Epitaxial growth of Ca(Ge1−xSnx)2 with group IV 2D layers on Si substrate

Two-dimensional (2D) material is drawing considerable attention as a promising thermoelectric material. This study establishes the formation method of renewed Ca-intercalated group IV 2D materials, Ca(Ge1−x Sn x )2 crystals including germanene-based 2D layers. The solid phase epitaxy allows us to form epitaxial Ca(Ge1−x Sn x )2 on Si. Atomic force microscopy reveals that the Ca(Ge1−x Sn x )2 has island structures. X-ray diffraction proved the epitaxial growth of the Ca(Ge1−x Sn x )2 island structures and the increase of the c-axis lattice constant with Sn content increase. The formation of this renewed intermetallic compound including group IV 2D layer opens an avenue for high performance thermoelectric generator/Si.

T he development of renewable energy sources and energy saving devices 1) has been a central target toward the realization of sustainable development goals.As one of the renewable energy sources, thermoelectric (TE) power generation is attracting much attention. 2,3)In particular, film TE generator on Si substrate can be a small and stand-alone power source for internet of things sensor. 4)[7][8][9][10][11][12][13][14] Therefore, a lot of researchers have worked on developing TE materials with high ZT.However, the dilemma of TE studies, the crucial trade-off relationship among three TE parameters, prevents us from increasing ZT.
][19][20] On the other hand, many researchers have been developing promising approaches of S 2 σ enhancement.][23] For example, the experimental study reported that graphene exhibited ultrahigh S 2 σ of ∼360 μWcm −1 K −2 , which is the maximum value at RT. 24) In 2021, it was theoretically reported that silicene, 2D material composed of Si, could exhibit S 2 σ of ∼300 μWcm −1 K −2 . 25)][28][29] However, the formation of silicene is difficult because of its instability in the air. 30)n 2022, we experimentally reported that epitaxial CaSi 2 (Ca-intercalated silicene) films, which are relatively-stable in the air, exhibited higher S 2 σ than CaSi 2 bulk because of the deformation of the buckling structure. 31,32)In this previous study, the compositional analysis revealed the formation of Si-rich CaSi 2 films, indicating that the deformation of the buckling structure could be related to non-stoichiometry. 33,34)herefore, controlling atom species, elemental composition and strain near the interface, etc., can largely deform the buckling structure, leading to ultrahigh S 2 σ.Among Caintercalated group IV 2D materials, CaGe 2 can exhibit higher TE performance than CaSi 2 because the theoretical study reported that germanene could have higher μ than silicene. 35)or the deformation of buckling structure in CaGe 2 , the atomic control such as substitution of Ge with Sn is worth studying.However, there have been no studies about the formation of Ca(Ge 1−x Sn x ) 2 .
In this study, we establish the epitaxial growth of Ca(Ge 1−x Sn x ) 2 , Ca-intercalated Ge-Sn 2D layers, on Si substrates [Fig.1(a)].Ca(Ge 1−x Sn x ) 2 crystals were epitaxially grown through atomic interdiffusion of Ca, Ge, and Sn.The Ca(Ge 1−x Sn x ) 2 had island structures with atomically-flat surfaces and a height of ∼100 ± 30 nm.It was also found that introducing Sn increased c-axis lattice constant of Ca(Ge 1−x Sn x ) 2 .The formation of this renewed intermetallic compound including group IV 2D layer opens an avenue for high performance TE generator on Si.
Epitaxial Ca(Ge 1−x Sn x ) 2 samples with x = 0, 0.05, and 0.10 were formed on Si(111) substrates using solid phase epitaxy.As for Ca(Ge 1−x Sn x ) 2 samples (x>0), before the deposition of Ca and Ge atoms, Sn films were formed on the Si substrates in the following processes.H-terminated Si substrates were introduced into the high vacuum chamber (base pressure of ∼1 × 10 −3 Pa) equipped with W filament basket containing Sn sources (99.999%).Sn films with the thicknesses of 5-10 monolayers (MLs) were formed on the H-terminated Si substrates at RT.
The H-terminated Si substrates or Sn films/Si substrates were introduced into an ultrahigh vacuum chamber (base pressure of ∼1 × 10 −8 Pa) equipped with Knudsen cells for Ca (99.99%) and Ge (99.999%).For Ca(Ge 1−x Sn x ) 2 samples (x>0), contaminations on the surfaces of the Sn films/Si substrates were desorbed by degassing at 500 °C for 6 h.For Ca(Ge 1−x Sn x ) 2 samples (x = 0), namely CaGe 2 samples, clean Si(111) surfaces were obtained by annealing at 750 °C for 10 min after degassing at 500 °C for 6 h.
98-108 ML of Ge was deposited on the clean Si(111) surfaces or the Sn films/Si substrates at RT, followed by the deposition of 54 ML of Ca at RT.By annealing the samples at 750 °C for 30 min, CaGe 2 or Ca(Ge 1−x Sn x ) 2 films were epitaxially grown on Si(111) substrates through atomic interdiffusion of Ca, Ge, and Sn.It is considered that the composition of (Ge-Sn)/Ca tends to be ∼2 in Ca-(Ge 1−x Sn x ) system with small x because CaGe 2 is easily formed in the Ca-Ge system. 33)n-situ reflection high energy electron diffraction (RHEED) observations were performed with a 13 keV electron beam incident in the <11 ̅ 2> Si direction.The surface morphologies of the samples were observed by atomic force microscopy (AFM) with Si cantilever.Ex-situ X-ray diffraction (XRD) measurements were performed with a Cu Kα line (wavelength: 0.15418 nm).The pole figure was measured under the condition that the tilt angle (α) is 0-75°and the rotation angle (β) is 0-360°.The in-plane compositional distribution was measured using ex-situ scanning electron microscopy (SEM) with energy dispersive X-ray spectroscopy (EDX).The compositional distribution along the direction of the depth z was measured using ex-situ secondary ion mass spectrometry (SIMS).
Figures 1(b) and 1(c) show RHEED patterns of the Ca(Ge 1−x Sn x ) 2 samples with x = 0 and 0.1, respectively.Both patterns displayed streaky one third diffraction spots, which are typical of CaGe 2 -based film surfaces, 36,37) indicating that the samples with atomically-flat reconstructed surfaces were epitaxially grown on the Si(111) substrates with the relationship of (001) CaGe2 || (111) Si with [1 ̅ 10] CaGe2 || [11 ̅ 2] Si .The surface morphologies of Ca(Ge 1−x Sn x ) 2 samples were observed using AFM.Figures 1(d) and 1(e) show AFM images of Ca(Ge 1−x Sn x ) 2 samples with x = 0 and 0.1, respectively.It was difficult to acquire an atomic resolution image due to the ex-situ AFM observation.Unlike CaGe 2 samples with relatively-flat surfaces, Ca(Ge 0.9 Sn 0.1 ) 2 samples had island structures with micrometer size.Ca(Ge 1−x Sn x ) 2 islands do not have a triangle shape unlike CaGe 2 domains.That root mean square of Ca(Ge 0.9 Sn 0.1 ) 2 island structure top is small (∼0.53 nm), which indicates that the islands have atomically-flat surfaces.Figures 1(f) and 1(g) show a height distribution histogram and a line profile along the dotted line in Fig. 1(e), respectively.Interestingly, the heights of the island structures were binarized to be ∼100 ± 20 nm and ∼180 ± 10 nm.From the line profile, a small island B with a height of ∼80 nm is sometimes on large island A with a height of 100 nm, resulting in two kinds of heights of ∼100 and 180 nm [Figs.1(f    , where it is difficult to distinguish between two crystal structures because of similar 2θ of diffraction peaks.It was also found that the sharpness of the diffraction spots coming from Ca(Ge 0.9 Sn 0.1 ) 2 (fullwidth at half maximum of 0.36°) indicates high crystallinity.To confirm that Ca(Ge 1−x Sn x ) 2 crystals were epitaxially grown, pole figure measurements were performed.[001] CaGe2 .Therefore, the appearance of the six-fold symmetric peaks indicates the formation of twin crystals rotated by 180°.Thus, it was found that Ca(Ge 0.9 Sn 0.1 ) 2 crystals with twin crystals were epitaxially grown on Si(111) substrates regardless of Sn amount within the x range of 0-0.1.The c-axis lattice constants of Ca(Ge 1−x Sn x ) 2 crystals were estimated from the XRD diffraction peaks of 0012 6R-Ca(Ge1−xSnx)2 or 004 2H-Ca(Ge1−xSnx)2 , as shown in Fig. 2(c).The c-axis lattice constant increased as Sn content increased.This implies the Sn substitution for Ge in CaGe 2 .The previous study reported that in the case of CaSi 2 and CaGe 2 , both crystals have almost the same c-axis lattice constant, although the atomic radius of Ge is much larger than that of Si. 38) On the other hand, in the present study, it should be noted that the Sn introduction increased the c-axis lattice constant of Ca(Ge 1−x Sn x ) 2 crystals although it remains unclear whether the increase of the c-axis lattice constant is attributed to the lattice mismatch strain between the crystal and substrate or the structure size change of the Snintroduced crystal.This mechanism of the increase of the c-axis lattice constant is left to future study because the scope of this study is the formation of Ca(Ge 1−x Sn x ) 2 crystals.
Figure 3(a) shows SEM observations of Ca(Ge 0.9 Sn 0.1 ) 2 samples.A closer look of the SEM image revealed the two kinds of contrasts in the island structures, indicating island A with a dark contrast (main) and island B with a bright contrast (minor), which is consistent with the AFM results.The in-plane compositional distributions of Ca(Ge 0.9 Sn 0.1 ) 2 were measured by SEM-EDX, as shown in Figs.3(b)-3(e).The Ca, Ge, and Sn contrasts were bright at the locations corresponding to the island structures (island A) in Fig. 3(a), whereas at the locations, the Si contrast was dark.On the other hand, in the regions that are not island structures, the Ca, Ge, and Sn contrasts were dark, and the Si contrast was bright, indicating the existence of Si surfaces.These results indicate the formation of isolated Ca(Ge 0.9 Sn 0.1 ) 2 island structures.As for minor island B, the Ca and Sn contrasts were bright, while the Si and Ge contrasts were dark in some region.These indicate island B is mainly composed of Ca and Sn, as shown Fig. 3(f).Island B is considered to be formed through easy surface segregation of Ca and Sn.However, we cannot rule out the possibility that island A and B included a small amount of Si atoms that diffused from the substrate.
The atomic concentration profile of the Ca(Ge 1−x Sn x ) 2 along the direction of the depth z was measured using SIMS [Fig.4(a)].Therein, the z = 0 nm corresponds to the sample surface.In the z range of 0-20 nm, the concentration profiles of all the atoms are artifacts near the surfaces.In the z range of >20 nm, each concentration of Ca and Sn atoms gradually decreased with small slope and that of Ge atoms was almost constant as z increased.The slopes of the concentration profiles of Ca, Ge, and Sn atoms changed at z = 90-120 nm; the magnitude of the slope of the concentration profile in the z range of >100 nm was larger than that in the z range of <100 nm.On the other hand, in the Si case, as z increased, the concentration  (2) In the z region of 90<z<120 nm, the islands structures are disappearing.This result is consistent with the islands structure height of ∼100 ± 30 nm. (3) In the z region of 20<z<90 nm, it is considered that Ca(Ge 1−x Sn x ) 2 with relatively-homogeneous compositions were formed because of the small slope of the concentration profile of Ca, Ge and Sn atoms.The concentration of the Si atoms in the range of 20<z<120 nm is attributed to the detection of Si atoms from the Si surface.The constant slope of the concentration profile of Ca, Ge and Sn atoms reveals that the formation of the Ca(Ge 1−x Sn x ) 2 (island A) was caused through Ca, Ge and Sn diffusion with certain diffusion coefficients.Furthermore, excessive surface diffusion of Ca and Sn causes island B composed of Ca and Sn on the island A. As for the origin of the island structure morphology, it is easily understood by considering the lattice mismatch as follows.The lattice mismatch between CaGe 2 and Si( 111) is reported to be ∼5%, 39) the value of which is close to the condition of Stranski-Krastanov growth mode.It is considered that the Sn introduction expanded the crystal structure of CaGe 2 , which can increase the lattice mismatch between Ca(Ge 1−x Sn x ) 2 and Si(111) substrates up to the condition of Volmer-Weber growth mode.This resulted in the formation of the island structures.
To form a continuous Ca(Ge 1−x Sn x ) 2 film for the thermoelectric application, the lattice mismatch should be reduced by inserting some strain-relaxed buffer layers between Ca(Ge 1−x Sn x ) 2 and Si.Here, the solid phase epitaxy allowed us to form epitaxial Ca(Ge 1−x Sn x ) 2 island structures with relatively-homogeneous compositions on Si(111) substrates [Fig.4(b)].
In conclusion, we formed epitaxial Ca(Ge 1−x Sn x ) 2 island structures including Ge-Sn 2D layer on Si substrates using solid phase epitaxy.RHEED and XRD 2θ-ω scan revealed the epitaxial growth of Ca(Ge 1−x Sn x ) 2 /Si(111) substrates.Unlike CaGe 2 films with flat surfaces, the Ca(Ge 0.9 Sn 0.1 ) 2 had island structures with atomically-flat surfaces and a height of ∼100 ± 30 nm.It was also found that introducing more of Sn (0⩽x⩽0.1)resulted in a larger c-axis lattice constant.The formation of this renewed intermetallic compounds including group IV 2D layer opens an avenue for high performance TE generator on Si.
) and 1(g)].It should be noted that there is such a magic number in the island structure height although the origin is unknown.XRD 2θ-ω scans of Ca(Ge 1−x Sn x ) 2 samples with x = 0, 0.05 and 0.1 [Fig.2(a)] displayed the diffraction peaks coming from CaGe 2 -based crystals in addition to the peaks coming from Si(111) substrates [111 Si Kβ (∼25.7°), 111 Si Kα (∼28.4°), and 222 Si (∼58.8°)].It is considered that the broad peak at ∼44.2°does not originate from Ca(Ge 1−x Sn x ) 2 because it also appears in Si substrate measurement in the present XRD systems [Fig.2(a)].There are two possible crystal structures of the stable 6R-Ca(Ge 1−x Sn x ) 2 or metastable 2H-Ca(Ge 1−x Sn x ) 2 .The observed diffraction peaks at

Figure 2
(b)  shows the pole figure of the representative Ca(Ge 0.9 Sn 0.1 ) 2 sample at the fixed planes of {107} 6R-Ca(Ge1−xSnx)2 or {102} 2H-Ca(Ge1−xSnx)2 .This pole figure displayed six-fold symmetric peaks at the α of ∼55°, in addition to the 111 Si peak at the α of 0°.The three-fold symmetric peaks should appear because CaGe 2 is three-fold symmetrical to