Growth of metastable 2H-CaSi2 films on Si(111) substrates with ultrathin SiO2 films by solid phase epitaxy

The Si-nano dot substrates formed using the ultrathin silicon oxide films were applied to fabricate CaSi2 films. The CaSi2 formed by this process was identified as the metastable phase 2H as the main component, and the 1H structure existed partially at the grains of the 2H phase. Although no experimental reports exist for the formation of 2H-CaSi2 crystal, the Si-nano dot substrates are considered as the high-entropy substrate to form the metastable phases. We experimentally determined the lattice parameter of the 2H phase by the annular dark field–scanning transmission electron microscopy observations using the Si as an internal standard sample.

Intriguingly, it was recently reported that CaSi 2 films with deformed 6R-structure were epitaxially grown on Si substrates by solid phase epitaxy (SPE), 3) where silicene buckling structure is deformed.Although 1H-CaSi 2 is a high-pressure phase, it has been reported that (110) oriented 1H-CaSi 2 films can be grown on SiO 2 glass substrates. 22)H-CaSi 2 crystal is difficult to grow experimentally because its Gibbs free energy is higher than that of other layered structures, but it has been observed as stacking faults near the surface of 6R-CaSi 2 epitaxial film. 21)The growth of metastable 1H-and 2H-CaSi 2 thin films, which cannot be achieved in bulk crystals, suggests a high-entropy influence of the growth substrate surface.
So far, the epitaxial silicide films have been grown on clean silicon surfaces.Recently, we developed a growth technique for well-controlled nanostructures in thin films, such as epitaxial Si nanodots (NDs), with several orders of nm in size. 23,24)This Si-ND film is expected to be a highentropy substrate with many distortions for growing CaSi 2 epitaxial films.In this study, we focused on growing CaSi 2 crystals by a method in which calcium is deposited on Si-NDs instead of on a clean silicon surface.The metastable 2H-CaSi 2 and high-pressure phase 1H-CaSi 2 crystals grew in epitaxial film (Fig. 1).Their crystal structures were determined by annular dark field-scanning transmission electron microscopy (ADF-STEM).
CaSi 2 thin films on Si(111) substrates were formed by SPE via the reaction between deposited Ca and Si-NDs, as shown in Fig. 2(a).First, an ultrathin SiO 2 film was formed on a Si (111) substrate, as previously reported. 23,24)Next, 25 Si monolayers (MLs) were deposited on the ultrathin SiO 2 films to form epitaxial Si-NDs with an ultrahigh density (>10 12 cm -2 ) at 450 °C.Then, 32 Ca MLs were deposited on Si-NDs at RT.By annealing the samples at 400 °C, CaSi 2 films were epitaxially grown via the inter-diffusion of Ca and Si atoms, which was confirmed by (2 × 1) reconstructed reflection high energy electron diffraction (RHEED) pattern of CaSi 2 surface.Finally, to prevent the oxidation of the sample, the CaSi 2 films were covered with an undoped amorphous Si layer (⩽10 nm).For comparison, a CaSi 2 epitaxial film was formed without ultrathin SiO 2 films via reacting the Si(111) substrate with Ca deposited.
The transmission electron microscopy (TEM) specimens were fabricated using a focused-ion beam machine.STEM   observations were conducted using JEM-2100F equipped with an aberration-corrected system at an accelerated voltage of 200 kV.20 ADF-STEM images were obtained with a resolution of 1024 × 1024 pixels and an image capturing dwell time of 2 μs/pixel.These series of images were obtained using the image processing utility (IPU) plug-in 25) and the acquired image series plug-in 26) (HREM Research Inc.) running on Gatan Microscopy Suite ver.2.3 (Gatan Inc.).To confirm the crystal structure of the CaSi 2 phase in the ADF-STEM images, the high-angle annular dark field (HAADF)-STEM images of 1H, 2H, and 6 R structures were simulated using software of Tempas (HULINKS Inc.).
Figure 2(b) shows RHEED patterns of Si-NDs and CaSi 2 thin films formed by the SPE process using Si-NDs with the ultrathin SiO 2 films, where the incident electron direction is 〈110〉 Si.The RHEED pattern of Si-NDs appears in clear spots.This suggests that the Si-NDs are epitaxially grown on the Si(111) substrate.While, the streaky spots and ring patterns were observed, in Fig. 2(b), indicating the CaSi 2 films were not epitaxially grown on Si-NDs on Si(111) substrate, but polycrystalline.
Figure 3(a) is the bright field (BF)-STEM image for the specimen fabricated by the SPE process using the Si-NDs with the ultrathin SiO 2 films.This shows a Ca-Si film thickness of ~10 nm on the Si(111) substrate.The Ca-Si film consists of several trapezoidal-shaped grains with a width of ~20 nm, and the edge of some grains seems to overlap. Figure 3(b) is the result of energy dispersive X-ray spectroscopy (EDX) analysis around the crystal grains with a width of approximately 20 nm, as determined by Fig. 3(a).When calculating the Ca to Si atomic ratio from the center (A) and the right (B) regions of the EDX mapping, region A, corresponding to the trapezoidal-shaped grain, has a Ca to Si ratio of 1:2; meanwhile, that ratio is 1:1.3 in region B. From this result, the grain of region A is considered the CaSi 2 phase, and region B would be a Ca-Si phase or a partial overlap between the CaSi 2 phase and protective coating of Si.
The oxygen distribution in EDX mapping is noteworthy, where oxygen is enriched at the interfaces of Si substrate/Ca-Si film and Ca-Si film/amorphous Si.The relationship between the uneven oxygen distribution and the crystal structure will be discussed later.structures, respectively.In another region, the CaSi 2 grain constituted only a 2H structure, although it is not shown here.Thus, the ADF-STEM observations reveal that the CaSi 2 film fabricated by the SPE process using the Si-NDs with the ultrathin SiO 2 films forms grains having a mixture of 2H and 1H structures or a single 2H structure.Although the 2H structure has been known for CaGe 2 , it has never been reported for CaSi 2 . 27,28)Hence, we confirm that the SPE process using the Si-NDs with the ultrathin SiO 2 films can produce 2H-CaSi 2 , although 1H-CaSi 2 is present as a minor phase.015501-3 © 2023 The Author(s).Published on behalf of The Japan Society of Applied Physics by IOP Publishing Ltd with the simulated HAADF-STEM image of the 6 R structure, which is well-known as the most stable phase.However, as mentioned above, the 2H and 1H structures formed by the SPE process using the Si-NDs with the ultrathin SiO 2 films are unreported phases at ambient pressure. 1,2)The 2H and 1H phases were formed from the Si-NDs used in the fabrication instead of the Si(111) substrate or the ultrathin oxide layer on the Si substrate that fabricated the Si-NDs.The former may mean that the Si-ND film acts as a high-entropy substrate with many distortions for growing the CaSi 2 epitaxial film.The latter has been reported that 1H-CaSi 2 is formed on the SiO 2 glass substrate. 22)As shown in Fig. 3(b) EDX results, the high oxygen concentration at the Si substrate/Ca-Si film and the CaSi 2 /amo.Si interfaces suggest that oxygen (Si-O-Si bonds) likely affects the formation of 1H.
The lattice parameters for the 1H and 2H phases were determined from the ADF-STEM images using Si as an internal standard material.These ADF-STEM images include both Si substrate and CaSi 2 phase, as shown in Figs.4(a) and 4(b).Two orthogonal vectors are selected using the mass center of the bright dots in each Si substrate and CaSi 2 phase.The two orthogonal vectors in the CaSi 2 phase are obliquely projected onto that in the Si, and the length of the two vectors in the CaSi 2 phase is calculated by the correlation of the internal standard Si sample.The lattice parameters of the 6 R structure were calculated to confirm the validity of the measurement mentioned above because 6 R is the stable phase.The lattice parameters of the 1H-, 2H-, 6R-, and 3R-CaSi 2 from present work and literature are listed in Table I. 22,[29][30][31][32] The number of measurements of the 2H, 1H, and 6 R phases is 25, 5, and 10, respectively, and the standard deviations are also indicated in Table I. A first, the difference between the experimental and theoretical lattice parameters of 6 R 29,30) is Δa = -0.2% and Δc = -0.8% with less than 1% errors for both.Thus, this procedure is valid as an estimation method for the lattice parameters.Meanwhile, the average error fraction in Table I is 1.2% for a 2H , and 0.6% for c 2H , which is approximately 1%.Also, the error of 1H is 2% for a 1H and 2.1% for c 1H ; both are higher than that for 2H.However, these 1H calculations are small values because the presence of 1H is limited to the upper region of the CaSi 2 film.
In the bulk crystal, the 6R-CaSi 2 phase transforms to 1H at RT under a pressure of 9.8 GPa. 2) In that case, the normalized volume (V 6R /6) has been reported to decrease from V 6R /6 = 0.0653 nm 3 to V 1H = 0.056827 nm 3 , which is 13%. 2)onversely, in the epitaxial film of this study, the normalized volumes do not show considerable differences like that in bulk (V 6R /6 = 0.0648, V 2H /2 = 0.0655, and V 1H = 0.0649 nm 3 ).Therefore, the mechanisms are different for the 6 R to 2H and 1H transformation by high-pressure treatment and the 1H and 2H grown by the SPE process using Si-NDs with the ultrathin SiO 2 films.
Then the lattice misfit at the interface of the CaSi 2 film and Si(111) substrate is discussed.The orientation relationship between the Si(111) substrate and the epitaxial CaSi 2 film (or grains) in the present work was (001) CaSi2 //(111) Si , [1-10] CaSi2 // [11-2] Si , and [001] CaSi2 //[111] Si .Thus, the lattice misfit is calculated from the lattice parameter, a, for each CaSi 2 phase and the d 110 lattice spacing of the Si substrate.The calculated lattice misfit values are 0.15% for 6 R, -0.08% for 2H, and -0.23% for 1H.indicating that the a-axis is compressed with respect to the Si substrate in both 1H and 2H.At the same time, 6R is expanded, as shown in Fig. 5.Alternatively, 6R-CaSi 2 grows on a clean Si surface in the SPE process without the ultrathin SiO 2 films, while 1H and  2H grow on a controlled Si-ND substrate formed on an ultrathin oxide film.This result suggests that the Si-O bond is 1.6 Å, whereas the Si-Si bond of diamond-type Si is 2.3 Å; thus, the SPE process used in this study introduces strain into the Si lattice, which is the starting point for film growth.Specifically, 1H preferentially grows when the oxygen concentration is high because the Si layer at the starting point of growth is greatly compressed.In contrast, 2H preferentially grows when the oxygen concentration is low.Furthermore, in the absence of oxygen, the Si-Si bonds do not shrink, and 6R is considered to grow.Although the 3 R phase was not formed by the SPE process in the present work, the lattice misfit of the 3 R is calculated to be -0.3%, showing a larger compression than that of the 2H and 1H.But the formation of the 3 R phase has been reported to be induced by tensile stress, 27) which is the opposite of the above result.Therefore, it is considered that the 3 R phase was not formed in the present work.
In conclusion, the present work revealed that using Si-ND, the SPE method formed CaSi 2 films, which consisted of the 2H and 2H/1H mixed grains.From the ADF-STEM observations, the lattice parameters of the 2H structure were determined experimentally with an error of ~1%.With respect to the growth mechanism, the ultrathin silicon oxide, which is formed to fabricate the Si-ND substrate, could act as a buffer layer, resulting in the formation of the 2H-and 1H-CaSi 2 phases.However, according to the results reported by Yaokawa et al., 27) the bulk 2H-CaSi 2 crystal cannot be obtained because of the metastable phase.Therefore, the preferential formation of the 2H phases in the present work is attributed to the ultrathin silicon oxide film and the specific substrate with many distortions caused by the Si-NDs.The detailed reaction mechanism is a subject for future study.

Fig. 1 .
Fig. 1.Various polymorphs of CaSi 2 formed by film formation via solid phase epitaxy (SPE) and bulk processes.

Fig. 2 .
Fig. 2. (a) Schematic diagram of the SPE process using Si-NDs with ultrathin oxide films and (b) RHEED images of the Si-NDs and CaSi 2 grains.

Figure 4 (
a) is the ADF-STEM image of the CaSi 2 film fabricated by the SPE process using the Si-NDs with the ultrathin SiO 2 films.The electron beam direction is Si [110].In the CaSi 2 film, the bright dots correspond to the Ca atomic columns, and the others correspond to the Si atomic columns.The Ca atom columns aligned along the Si(111) of the lattice form the Ca layer.Two Si atomic columns form zigzag Si layers between two Ca layers.The Ca and Si layers are aligned along the [111] direction of the Si lattice, forming a twofold periodic structure (ABAB sequence).A onefold periodic structure (AA sequence) partially appears at the upper region of that area.The two insets in Fig. 4(a) are the simulated HAADF-STEM images of the 1H and 2H structures.The lower and upper regions of the CaSi 2 film in the ADF-STEM image agree with the simulated 2H and 1H

Figure 4 (
b) is the ADF-STEM image for the CaSi 2 epitaxial film fabricated by the SPE process by reacting the Si(111) substrate with the deposited Ca, showing the interface between the Si substrate and the CaSi 2 epitaxial film.The CaSi 2 epitaxial film has a sixfold periodic structure (AABBCC sequence).The ADF-STEM image is consistent

Fig. 3 .
Fig. 3. STEM micrographs and EDX analysis of the CaSi 2 films formed by the SPE process using the Si-NDs with the ultrathin SiO 2 films.(a) BF-STEM image and (b) EDX maps and overlayed maps of O, Si, and Ca, showing the EDX spectra for regions A and B in the ADF-STEM image.

Fig. 4 .
Fig. 4. ADF-STEM images of the CaSi 2 films formed by the SPE process (a) using the Si-NDs with the ultrathin SiO 2 films and (b) without the SiO 2 films with the insets of the HAADF-STEM images simulated from the 1H-, 2H-, and 6R-CaSi 2 structures.

Table I .
Lattice parameters of 1H-, 2H-, 6R-, and 3R-CaSi 2 from present work and literature.The number in brackets is the standard deviation.
©2023The Author(s).Published on behalf of The Japan Society of Applied Physics by IOP Publishing Ltd