A comprehensive study on Li4Si1−xTixO4 ceramics for advanced tritium breeders

Hetero-element doped lithium orthosilicates have been considered as advanced tritium breeders due to the superior performances. In this work, Li4Si1−xTixO4 ceramics were prepared by proprietary hydrothermal process and multistage reactive sintering. The reaction mechanism of Li4Si1−xTixO4 was put forward. XRD and SEM analyses indicate that insertion of Ti leads to lattice expansion, which promotes the grain growth and changes the fracture mode. The compressive tests show that the crush load increases almost four times by increasing x from 0 to 0.2. However, the thermal conductivity and ionic conductivity are the best when x = 0.05 and x = 0.1, respectively. Thermal cycling stability of Li4Si1−xTixO4 pebbles was further appraised through investigating the changes of microstructure and crush load. After undergoing thermal cycling, the Li4Si1−xTixO4 still show higher crush load compared with Li4SiO4, despite Ti segregation in some samples. The x = 0.05 sample exhibits excellent thermal cycling stability. In summary, proper amount of Ti doping can improve the crush load, thermal and ionic conductivity, and thermal cycling stability of Li4SiO4.


Introduction 
Tritium breeding materials are taken as one of the key functional materials for fusion reactor blanket. Neutrons generated from the reaction between deuterium (D) and tritium (T) can react with Li atom to produce tritium, thus achieving tritium self-sustaining of a D-T and hetero-element doped ceramics (notably Li 4+x Si 1-x Al x O 4 and Li 2 Ti 1-x Zr x O 3 ) [17][18][19][20] have been prepared in recent years. Previous studies reported that Al doped lithium orthosilicate exhibited the enhanced crushing strength [17,19,20], thermal and ionic conductivity [17,[20][21][22], and tritium release performance [18]. With respect to fusion reactor maintenance and waste disposal, it is hoped that the elements constituting the tritium breeding material have low neutron activation and short-lived radionuclides. But 26 A1 has a half-life of 7.2×10 5 years [23].
Since tritium breeding materials should endure long periods of harsh operating conditions (high temperature, thermal gradient, irradiation by neutron and energetic particles), thermal and irradiation stability are two important parameters and deserve more attention. However, the influence of doping elements on the thermal cycling stability of tritium breeding materials has not been investigated so far. Meanwhile, fine-grained ceramics are identified as the most promising, regarding tritium release behavior and mechanical strength [24]. But with the grain refinement, surface energy of ceramic particles increases, and thus the structural and functional instability of materials may get more acute in high temperature environment. Therefore, it is significant to fabricate fine-grained lithium ceramic solid solution and investigate its thermal cycling stability.
The aim of this work is to prepare Li 4 Si 1-x Ti x O 4 by proprietary hydrothermal process and multistage reactive sintering (Ti has lower neutron activation than Al), and to investigate the influences of Ti doping on the microstructure, mechanical and physical properties of Li 4 SiO 4 . Furthermore, the thermal cycling stability of Li 4 Si 1-x Ti x O 4 pebbles is further studied. These results can cast light for future development of advanced tritium breeding materials.

1 Preparation of precursor powders
The precursor powders were hydrothermally synthesized using LiOH·H 2 O, fumed SiO 2 , and TiO 2 nanopowders with a ratio of 4:(1-x):x (x = 0, 0.05, 0.1, 0.2). Briefly, LiOH·H 2 O (0.14 mol) was thoroughly dissolved in 70 mL of deionized water under magnetic stirring. Meanwhile, fumed SiO 2 (0.035(1-x) mol) was uniformly dispersed in 70 mL of ethanol. Subsequently, the above two solutions were mixed under magnetic stirring at room temperature. After that, TiO 2 nanopowders (0.035x mol) were added, and continued stirring for 30 min. Then, the mixed solution was transferred into a Teflon-lined stainless-steel autoclave with a capacity of 200 mL and performed at 180 ℃ for 12 h. Finally, hydrothermal products were dried at 80 ℃ for overnight, and ground in an agate mortar to obtain precursor powders.

2 Fabrication of Li 4 Si 1-x Ti x O 4 pebbles
The as-prepared precursor powders and deionized water were mixed in a mass ratio of 5:4 to form the slurry, and then the green spheres could be produced by dropping the slurry through a nozzle (0.7 mm in diameter) into a container of liquid nitrogen. By freezing for more than 15 min, the green spheres were salvaged and placed on filter papers, dried in air for 30 min and then in drying oven of 70 ℃ for overnight. Li 4 Si 1-x Ti x O 4 pebbles could be finally obtained by sintering the dried green spheres in a box-type resistance furnace. To be specific, the samples were heated to 420 ℃, dwelling for 1 h, then heated to 710 ℃, dwelling for 1 h, and finally sintered at 800 ℃ for 4 h. The heating rate was 5 ℃/min.

3 Characterization
Thermal behavior of the precursor powders was studied by thermogravimetry and differential scanning calorimetry (TG/DSC, NETZSCH 409 PC) in air at a constant heating rate of 10 ℃/min. The phase composition and crystal structure were investigated by X-ray diffractometry (XRD-7000, Shimadzu, Japan), and the cell parameters were refined via Jade 6.5 software. The morphology and structural studies were conducted on a scanning electron microscope (SEM, Model S-4800, Hitachi, Japan) attached with an energy-dispersive X-ray spectroscopy (EDS). Grain size was measured by Nano Measure 1.2 software from SEM images. The density was measured by an electronic density balance using ethyl alcohol as the immersion medium. The crush load was tested by a universal material strength-testing machine with a 5 kN load cell and cross-head speed of www.springer.com/journal/40145 0.1 mm/min (HT-2402, Hungta). To minimize the influence caused by the pebble size, the pebbles with a diameter of 1.2-1.3 mm were used for the test. The average crush load was estimated based on the results of no less than ten pebbles. For thermal and ionic conductivity tests, the precursor powders were pressed into pellets (12.7 mm in diameter and about 1.5 mm in thickness) and sintered at the same sintering parameters. The thermal conductivity was measured by LFA 457 Micro Flash Analyzer of NETZSCH. The ionic conductivity was determined by alternating current impedance spectra (ACIS) measured using an Agilent E4980A impedance analyzer in the frequency range from 100 to 10 6 Hz, and analyzed by fitting the equivalent circuit model using the ZView software. For thermal cycling tests, the ceramic pebbles were heated to 800 ℃ in a box-type resistance furnace, dwelling for 4 h, then cooled down to room temperature (set the heating and cooling rate to 5 ℃/min). After each three cycles, a batch of ceramic pebbles were fetched out for microstructure characterization and compressive strength testing.  hydrothermal products in a forced-air convection oven (

1 Phase composition of precursor powders
). The diffraction peaks of the Ti doped samples are basically the same as that of the undoped one, except the emergence of LiOH (JCPDS No. 85-0736) and Li 2 TiO 3 (JCPDS No. 03-1024) peaks. Moreover, with the increase of Ti content, the proportion of Li 2 CO 3 decreases. It reveals there are two competitive reactions.

2 TG/DSC analyses
The TG/DSC curves of precursor powders with varying Ti doping amount are illustrated in Fig. 2. The weight loss below 100 ℃, accompanied by an endothermic peak around 100 ℃, is ascribed to the removal of adsorbed water. The slight weight loss and endothermic peak around 425 ℃ can be ascribed to the reaction of residual SiO 2 /TiO 2 with Li 2 CO 3 [25]. Major weight loss occurs in the range of 650-750 ℃, accompanied by a sharp endothermic peak at 709 ℃, which is attributed to the formation of Li 4 Si 1-x Ti x O 4 . The weight loss decreases with increasing Ti content due to the reduced lithium carbonate in the samples, in accordance with XRD results. Compared with the undoped sample, the endothermic peak of Ti doped samples shifts to lower temperature, indicating Ti doping can reduce the reaction temperature and raise the reaction efficiency. This is due to the lattice distortion caused by the substitution of Ti 4+ for Si 4+ , which may reduce the activation energy [26]. No obvious weight changes are found above 800 ℃, suggesting the synthesis process is finished. Moreover, considering the weight losses around 425 and 710 ℃, multistage reactive sintering is adopted to reduce the pores and impurities (i.e., 420 ℃ for 1 h, 710 ℃ for 1 h, and 800 ℃ for 4 h). Figure 3 shows the XRD patterns of the Li Figure 4 shows the cross-section SEM micrographs of    be found in some regions (marked with red circles). Combined with the XRD results, it can be deduced the small-sized Li 2 TiO 3 particles exist as second phase.

5 Crush load, density, and grain size
The ceramic pebbles need high crush load to prevent the breaking and fragmentation, which lead to plugging of purge circuits and diminished heat transfer [24]. The dependence of crush load, density, and grain size on Ti doping amount is illustrated in Fig. 6. Grain size and porosity are critical factors affecting the strength of ceramics, which can be expressed by the empirical equation [34]: S = kG -a e -bP , where S is the strength, G is the grain size, P is the porosity, b is a constant related to pore shape, and k, a are positive constants. The Li 4 Si 1-x Ti x O 4 samples have a comparable density and larger grain size, but higher crushing strength, compared to the Li 4 SiO 4 sample. The reasons for this phenomenon are, on one side, with increasing Ti content, the lattice distortion increases and the deformation resistance of the matrix is increased, Fig. 6 Dependence of crush load, density, and grain size on Ti doping content. thereby resulting in a more significant solid solution strengthening effect. On the other side, second phase Li 2 TiO 3 contained in the Li 4 Si 1-x Ti x O 4 samples also makes a huge contribution for the enhanced strength as the crushing strength of Li 2 TiO 3 is better than that of Li 4 SiO 4 . It also should be noted that strength of ceramic pebbles is also affected by flaws, sphericity, and impurities.

6 Thermal conductivity
Thermal conductivity k can be calculated by multiplying the bulk density ρ, the thermal diffusivity α, and the specific heat c p . (2) Generally, thermal conductivity decreases with increasing porosity and is very sensitive to impurities. Hence, the presence of pores and Li 2 TiO 3 may be detrimental to the thermal conductivity. Even so, the thermal conductivity of Li 4 SiO 4 is obviously improved through the substitution of Ti (the best performance is not necessarily the most doped sample), it is foreseen that solid solution ceramics should be a good candidate for advanced tritium breeders.

7 Ion conductivity
Tritium diffusion process can be envisaged as lithium-ion migration which acts as the charge carrier in the ceramics [22,35]; the tritium release behavior can be evaluated by measuring the conductivity of breeder materials. Figure 8 is the impedance spectra of the samples recorded at room temperature in the frequency range of 100-10 6 Hz. The equivalent circuit is composed of a resistance R1 in series with a component consisting of another resistance R2 in parallel to a constant phase element (CPE). The depressed semicircle is present in the plot, the right intercept of the semicircle with the real axis corresponds to the bulk resistance (grain interior resistance plus grain boundary resistance). The ion conductivity can be calculated in accordance with the relation: where σ is the ion conductivity, L is the sample thickness, R is the bulk resistance, and S is the area of the electrode. The calculated ion conductivity is illustrated in Table 2. It can be seen that substitution of Ti for Si causes an increase in ionic conductivity, possibly indicating the improvement in tritium release performance. The insertion of Ti 4+ enlarges the lattice size of the Li 4 SiO 4 -type structure (Li 4 SiO 4 with a monoclinic structure (space group P2/1) contains SiO 4 tetrahedral elements, in which Li floats around the silicon-oxygen tetrahedron), i.e., the size of migration channels of Li + increases [36]. Moreover, since Ti-O bond is stronger than Si-O bond, the Li-O bond interaction in Li 4 SiO 4 structure is decreased when the Si atom is replaced by Ti atom, and thus the ionic conductivity is improved [37]. However, the ionic conductivity decreases with increasing x from 0.1 to 0.2, and this is probably due to the increased proportion of second-phase Li 2 TiO 3 . According to the research by Tanigawa et al. [38], lithium orthosilicate has better conductivity than lithium titanate, and the electrical conductivity of Li 4 SiO 4 is about two orders larger than that of Li 2 TiO 3 at 702 ℃.

8 Thermal cycling
It is important to study the microstructure and crushing strength changes during the thermal cycling tests. As shown in Figs. 9 and 10, the grain size of Li 4 SiO 4 sample increases following thermal cycles, small-sized pores merge with each other, and coarse grains with a transgranular fracture morphology can be observed. EDS analyses indicate both fine grains and coarse grains in Fig. 10(a2) are Li 4 SiO 4 (Figs. S4 and S5 in the ESM), confirming the secondary growth of grains. The crush load increases slightly after 9 cycles ( Fig.  11(a)), which is ascribed to the change of fracture mode and the decrease of pores. However, due to the presence of intracrystalline pores and cracks, the crush load decreases after 12 cycles. For the Li 4 Si 0.95 Ti 0.05 O 4 sample, no obvious changes in the microstructure and the crush load after 6 thermal cycles ( Fig. 11(b)). Many tiny particles precipitate on the grain boundaries and the surface of grains ( Fig.  10(b3)). EDS analysis (see Fig. S6 in the ESM) shows that Si/Ti ratio is lower than the designed value (viz. 8.3 vs. 19), suggesting the segregation of Ti toward the grain surface during long-time thermal cycling. This can lead to the slight decrease of strength after 9 cycles. In general, the crush load remains the same within the accuracy.
With prolonging the thermal cycling, the microstructure of Li 4 Si 0.9 Ti 0.1 O 4 sample changes significantly and the strength decreases rapidly (Fig. 11(c)). EDS analysis confirms the tiny particles as titanium-rich (Fig. S7 in the ESM). The atom ratio of Si:Ti is close to 2.5:1 which is much lower than the designed value. It is reasonable to assume that the segregation becomes more serious with the increase of Ti doping content. Furthermore, a great amount of intracrystalline pores  can be observed after 12 cycles. As a result, the crush load decreases to less than 20 N, revealing inferior durability of the Li 4 Si 0.9 Ti 0.1 O 4 sample.
For the Li 4 Si 0.8 Ti 0.2 O 4 sample, laminate structure mainly composed of Ti and O can be seen clearly after 9 cycles (Fig. S8 in the ESM). EDS mapping also indicates Ti with high concentration within the laminate grains (Fig. S9 in the ESM). It suggests the formation of titanate. Although the microstructure of the Li 4 Si 0.8 Ti 0.2 O 4 sample changes slightly after the thermal cycling, it still exhibits a high crush load (Fig. 11(d)). Since many factors can affect the strength, such as grain size, porosity, pore size and distribution, sphericity, the content and size of second-phases, humidity, etc., the variation in strength with thermal cycling periods is irregular.
To sum up, the moderate amount of Ti doping can improve the mechanical and physical properties and thermal cycling stability of Li 4 SiO 4 . Compared with Al-doped Li 4 SiO 4 [17,19,20], Ti doping plays a more significant role in enhancing crushing strength, whilst Al doping contributes more to the improvement of conductivity. Tritium release experiments conducted by Zhao et al. [18] revealed that Li 4.2 Si 0.8 Al 0.2 O 4 had lower tritium release temperature and potentially better irradiation resistance compared to Li 4 SiO 4 . The tritium release performance and irradiation resistance of Li 4 Si 1-x Ti x O 4 solid solution are also worthy of expectation (Ti has lower neutron activation than Al). Since the thermal cycling stability of Li 4+x Si 1-x Al x O 4 pebbles has not been reported, we cannot make a comparison yet. View from the current studies, solid solution materials should have unique advantages in advanced tritium breeding materials, although more deep studies are still required to further understand the influence of various doping elements.  With the increase of Ti doping content, the segregation becomes more serious. In terms of thermal stability, the optimal Ti doping amount should be x = 0.05. As titanium has low neutron activation behavior, Li 4 Si 1-x Ti x O 4 solid solution may have good application prospect in the field of solid tritium breeders.