Tritium has many important technological applications. It is common to combine tritium with a metal to form metal tritides for stable, high-density storage of tritium. ³He is produced by the radioactive decay of tritium. It is well known that at least initially, most of the ³He remains in the solid, and only less than 1% of the generated ³He is released from metal tritides.^[1,2] Snow et al. showed that the total amount of helium released early in life was independent of the film thickness, which suggested that the helium released early in life came from the near-surface region.^[3] When the helium concentration in metal tritides reached a critical value, helium was released at rates equal to or sometimes exceeding the generation rate. The onset of critical release was found to vary for different metals, and it occurred around 0.3 (He:metal ratio) for erbium tritides and titanium tritides, and around 0.40 for zirconium tritides.^[2] Helium accumulation in metal tritides would cause considerable changes in the structure and properties of these materials, such as volume expansion and the decrease of the plateau pressure.

The recoil energy of ³He created by tritium decay was estimated to be ∼ 1.03 eV, which was too low to make it leave the tetrahedral site in which it was created.^[4] It was found that helium atoms could become weakly bound to either lattice defects or other helium atoms they encountered as they migrated.^[5,6] It has also been proved that the interstitial helium can strongly bind to the grain boundary.^[7] Helium bubbles in aging metal tritides were commonly seen in the lattice and grain boundary by the transmission electronic microscopy (TEM).^[8–12] Helium bubble evolution in metal tritides is a complicated process, the study of which requires modeling at many different levels including atomic properties of ³He atoms in metal lattices, diffusion mechanisms, kinetics of bubble nucleation and growth, and the relationship between the microstructural evolution and the change in mechanical properties.^[13,14]

The microstructure of the film was believed to have a strong effect on the helium migration and has been studied extensively.^[3,15–18] The structure of films could involve a number of characteristics, such as grain size, strain, crystallographic orientation, and so on. It was shown that the film structure was strongly affected by the preparation procedures and deposition condition, such as substrate temperature,^[19,20] deposition rate,^[21] substrate materials,^[22] impurity contamination, and film thickness.^[23–25]

The purpose of this paper is to study ³He migration in TiT_1.9 films. We prepared two kinds of Ti tritide films with different grain morphologies, and presented the results of the helium release measurement and TEM analysis of these two kinds of Ti tritide films during aging. The decisive factors in microstructure and the roles of these factors on the helium bubble distribution and helium release were clarified.

A large number of measurements have been made on helium emission from these two kinds of Ti tritide films in the past.^[2] Figure 1 shows the normalized ³He release rate fraction (release rate/generation rate) from these two kinds of films as a function of the generated ³He concentration. Helium release can be divided into two regions: an “early release” period characterized by a constant and small release fraction, and a “critical release” period where the helium is released at rates equal to and sometimes exceeding the generation rate.^[1] The onset of the critical release occurs below 0.28 (He:Ti ratio) for sample C and around 0.3 or more for sample E. The early release fraction of Ti tritides is similar to that of Zr tritides, but lower than Er tritides by more than one order of magnitude.^[2]

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. (color online) Normalized He release rate from Ti tritide samples as a function of the generated He concentration.

The crystal structures of these two kinds of films were studied by x-ray diffraction (XRD), and the results are shown in Fig. 2. The peak broadening and overlap with the peaks of the Mo substrate make it difficult to find the differences. There are both (110) out of plane and (200) out of plane rolling texture in our Mo substrate. Because the {110} and {200} spacing for Mo is close to the {200} and {220} spacing for TiH_1.9, respectively, these TiT_1.9 films show both (200) out of plane and (220) out of plane textures.^[26] More studies will be carried out in the future on the pole figure for these two kinds of films.

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. (color online) Diffraction peaks of the two kinds of films. He:Ti ratios of sample C and E are 0.299 and 0.356, respectively.

Figures 3 and 4 show the grain structures of the two kinds of films before and after tritiation. Obviously the process of hydriding does not change the microstructure of the two kinds of films. The cross-section of sample C shows a multi-layered columnar grain structure. The upper layer is obviously made up of stacked flaky crystals perpendicular to the surface. Savaloni et al. studied the growth of erbium films on Mo substrates with different roughness, different substrate temperatures, different deposition rates, and under three different ultra-high vacuum (UHV) conditions, namely standard, semi-rigorous, and rigorous.^[27] They observed that for substrate temperatures up to 775 K, all the films produced under the rigorous UHV condition, in contrast to those produced under the standard UHV condition, were free from growth steps and layering. The growth steps and the corresponding layered cross-section were the results of film oxidation, which was due to the poor vacuum condition. The decreased mobility of contaminated grain boundaries limited further grain growth.

Fig. 3.

	Figure Option ViewDownloadNew Window
	Fig. 3. Cross-sections of the two kinds of films before tritiation. (a) Columnar grain structure of sample C. (b) Equiaxed grain structure of sample E.

Fig. 4.

Figure Option ViewDownloadNew Window

Fig. 4. Grain structure and 3He-filled cracks in the two kinds of films at the beginning of accelerated release. (a) Columnar grain structure of sample C at He:Ti ratio of 0.299. (b)–(c) Equiaxed grain structure of sample E at He:Ti ratio of 0.356. The circled areas in (b) and (c) show the hidden right part of the crack which is growing along one of the {111} planes of the grain below as indicated by the corresponding selected-area diffraction pattern (close to [110] zone axis).

The grain in sample E is a typical equiaxed three-dimensional structure. The film of sample E was deposited at a high deposition rate by resistive heating, and the grains grew faster, thereby favoring grain coarsening and restructuration, as well as less oxygen contamination.

Figure 4 shows that cracks are developed along the grain boundary of both samples at the beginning of the helium accelerated release stage. Most of these cracks are nearly parallel to the film surface. Almost all these cracks occur in regions of grain boundary where the boundary is approximately parallel to one of the equivalent sets of {111} planes in at least one of the adjacent grains. This can be confirmed by the appearance of the hidden part of the crack near the film surface when the sample is tilted to allow the {111} planes of the grain below to become parallel to the electron beam, as shown in Figs. 4(b) and 4(c).

More cracks are found in sample C, and these cracks become wider as they approach the film surface, as shown in Fig. 4(a). The cracks in sample E are few and narrow. It can be observed that in both samples, cracks extend to the film surface. The sudden connection of such cracks to the outside may bring about large bursts of ³He into the atmosphere.

The helium content of sample C is lower than that of sample E, but more cracks developed in sample C, which implies that there are more helium atoms migrating from the interior grain to the grain boundary in sample C than in sample E.

Figure 5 illustrates the helium bubble distribution in the grains of these two kinds of films. Compared with sample E, sample C shows a strong bubble linkage and orientation distribution. Many bubbles congregate as ribbon-like features in Fig. 5(a). Nearly all these ribbon-like features are parallel to the film surface. Some helium bubbles are aggregated as loops in sample C. The sizes of these loops are less than 5 nm. Large loops tend to transform into a ribbonlike structure. The bubbles aggregated as loops are more commonly seen on the thin fringe of the samples with low helium content because of the stress release in these regions. Accordingly, there may be stress release in the grains of sample C. In sample E, many interconnected bubble strings are observed in a uniform distribution in the grains.

Fig. 5.

	Figure Option ViewDownloadNew Window
	Fig. 5. Helium bubble distribution inside grains of the two kinds of films. (a) Sample C at He:Ti ratio of 0.299. (b) Sample E at He:Ti ratio of 0.356.

Figure 6(a) shows the crack along the grain boundaries in sample C. These cracks develop along a strip feature at the grain boundary. The strip features can be found in the grain or boundaries of both kinds of films. Figure 6(b) shows the strip feature distributed in the grain of sample E. Similar inclusions were also found in almost all Er tritide samples, which was confirmed to be oxide particles.^[3] One possibility is that the strip feature is formed by the enrichment of oxygen on the grain surface when preparing the film. No helium bubble is distributed in these strip features while a high concentration of helium is distributed at the boundaries of these inclusions.

Fig. 6.

	Figure Option ViewDownloadNew Window
	Fig. 6. (a) A big crack along the strip feature at the grain boundaries of sample C. (b) The strip feature distributed in the grain of sample E.

Figure 7 shows that a few helium bubbles are observed near the surface region and many large-size black loops have developed near the surface. These tiny bubbles distributed near the surface are dispersed compared with the interconnected bubbles in the inner films. This zone termed a “denuded zone” or “depleted zone” has been reported in other materials.^[28–30] The thickness of the depleted zone is ∼ 25 nm. The origin of this region can be understood by assuming that most of the helium created in this region diffuses to the surface and thus the bubble density is reduced.^[31,32] Snow et al. observed ∼ 15 nm of surface region void of helium bubbles in Er(D,T)₂ films.^[3] Their samples were not at the helium accelerated release stage, and no such loops were observed in Er(D,T)₂ films.

Fig. 7.

	Figure Option ViewDownloadNew Window
	Fig. 7. Helium bubble distribution near the film surface. (a) Sample C at He:Ti ratio of 0.299. (b) Sample E at He:Ti ratio of 0.356.

As the helium content in the film increases, the stress increases, which would result in a large difference in stress between the inner films and the surface. These loops are likely to be produced by the high internal stress. The number and the size of these loops in sample E are higher than those in sample C. The increase of the stress with helium content in the samples would generate more dislocations in the surface layer and drive the growth of these dislocations.

The film structure of sample C is a multi-layered columnar grain. In the model of Barna et al., the formation of polycrystalline films is considered as the growth of many single crystallites through the growth of characteristic layers, which occurs on their individual crystallographic faces.^[33] The presence of foreign atoms or molecules directly influences these processes. The adsorbed oxygen continuously accumulates at growth steps. Consequently, two-dimensional oxide films are formed, obstructing layer growth.^[21] This can be confirmed by the strip inclusion distributed on the boundaries of two-layered columnar grains, as shown in Fig. 6(a). Sample C was prepared at a low deposition rate. Therefore, more oxygen was introduced during the film growth. The layer growth was then seriously interrupted, and finally a stacked flaky crystal was generated. The crack at grain boundaries in sample C broadens when it is close to the surface, indicating the low interfacial energy between the grains. The weak binding energy or the high density of foreign atoms or molecules at the boundary between these mismatched grains accelerates the migration of the surrounding helium atoms to the boundaries and the development of cracks.

Inclusions are generally observed in all Ti tritide samples. Some of them are distributed inside grains, but most of them are distributed at grain boundaries. The inclusions in Ti tritides appear as long strips, which are different from the particle-shaped oxides in Er(D,T)_{2 − x} ³He_x.^[3] The energy-dispersive x-ray spectroscopy (EDS) analyses of these inclusions did not show the signal of oxygen. Parish et al. studied the oxygen contamination in ErD₂ thin films and indicated that due to oxygen’s low x-ray and the easy absorption of soft oxygen x-rays, EDS was not expected to yield high signal-to-noise ratios for oxygen.^[16] However, energy-filtered TEM showed thick near-surface oxides and intrafilm oxide particles in the D-loaded ErD₂ films.^[16] Thus the strip feature in Ti tritide films is probably flaky and has a low oxygen distribution.

Snow et al. studied helium release in Er(D,T)_{2 − x} ³He_x films and showed that the amount of helium released as a function of film thickness was relatively constant, suggesting that helium was being released only from the near-surface region and did not diffuse into the surface from the interior of the film.^[3] The surface of Ti tritide films forms a thicker depleted zone than Er tritides, but the early release fraction for Ti tritides is lower by more than one order of magnitude.^[2] There are more bubbles observed in the depleted zone of the Ti tritide films. Accordingly, the difference of helium diffusion rate in the depleted zone of the Ti and Er tritide films is obvious. The slow helium diffusion rate in the depleted zone of the Ti tritide films could be responsible for the observation of more helium bubbles and the low early release fraction from the film surface.

Compared with sample E, sample C shows strong helium bubble linkage and orientation distribution in the grains. The bubble coalescence in the grains is probably the result of the stress distribution in the film. Rodriguez et al. made a detailed XRD analysis of the ErT₂ film, which revealed significant in-plane compressive macro-strain due to ³He bubble formation/growth.^[34] The different size and number of dislocation loops ejected on these two kinds of film surfaces could be understood by the different stress distributions in these two films. It seems that a higher stress is distributed in sample E and relatively low stress is distributed in sample C.

Before the bubble interconnection, it is common to observe helium bubbles distributed along {111} planes, as shown in Fig. 8, after which in sample C almost all of these bubbles are linked as a ribbon-like feature parallel to the film surface. However, few bubbles are linked perpendicular to the film surface, as shown in Fig. 5(a). Many metal tritides showed a steady expansion rate during their storage since the ³He atoms required more space than the tritium atoms.^[35–37] Obviously, the film swelling would release the stress perpendicular to the film surface to a certain extent, which may impede the bubble interconnection in that direction. Helium bubbles tended to distribute with the orientation of high stress. The crack widening near the surface can also be understood by the film swelling in sample C.

Fig. 8.

	Figure Option ViewDownloadNew Window
	Fig. 8. Bright-field images of samples at He:Ti ratios of 0.109. (a) Underfocus; (b) the corresponding diffraction pattern of the selected-area (close to [110] zone axis).

At He/Ti = 0.27, the He release rate for sample E begins to increase rapidly from a constant value, much later than sample C in Fig. 1. The rapid increase of He release rate has been widely seen in Ti, Zr, and Er tritides when the sample is coming in the critical release period. Obviously this change comes from the depleted zone of the film surface. The interconnected bubbles beneath the depleted zone (Fig. 7) may not contribute to the critical release of helium. Mitchell et al. observed that the ³He release rate from erbium and zirconium ditritide films exhibited some differences and appeared to be associated with the spontaneous release of helium in bursts of about 10⁹ atoms.^[38] This irregular ³He release should only come from the cracks extending to the film surface.

The helium migration is greatly affected by the grain structure of the film. The migration and accumulation of helium at the grain boundary would be accelerated in the film consisting of columnar grains due to its small flaky grain structure and grain orientation distribution, compared with the film consisting of equiaxed grain structure. Columnar grains are weakly bonded and easily separated by the accumulation of helium on the grain boundaries. Cracks develop along the grain boundary in both films, most of which are nearly parallel to the film surface and widen near the surface. Bubble linkage as a ribbonlike feature develops parallel to the film surface in the film consisting of columnar grains. Inclusions as a strip feature are found in both kinds of films, at the boundary of which a high concentration of helium is observed. The critical ³He release from the surface comes from the cracks developed at grain boundaries extending to the film surface.