Hanrahan, C.J. Wetteland, Uranium Hydride Nucleation Kinetics: Effects of Oxide Thickness and Vacuum Outgassing, LANL Technical Report, 2001. [16] R.M. Harker, The influence of oxide thickness on the early stages of the massive uraniumehydrogen reaction, J. Alloys Compd. 426 (1) (2006) 106e117. [17] C.A. Colmenares, Oxidation mechanisms and catalytic properties of the actinides, Prog. Solid St. Chem. 15 (48) (1985) 257e364. [18] Z. Chernia, Reflectance spectroscopy in analysis of UO2 scale: derivation of a kinetic model of uranium oxidation, Phys. Chem. Chem. Phys. 11 (11) (2009) 1729e1739. [19] S. Lin, X. Lai, X. Lv, H. Zhang, Study of the initial oxidation characteristics of uranium with pure oxygen below 100 C by spectroscopic ellipsometry, Surf. Interface Anal. 40 (3e4) (2008) 645e648. [20] S.G. Bazley, J.R. Petherbridge, J. Glascott, The influence of hydrogen pressure and reaction temperature on the initiation of uranium hydride sites, Solid State Ionics 211 (211) (2012) 1e4. [21] V.J. Wheeler, The diffusion and solubility of hydrogen in uranium dioxide single crystals, J. Nucl. Mater. 40 (40) (1971) 189e194. [22] L.W. Owen, R.A. Scudamore, A microscope study of the initiation of the hydrogen-uranium reaction, Corros. Sci. 6 (11e12) (1966) 461e468. [23] R. Arkush, A. Venkert, M. Aizenshtein, S. Zalkind, D. Moreno, M. Brill, M.H. Mintz, N. Shamir, Site related nucleation and growth of hydrides on uranium surfaces, J. Alloys Compd. 244 (1e2) (1996) 197e205.
Contents lists available at ScienceDirect
Journal of Nuclear Materials journal homepage: www.elsevier.com/locate/jnucmat
The influence of Arþ sputtering on the hydriding behavior of uranium Ren Bin a, Guangfeng Zhang a, Hefei Ji b, Lizhu Luo a, Peng Shi b, Xiaolin Wang c, * a
Science and Technology on Surface Physics and Chemistry Laboratory, PO Box 9-35, Jiangyou 621908, Sichuan, China Institute of Materials, China Academy of Engineering Physics, PO Box 9-13, Mianyang 621900, Sichuan, China c China Academy of Engineering Physics, Mianyang 621900, Sichuan, China b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 22 December 2016 Received in revised form 7 July 2017 Accepted 7 July 2017 Available online 8 July 2017
The influence of superficial defects (induced by surface sputtering) on uranium oxidation and hydriding behaviors has been studied. Depleted uranium surface was etched by Arþ sputtering, the etched surface morphology was investigated by scanning electron microscope. The initial oxidation kinetics of sputtered uranium surface and mechanical polished uranium surface in ambient atmosphere were characterized using spectroscopic ellipsometry. As-polished and sputtered samples, as well as sequence sputtered samples ambient oxidized for different time, were compared in hydriding kinetics to investigate the coinfluence of surface oxidation and superficial defects induced by Arþ ion bombardment. The morphologic characteristics of hydride sites in a few grains were also revealed and enhanced by sputtering, showing a network of fractures for fast hydride transportation. © 2017 Published by Elsevier B.V.
Keywords: Depleted uranium Hydrogen corrosion Induction time Arþ sputtering
1. Introduction Uranium is an important material in military and energy industries. However, the storage and usage of uranium or uranium alloys components are at risk due to the high reactivity between uranium and ambient oxygen, water and hydrogen. Hydrogen corrosion is one of the most serious corrosion problems for damaging the integrity or decaying the mechanic properties of uranium. Due to the remarkable density mismatch between U (19.05 g/cm3) and UH3 (10.92 g/cm3), when constrained by metal the growth of hydride precipitates would cause a 75% expansion in lattice, force the surrounding metal to plastic deform or even fracture, and inevitably cause damage or failure to components made of uranium or uranium alloys. What's worse, the pyrophoric uranium hydride is dangerous if exposed to ambient and also threatening the environment. Therefore, research on uraniumhydrogen (U-H) reaction is a persisting focus since Manhattan Project in 1940s, covering many aspects like reaction kinetics, metallography and surface chemical analysis, etc. Fruitful achievements have provided us a lot of fundamental understandings and many methods are developed to reduce the surface hydrogen corrosion. However, the initial hydriding kinetics of U-H reaction,
* Corresponding author. E-mail address: [email protected] (X. Wang). http://dx.doi.org/10.1016/j.jnucmat.2017.07.017 0022-3115/© 2017 Published by Elsevier B.V.
especially for the key factors of hydride nucleation, is still ambiguous. One critical problem is the discrepancies in nucleation preference with just a little difference in experiment setup, or even the same setup would yield very different results. A popular idea is that clean (oxide free) uranium surface would react with pure hydrogen rapidly and entirely in an ideal assumption, the localization of hydride nucleation is attributed to the discontinuity in the semiprotective surface oxide film. Glascott et al. have proposed a model for the initial reaction assuming an enhanced transportation through thin areas on surface oxide [1], and their recent report has confirmed that the abundant grain boundaries in nm-sized polycrystalline air grown oxide contribute to the high diffusion coefficient in gaseous diffusion model [2]. Furtherly, Bloch et al. [3e5] proposed that not only surface oxide, but also the layer beneath the oxide which is saturated of water, hydroxyl clusters or organic absorbents, as well as strain defects and residual stress induced by mechanical processing, have critical influence in the induction time and localization for hydriding nucleation. The combination of surface oxide layer and underlying mechanically modified layer is called surface passivation layer (SPL). The characteristics of SPL changes with different mechanical processing methods and parameters, may consequently causes the inconsistencies in the hydriding behavior reported in different studies. To clarify the role of SPL playing in hydriding kinetics and nucleation behavior, preparing a surface without SPL, or making it uniform and known, is essential to reduce the inconsistencies and
56
R. Bin et al. / Journal of Nuclear Materials 494 (2017) 55e60
yield better understandings. Several pretreatments had been adopted to achieve this aim. The most common one is vacuum heat-treatment prior to hydriding experiment [6,7]. This pretreatment shows an ‘activation’ to the uranium surface, namely shortening the induction period and accelerating the nucleation, by removing chemisorbed water or hydroxyl group and release active sites for hydrogen's accumulation. This pretreatment is commonly adopted by researchers before conducting U-H reaction, and the consistency is generally better. However, with this method the oxide on uranium surface was not removed or even thickened by reaction of uranium and absorbed water or residual oxygen during heat treatment, the uniformity is also not improved. Another method is electropolishing [8]. Surface oxide and strained layer induced by mechanical polishing is electrochemically etched and the surface becomes uniform (not considering the inclusions) and covered by a protective oxide layer. However, the electropolishing process is usually conducted in aqueous solution so the absorbed water or hydroxyl are not negligible, and the newly formed oxide layer will conceal some characteristics which are important in U-H reaction, for example an intermediate layer of oxide along carbidemetal interface is formed under the driven of applied electric field, thus the preferred nucleation sites around inclusions would be blocked [9]. Despite of previous efforts, a more appealing approach is highly desired for making a uniform surface to deeply understand its reaction behavior. The technique of Arþ ion sputtering is used customarily for the processing and analysis of solid surfaces. The bombardment of Arþ ions with kinetic energy ‘physically’ removes the surface atoms layer by layer. Zhang et al. reported that the Arþ ion beam sputtering has the fine polishing effect on uranium metal surface [10]. Ion beams could also reproducibly generate atomic level defects and nanometer-scale vacancy clusters in a controlled manner [11], and induces changes in the composition and chemical state [12]. Compared with heat treatment and electropolishing, Arþ sputtering can almost avoid the adsorbed water, hydroxyl or other impurities. For the reasons mentioned above, it would be attractive to apply this approach to modify the metallic uranium surface. In this paper, uranium surface was modified via Arþ ion sputtering technique. The effect of sputtering on the morphology of uranium surface was investigated, as well as the oxidation and hydriding behavior, comparing with the samples just mechanical polished and vacuum heat treated. The difference between sputtered and mechanical polished samples in oxidation and hydriding kinetics could be referred as the effect of surface layer modified by mechanical processing, and the role of metal substrate plays in hydride nucleation was also partially revealed. 2. Experimental 2.1. Sample preparation Depleted uranium (DU) coupons (4 5 mm 1.5 mm, with impurities carbon <100 wppm, nitrogen <50 wppm, calcium 70 wppm, tungsten 90 wppm, others < 120 wppm) were cut from an ingot prepared by arc melting. Each sample was polished, rinsed with distilled water and alcohol, and then ultrasonic rinsed in acetone for 5 min. The sputtering was conducted with a high-flux ion beam sputtering system, in which samples were etched with Arþ ion beam of 600 eV, 24 mA for 30 min. The base vacuum and working argon pressure were 1 104 Pa and 2.2 102 Pa, respectively. 2.2. Oxidation Following preparation, the oxidation of sputter DU was
conducted in ambient air at a temperature of 25 C and RH 40%e 60%. The refraction spectra on oxidized surface were measured with spectroscopic ellipsometer (model Woollam M2000). Oxide thickness is obtained by fitting each spectrum with Lorentz model, and the value derived from first spectrum was subtracted as an assumption that the initial thickness at the first measurement is zero. Actually, there is an intrinsic oxide layer because the samples had been exposed to air for about 30 min before the spectrometer was calibrated and the tests began. Sequence of thickness data was obtained periodically and in situ. Note that the beam spot size is in mm scale thus the oxide thickness measured is a statistical average of many grains. 2.3. Hydriding Reactions of hydrogen and DU specimen polished or Arþ sputtered were carried out in a visual reaction chamber. A vacuum heat pretreatment was conducted by gradually heating to 160 C (about 5 C/min), kept for 1 h and then naturally cooled down to 70 C. Before and during heat-treatment the chamber was evacuated with turbo-pump to make sure the pressure was below 0.1 Pa. After the temperature stabilized, high purity hydrogen (99.999% and purified with a LaNi5 bed) was introduced to reaction chamber, making the initial pressure around 1 105 Pa and began the reaction. The pressure during reaction was closely observed and recorded, by about 2% dropping of the initial pressure, or the macroscopic nuclei appeared on sample surface, the chamber was immediately evacuated again and the reaction was halted. The microscopic morphologies before and after U-H reaction were observed with scanning electron microscope (model Helios Nanolab 600i). After hydriding experiment uranium surface was covered by a thick oxide layer due to the long exposure in ambient. To reveal the distribution of hydride nucleation sites and the morphology of hydride underlying the oxide layer, sample surfaces were gently polished, and etched to remove the oxide layer formed due to the exposure after reaction and clean the surface. 3. Results and discussion 3.1. Surface morphology of sputtered DU It has been reported that due to the extra low solubility of hydrogen in UO2 lattice, integrated UO2 layer can act as a good physical barrier for hydrogen diffusion theoretically. However, in real conditions, hydrogen always penetrates the oxide layer through its defects including dislocations, impurity atoms and vacancy clusters, formed intrinsically or extrinsically, and their assemblies like grain boundaries and inclusion-substrate interfaces. In addition, the solubility of hydrogen in uranium metal are also very low and far below the critical concentration to form UH3 [13], it's not likely to nucleate in a defect-free metal lattice [14]. The preferred nucleation sites are located where a lot of defects exist thus capable to accumulate supersaturate H atoms. Hence, it's essential to understand the relationship between preferred hydride nucleation sites and surface layer, which for several microns in depth is modified by surface processing while preparing. To avoid the influence of surface processing, the surface layer was removed completely by sputtering for sufficient long time, and keep the process in vacuum to avoid the ambient agents. Fig. 1 shows the micrographs of DU surface with or without etching. The surface only mechanical polished shows a flat and homogenous surface except for some inclusions and scratches, lacking microstructure details like grain boundaries (Fig. 1a). After etching for 30 min the difference in height for atomic planes of grains with different orientations are clearly observed, showing a
R. Bin et al. / Journal of Nuclear Materials 494 (2017) 55e60
57
Fig. 1. Micrographs of DU surface with different sputtering time. (a) As-polished; (b) 30 min sputtering. The square inclusions shown in are uranium nitride (see in supplementary material).
preferential sputtering by difference in sputter rate on grains with different orientations (Fig. 1b). The surface micrographs contain more details than on as-polished surface, similar to the effect of electrochemical polishing. The roughness is in nanoscale and varies on different grains. Inclusions on the surface are mainly nitride and oxide examined by EDS, they have changed little in height and roughness after being sputtered, reveals that the sputter rate on carbide and nitride are much lower than on metal uranium due to a higher hardness. 3.2. Oxidation In U-H reaction, oxide layer on uranium surface is unavoidable and always playing an important role [15,16]. It's necessary for us to investigate the oxidation behavior of sputtered uranium before conducting hydriding reaction, on this purpose the spectroscopic ellipsometry was applied. According to literature [17e19], in the early stage of uranium oxidization the diffusion of oxygen in UO2 lattice is the rate determining step at moderate temperature and moisture. The relationship between the oxide thickness s and the exposure time t is parabolic
s ¼ a þ kt 1=2
(1)
where k is the rate constant. The intercept a is none-zero due to three reasons: the error of fitting, the kinetics of very early stage of oxidation does not obey the time square-root law or the recorded t is not exactly the time oxidation begins. There is about 30 min prior to the beginning of the spectroscopic ellipsometry experiments, thus an initial oxidation on uranium surface is already formed due to the high reactivity of uranium and oxygen. The growth of oxide thicknesses on sputtered surface, as well as that of as-polished, is shown in Fig. 2. The fluctuation of these two curves is due to the drift in environmental temperature and moisture. As shown in Fig. 2, both curves can be split into two segments and linear fitted separately. On as-polished surface, the parabolic rate constant k is 3.66 nm h1/2 at the initial stage, and for the stage that the oxidation is deep than 15 nm, k is 2.20 nm h1/2. As for sputtered surface, the rate constant k for the initial stage and deep oxidation stage is 6.43 nm h1/2 and 1.93 nm h1/2, respectively. Given that the rate constant k is related to the diffusion barrier at a certain temperature, the fitting result indicates that the activation energy for oxygen diffusing in the oxide layer within 15 nm is much lower than the oxide layer deeper. On sputtered surface k is
Fig. 2. The oxidation on as-polished and sputtered uranium surfaces. Both curves are divided into two regions and fitted following equation (1), the y-axis represents the thickness of surface oxide, and the x-axis are square root of time to reveal the linear relationship between s and t1/2.
markedly higher than on the as-polished surface, which implies that a significant decrease in activation energy for uranium oxidation is caused by sputtering. The difference between the surface layer of as-polished and sputtered surface with oxide thickness less than 15 nm, is mainly density of defects. In close packed metal, especially for uranium with a large atomic radius, migrating through interstitials and vacancies is the dominating diffusion mechanism. In this mechanism, the density of vacancies is critical for the diffusion barrier. On sputtered surface the defects are much denser in a very thin (15 nm) surface layer due to heavy dose Arþ bombardment. In this layer, the saturated solubility, as well as chemical potential of oxygen atoms are higher than in the surface layer of as-polished sample, thus it shows a fast oxidation in the initial stage. Yet on as-polished surface the density of defects is also slightly higher than in intrinsic bulk due to the damage of mechanic processing, hence the oxidation rate is also higher than the inner layer. In the depth of sputtered or as-polished sample, the density of vacancies is very close to intrinsic level and the diffusion barrier is higher than sputtered and mechanical modified layer.
58
R. Bin et al. / Journal of Nuclear Materials 494 (2017) 55e60
Fig. 3. The pressure drop of hydrogen in reaction with sputtered and polished surface. The initial pressures are normalized to 1.0 105 Pa.
3.3. Hydriding Both as-polished and sputtered samples are ambient exposed for less than 10 min before they are put into reaction chamber and evacuated, hence the oxide layer developed on sample surface is relatively thin. The hydrogen consumption curves for both aspolished and Arþ sputtered specimen is illustrated in Fig. 3, which difference reveals the influence of surface defective layer. On as-polished sample, the reaction is delayed and the hydrogen pressure does not drop immediately after the initial hydrogen exposure, this delay on pressure-time curve is referred as the induction period. On sputtered uranium surface the induction period is negligible, and the consumption of hydrogen begins immediately after the initial exposure of hydrogen. Induction period are usually regarded as the time consumed for hydrogen to penetrate surface oxide layer, reach the metal-oxide interface and then exceed the critical concentration for nucleation. It is commonly accepted that
induction period is mainly due to the diffusion of hydrogen in ‘adhesive’ oxide layer on uranium surface, supported by the similarity in activation energy of hydriding reaction by Arrhenius fitting of the nucleation rate constants [20] and the diffusion barrier of hydrogen in single crystal uranium dioxide [21]. The absence of induction period is mainly due to the removal of residual surface oxide and introduction of abundant superficial defects. According to the case of oxidation discussed in section 3.2, for a sputtered sample with a short period exposure to air, the defects induced by the bombardment of Arþ ions would distribute in both the surface oxide layer and the top of underlying metal. These defects act as enhanced diffusion paths and hydrogen accumulation sites. The time consumed for hydrogen to penetrate the surface oxide layer and exceeds the critical concentration are greatly truncated, comparing to a regular oxidized surface. After induction period hydrogen pressure are dropped rapidly for both samples. The reaction rate for sputtered sample is slightly higher than as-polished one, probably this is also due to the superficial defects. The hydriding induction periods and consumption rates for sputtered sample, which are of different oxidation states before UH reaction, are listed in Table 1. Here the induction period is defined as the period for first 0.1 kPa drop of hydrogen pressure while the hydrogen consumption rate is the slope of the linear region in hydrogen consumption curve. The prolonging of induction period with increasing oxide thickness is observed for sputtered samples with longer ambient exposure, meanwhile the hydrogen consumption rates in linear region are progressively slowdown. For the sputtered samples exposed to ambient for 0.5 h and 2 h, there is no evident induction period observed because of the enhanced hydrogen diffusion and accumulation in defective layer, as discussed above. Stored in ambient air for 12 h or longer, the induction period appeared again, indicates that the defective layer is completely oxidized and the enhancement by defects vanishes. The relationship between surface defective layer induced by Arþ bombardment, surface oxide layer and induction time for U-H reaction is shown in Fig. 4 schematically. With short exposure (Fig. 4a), the surface layer with abundant defects is not completely oxidized and still provides the benefits for hydrogen diffusion and
Table 1 Induction time and hydrogen consumption rates on sputtered uranium surface with different oxidizing time. Exposure time (h)
Oxide thicknessa (nm)
Induction period (s)
Hydrogen consumption rate (Pa/s)
0.5 2 12 24 42 118
3.0 7.9 16.4 19.9 20.5 31.6b
51 36 245 290 405 461
8.85 6.85 6.56 5.66 4.82 4.00
a b
The oxide thicknesses are read from spectroscopic ellipsometry data. Oxide thickness for 118 h exposed surface is evaluated by parabolic fitting and extrapolating of spectroscopic ellipsometry data.
Fig. 4. Scheme of hydride nucleation on sputtered surface on situation that (a) the oxide layer is shallower than the defective layer, or (b) the oxidation exceeds the defective layer.
R. Bin et al. / Journal of Nuclear Materials 494 (2017) 55e60
accumulation, thus the induction time is eliminated or greatly shortened. As the oxidation exceeds the defective region (Fig. 4b), the integrated oxide layer formed underlying the defective layer act as a barrier for hydrogen diffusion, and its thickness would affect how soon hydride nucleation at oxide-metal interface would begin. In addition, the compression induced by surface oxide layer would also suppress the expansion of underlying nuclei, reduce the growth rate of hydride too, that is another reason for the slowdown with increasing oxide thickness. 3.4. Hydride morphology According to Owen et al. [3,22], the hydride nuclei can be classified into two families (or four types according to Arkush et al. [23]): One is high in nucleation density and low in growth rate blister-like nucleus with size limited to several microns. Another is much lower in density but growing much faster with an unlimited final size. In present work, the latter type of nucleus is focus on. Fig. 5 shows the microscopy of hydride nuclei on as-polished and sputtered sample surface. As shown in Fig. 5a, the hydride nucleation on polishing modified surface occurs densely and the hydrides expand laterally beneath surface layer, their growth are suppressed by overlying surface layer. As hydride nuclei grow, the stress induced by hydride expansion causes the crack of surface
59
layer, exposing a porous hydride surface and then the reaction is enhanced by the direct contact of hydrogen and bulk uranium. On sputtered surface the compression is less intense due to the integrity of surface layer is damaged by superficial defects, so the surface cracks easily and the hydrogen attack exhibits a ‘explosive’ look (Fig. 5b). It's natural that strong compression of thick, integrated and adhesive oxide layer has suppressed the lateral growth and expansion of uranium hydride, thus the hydride is more likely to develop inward. When the suppression is strong enough, if there is no possibility to accommodate the hydride newly formed, or the substrate is hard to deform and release the stress induced by hydride expansion, the hydride growth would be inhibited before grown up. Conversely, if there are ways to accommodate hydride growth like grain boundaries or interfaces between inclusions and substrate, or release of stress induced by hydride expansion (structure deformation like twinning and fracturing), hydride nuclei growth would persist and the final size is much larger than in the former case. From this point, the classification of hydride nuclei evaluating with the difference of size and growing behavior is in correspondence with not only the integration of surface oxide layer but also the properties of metal substrate. Fig. 6 shows the hydride sites on an as-polished sample, with the surface slightly polished and etched again after hydriding experiment. The sputtering process is very helpful not only to
Fig. 5. Surface morphologies of hydride corrosion sites on (a) as-polished surface and (b) sputtered surface.
Fig. 6. The preferential hydriding on as-polished surface. (a) The network of fractures; (b) details of the fractures that connect the hydride sites. Remind that (a) and (b) are polished and sputtered again after hydrogen attack.
60
R. Bin et al. / Journal of Nuclear Materials 494 (2017) 55e60
enhance the detailed surface structures like grain boundaries and fracture edges, but also revealed the trace of hydride propagation in the metal bulk. The hydride would oxidize immediately after exposure to ambient. Compares to uranium substrate, uranium oxide is higher in hardness hence the sputtering rate is relatively lower and the oxide would protrude from the sputtered surface. In Fig. 6 the oxide clusters exist in the cracks around hydride craters, which implies that the cracking occurred during U-H reaction, induced by expansion of hydride, and have formed a network among individual hydride sites, provide ‘highways’ for hydrogen or hydride to transport from one site to another. It occurs only in a few grains, as we observed the hydriding sites are sparse and inhomogeneous. However, for these specific grains the corrosion is serious and the integrity is almost destroyed. If the attack persists, the mechanical and structure properties of the whole sample would be progressively ruined. Remind that the network is beneath the surface layer and only a few reaction sites are visible without polishing and sputtering, so actually what it really happens would be worse than what it shows from the surface. 4. Conclusion Surface passivation layers on uranium are removed by Arþ sputtering and the oxidizing and hydriding behaviors of sputtered surface are investigated. The surface layer modified by mechanical polishing is removed by sputtering. The kinetics of initial oxidation on sputtered surface is parabolic indicates a diffusion controlled mechanism. The bombardment of Arþ introduces abundant superficial defects, which act as enhanced oxygen diffusion paths, thus the parabolic oxidizing scheme exhibits a transition in rate constant at the oxide thickness exceeds 15 nm. In hydriding kinetics of sputtered sample, the diffusion and accumulation is also enhanced while the initial oxide thickness is less than the thickness of defective layer, hence no evident induction period observed; when the oxidation is deeper than the defective layer, the diffusion of hydrogen is impeded by the integrated oxide lattice thus the induction period appears again and prolongs with the increase of oxide thickness. As compared in morphology, the oxide layer not only act as the diffusion barrier to hydrogen, but also inhibits the growth of underlying hydride. By polishing and sputtering it becomes visible that hydrogen attack mainly occurs in a few grains densely and the hydride sites are connected with a network of fractures, which provides pathways for hydride propagation. Acknowledgements The authors acknowledge Yongqiang Zhong, Xiandong Meng, Chao Lu, and Dongli Zou for their assistance in experimental measurement, and Xiaofang Wang for her valuable advice. This work is supported by Foundations for Development of Science and Technology of China Academy of Engineering Physics (2015B0307069), National Natural Science Foundation of China (11404295) and Major Scientific Equipment Development Project of China (2012YQ130125).
Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jnucmat.2017.07.017. References [1] J. Glascott, A model for the initiation of reaction sites during the uraniumehydrogen reaction assuming enhanced hydrogen transport through thin areas of surface oxide, Philos. Mag. 94 (3) (2014) 221e241. [2] C.P. Jones, J.R. Petherbridge, S.A. Davis, J.A. Jones, T.B. Scott, The crystallographic structure of the air-grown oxide on depleted uranium metal, Corros. Sci. 111 (2016) 486e493. [3] J. Bloch, M.H. Mintz, Kinetics and mechanisms of metal hydrides formationda review, J. Alloys Compd. 253e254 (6) (1997) 529e541. [4] C.P. Jones, T.B. Scott, J.R. Petherbridge, J. Glascott, A surface science study of the initial stages of hydrogen corrosion on uranium metal and the role played by grain microstructure, Solid State Ionics 231 (231) (2013) 81e86. [5] C.A. Stitt, C. Paraskevoulakos, N.J. Harker, C.P. Jones, T.B. Scott, The effects of metal surface geometry on the formation of uranium hydride, Corros. Sci. 98 (2015) 63e71. [6] A. Danon, J.E. Koresh, M.H. Mintz, Temperature programmed desorption characterization of oxidized uranium Surfaces: relation to some gasuranium reactions, Langmuir 15 (18) (1999) 5913e5920. [7] J. Bloch, M.H. Mintz, The effect of thermal annealing on the hydriding kinetics of uranium, J. Less Common Met. 166 (2) (1990) 241e251. [8] M.A. Hill, R.K. Schulze, J.F. Bingert, R.D. Field, R.J. Mccabe, P.A. Papin, Filiformmode hydride corrosion of uranium surfaces, J. Nucl. Mater. 442 (1e3) (2013) 106e115. [9] N.J. Harker, T.B. Scott, C.P. Jones, J.R. Petherbridge, J. Glascott, Altering the hydriding behaviour of uranium metal by induced oxide penetration around carbo-nitride inclusions, Solid State Ionics 241 (21) (2013) 46e52. [10] H. Zhang, X. Lv, D. Ren, R. Li, X. Lai, T. Zhao, Influence of Arþ ion beam sputter on surface roughness of uranium film, Rare Metal Mater. Eng. 39 (5) (2010) 889e891. [11] P. Karmakar, G.F. Liu, J.A. Yarmoff, Sputtering-induced vacancy cluster formation on TiO2(110), Phys. Rev. B 76 (19) (2007) 3410. [12] N.Q. Lam, G.K. Leaf, H. Wiedersich, Sputter-induced surface composition changes in alloys, J. Nucl. Mater. 88 (2e3) (1980) 289e298. [13] A. Loui, The Hydrogen Corrosion of Uranium: Identification of Underlying Causes and Proposed Mitigation Strategies, LLNL Technical Report, 2012. [14] C.D. Taylor, T. Lookman, R.S. Lillard, Ab initio calculations of the uraniumehydrogen system: thermodynamics, hydrogen saturation of a-U and phase-transformation to UH3, Acta Mater. 58 (3) (2010) 1045e1055. [15] D.F. Teter, R.J. Hanrahan, C.J. Wetteland, Uranium Hydride Nucleation Kinetics: Effects of Oxide Thickness and Vacuum Outgassing, LANL Technical Report, 2001. [16] R.M. Harker, The influence of oxide thickness on the early stages of the massive uraniumehydrogen reaction, J. Alloys Compd. 426 (1) (2006) 106e117. [17] C.A. Colmenares, Oxidation mechanisms and catalytic properties of the actinides, Prog. Solid St. Chem. 15 (48) (1985) 257e364. [18] Z. Chernia, Reflectance spectroscopy in analysis of UO2 scale: derivation of a kinetic model of uranium oxidation, Phys. Chem. Chem. Phys. 11 (11) (2009) 1729e1739. [19] S. Lin, X. Lai, X. Lv, H. Zhang, Study of the initial oxidation characteristics of uranium with pure oxygen below 100 C by spectroscopic ellipsometry, Surf. Interface Anal. 40 (3e4) (2008) 645e648. [20] S.G. Bazley, J.R. Petherbridge, J. Glascott, The influence of hydrogen pressure and reaction temperature on the initiation of uranium hydride sites, Solid State Ionics 211 (211) (2012) 1e4. [21] V.J. Wheeler, The diffusion and solubility of hydrogen in uranium dioxide single crystals, J. Nucl. Mater. 40 (40) (1971) 189e194. [22] L.W. Owen, R.A. Scudamore, A microscope study of the initiation of the hydrogen-uranium reaction, Corros. Sci. 6 (11e12) (1966) 461e468. [23] R. Arkush, A. Venkert, M. Aizenshtein, S. Zalkind, D. Moreno, M. Brill, M.H. Mintz, N. Shamir, Site related nucleation and growth of hydrides on uranium surfaces, J. Alloys Compd. 244 (1e2) (1996) 197e205.