While accommodating on substrates, silicene quickly degrades from its top face if exposed to environmental conditions. Degradation results in the silicon oxidation arising from direct incorporation of oxygen atoms in between the silicon bonds or amorphization-mediated oxygen reactions [13]. Both cases are related to the higher reactivity of the remnant sp³ hybrid bonds that are constitutional of the buckled silicene structure. Sequential encapsulation with a nonreactive capping layer of amorphous Al₂O₃ is an effective way to prevent silicon from degradation and preserve its own structure. Al₂O₃ films are grown from RT reactive deposition of aluminum atomic flux in O₂-rich overpressure. Predeposition of an aluminum ultrathin layer prevents silicon from getting locally oxidized and serves as precursor for the successive Al₂O₃ layers to grow up. Protection and stabilization of the silicene layer are validated by in situ X-ray photoemission spectroscopy of the Si 2p core level (before and after encapsulation) and ex situ Raman spectroscopy of the encapsulated structure. The former one makes sure that the chemical nature of the silicene bonding is not affected by Al₂O₃ capping [13], while the latter one appears as a quick probe of the characteristic Raman spectrum of silicene [14]. This encapsulation strategy is pictorially illustrated in Figure 2 for the case of epitaxial silicene grown on a delaminable substrate, for instance epi-Ag(111)-on-mica. It can be applied to each silicene configuration, therein proving a general environmental stabilization strategy whenever the encapsulated layer is taken out from the vacuum. Moreover, the effectiveness of the same can be extended potentially to other silicene-like X-enes like stanene or epitaxial phosphorene [15, 16].

Silicene on Ag(111) represented a decisive milestone in the development of the X-enes beyond graphene [4]. It was the first time that silicon was demonstrated to arrange in a nonconventional crystal lattice when shrunk down in 2D, with silicon bonds showing a mixed hybridization in between the sp² of flat graphene and the sp³ of diamond. Many experimental studies corroborated by theoretical support constitute now the advanced knowledge on this artificial 2D crystal. Hitherto, Ag(111) is probably the most widely used template for silicene. Therefore for an extended discussion of the physical and chemical properties of silicene on Ag(111), we refer to detailed topic reviews [3, 17]. Here, we will focus on the specific route implemented to bypass the role of Ag(111) substrate for integrating silicene into electronic devices despite its pristine growth on a metal substrate. This route is therefore specific for silicene grown by MBE on metal substrates, even if it might be also adapted to silicene achieved by intercalation or segregation methods (see Figure 1) with appropriate measures [3]. On Ag(111), silicene cannot be grown in its freestanding fashion, i.e., with silicon atoms alternatively up and down, but lies down by accommodating its silicon atoms on top or in between the silver ones. The resulting silicene sheets, i.e., 3\times 3, \surd 7\times \surd 7, and \surd 7\times \surd 7 freestanding silicene reconstructions (or equivalently 4\times 4, \surd 13\times \surd 13, and 2\surd 3\times 2\surd 3 silver superstructures), constitute a phase diagram on monocrystalline silver substrate that can be equally resumed on thin silver epitaxial films displaying (111) termination [18]. The silver substrate plays the same key role in dictating the details of the multilayer silicene growth, namely, the deposition of a multiple silicene layer on top of the pristine one through a terrace growth mode [5]. Indeed, only when the single layer silicene acts as a seeding layer for the additional silicene layers in a three-dimensional (3D) growth regime, the multilayer silicene shows up with both clear \surd 3\times \surd 3 termination and a characteristic Raman shift markedly different from those of conventional crystalline and amorphous silicon [19]. Interestingly, the oxygen intercalation through bilayer silicene is a viable route to achieve quasi-freestanding silicene decoupled from the silver substrate which survives in the ambient environment for at least 120 hours [20]. The thickness reduction of the silver template down to thin films, either in the form of epitaxial silver on mica [21] or on Si(111) [18] substrates, appeared as the enabling step for the exploitation of silicene in applicative directions like the integration into a FET device as detailed in the following. For instance, thin epitaxial silver on mica bypasses the use of highly expensive monocrystal substrates and allows for the delamination of fully encapsulated silicene heterostack out of the solid mica substrate. Even reduced to the nanometer thickness, silver still presents two technical hurdles. The former is related to the removal of the rigid template, e.g., silicon or mica, and the latter is concerned with the unavoidable contribution to the electronic transport of a thin silver film overwhelming silicene, when integrated as a FET channel. Mica can be mechanically delaminated from the silver thin film by means of a cutter-assisted two-tape method allowing for a large area (few cm²) transfer of the thin silver film (Figure 3(a)). All along this way, a thin membrane where silicene is sandwiched in between the residual silver at the bottom and the Al₂O₃ stabilizer (Figure 3(b)). Then, the so-achieved flexible membrane can be faced downward onto a new supporting substrate, e.g., a device substrate like SiO₂/Si⁺⁺ (Figures 3(c) and 3(d)). This flip exposes the Al₂O₃ in contact with device substrate and the silver layer to air. A key point at this stage is to recycle the native silver layer in source and drain contacts by taking benefit from the known hybridization between it and the originally supported silicene (see Figure 3(e)). By means of electron beam lithography, the silver native layer can be patterned in a FET configuration by designing silver-made contacts and silver-free active channel. In addition to silver-made source and drain contacts, the shrewdness of using silicon oxide on highly doped silicon readily serves as bottom gate. Remarkably, the same process flow equivalently stands out both for single and for multilayer silicene. Single and multilayer silicene-based FETs in low-field (V_{\text{d}}=20\text{mV}) conditions show an ambipolar transport behavior at RT (Figures 3(f) and 3(g)) [19, 21]. In order to numerically characterize these FETs, the well-accepted ambipolar model used for graphene FET [22] has been used as well. Fitting the resistance (R=V_{\text{d}}/I_{\text{d}}) plots, the best mobility values achieved for mono- and multilayer silicene FETs are 100 and 200 cm²/Vs, respectively. Although below the maximum value predicted by the theory for freestanding/ideal silicene, i.e., 1000 cm²/Vs [23], these mobility values are however encouraging in perspective as they are prone to further optimization. On the one hand, silicene FETs suffer from stability issues that have to be fixed by means of a second stage encapsulation on the silver-free silicene channel (see Figure 3(e)), possibly leading to enhanced mobility through dielectric screening effects [24]. On the other hand, silicene itself can be perfectioned in its inherent structure by reducing the density of grain boundaries or point defects (namely, unavoidable by-products of the epitaxy) that act as mobility limiter. Despite these further and necessary improvements, silicene FETs turn out to be a disruptive proof-of-concept paving the way to viable process schemes towards reliable silicene integration.

Structures composed by alternating semimetallic and semiconducting (or insulating) 2D materials are very interesting for use in nanoelectronic devices as they may work as a prototypical device structure triggering different electronic and optoelectronic characteristics [25]. These so-called “van der Waals (vdW) heterostructures” are composed by 2D layers in which the strong covalent bonds provide the in-plane stability of the 2D layers, whereas relatively weak (vdW) forces are sufficient to keep the layers together [26]. Inspired by this material concept, here, we outline the case of the silicene epitaxy on MoS₂ surface as derived from rheological pieces and mechanically exfoliated flakes. MoS₂ has become extremely popular as the first representative of layered transition metal dichalcogenides with a semiconducting character. Computational outputs based on density functional theory select three different silicene configurations that are commensurate with the MoS₂ surface lattice in terms of silicon atom accommodation as sketched in Figure 4(a) [9]. These include one-by-one atom match (AAA stacking), alternate atom match (ABA stacking), and an intermediate positioning of the silicene lattice. Match with the experimental facts results in a highly buckled silicene lattice with AAA stacking. Basically, this picture comes from the positional correspondence of silicon atoms in the silicene lattice with protruding sulphur atoms of the MoS₂ top layer as deduced from scanning tunneling microscopy (STM) imaging (see Figure 4(b)). In Figure 4(b), the top one can clearly discriminate MoS₂ regions from silicene ones while ruling out any amorphous silicon formation as from the electron diffraction figure in the inset. A more detailed insight into atomically resolved topography (Figure 4(b), B) reveals a one-by-one positional correspondence between top Si atoms and substrate surface atoms and two characteristic heights at the MoS₂-silicene step-edge, 3 and 5 Å. Both facts are consistent with highly stretched silicon bonding with a AAA stacking (Figure 4(b), C). In agreement with the computational modelling of the electronic bands, so-grown silicene proves to be metallic in character as resulting from direct evidence of the local density of states [9]. Recently, this picture has been objected by figuring out silicon intercalation in between the upmost adjacent MoS₂ layers [27]. Intercalation of silicon atoms inside layered materials to form a self-organized silicene layer is an emerging topic of study in a number of combinations; however, many factors in the silicon growth processing as well as in the affinity of silicon with underlying substrate may affect the details or the emergence of the intercalation. This scenario is well-described by the case of silicene intercalated in between graphene and ruthenium by silicon deposition on top where the growth temperature triggers silicon atoms from being arranged as bubble to a silicene lattice below the graphene layer [28]. When grown on freshly recovered MoS₂ flakes at 200°C, the highly buckled silicene lattice proves super-surface positioning based on compositional depth sensitive diagnostic. Subsequently, it was capped by Al₂O₃ stabilizer and integrated into a FET structure patterned on a Al₂O₃/silicene/MoS₂ heterosheet. This heterosheet was in turn supported by a SiO₂/Si⁺⁺ substrate serving as a bottom gate electrode as well as for the above-mentioned silicene FET [29]. Despite (high-buckled) silicene being metallic, the transfer characteristics display a field-modulated double transconductance peak as shown in Figure 4(c). The twofold feature was ascribed to two separate transport channels, one driven by the field drop at the MoS₂/SiO₂ interface and another at the silicene/MoS₂ interface [29]. The feasibility of such silicene-based heterosheet transistor paves the way to other device analogues where appropriate layered materials can be used to bear silicene without substrate decoupling therein addressing a stability issue in the process flow.

The proved dramatic interaction between the metal substrate and silicene drove the theoretical efforts towards a refined survey of template compatible with silicene [30]. In this sense, not only lattice-match conditions (verified by relaxing the honeycomb structure) are taken into account but also the band alignment between the silicon overlayer and the substrate. Silicene (also germanene) can be proved to be stabilized by Al₂O₃(0001) as substrate in a reconstructed fashion with respect to the freestanding lattice [30]. First principle calculations prove that silicene is stable when formed on the Al-terminated surface exhibiting a low-buckled honeycomb structure with gapped Dirac cones at K point. However, even on Al₂O₃(0001), the role of interaction turns out to tune the electronic band structure [11]. Indeed, among the manifold minima of the Born-Oppenheimer energy, it is possible to identify two groups of silicene geometries that differentiate each other about the degree of interaction with the Al₂O₃(0001) surface. Although their common structure is a metastable \surd 13\times \surd 13 silicene lattice on the 3\times 3 Al₂O₃(0001) substrate (Figure 5(a)), accounting for a strain of about 3.3% because of the lattice mismatch, when the average distance between the silicene and substrate is about 2.8 Å (3.3 Å), the interaction is strong (weak). In both cases, being a reconstruction of the freestanding lattice, either weak (W) or strong interacting silicene (SIS) lose its pristine D_3d symmetry. Although WIS and SIS configurations are separated by a small energy difference, their electronic and optical properties are remarkably different. Figure 5(b) shows the electronic band structure of the WIS geometry showing small bandgap opening in the Dirac cone of about 0.05 eV at K point retaining the linear band behavior of freestanding silicene close to the Fermi level. Conversely, the SIS geometry is characterized by a larger (indirect) bandgap opening without features of the Dirac energy dispersion (not shown). At variance with metal substrate, e.g., Ag(111) [31], the Al₂O₃(0001) substrate intriguingly paves the way for the first time optical characterization of silicene [11]. Opposed to relatively cold growth of silicene-on-silver, silicene is grown on a Al₂O₃(0001) surface at 670°C. Generally, silicon nanosheets grown on Al₂O₃(0001) exhibit a peculiar optical behavior as demonstrated in Figure 5(c). Similar to previously described cases, air-instability plagues the survival of silicene and thicker silicon film on Al₂O₃(0001); therefore, encapsulation with the Al₂O₃ is still necessary as described for the two previous cases in point while unaffecting the optical response of silicon due to its transparency in a large range of the electromagnetic spectrum. Hence, the silicon nanosheets are sandwiched in between two optical transparent layers for months without chemical intermixing [11]. By making tailored samples with variable silicon thickness along one substrate direction (see sketch in Figure 5) with a wedge-like shape, the overall silicon nanosheet thickness scrutinized in our survey ranges between 0.5 and 7 nm. The real part of the optical conductivity {\sigma }_1, extracted from transmittance measurements through Kramers-Kronig transformations, is shown in Figure 5(c). In the photon range 0.25-4.5 eV, the optical conductivity {\sigma }_1 shows a thickness-dependent behavior. At the lowest thickness investigated at the pure 2D limit, i.e., 0.5 nm, the black curve of Figure 5(c) is characterized by two spectral features at ~1.4 and 4.5 eV. These peaks closely resemble those arising from the resonant interband transitions \pi \to {\pi }^{\ast } and \sigma \to {\sigma }^{\ast } occurring at the M and Γ points of the Brilluoin zone of freestanding silicene due to the van Hove singularities of the joint density of states [32]. These spectral features are the characteristic hallmarks of the X-enes distinguishing graphene, silicene, and germanene, while they share the low-energy part of the spectrum with the same absorption behavior depending only on πα, where \alpha is the Sommerfeld fine-structure constant, linking the quantum electrodynamics to the condensed matter physics [33]. Noticeably, these interband transitions are demonstrated to occur in the WIS geometry. In particular, the pristine \pi \to {\pi }^{\ast } transition of silicene is almost unaffected by the presence of Al₂O₃(0001) substrate, at variance with Ag(111) [31], and by the strain, thus being very close to the experimental one. When the silicon nanosheet thickness is increased up to the maximum thickness of 7 nm, the optical behavior progressively deviates from that of freestanding silicene but still remains unconventional and does not recover the behavior of bulk amorphous or crystalline silicon because of the absence of an optical gap. On the other hand, the silicene-like nature of the thinnest silicon nanosheet investigated is further corroborated by an additional evidence. The low-energy part of the {\sigma }_1 reported in Figure 5(c) can be demonstrated to be quantized as a multiple integer of \pi {\text{e}}^2/2h, i.e., the universal optical conductance G₀ [34], where h is the Planck constant. In other words, the two silicon nanosheets with thickness of 0.5 and 1.5 nm show a low-energy optical conductance equal to G₀ and 2G₀, respectively. This is an additional evidence of the presence of massless Dirac electrons as already reported on single and few-layer graphene [35]. With the exception of the inelastic scattering associated with phonons, i.e., Raman spectroscopy, metal substrates hamper to probe the optical properties of silicene. Therefore, the choice of the transparent Al₂O₃(0001) for the theoretically predicted weakly interacting silicene allowed for the first time measured optical properties by fabricating and encapsulating silicon nanosheets with variable thickness on Al₂O₃(0001). A Dirac-like absorption is observed in the infrared range of the optical conductivity at the 2D limit. This optical conductivity shows an overall behavior similar to that expected from the ideal silicene with a clear feature related to the \pi \to {\pi }^{\ast } interband transition that is corroborated by the evidence of a quantized optical conductance for the first two layers. This outcome opens up the opportunity to exploit graphene-like silicon in a silicon photonics directions operating in the THz regime.