Progress in the materials science of silicene

While at a first glance, the crystal structure of silicene is similar to graphene, it is not as simple as it seems. In the case of graphene, the honeycomb lattice is planar and the lattice constant remains hardly modified. On the other hand, over a large range of in-plane lattice constants, freestanding silicene is predicted to be stable in the so-called 'low-buckled' or 'regularly buckled' structure [9, 10], shown in figure 2(a), in which two sublattices of the bipartite lattice are at different heights [6, 9]. Even if regularly buckled, the atomic orbital mixing is close to that of the sp² hybridization such that freestanding silicene is expected to display electronic properties similar to those of graphene [6, 9, 10]. As it can be recognized in the band structure shown in figure 3(a), in particular, most prominent is the absence of the opening of a recognizable gap in π bands at the Dirac point located at K_Si(1 × 1). This indicates that the buckling is not associated with a gain in band energy [10] and the stability of the buckled silicene over the planar one may instead relate to an instability in the phonon part that involves the lattice repulsive potential and the response of electrons to the lattice vibration [6, 9, 10]. Of course, due to the smaller size of interatomic overlap integrals, the overall π band width is reduced by a factor of about 3 [5, 6, 9, 10].

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Under out-of-equilibrium conditions, the degree of buckling of the honeycomb structure of freestanding silicene does vary widely as a function of the in-plane lattice constants [9, 10]. Additionally, the buckling could vary locally on the atomic length scale. This can be described as a reconstruction of the 'original' honeycomb, or (1 × 1) lattice. In particular, for a certain range of lattice constants, a (√3 × √3)-reconstructed phase, called the 'planar-like' phase [10], turns out to be quite stable. Its crystal and band structures (unfolded [25] into the (1 × 1) unit cell) are shown in figures 2(b) and 3(b), respectively.

Note that the importance of this phase lies in its relevance for epitaxial silicene: while for freestanding layers, the planar-like phase is less stable than the regularly buckled form of silicene [10], in its form with stripes [26], it becomes the ground state on the ZrB₂ surface [10, 27]. Note that it has also been calculated to form on the Ag(111) surface [28].

In this planar-like phase, all but one of the Si atoms per hexagon reside in a single plane. In this way, it is able to sustain a longer in-plane bond length as compared to the planar phase such that the bond length becomes closer to that of the regularly buckled phase [10]. Due to the (√3 × √3) reconstruction, the symmetry is broken which causes back-folding of electronic states into the reduced Brillouin zone and the lifting of the degeneracy of bands. As can be recognized in figure 3(b), some resemblance to the band structure of regularly buckled silicene can be found. For instance, cone-like band dispersions are still noticeable even under the (√3 × √3) reconstruction, which, however, leads to the opening of a small gap and an up-shift of the corresponding features [10]. This can be visualized at the K_Si(1 × 1) point via unfolding, as shown in figure 3(b).

Previously, the search for the lowest-energy structures of freestanding 2D Si allotropes has been focused on honeycomb lattices that relate to either regularly-buckled, unreconstructed [5, 6, 9, 10] or (√3 × √3)-reconstructed [10] silicene phases. However, very recently, it has been shown that the addition of Si adatoms to silicene results in the formation of a dumbbell structure with a lower energy per atom [29, 30]. Towards the complete coverage, the periodic dumbbells can be recognized to form the structure of a well-known single layer of MoS₂ that possesses a lower total energy per atom than regularly buckled silicene [31]. Although hypervalent Si atoms in heteroatomic molecules have been known since the 19th century [32, 33], this came as a surprise to us since by considering the 4-fold coordination realized in the sp³ bonding of Si atoms, it is difficult to understand why bonding with 6-fold coordination could be formed by Si atoms in the MoS₂ structure. Here, the new form of σ bonding expressed by three cigar-shaped orbitals coexist with an extended π electronic structure. The direction of these σ orbitals has changed from the typical in-plane direction of the orbitals in the sp² hybridization (see figure 1(b)) to the out-of-plane direction to form cigar-shaped, so-called 'nematic' orbitals [31]. Quite clearly, such a previously unimagined bonding configuration and structure must be considered as a candidate for epitaxial 2D Si layers, as just predicted for the (√3 × √3) phase on Ag(111) [34].

Honeycomb Si structures occur in disilicides [35–37], in Si sheets chemically exfoliated from calcium disilicides [38], and can even exist as a single sheet on an erbium-covered Si(111) single-crystal surface [39]. Such Si sheets in disilicides are characterized by charge transfer from the metals and a substantial amount of hybridization between the electronic states derived from the two subsystems.

More recently, Si nanoribbons with an internal honeycomb structure have been obtained by the deposition of Si atoms on the Ag(110) surface [40–44]. The epitaxial Si nanoribbons observed by scanning tunneling microscopy (STM) are uniformly 1.6 nm in width and 0.2 nm in height [40]. Density functional theory (DFT) calculation of a structure model with Si honeycombs based on STM observations resulted in an average Si–Si distance of 2.24 Å [42], a value which is close to that of freestanding silicene [6]. Along the long direction of the ribbons, dispersive states reminiscent of π electronic states of graphene have been observed [43]. These states are cone-like and split with a gap of 0.5 eV, centered at 0.6 eV below the Fermi level (E_F) at the high symmetry point corresponding to the silicene K_Si(1 × 1) point. In the short direction, states are localized revealing the one-dimensional (1D) character of this nanostructure [43].

The Ag(110) surface has a rectangular lattice which is ideal for the template of parallel 1D nanoribbons. On the other hand, in order to form 2D sheets with honeycomb structures, hexagonal lattices are the natural choice. Following the preparation and study of silicene nanoribbons, in 2010, the highly contested possible formation of a silicene sheet on Ag(111) surface has been reported [45]. In 2012, several groups described the formation of better-characterized Si honeycomb structures on Ag(111) [15–18] and ZrB₂(0001) thin film [4] surfaces, of which both have hexagonal symmetry.

The deposition of Si atoms on Ag(111) single crystal surfaces under UHV conditions and in the typical substrate temperature range between 250 °C and 300 °C leads to a number of surface reconstructions depending on the amount of Si atoms and the temperature during deposition. Since independently well-calibrated Si sources are needed to precisely establish the amount of Si atoms deposited onto the surface, the interpretation of the obtained surface reconstructions differs among these reports. As such, (2√3 × 2√3) [45] and (4 × 4) [15–17] reconstructions of the Ag(111) surface, and a reconstruction corresponding to a (√3 × √3)-reconstructed silicene lattice [18] have been reported as silicene sheets on the Ag(111) surface. These reconstructions may represent various phases of epitaxial silicene or alternatively, as has been discussed, of Si networks that consist of incomplete honeycombs [18] or possibly bilayer [46] structures.

It is obvious that control of the source conditions and of the temperature is key to the formation of silicene phases by a deposition process. The advantage of this process is that different substrate materials can be easily tested. This lead to the reports on Ir(111) [47] and ZrC(111) [48] as possible substrates.

On the other hand, epitaxial silicene on ZrB₂(0001) thin films forms by a completely different process: surface segregation at elevated temperatures. We found that on the surface of oxide-free, single-crystalline ZrB₂ thin films grown epitaxially on Si(111) wafers, epitaxial silicene forms spontaneously [4]. The thin films of metallic ceramic zirconium diboride, with the thickness of about 15–30 nm, have been grown by UHV-chemical vapor epitaxy and exhibit the epitaxial relationship described by ZrB₂(0001)//Si(111) and ZrB₂[100]//Si[11] [49].

In the case of surface segregation, the source of Si atoms is the silicon substrate of the diboride film. The process is governed by thermodynamic equilibrium conditions and thus self-terminating. Following the transfer under ambient conditions, native oxides including those containing Si atoms formed upon exposure to air are easily removed by heating to 800 °C under UHV conditions such that the diboride surface is again spontaneously and uniformly covered with epitaxial silicene. Importantly, under optimal conditions, this procedure highly reproducibly leads to samples with more than 99.5% of the surface covered with ZrB₂(0001) terraces, and thus with single-crystalline-like silicene [49, 50]. Different from the deposition technique, surface segregation processes are materials specific and are thus not applicable to any kind of substrate. But for example, it is known that on AlN(0001) epilayers grown with Si-doping or on Si-containing substrates, Si-induced surface reconstructions are observed [51]. Here, surface-segregated Si atoms work as a surfactant [51] just like in the case of ZrB₂ thin films grown on Si wafers. If a comprehensive characterization would be applied to this surface, one may possibly find epitaxial silicene on an insulating substrate.

As many controversial discussions of early results have shown, evidence for epitaxial silicene should come from a combination of comprehensive experimental characterization techniques and first-principles calculations covering both structural and electronic properties. STM performed in UHV is the most frequently used characterization method for epitaxial silicene [4, 15–18, 45, 47], and in the most cases [15, 16, 18, 45, 47], combined with an electron diffraction technique, the only characterization method. But clearly, STM images alone are not enough to claim the formation of silicene since it provides limited information about the electronic structure of the surface layer.

For instance, in our original [4] and subsequent studies [27, 52–54], structural evidence for epitaxial silicene on diboride thin films grown on Si wafers has been obtained from atomic-resolution surface imaging using STM, chemical information such as on the elemental composition and the chemical environment of Si atoms from high-resolution, core-level photoelectron spectra measured at a synchrotron radiation facility, and band structure-related information from the combination of angle-resolved photoelectron (ARPES) spectra and DFT calculations. Complementary and additional structural information might be obtained for instance by reflection high-energy electron diffraction (RHEED) [49], low-energy electron diffraction (LEED) [52] and from the analysis of incident electron beam energy dependent intensities (I–V) of LEED spots [55].

Why so many characterization techniques are needed is because (i) there is no single, reliable identification method for epitaxial silicene, and (ii) epitaxial silicene phases are always in a kind of hybrid state with the surface of the respective substrates and thus not well-defined in terms of isolated, freestanding or 'ideal' silicene. Since epitaxial silicene is not stable under ambient conditions, the characterization must be performed in situ inside the UHV set-up accompanying the preparation of silicene. In order to then obtain the required variety of meaningful data, the samples should reproducibly be synthesized in different experimental set-ups. This makes experimental research very challenging and that is why the number of groups working on the synthesis and characterization of epitaxial silicene is small compared to those who are carrying out theoretical studies.

Figure 4(a) shows the typical large-scale STM image of epitaxial silicene observed as a (2 × 2) surface reconstruction of ZrB₂(0001) thin films grown on Si(111) wafers [4]. For the first time, this reconstruction has been observed by us nearly 10 years ago [56], but it was not until recently and only in combination with the use of additional characterization methods and calculations that we were able to relate this surface structure to an atom-thick, buckled honeycomb structure made of Si atoms sitting on top of the diboride surface.

Zoom In Zoom Out Reset image size

Under good experimental conditions, fine details can be recognized in constant-current STM images as shown in figure 4(b) [4]. The fine details are related to both the structure of the layer [4] and electronic properties [53]. It is understood that the layer corresponds to a honeycomb mesh with the lattice constant of about 3.65 Å. The layer is compressed by 5% with respect to a bulk Si(111) bilayer, and is thus similarly compressed with respect to hypothetical, freestanding silicene. It is (√3 × √3)-reconstructed such that its unit cell is adjusted to that of the ZrB₂(0001)-(2 × 2) unit cell [4].

Apart from the surface reconstruction, stripe domains that are offset with respect to each other are recognized [4]. The repetition of the spacing between boundaries and the alternation of the direction of the offsets are signature of the spontaneous formation of stress domains as a result of large-scale interactions within a 2D layer of these adatoms.

Very recently, the origin of the large-scale stripe pattern has been suggested by first-principles calculations [26]. In short, without stripes, the (√3 × √3)-reconstructed, one-atom-thick Si layer has been found to exhibit a 'zero-frequency' phonon instability at the M point. In order to avoid a divergent response, the relevant phonon mode triggers the spontaneous formation of a new phase with the observed particular stripe pattern offering a way to lower both the atomic surface density and the total energy of silicene on the particular substrate.

While STM images of the (2 × 2)-reconstructed ZrB₂(0001) surface provide evidence for the presence of surface ad-atoms, proof for their elemental and chemical nature has been derived from Si 2p core-level photoelectron spectra obtained at various photon energies, hν, at beamlines 18 and 13 at the KEK-PF (photon factory) synchrotron radiation facility, located in Tsukuba, Japan. The surface-sensitive spectrum obtained with hν = 130 eV and in the normal emission geometry, shown in figure 5(a) shows the Si 2p doublet in which the spin–orbit splitting amounts to 600 ± 5 meV and in which each of the main lines consists of two peaks with binding energies that relate to distinct chemical environment of the Si atoms [4, 54].

Zoom In Zoom Out Reset image size

The bulk 2p_3/2 component of the clean Si(111)-(7 × 7) surface as assigned in [57, 58] appears at a binding energy of about 99.43 eV. The peaks associated with the 2p_3/2 line of silicene are shifted by about 450–710 meV towards lower binding energy [4, 54].

The observed binding energy difference is reminiscent of the one observed between the C 1s electrons of diamond and graphene. Note that the C 1s binding energy of sp³-hybridized carbon atoms in the bulk of diamond (∼285.0 eV [59]) is about 0.8 eV higher than that of sp²-hybridized atoms in graphene (∼284.2 eV [60]). This suggests that the low Si 2p binding energy is related to a certain degree of sp² hybridization for all of the Si atoms within the buckled honeycomb lattice of epitaxial silicene.

In initial attempts to identify components related to chemically distinct Si atoms in a spectrum related to chemical environments obtained with a lower experimental resolution, three components labeled α, β and γ have been identified by a peak fitting procedure and assigned to three chemical environments of the Si adatoms [4], denoted A, B and C in figure 5(c).

While in a wide range of in-plane lattice constants, freestanding silicene is predicted to be stable only with a single type of buckling, until now, experimentally found epitaxial silicene phases are all 'reconstructed' in terms of the (1 × 1) lattice. This means that the buckling varies locally on the atomic length scale in order to find 'epitaxial' conditions with the respective substrates. As such, the phases formed on ZrB₂(0001) and Ag(111) surfaces differ from each other in lattice constants and reconstructions.

For silicene on ZrB₂(0001), the existence of the three chemical environments α, β and γ as derived from the surface-sensitive Si 2p spectrum is consistent with an in-plane structure model that has been concluded from STM images (figures 4(b) and 5(b)) for the center of the stripe domains [4] accommodating the (√3 × √3)-reconstructed Si honeycomb lattice that matches the (2 × 2) unit cell of the ZrB₂(0001) thin film surface. For this model (figure 5(c)), the Si atomic ratio is 2 : 3 : 1, among the A, B and C atomic sites. Two Si_A atoms per unit cell are sitting on hollow sites of the Zr lattice, three Si_B atoms are located at the intermediate position between top and bridge sites, or so called 'near-bridge' sites, and one Si_C atom is on top of a Zr atom.

Note that from our later work using spectra obtained with a higher experimental resolution, it has clearly been suggested that additional chemical states must be present that cannot be properly resolved by the fitting procedure [54]. The corresponding atoms may be located in extended boundary regions separating stripe domains. In particular, since on-top positions are the energetically least favorable ones [10], at the boundaries, Si_C atoms might be shifted away from on-top sites towards near-bridge sites [54].

While the in-plane structure model for the center of the stripe domains could be derived at an early stage of the study, the out-of-plane atomic positions related to the buckling have been proven difficult to determine experimentally e.g. from photoelectron diffraction effects observed as strong angle- [4] and photon-energy-dependent [54] intensity variations of the Si 2p core-level components.

While initially [4, 53], our interpretation related to the buckling of epitaxial silicene on the ZrB₂(0001) thin film surface leaned towards a metastable, so-called 'regularly-buckled-like' phase, the controversial issue in favor of the theoretically preferred 'planar-like' structural modification [10], shown in figures 5(d) and (e), in the top and side views, respectively, has been only resolved in very recent work [27] by finding substantial agreement between the results of DFT calculations and ARPES data obtained in a wide energy range. That is, it turns out that the substrate–silicene hybrid electronic band structure can be the most useful fingerprint of the structural configuration of adsorbed Si layers, a fact that is due to the high sensitivity of electronic to structural properties. As our work shows, the agreement between calculations and the ARPES spectra has in part been achieved by slightly increasing the in-plane lattice constant. It has been pointed out that this has been necessary in order likely to account for a misestimate of the exchange-correlation energy in the generalized gradient approximation and to simulate the effect caused by the larger lattice constant of ZrB₂ thin films and the expected lower surface density of the Si atoms induced to avoid epitaxial strain [27].

The results of the calculations and the agreement with the experimental ARPES data provide the strongest evidence so far for the presence of the so-called (√3 × √3)-reconstructed, 'planar-like' silicene phase on the ZrB₂(0001) thin film surface.

Since the electronic structure of epitaxial silicene is expected to be modified from that of the predicted one of freestanding silicene in terms of a modified crystal structure (that is by a different degree of buckling, by the reconstruction and different lattice constants) and by hybridization with the electronic states of the substrate surface, and also because bulk electronic states of substrate are measured as well, the interpretation of ARPES data is not so straightforward. First-principles calculations are necessary to understand the band dispersions; and to be able to calculate, one needs a structure model as a starting point.

Figures 6(a) and (b) show the measured ARPES spectra along the $\bar{\Gamma }$–K_Si direction of epitaxial silicene on ZrB₂ thin film as a function of the in-plane wave number k_|| [27]. Note that due to the (√3 × √3) reconstruction of the Si honeycomb layer, the K_Si and M_Si points of unreconstructed, hypothetical, freestanding silicene (with a unit cell containing two Si atoms) coincide with the $\bar{\Gamma }$ and ${\rm \bar{M}}$ points of the repeated Brillouin zone of the reconstructed surface, respectively.

Zoom In Zoom Out Reset image size

Features denoted 'S₁' and 'S₂' have been assigned [49] to surface states characteristic for the unreconstructed Zr-terminated ZrB₂(0001) surface [61]. Since the Zr-derived surface states are robust, the outermost Zr layer must be considered structurally intact [4]. Other features, among those denoted 'X₁', 'X₂', 'X₂' and 'X₃', do not have a counterpart in the calculations for the unreconstructed Zr-terminated ZrB₂(0001) surface [61] and are therefore related to the presence of silicene [4]. In particular, the intense features X₂ and X₃ approach E_F by up to about 250 meV at the K_Si point and, owing to back-folding, are mirrored as 'X₂' with weak intensity in the first Brillouin zone [4, 27]. The upward curvature of X₂ bears some resemblance to the predicted Dirac cone of π bands of freestanding, nonreconstructed silicene at K_Si [4]. The intensity changes have been discussed to be related to the sensitivity of the ARPES cross section to the selected Brillouin zone. These changes as well as the energy and dispersion of the spectral features could be represented well by DFT calculations [27].

Even if the actual stripe pattern is not accounted for, the comparison between the ARPES data and the results of the calculations shown in figure 6(c) provide a conclusive interpretation of the nature and orbital character of the states at the silicene/ZrB₂(0001) interface [27]. It has been confirmed that all silicene-derived bands are hybridized to some extent with Zr d electronic states, which is consistent with non-negligible interactions at the interface. While the upward curved bands in the vicinity of E_F at K_Si, denoted p₂ and p₃ (that correspond to features X₂ and X₃), are of partial π character indeed, they are actually formed by a hybridization of Si s, p_x, p_y and p_z and Zr d orbitals. This hybridization reflects the intermediate sp²/sp³ hybridization of epitaxial silicene and its interactions with the metallic substrate [27].

Because of the presence of Zr-derived surface states and because of the hybridization of Si-derived orbitals with Zr d states, the surface involving silicene is metallic. However, band edges of the occupied, upward curved p₂ and p₃ bands, on one side, and of the downward curved band p₅, on the other, provide a high density of states that has strong contributions within the silicene layer. Since using the ARPES technique, only the filled electronic states can be probed, the energetic separation between the band edges and thus the size of the silicene-related gap should be measured with other techniques. Scanning tunneling spectroscopy performed at temperatures as low as 5.5 K revealed a 350 meV gap, with its center shifted 60 meV below the Fermi level [53]. Although the tunneling conductance was not zero inside this gap, the observed position of band edges is consistent with the silicene-related character of the p₂, p₃ and p₅ states and the close proximity of X₅ and the Fermi level that has been observed by ARPES in doping experiments [52].

Like for graphene [62, 63], in a rigid band picture, the reaction of freestanding silicene with strong donors like alkali and alkaline earth metal atoms is expected to shift E_F with respect to the π bands such rendering semi-metallic silicene metallic [64, 65]. According to the predictions, potassium (K) atoms adsorb preferentially on hollow sites [64, 65]. As we have already discussed for the case of silicene on ZrB₂(0001) thin films, the low-energy band structure of epitaxial silicene phases is different from that predicted for freestanding silicene but it is still expected that upon alkali metal adsorption, charge is donated to originally unoccupied electronic states with silicene π or σ contributions.

Recently, we have performed an experiment where small amounts of K atoms (with up to about 0.18 K atoms per Si atoms) have been deposited on epitaxial silicene formed on a ZrB₂ thin film [52]. As the LEED pattern and even the LEED I–V have barely changed upon adsorption, it has been concluded that the type of reconstruction as well as the degree of buckling of the Si honeycomb layer remains close to that of the original one [52]. This is consistent with the prediction made for freestanding silicene in which it has been found that K adsorption does not lead to major structural changes [64].

Potassium adsorption leads to charge donation to the silicene lattice, thus n-type doping, and to the partial filling [52] of a previously almost unoccupied band, denoted 'X₅', with contributions from Si p_z and Zr d orbitals [27]. Due to the charge donation, occupied states with a partial π character in the vicinity of the K_Si(1 × 1) point corresponding to the unreconstructed silicene lattice shift by about 100 meV towards higher binding energy. According to a rigid band picture and n-type doping, this confirms their character as silicene-derived states.

On the other hand, while not additionally filled, a diboride surface state is affected as well which indicates an enhancement of interactions at the silicene–diboride interface upon K adsorption [52].

Since n-type doping can approximately be described by a rigid band picture, it can be expected that electron withdrawal by acceptors, e.g. iodine molecules, will lead to an upward shift of bands with (partial) π character.

Since epitaxial silicene phases are stabilized and since their electronic properties are determined by interactions with their respective substrates, the key for any advance towards a future silicene nanoelectronics is in the understanding of these essential interactions. So far, there are still very few cases in which epitaxial Si honeycomb lattices could be created and characterized. Most of them had been found by trial-and-error approaches or during work on unrelated topics. A more systematic approach would rely on the knowledge of the fundamental principles behind the formation mechanisms of silicene.

As already mentioned in the previous section, Si honeycomb structures have been known to exist in disilicides [35–37]. It should therefore be a matter of the strength of interactions between these honeycomb structures and the metal layers in contact if these 2D layers exhibit freestanding silicene-like properties or not. While the role of metal atoms (and in particular of d electronic states) for the stabilization of Si honeycomb lattices is still not fully understood, it might be expected that the change from a 'disilicide' picture to an 'epitaxial silicene' picture is gradual.

One way the interaction strength is expressed is in the hybridization between silicene and substrate electronic states. Another expression lies in differences in the thermal stability. Silicene on ZrB₂ thin films is destabilized at temperatures above 650 °C which is observed as the ZrB₂(0001)-(2 × 2) to (1 × 1) phase transition [66] in which fractional streaks or spots corresponding to the silicene-(√3 × √3) pattern disappear upon heating. This transition is reversible, and upon cooling, silicene 'crystallizes' again. A similar phase transition is observed for silicene on ZrC(111) at about 730 °C [48]. On the other hand, a silicene phase on Ag(111) observed as a Ag(111)-(2√3 × 2√3) reconstruction is known to be irreversibly destroyed upon heating beyond 330 °C [18] and also by the deposition of additional Si atoms causing a phase transition to a structure involving sp³-hybridized Si atoms and the simultaneous exposure of the Ag(111) substrate surface [67]. Obviously, interactions of silicene with the Ag(111) surface are weaker as compared to those with the outermost Zr layers of ZrB₂(0001) or ZrC(111). This may actually be expected with regard to the tendency of zirconium forming a number of crystalline Si–Zr phases [68] while silver forms only metastable ones [69].

In the previous section, we introduced the results of K adsorption in which charge donation from K atoms to silicene has been observed. This charge donation increased the hybridization between the silicene-related and diboride-related states resulting in stronger interactions at the interface [52]. If charge donation results in stronger interaction, charge withdrawal may result in the weakening of the interactions. Further experimental efforts to engineer the interface property of epitaxial silicene by foreign atom adsorption and/or intercalation may be the key to realize silicene with properties close to those of the hypothetical, freestanding one.

Like for silicene nanoribbons [70] and silicene sheets [23], both prepared on Ag surfaces, silicene on zirconium diboride substrates resists oxidation to some extent as well. As such, while due to the presence of dangling bonds, Si(111) and Si(001) are easily oxidized, the Si 2p spectra of silicene on ZrB₂(0001) thin films are hardly affected upon exposure to 4500 Langmuir (L) of molecular oxygen (O₂) [24]. These spectra, shown in figure 7(a), have been measured at beamline D1011 of the MAX-Lab synchrotron radiation facility located in Lund, Sweden.

Zoom In Zoom Out Reset image size

Still, epitaxial silicene is immediately oxidized once removed from the UHV environment. In order to be able to handle it in air, a capping layer is a must. Such layers should ideally be insulating and non-interacting, if the silicene sheet should be employed in electronics applications. Alternatively, temporary metallic capping layer could serve as protection during the transport from one UHV system to another, as long the layer can be removed at moderate temperatures.

Recently, it has been suggested that aluminum oxide (Al₂O₃) could be a suitable, non-destructive encapsulation material for silicene on Ag(111) [23]. However, as our study of silicene on the ZrB₂(0001) thin film surface covered with a submonolayer of Al atoms shows, upon exposure to 4500 L, a large fraction of the Si atoms becomes oxidized (figure 7(b)) [24]. This is attributed to dissociative chemisorption of O₂ molecules by Al atoms at the surface, producing reactive atomic oxygen species that cause the oxidation. It is concluded that aluminum oxide overlayers prepared in this fashion on silicene on ZrB₂(0001) thin films are not suitable for encapsulation since they do not prevent but actually enhance the degradation of silicene [24].

In one of our yet unpublished studies [71], it has been found that non-destructive encapsulation of silicene on the ZrB₂(0001) thin films by aluminum nitride deposited at elevated temperatures using trimethylaluminum and ammonia precursors is not achieved either.

At present, silicon-based semiconductor and organic electronics are largely separated fields. In order to achieve an integration of the two technological areas, hybrid organic/semiconductor junctions shall be created in which interfaces must be well defined and in which each of the thin films shall be crystalline. Since Si atoms favor sp³ hybridization and the formation of the diamond crystal structure, at the related semiconductor surfaces, the presence of reactive, dangling bonds leads to chemisorption of organic molecules bound with strong covalent bonds [72]. This often prevents the formation of ordered adlayers.

In our recent, yet unpublished work, the electronic and crystal structures of thin anthracene (Ac) films grown on epitaxial silicene formed on single-crystalline ZrB₂ thin films have been studied using RHEED and ultraviolet photoelectron spectroscopy [73]. At 140 K, Ac multilayer films grow with the orientation of the long molecular axis aligned parallel to the surface. The films consist of layers in which the herringbone-stacked molecules have alternating tilt angles. The stacking distance is close to that between bc-planes in the Ac bulk crystal under ambient conditions [74]. The lattices of the organic film and the substrate have a commensurate relationship indicating the presence of an epitaxial relationship. This result shows that ordered multilayer films of an oligoacene molecule can be grown epitaxially on silicene.

With a low, possibly monolayer coverage, the Ac molecules also adsorb on silicene even at room temperature. This fact indicates that the interactions between Ac molecules and silicene are stronger than those on the graphite (0001) [75, 76] surface and enhanced by the atomic-site specific charge distribution associated with the buckling of silicene.

On the other hand, the chemical properties of epitaxial silicene are also quite different from those on reactive [72] Si surfaces. This enables the formation of a sharp, well-defined interface between organic molecules and silicene without the formation of covalent bonds.

In summary, the atomic-scale buckling of silicene does not prevent the growth of highly ordered organic multilayer films provided that epitaxial conditions are met. If in this way epitaxial conditions are obtained throughout device structures, highly ordered organic films can be prepared on both ultimately thin honeycomb Si layers and on widely used silicon wafers via buffer layers. The growth of ordered organic thin films on silicene may then represent a promising bottom-up approach for the integration of silicon-based and organic electronics under highly controlled conditions at various length scales.